OpenCV version++ (3.4.5)

OpenCV 3.4.5

3rdparty/carotene/hal/CMakeLists.txt

@@ -53,7 +53,7 @@ endif()
 set(CAROTENE_NS "carotene_o4t" CACHE STRING "" FORCE)
 
 function(compile_carotene)
-  if(ENABLE_NEON)
+  if(";${CPU_BASELINE_FINAL};" MATCHES ";NEON;")
     set(WITH_NEON ON)
   endif()
 

CMakeLists.txt

@@ -220,74 +220,212 @@ OCV_OPTION(BUILD_ITT                "Build Intel ITT from source"        (NOT MI
 
 # Optional 3rd party components
 # ===================================================
-OCV_OPTION(WITH_1394           "Include IEEE1394 support"                    ON   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_AVFOUNDATION   "Use AVFoundation for Video I/O (iOS/Mac)"    ON   IF APPLE)
-OCV_OPTION(WITH_CARBON         "Use Carbon for UI instead of Cocoa (OBSOLETE)"                               OFF IF APPLE )
-OCV_OPTION(WITH_CAROTENE       "Use NVidia carotene acceleration library for ARM platform"                   ON  IF (ARM OR AARCH64) AND NOT IOS AND NOT (CMAKE_VERSION VERSION_LESS "2.8.11"))
-OCV_OPTION(WITH_CPUFEATURES    "Use cpufeatures Android library"             ON   IF ANDROID)
-OCV_OPTION(WITH_VTK            "Include VTK library support (and build opencv_viz module eiher)"             ON  IF (NOT ANDROID AND NOT IOS AND NOT WINRT AND NOT CMAKE_CROSSCOMPILING) )
-OCV_OPTION(WITH_CUDA           "Include NVidia Cuda Runtime support"                                         OFF IF (NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_CUFFT          "Include NVidia Cuda Fast Fourier Transform (FFT) library support"            ON  IF (NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_CUBLAS         "Include NVidia Cuda Basic Linear Algebra Subprograms (BLAS) library support" ON IF (NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_NVCUVID        "Include NVidia Video Decoding library support"                               ON IF (NOT IOS AND NOT APPLE) )
-OCV_OPTION(WITH_EIGEN          "Include Eigen2/Eigen3 support"               (NOT CV_DISABLE_OPTIMIZATION)       IF (NOT WINRT AND NOT CMAKE_CROSSCOMPILING) )
-OCV_OPTION(WITH_VFW            "Include Video for Windows support (deprecated, consider using MSMF)"  OFF  IF WIN32 )
-OCV_OPTION(WITH_FFMPEG         "Include FFMPEG support"                      ON   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_GSTREAMER      "Include Gstreamer support"                   ON   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_GSTREAMER_0_10 "Enable Gstreamer 0.10 support (instead of 1.x)"                              OFF )
-OCV_OPTION(WITH_GTK            "Include GTK support"                         ON   IF (UNIX AND NOT APPLE AND NOT ANDROID) )
-OCV_OPTION(WITH_GTK_2_X        "Use GTK version 2"                           OFF  IF (UNIX AND NOT APPLE AND NOT ANDROID) )
-OCV_OPTION(WITH_IPP            "Include Intel IPP support"                   (NOT MINGW AND NOT CV_DISABLE_OPTIMIZATION)  IF (X86_64 OR X86) AND NOT WINRT AND NOT IOS )
-OCV_OPTION(WITH_HALIDE         "Include Halide support"                      OFF)
-OCV_OPTION(WITH_INF_ENGINE     "Include Intel Inference Engine support"      OFF)
-OCV_OPTION(WITH_JASPER         "Include JPEG2K support"                      ON   IF (NOT IOS) )
-OCV_OPTION(WITH_JPEG           "Include JPEG support"                        ON)
-OCV_OPTION(WITH_WEBP           "Include WebP support"                        ON   IF (NOT WINRT) )
-OCV_OPTION(WITH_OPENEXR        "Include ILM support via OpenEXR"             ON   IF (NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_OPENGL         "Include OpenGL support"                      OFF  IF (NOT ANDROID AND NOT WINRT) )
-OCV_OPTION(WITH_OPENVX         "Include OpenVX support"                      OFF)
-OCV_OPTION(WITH_OPENNI         "Include OpenNI support"                      OFF  IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_OPENNI2        "Include OpenNI2 support"                     OFF  IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_PNG            "Include PNG support"                         ON)
-OCV_OPTION(WITH_GDCM           "Include DICOM support"                       OFF)
-OCV_OPTION(WITH_PVAPI          "Include Prosilica GigE support"              OFF   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_GIGEAPI        "Include Smartek GigE support"                OFF   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_ARAVIS         "Include Aravis GigE support"                 OFF   IF (NOT ANDROID AND NOT IOS AND NOT WINRT AND NOT WIN32) )
-OCV_OPTION(WITH_QT             "Build with Qt Backend support"               OFF  IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_WIN32UI        "Build with Win32 UI Backend support"         ON   IF WIN32 AND NOT WINRT)
-OCV_OPTION(WITH_QUICKTIME      "Use QuickTime for Video I/O (OBSOLETE)"      OFF  IF APPLE )
-OCV_OPTION(WITH_QTKIT          "Use QTKit Video I/O backend"                 OFF  IF APPLE )
-OCV_OPTION(WITH_TBB            "Include Intel TBB support"                   OFF  IF (NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_OPENMP         "Include OpenMP support"                      OFF)
-OCV_OPTION(WITH_CSTRIPES       "Include C= support"                          OFF  IF (WIN32 AND NOT WINRT)  )
-OCV_OPTION(WITH_PTHREADS_PF    "Use pthreads-based parallel_for"             ON   IF (NOT WIN32 OR MINGW) )
-OCV_OPTION(WITH_TIFF           "Include TIFF support"                        ON   IF (NOT IOS) )
-OCV_OPTION(WITH_UNICAP         "Include Unicap support (GPL)"                OFF  IF (UNIX AND NOT APPLE AND NOT ANDROID) )
-OCV_OPTION(WITH_V4L            "Include Video 4 Linux support"               ON   IF (UNIX AND NOT ANDROID AND NOT APPLE) )
-OCV_OPTION(WITH_LIBV4L         "Use libv4l for Video 4 Linux support"        OFF  IF (UNIX AND NOT ANDROID AND NOT APPLE) )
-OCV_OPTION(WITH_DSHOW          "Build VideoIO with DirectShow support"       ON   IF (WIN32 AND NOT ARM AND NOT WINRT) )
-OCV_OPTION(WITH_MSMF           "Build VideoIO with Media Foundation support" ON  IF WIN32 )
-OCV_OPTION(WITH_XIMEA          "Include XIMEA cameras support"               OFF  IF (NOT ANDROID AND NOT WINRT) )
-OCV_OPTION(WITH_XINE           "Include Xine support (GPL)"                  OFF  IF (UNIX AND NOT APPLE AND NOT ANDROID) )
-OCV_OPTION(WITH_CLP            "Include Clp support (EPL)"                   OFF)
-OCV_OPTION(WITH_OPENCL         "Include OpenCL Runtime support"              (NOT ANDROID AND NOT CV_DISABLE_OPTIMIZATION)  IF (NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_OPENCL_SVM     "Include OpenCL Shared Virtual Memory support" OFF ) # experimental
-OCV_OPTION(WITH_OPENCLAMDFFT   "Include AMD OpenCL FFT library support"      ON   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_OPENCLAMDBLAS  "Include AMD OpenCL BLAS library support"     ON   IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_DIRECTX        "Include DirectX support"                     ON   IF (WIN32 AND NOT WINRT) )
-OCV_OPTION(WITH_INTELPERC      "Include Intel Perceptual Computing support"  OFF  IF (WIN32 AND NOT WINRT) )
-OCV_OPTION(WITH_VA             "Include VA support"                          OFF  IF (UNIX AND NOT ANDROID) )
-OCV_OPTION(WITH_VA_INTEL       "Include Intel VA-API/OpenCL support"         OFF  IF (UNIX AND NOT ANDROID) )
-OCV_OPTION(WITH_MFX            "Include Intel Media SDK support"             OFF   IF ((UNIX AND NOT ANDROID) OR (WIN32 AND NOT WINRT AND NOT MINGW)) )
-OCV_OPTION(WITH_GDAL           "Include GDAL Support"                        OFF  IF (NOT ANDROID AND NOT IOS AND NOT WINRT) )
-OCV_OPTION(WITH_GPHOTO2        "Include gPhoto2 library support"             OFF  IF (UNIX AND NOT ANDROID AND NOT IOS) )
-OCV_OPTION(WITH_LAPACK         "Include Lapack library support"              (NOT CV_DISABLE_OPTIMIZATION)  IF (NOT ANDROID AND NOT IOS) )
-OCV_OPTION(WITH_ITT            "Include Intel ITT support"                   ON   IF (NOT APPLE_FRAMEWORK) )
-OCV_OPTION(WITH_PROTOBUF       "Enable libprotobuf"                          ON )
-OCV_OPTION(WITH_IMGCODEC_HDR   "Include HDR support"                         ON)
-OCV_OPTION(WITH_IMGCODEC_SUNRASTER "Include SUNRASTER support"               ON)
-OCV_OPTION(WITH_IMGCODEC_PXM   "Include PNM (PBM,PGM,PPM) and PAM formats support" ON)
-OCV_OPTION(WITH_QUIRC          "Include library QR-code decoding"            ON)
+OCV_OPTION(WITH_1394 "Include IEEE1394 support" ON
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_DC1394)
+OCV_OPTION(WITH_AVFOUNDATION "Use AVFoundation for Video I/O (iOS/Mac)" ON
+  VISIBLE_IF APPLE
+  VERIFY HAVE_AVFOUNDATION)
+OCV_OPTION(WITH_CARBON "Use Carbon for UI instead of Cocoa (OBSOLETE)" OFF
+  VISIBLE_IF APPLE
+  VERIFY HAVE_CARBON OR HAVE_COCOA)
+OCV_OPTION(WITH_CAROTENE "Use NVidia carotene acceleration library for ARM platform" ON
+  VISIBLE_IF (ARM OR AARCH64) AND NOT IOS AND NOT (CMAKE_VERSION VERSION_LESS "2.8.11"))
+OCV_OPTION(WITH_CPUFEATURES "Use cpufeatures Android library" ON
+  VISIBLE_IF ANDROID
+  VERIFY HAVE_CPUFEATURES)
+OCV_OPTION(WITH_VTK "Include VTK library support (and build opencv_viz module eiher)" ON
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT AND NOT CMAKE_CROSSCOMPILING
+  VERIFY HAVE_VTK)
+OCV_OPTION(WITH_CUDA "Include NVidia Cuda Runtime support" OFF
+  VISIBLE_IF NOT IOS AND NOT WINRT
+  VERIFY HAVE_CUDA)
+OCV_OPTION(WITH_CUFFT "Include NVidia Cuda Fast Fourier Transform (FFT) library support" WITH_CUDA
+  VISIBLE_IF WITH_CUDA
+  VERIFY HAVE_CUFFT)
+OCV_OPTION(WITH_CUBLAS "Include NVidia Cuda Basic Linear Algebra Subprograms (BLAS) library support" WITH_CUDA
+  VISIBLE_IF WITH_CUDA
+  VERIFY HAVE_CUBLAS)
+OCV_OPTION(WITH_NVCUVID "Include NVidia Video Decoding library support" WITH_CUDA
+  VISIBLE_IF WITH_CUDA
+  VERIFY HAVE_NVCUVID)
+OCV_OPTION(WITH_EIGEN "Include Eigen2/Eigen3 support" (NOT CV_DISABLE_OPTIMIZATION)
+  VISIBLE_IF NOT WINRT AND NOT CMAKE_CROSSCOMPILING
+  VERIFY HAVE_EIGEN)
+OCV_OPTION(WITH_VFW "Include Video for Windows support (deprecated, consider using MSMF)" OFF
+  VISIBLE_IF WIN32
+  VERIFY HAVE_VFW)
+OCV_OPTION(WITH_FFMPEG "Include FFMPEG support" ON
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_FFMPEG)
+OCV_OPTION(WITH_GSTREAMER "Include Gstreamer support" ON
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_GSTREAMER AND GSTREAMER_BASE_VERSION VERSION_GREATER "0.99")
+OCV_OPTION(WITH_GSTREAMER_0_10 "Enable Gstreamer 0.10 support (instead of 1.x)" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_GSTREAMER AND GSTREAMER_BASE_VERSION VERSION_LESS "1.0")
+OCV_OPTION(WITH_GTK "Include GTK support" ON
+  VISIBLE_IF UNIX AND NOT APPLE AND NOT ANDROID
+  VERIFY HAVE_GTK)
+OCV_OPTION(WITH_GTK_2_X "Use GTK version 2" OFF
+  VISIBLE_IF UNIX AND NOT APPLE AND NOT ANDROID
+  VERIFY HAVE_GTK AND NOT HAVE_GTK3)
+OCV_OPTION(WITH_IPP "Include Intel IPP support" (NOT MINGW AND NOT CV_DISABLE_OPTIMIZATION)
+  VISIBLE_IF (X86_64 OR X86) AND NOT WINRT AND NOT IOS
+  VERIFY HAVE_IPP)
+OCV_OPTION(WITH_HALIDE "Include Halide support" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_HALIDE)
+OCV_OPTION(WITH_INF_ENGINE "Include Intel Inference Engine support" OFF
+  VISIBLE_IF TRUE
+  VERIFY INF_ENGINE_TARGET)
+OCV_OPTION(WITH_JASPER "Include JPEG2K support" ON
+  VISIBLE_IF NOT IOS
+  VERIFY HAVE_JASPER)
+OCV_OPTION(WITH_JPEG "Include JPEG support" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_JPEG)
+OCV_OPTION(WITH_WEBP "Include WebP support" ON
+  VISIBLE_IF NOT WINRT
+  VERIFY HAVE_WEBP)
+OCV_OPTION(WITH_OPENEXR "Include ILM support via OpenEXR" ON
+  VISIBLE_IF NOT IOS AND NOT WINRT
+  VERIFY HAVE_OPENEXR)
+OCV_OPTION(WITH_OPENGL "Include OpenGL support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT WINRT
+  VERIFY HAVE_OPENGL)
+OCV_OPTION(WITH_OPENVX "Include OpenVX support" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_OPENVX)
+OCV_OPTION(WITH_OPENNI "Include OpenNI support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_OPENNI)
+OCV_OPTION(WITH_OPENNI2 "Include OpenNI2 support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_OPENNI2)
+OCV_OPTION(WITH_PNG "Include PNG support" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_PNG)
+OCV_OPTION(WITH_GDCM "Include DICOM support" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_GDCM)
+OCV_OPTION(WITH_PVAPI "Include Prosilica GigE support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_PVAPI)
+OCV_OPTION(WITH_GIGEAPI "Include Smartek GigE support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_GIGE_API)
+OCV_OPTION(WITH_ARAVIS "Include Aravis GigE support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT AND NOT WIN32
+  VERIFY HAVE_ARAVIS_API)
+OCV_OPTION(WITH_QT "Build with Qt Backend support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_QT)
+OCV_OPTION(WITH_WIN32UI "Build with Win32 UI Backend support" ON
+  VISIBLE_IF WIN32 AND NOT WINRT
+  VERIFY HAVE_WIN32UI)
+OCV_OPTION(WITH_QUICKTIME "Use QuickTime for Video I/O (OBSOLETE)" OFF
+  VISIBLE_IF APPLE
+  VERIFY HAVE_QUICKTIME)
+OCV_OPTION(WITH_QTKIT "Use QTKit Video I/O backend" OFF
+  VISIBLE_IF APPLE
+  VERIFY HAVE_QTKIT)
+OCV_OPTION(WITH_TBB "Include Intel TBB support" OFF
+  VISIBLE_IF NOT IOS AND NOT WINRT
+  VERIFY HAVE_TBB)
+OCV_OPTION(WITH_OPENMP "Include OpenMP support" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_OPENMP)
+OCV_OPTION(WITH_CSTRIPES "Include C= support" OFF
+  VISIBLE_IF WIN32 AND NOT WINRT
+  VERIFY HAVE_CSTRIPES)
+OCV_OPTION(WITH_PTHREADS_PF "Use pthreads-based parallel_for" ON
+  VISIBLE_IF NOT WIN32 OR MINGW
+  VERIFY HAVE_PTHREADS_PF)
+OCV_OPTION(WITH_TIFF "Include TIFF support" ON
+  VISIBLE_IF NOT IOS
+  VERIFY HAVE_TIFF)
+OCV_OPTION(WITH_UNICAP "Include Unicap support (GPL)" OFF
+  VISIBLE_IF UNIX AND NOT APPLE AND NOT ANDROID
+  VERIFY HAVE_UNICAP)
+OCV_OPTION(WITH_V4L "Include Video 4 Linux support" ON
+  VISIBLE_IF UNIX AND NOT ANDROID AND NOT APPLE
+  VERIFY HAVE_CAMV4L OR HAVE_CAMV4L2 OR HAVE_VIDEOIO)
+OCV_OPTION(WITH_LIBV4L "Use libv4l for Video 4 Linux support" OFF
+  VISIBLE_IF UNIX AND NOT ANDROID AND NOT APPLE
+  VERIFY HAVE_LIBV4L)
+OCV_OPTION(WITH_DSHOW "Build VideoIO with DirectShow support" ON
+  VISIBLE_IF WIN32 AND NOT ARM AND NOT WINRT
+  VERIFY HAVE_DSHOW)
+OCV_OPTION(WITH_MSMF "Build VideoIO with Media Foundation support" NOT MINGW
+  VISIBLE_IF WIN32
+  VERIFY HAVE_MSMF)
+OCV_OPTION(WITH_MSMF_DXVA "Enable hardware acceleration in Media Foundation backend" WITH_MSMF
+  VISIBLE_IF WIN32
+  VERIFY HAVE_MSMF_DXVA)
+OCV_OPTION(WITH_XIMEA "Include XIMEA cameras support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT WINRT
+  VERIFY HAVE_XIMEA)
+OCV_OPTION(WITH_XINE "Include Xine support (GPL)" OFF
+  VISIBLE_IF UNIX AND NOT APPLE AND NOT ANDROID
+  VERIFY HAVE_XINE)
+OCV_OPTION(WITH_CLP "Include Clp support (EPL)" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_CLP)
+OCV_OPTION(WITH_OPENCL "Include OpenCL Runtime support" (NOT ANDROID AND NOT CV_DISABLE_OPTIMIZATION)
+  VISIBLE_IF NOT IOS AND NOT WINRT
+  VERIFY HAVE_OPENCL)
+OCV_OPTION(WITH_OPENCL_SVM "Include OpenCL Shared Virtual Memory support" OFF
+  VISIBLE_IF TRUE
+  VERIFY HAVE_OPENCL_SVM) # experimental
+OCV_OPTION(WITH_OPENCLAMDFFT "Include AMD OpenCL FFT library support" ON
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_CLAMDFFT)
+OCV_OPTION(WITH_OPENCLAMDBLAS "Include AMD OpenCL BLAS library support" ON
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_CLAMDBLAS)
+OCV_OPTION(WITH_DIRECTX "Include DirectX support" ON
+  VISIBLE_IF WIN32 AND NOT WINRT
+  VERIFY HAVE_DIRECTX)
+OCV_OPTION(WITH_INTELPERC "Include Intel Perceptual Computing support" OFF
+  VISIBLE_IF WIN32 AND NOT WINRT
+  VERIFY HAVE_INTELPERC)
+OCV_OPTION(WITH_VA "Include VA support" OFF
+  VISIBLE_IF UNIX AND NOT ANDROID
+  VERIFY HAVE_VA)
+OCV_OPTION(WITH_VA_INTEL "Include Intel VA-API/OpenCL support" OFF
+  VISIBLE_IF UNIX AND NOT ANDROID
+  VERIFY HAVE_VA_INTEL)
+OCV_OPTION(WITH_MFX "Include Intel Media SDK support" OFF
+  VISIBLE_IF (UNIX AND NOT ANDROID) OR (WIN32 AND NOT WINRT AND NOT MINGW)
+  VERIFY HAVE_MFX)
+OCV_OPTION(WITH_GDAL "Include GDAL Support" OFF
+  VISIBLE_IF NOT ANDROID AND NOT IOS AND NOT WINRT
+  VERIFY HAVE_GDAL)
+OCV_OPTION(WITH_GPHOTO2 "Include gPhoto2 library support" OFF
+  VISIBLE_IF UNIX AND NOT ANDROID AND NOT IOS
+  VERIFY HAVE_GPHOTO2)
+OCV_OPTION(WITH_LAPACK "Include Lapack library support" (NOT CV_DISABLE_OPTIMIZATION)
+  VISIBLE_IF NOT ANDROID AND NOT IOS
+  VERIFY HAVE_LAPACK)
+OCV_OPTION(WITH_ITT "Include Intel ITT support" ON
+  VISIBLE_IF NOT APPLE_FRAMEWORK
+  VERIFY HAVE_ITT)
+OCV_OPTION(WITH_PROTOBUF "Enable libprotobuf" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_PROTOBUF)
+OCV_OPTION(WITH_IMGCODEC_HDR "Include HDR support" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_IMGCODEC_HDR)
+OCV_OPTION(WITH_IMGCODEC_SUNRASTER "Include SUNRASTER support" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_IMGCODEC_SUNRASTER)
+OCV_OPTION(WITH_IMGCODEC_PXM "Include PNM (PBM,PGM,PPM) and PAM formats support" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_IMGCODEC_PXM)
+OCV_OPTION(WITH_QUIRC "Include library QR-code decoding" ON
+  VISIBLE_IF TRUE
+  VERIFY HAVE_QUIRC)
 
 # OpenCV build components
 # ===================================================
@@ -299,8 +437,8 @@ OCV_OPTION(BUILD_ANDROID_EXAMPLES   "Build examples for Android platform"
 OCV_OPTION(BUILD_DOCS               "Create build rules for OpenCV Documentation" OFF  IF (NOT WINRT AND NOT APPLE_FRAMEWORK))
 OCV_OPTION(BUILD_EXAMPLES           "Build all examples"                          OFF )
 OCV_OPTION(BUILD_PACKAGE            "Enables 'make package_source' command"       ON  IF NOT WINRT)
-OCV_OPTION(BUILD_PERF_TESTS         "Build performance tests"                     ON  IF (NOT APPLE_FRAMEWORK) )
-OCV_OPTION(BUILD_TESTS              "Build accuracy & regression tests"           ON  IF (NOT APPLE_FRAMEWORK) )
+OCV_OPTION(BUILD_PERF_TESTS         "Build performance tests"                     NOT INSTALL_CREATE_DISTRIB  IF (NOT APPLE_FRAMEWORK) )
+OCV_OPTION(BUILD_TESTS              "Build accuracy & regression tests"           NOT INSTALL_CREATE_DISTRIB  IF (NOT APPLE_FRAMEWORK) )
 OCV_OPTION(BUILD_WITH_DEBUG_INFO    "Include debug info into release binaries ('OFF' means default settings)" OFF )
 OCV_OPTION(BUILD_WITH_STATIC_CRT    "Enables use of statically linked CRT for statically linked OpenCV" ON IF MSVC )
 OCV_OPTION(BUILD_WITH_DYNAMIC_IPP   "Enables dynamic linking of IPP (only for standalone IPP)" OFF )
@@ -327,7 +465,6 @@ OCV_OPTION(ENABLE_PROFILING           "Enable profiling in the GCC compiler (Add
 OCV_OPTION(ENABLE_COVERAGE            "Enable coverage collection with  GCov"                    OFF  IF CV_GCC )
 OCV_OPTION(ENABLE_OMIT_FRAME_POINTER  "Enable -fomit-frame-pointer for GCC"                      ON   IF CV_GCC )
 OCV_OPTION(ENABLE_POWERPC             "Enable PowerPC for GCC"                                   ON   IF (CV_GCC AND CMAKE_SYSTEM_PROCESSOR MATCHES powerpc.*) )
-OCV_OPTION(ENABLE_VSX                 "Enable POWER8 and above VSX (64-bit little-endian)"       ON   IF ((CV_GCC OR CV_CLANG) AND PPC64LE) )
 OCV_OPTION(ENABLE_FAST_MATH           "Enable -ffast-math (not recommended for GCC 4.6.x)"       OFF  IF (CV_GCC AND (X86 OR X86_64)) )
 if(NOT IOS)  # Use CPU_BASELINE instead
 OCV_OPTION(ENABLE_NEON                "Enable NEON instructions"                                 (NEON OR ANDROID_ARM_NEON OR AARCH64) IF (CV_GCC OR CV_CLANG) AND (ARM OR AARCH64 OR IOS) )
@@ -348,6 +485,7 @@ OCV_OPTION(CV_ENABLE_INTRINSICS       "Use intrinsic-based optimized code" ON )
 OCV_OPTION(CV_DISABLE_OPTIMIZATION    "Disable explicit optimized code (dispatched code/intrinsics/loop unrolling/etc)" OFF )
 OCV_OPTION(CV_TRACE                   "Enable OpenCV code trace" ON)
 OCV_OPTION(OPENCV_GENERATE_SETUPVARS  "Generate setup_vars* scripts" ON IF (NOT ANDROID AND NOT APPLE_FRAMEWORK) )
+OCV_OPTION(ENABLE_CONFIG_VERIFICATION "Fail build if actual configuration doesn't match requested (WITH_XXX != HAVE_XXX)" OFF)
 
 OCV_OPTION(ENABLE_PYLINT              "Add target with Pylint checks"                            (BUILD_DOCS OR BUILD_EXAMPLES) IF (NOT CMAKE_CROSSCOMPILING AND NOT APPLE_FRAMEWORK) )
 OCV_OPTION(ENABLE_FLAKE8              "Add target with Python flake8 checker"                    (BUILD_DOCS OR BUILD_EXAMPLES) IF (NOT CMAKE_CROSSCOMPILING AND NOT APPLE_FRAMEWORK) )
@@ -470,6 +608,7 @@ else()
     endif()
   endif()
   ocv_update(OPENCV_INCLUDE_INSTALL_PATH "include")
+  #ocv_update(OPENCV_PYTHON_INSTALL_PATH  "python")  # no default value, see https://github.com/opencv/opencv/issues/13202
 endif()
 
 ocv_update(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${OPENCV_LIB_INSTALL_PATH}")
@@ -1342,6 +1481,7 @@ endif()
 
 if(WITH_MSMF OR HAVE_MSMF)
   status("    Media Foundation:" HAVE_MSMF    THEN YES                                        ELSE NO)
+  status("      DXVA:"           HAVE_MSMF_DXVA    THEN YES    ELSE NO)
 endif()
 
 if(WITH_XIMEA OR HAVE_XIMEA)
@@ -1494,7 +1634,7 @@ if(BUILD_opencv_python2)
     status("    Libraries:"   HAVE_opencv_python2  THEN  "${PYTHON2_LIBRARIES}"                                      ELSE NO)
   endif()
   status("    numpy:"         PYTHON2_NUMPY_INCLUDE_DIRS THEN "${PYTHON2_NUMPY_INCLUDE_DIRS} (ver ${PYTHON2_NUMPY_VERSION})" ELSE "NO (Python wrappers can not be generated)")
-  status("    packages path:" PYTHON2_EXECUTABLE         THEN "${PYTHON2_PACKAGES_PATH}"                                    ELSE "-")
+  status("    install path:"  HAVE_opencv_python2  THEN "${__INSTALL_PATH_PYTHON2}"                            ELSE "-")
 endif()
 
 if(BUILD_opencv_python3)
@@ -1507,7 +1647,7 @@ if(BUILD_opencv_python3)
     status("    Libraries:"   HAVE_opencv_python3  THEN  "${PYTHON3_LIBRARIES}"                                      ELSE NO)
   endif()
   status("    numpy:"         PYTHON3_NUMPY_INCLUDE_DIRS THEN "${PYTHON3_NUMPY_INCLUDE_DIRS} (ver ${PYTHON3_NUMPY_VERSION})" ELSE "NO (Python3 wrappers can not be generated)")
-  status("    packages path:" PYTHON3_EXECUTABLE         THEN "${PYTHON3_PACKAGES_PATH}"                                    ELSE "-")
+  status("    install path:"  HAVE_opencv_python3  THEN "${__INSTALL_PATH_PYTHON3}"                            ELSE "-")
 endif()
 
 status("")
@@ -1542,6 +1682,10 @@ status("")
 
 ocv_finalize_status()
 
+if(ENABLE_CONFIG_VERIFICATION)
+  ocv_verify_config()
+endif()
+
 ocv_cmake_hook(POST_FINALIZE)
 
 # ----------------------------------------------------------------------------

cmake/OpenCVCompilerOptimizations.cmake

@@ -5,6 +5,10 @@
 # AVX / AVX2 / AVX_512F
 # FMA3
 
+# ppc64le arch:
+# VSX  (always available on Power8)
+# VSX3 (always available on Power9)
+
 # CPU_{opt}_SUPPORTED=ON/OFF - compiler support (possibly with additional flag)
 # CPU_{opt}_IMPLIES=<list>
 # CPU_{opt}_FORCE=<list> - subset of "implies" list
@@ -26,10 +30,12 @@
 # CPU_DISPATCH_FINAL=<list> - final list of dispatched optimizations
 #
 # CPU_DISPATCH_FLAGS_${opt} - flags for source files compiled separately (<name>.avx2.cpp)
+#
+# CPU_{opt}_ENABLED_DEFAULT=ON/OFF - has compiler support without additional flag (CPU_BASELINE_DETECT=ON only)
 
 set(CPU_ALL_OPTIMIZATIONS "SSE;SSE2;SSE3;SSSE3;SSE4_1;SSE4_2;POPCNT;AVX;FP16;AVX2;FMA3;AVX_512F;AVX512_SKX")
 list(APPEND CPU_ALL_OPTIMIZATIONS NEON VFPV3 FP16)
-list(APPEND CPU_ALL_OPTIMIZATIONS VSX)
+list(APPEND CPU_ALL_OPTIMIZATIONS VSX VSX3)
 list(REMOVE_DUPLICATES CPU_ALL_OPTIMIZATIONS)
 
 ocv_update(CPU_VFPV3_FEATURE_ALIAS "")
@@ -81,7 +87,7 @@ ocv_optimization_process_obsolete_option(ENABLE_FMA3 FMA3 ON)
 ocv_optimization_process_obsolete_option(ENABLE_VFPV3 VFPV3 OFF)
 ocv_optimization_process_obsolete_option(ENABLE_NEON NEON OFF)
 
-ocv_optimization_process_obsolete_option(ENABLE_VSX VSX OFF)
+ocv_optimization_process_obsolete_option(ENABLE_VSX VSX ON)
 
 macro(ocv_is_optimization_in_list resultvar check_opt)
   set(__checked "")
@@ -289,14 +295,24 @@ elseif(ARM OR AARCH64)
     set(CPU_BASELINE "NEON;FP16" CACHE STRING "${HELP_CPU_BASELINE}")
   endif()
 elseif(PPC64LE)
-  ocv_update(CPU_KNOWN_OPTIMIZATIONS "VSX")
+  ocv_update(CPU_KNOWN_OPTIMIZATIONS "VSX;VSX3")
   ocv_update(CPU_VSX_TEST_FILE "${OpenCV_SOURCE_DIR}/cmake/checks/cpu_vsx.cpp")
+  ocv_update(CPU_VSX3_TEST_FILE "${OpenCV_SOURCE_DIR}/cmake/checks/cpu_vsx3.cpp")
+
+  if(NOT OPENCV_CPU_OPT_IMPLIES_IGNORE)
+    ocv_update(CPU_VSX3_IMPLIES "VSX")
+  endif()
 
   if(CV_CLANG AND (NOT ${CMAKE_CXX_COMPILER} MATCHES "xlc"))
     ocv_update(CPU_VSX_FLAGS_ON "-mvsx -maltivec")
+    ocv_update(CPU_VSX3_FLAGS_ON "-mpower9-vector")
   else()
     ocv_update(CPU_VSX_FLAGS_ON "-mcpu=power8")
+    ocv_update(CPU_VSX3_FLAGS_ON "-mcpu=power9 -mtune=power9")
   endif()
+
+  set(CPU_DISPATCH "VSX3" CACHE STRING "${HELP_CPU_DISPATCH}")
+  set(CPU_BASELINE "VSX" CACHE STRING "${HELP_CPU_BASELINE}")
 endif()
 
 # Helper values for cmake-gui
@@ -331,6 +347,7 @@ macro(ocv_check_compiler_optimization OPT)
           ocv_check_compiler_flag(CXX "${CPU_BASELINE_FLAGS}" "${_varname}" "${CPU_${OPT}_TEST_FILE}")
           if(${_varname})
             list(APPEND CPU_BASELINE_FINAL ${OPT})
+            set(CPU_${OPT}_ENABLED_DEFAULT ON)
             set(__available 1)
           endif()
         endif()
@@ -448,7 +465,7 @@ foreach(OPT ${CPU_KNOWN_OPTIMIZATIONS})
       if(NOT ";${CPU_BASELINE_FINAL};" MATCHES ";${OPT};")
         list(APPEND CPU_BASELINE_FINAL ${OPT})
       endif()
-      if(NOT CPU_BASELINE_DETECT)  # Don't change compiler flags in 'detection' mode
+      if(NOT CPU_${OPT}_ENABLED_DEFAULT)  # Don't change compiler flags in 'detection' mode
         ocv_append_optimization_flag(CPU_BASELINE_FLAGS ${OPT})
       endif()
     endif()

cmake/OpenCVDetectApacheAnt.cmake

@@ -1,3 +1,6 @@
+set(OPENCV_JAVA_SOURCE_VERSION "" CACHE STRING "Java source version (javac Ant target)")
+set(OPENCV_JAVA_TARGET_VERSION "" CACHE STRING "Java target version (javac Ant target)")
+
 file(TO_CMAKE_PATH "$ENV{ANT_DIR}" ANT_DIR_ENV_PATH)
 file(TO_CMAKE_PATH "$ENV{ProgramFiles}" ProgramFiles_ENV_PATH)
 

cmake/OpenCVDetectCUDA.cmake

@@ -52,7 +52,7 @@ if(CUDA_FOUND)
 
   message(STATUS "CUDA detected: " ${CUDA_VERSION})
 
-  set(_generations "Fermi" "Kepler" "Maxwell" "Pascal" "Volta")
+  set(_generations "Fermi" "Kepler" "Maxwell" "Pascal" "Volta" "Turing")
   if(NOT CMAKE_CROSSCOMPILING)
     list(APPEND _generations "Auto")
   endif()
@@ -107,7 +107,7 @@ if(CUDA_FOUND)
                        ERROR_QUIET OUTPUT_STRIP_TRAILING_WHITESPACE)
       if(NOT _nvcc_res EQUAL 0)
         message(STATUS "Automatic detection of CUDA generation failed. Going to build for all known architectures.")
-        set(__cuda_arch_bin "5.3 6.2 7.0 7.5")
+        set(__cuda_arch_bin "5.3 6.2 7.2")
       else()
         set(__cuda_arch_bin "${_nvcc_out}")
         string(REPLACE "2.1" "2.1(2.0)" __cuda_arch_bin "${__cuda_arch_bin}")

cmake/OpenCVDetectInferenceEngine.cmake

@@ -78,9 +78,9 @@ endif()
 
 if(INF_ENGINE_TARGET)
   if(NOT INF_ENGINE_RELEASE)
-    message(WARNING "InferenceEngine version have not been set, 2018R3 will be used by default. Set INF_ENGINE_RELEASE variable if you experience build errors.")
+    message(WARNING "InferenceEngine version have not been set, 2018R5 will be used by default. Set INF_ENGINE_RELEASE variable if you experience build errors.")
   endif()
-  set(INF_ENGINE_RELEASE "2018030000" CACHE STRING "Force IE version, should be in form YYYYAABBCC (e.g. 2018R2.0.2 -> 2018020002)")
+  set(INF_ENGINE_RELEASE "2018050000" CACHE STRING "Force IE version, should be in form YYYYAABBCC (e.g. 2018R2.0.2 -> 2018020002)")
   set_target_properties(${INF_ENGINE_TARGET} PROPERTIES
     INTERFACE_COMPILE_DEFINITIONS "HAVE_INF_ENGINE=1;INF_ENGINE_RELEASE=${INF_ENGINE_RELEASE}"
   )

cmake/OpenCVFindLibsVideo.cmake

@@ -274,14 +274,15 @@ endif(WITH_DSHOW)
 ocv_clear_vars(HAVE_MSMF)
 if(WITH_MSMF)
   check_include_file(Mfapi.h HAVE_MSMF)
-  check_include_file(D3D11.h D3D11_found)
-  check_include_file(D3d11_4.h D3D11_4_found)
-  if(D3D11_found AND D3D11_4_found)
-    set(HAVE_DXVA YES)
-  else()
-    set(HAVE_DXVA NO)
+  set(HAVE_MSMF_DXVA "")
+  if(WITH_MSMF_DXVA)
+    check_include_file(D3D11.h D3D11_found)
+    check_include_file(D3d11_4.h D3D11_4_found)
+    if(D3D11_found AND D3D11_4_found)
+      set(HAVE_MSMF_DXVA YES)
+    endif()
   endif()
-endif(WITH_MSMF)
+endif()
 
 # --- Extra HighGUI and VideoIO libs on Windows ---
 if(WIN32)

cmake/OpenCVGenSetupVars.cmake

@@ -43,7 +43,25 @@ else()
 endif()
 file(RELATIVE_PATH OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG
     "${CMAKE_INSTALL_PREFIX}/${OPENCV_SETUPVARS_INSTALL_PATH}/" "${CMAKE_INSTALL_PREFIX}/")
-ocv_path_join(OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG "${OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG}" "python_loader")  # https://github.com/opencv/opencv/pull/12977
+if(DEFINED OPENCV_PYTHON_INSTALL_PATH)
+  set(__python_path "${OPENCV_PYTHON_INSTALL_PATH}")
+elseif(DEFINED OPENCV_PYTHON_INSTALL_PATH_SETUPVARS)
+  set(__python_path "${OPENCV_PYTHON_INSTALL_PATH_SETUPVARS}")
+endif()
+if(DEFINED __python_path)
+  if(IS_ABSOLUTE "${__python_path}")
+    set(OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG "${__python_path}")
+    message(WARNING "CONFIGURATION IS NOT SUPPORTED: validate setupvars script in install directory")
+  else()
+    ocv_path_join(OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG "${OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG}" "${__python_path}")
+  endif()
+else()
+  if(DEFINED OPENCV_PYTHON3_INSTALL_PATH)
+    ocv_path_join(OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG "${OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG}" "${OPENCV_PYTHON3_INSTALL_PATH}")
+  else()
+    set(OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG "python_loader_is_not_installed")
+  endif()
+endif()
 configure_file("${OpenCV_SOURCE_DIR}/cmake/templates/${OPENCV_SETUPVARS_TEMPLATE}" "${CMAKE_BINARY_DIR}${CMAKE_FILES_DIRECTORY}/install/${OPENCV_SETUPVARS_FILENAME}" @ONLY)
 install(FILES "${CMAKE_BINARY_DIR}${CMAKE_FILES_DIRECTORY}/install/${OPENCV_SETUPVARS_FILENAME}"
     DESTINATION "${OPENCV_SETUPVARS_INSTALL_PATH}"

cmake/OpenCVModule.cmake

@@ -909,7 +909,11 @@ macro(_ocv_create_module)
       source_group("Src" FILES "${_VS_VERSION_FILE}")
     endif()
   endif()
-  if(WIN32 AND NOT ("${the_module}" STREQUAL "opencv_core" OR "${the_module}" STREQUAL "opencv_world")
+  if(WIN32 AND NOT (
+          "${the_module}" STREQUAL "opencv_core" OR
+          "${the_module}" STREQUAL "opencv_world" OR
+          "${the_module}" STREQUAL "opencv_cudev"
+      )
       AND (BUILD_SHARED_LIBS AND NOT "x${OPENCV_MODULE_TYPE}" STREQUAL "xSTATIC")
       AND NOT OPENCV_SKIP_DLLMAIN_GENERATION
   )

cmake/OpenCVUtils.cmake

@@ -508,7 +508,7 @@ macro(ocv_warnings_disable)
       foreach(var ${_flag_vars})
         foreach(warning ${_gxx_warnings})
           if(NOT warning MATCHES "^-Wno-")
-            string(REGEX REPLACE "${warning}(=[^ ]*)?" "" ${var} "${${var}}")
+            string(REGEX REPLACE "(^|[ ]+)${warning}(=[^ ]*)?([ ]+|$)" " " ${var} "${${var}}")
             string(REPLACE "-W" "-Wno-" warning "${warning}")
           endif()
           ocv_check_flag_support(${var} "${warning}" _varname "")
@@ -571,14 +571,21 @@ endmacro()
 # Provides an option that the user can optionally select.
 # Can accept condition to control when option is available for user.
 # Usage:
-#   option(<option_variable> "help string describing the option" <initial value or boolean expression> [IF <condition>])
+#   option(<option_variable>
+#          "help string describing the option"
+#          <initial value or boolean expression>
+#          [VISIBLE_IF <condition>]
+#          [VERIFY <condition>])
 macro(OCV_OPTION variable description value)
   set(__value ${value})
   set(__condition "")
+  set(__verification)
   set(__varname "__value")
   foreach(arg ${ARGN})
-    if(arg STREQUAL "IF" OR arg STREQUAL "if")
+    if(arg STREQUAL "IF" OR arg STREQUAL "if" OR arg STREQUAL "VISIBLE_IF")
       set(__varname "__condition")
+    elseif(arg STREQUAL "VERIFY")
+      set(__varname "__verification")
     else()
       list(APPEND ${__varname} ${arg})
     endif()
@@ -605,17 +612,55 @@ macro(OCV_OPTION variable description value)
       option(${variable} "${description}" ${__value})
     endif()
   else()
-    if(DEFINED ${variable} AND NOT OPENCV_HIDE_WARNING_UNSUPPORTED_OPTION)
+    if(DEFINED ${variable} AND "${${variable}}"  # emit warnings about turned ON options only.
+        AND NOT (OPENCV_HIDE_WARNING_UNSUPPORTED_OPTION OR "$ENV{OPENCV_HIDE_WARNING_UNSUPPORTED_OPTION}")
+    )
       message(WARNING "Unexpected option: ${variable} (=${${variable}})\nCondition: IF (${__condition})")
     endif()
     if(OPENCV_UNSET_UNSUPPORTED_OPTION)
       unset(${variable} CACHE)
     endif()
   endif()
+  if(__verification)
+    set(OPENCV_VERIFY_${variable} "${__verification}") # variable containing condition to verify
+    list(APPEND OPENCV_VERIFICATIONS "${variable}") # list of variable names (WITH_XXX;WITH_YYY;...)
+  endif()
   unset(__condition)
   unset(__value)
 endmacro()
 
+
+# Check that each variable stored in OPENCV_VERIFICATIONS list
+# is consistent with actual detection result (stored as condition in OPENCV_VERIFY_...) variables
+function(ocv_verify_config)
+  set(broken_options)
+  foreach(var ${OPENCV_VERIFICATIONS})
+    set(evaluated FALSE)
+    if(${OPENCV_VERIFY_${var}})
+      set(evaluated TRUE)
+    endif()
+    status("Verifying ${var}=${${var}} => '${OPENCV_VERIFY_${var}}'=${evaluated}")
+    if (${var} AND NOT evaluated)
+      list(APPEND broken_options ${var})
+      message(WARNING
+        "Option ${var} is enabled but corresponding dependency "
+        "have not been found: \"${OPENCV_VERIFY_${var}}\" is FALSE")
+    elseif(NOT ${var} AND evaluated)
+      list(APPEND broken_options ${var})
+      message(WARNING
+        "Option ${var} is disabled or unset but corresponding dependency "
+        "have been explicitly turned on: \"${OPENCV_VERIFY_${var}}\" is TRUE")
+    endif()
+  endforeach()
+  if(broken_options)
+    string(REPLACE ";" "\n" broken_options "${broken_options}")
+    message(FATAL_ERROR
+      "Some dependencies have not been found or have been forced, "
+      "unset ENABLE_CONFIG_VERIFICATION option to ignore these failures "
+      "or change following options:\n${broken_options}")
+  endif()
+endfunction()
+
 # Usage: ocv_append_build_options(HIGHGUI FFMPEG)
 macro(ocv_append_build_options var_prefix pkg_prefix)
   foreach(suffix INCLUDE_DIRS LIBRARIES LIBRARY_DIRS)

cmake/OpenCVVersion.cmake

@@ -15,7 +15,7 @@ set(OPENCV_LIBVERSION "${OPENCV_VERSION_MAJOR}.${OPENCV_VERSION_MINOR}.${OPENCV_
 
 # create a dependency on the version file
 # we never use the output of the following command but cmake will rerun automatically if the version file changes
-configure_file("${OPENCV_VERSION_FILE}" "${CMAKE_BINARY_DIR}/junk/version.junk" COPYONLY)
+configure_file("${OPENCV_VERSION_FILE}" "${CMAKE_BINARY_DIR}${CMAKE_FILES_DIRECTORY}/opencv_junk/version.junk" COPYONLY)
 
 ocv_update(OPENCV_VS_VER_FILEVERSION_QUAD "${OPENCV_VERSION_MAJOR},${OPENCV_VERSION_MINOR},${OPENCV_VERSION_PATCH},0")
 ocv_update(OPENCV_VS_VER_PRODUCTVERSION_QUAD "${OPENCV_VERSION_MAJOR},${OPENCV_VERSION_MINOR},${OPENCV_VERSION_PATCH},0")

cmake/checks/cpu_vsx.cpp

@@ -1,8 +1,12 @@
-# if defined(__VSX__)
-#   include <altivec.h>
-# else
-#   error "VSX is not supported"
-# endif
+#if defined(__VSX__)
+    #if defined(__PPC64__) && defined(__LITTLE_ENDIAN__)
+        #include <altivec.h>
+    #else
+        #error "OpenCV only supports little-endian mode"
+    #endif
+#else
+    #error "VSX is not supported"
+#endif
 
 int main()
 {

cmake/checks/cpu_vsx3.cpp

@@ -0,0 +1,17 @@
+#if defined(__VSX__)
+    #if defined(__PPC64__) && defined(__LITTLE_ENDIAN__)
+        #include <altivec.h>
+    #else
+        #error "OpenCV only supports little-endian mode"
+    #endif
+#else
+    #error "VSX3 is not supported"
+#endif
+
+int main()
+{
+    __vector unsigned char a = vec_splats((unsigned char)1);
+    __vector unsigned char b = vec_splats((unsigned char)2);
+    __vector unsigned char r = vec_absd(a, b);
+    return 0;
+}

cmake/templates/setup_vars_win32.cmd.in

@@ -1,18 +1,36 @@
 @ECHO OFF
-SETLOCAL EnableDelayedExpansion
 
 SET "SCRIPT_DIR=%~dp0"
 
 IF NOT DEFINED OPENCV_QUIET ( ECHO Setting vars for OpenCV @OPENCV_VERSION@ )
-SET "PATH=!SCRIPT_DIR!\@OPENCV_LIB_RUNTIME_DIR_RELATIVE_CMAKECONFIG@;%PATH%"
+SET "PATH=%SCRIPT_DIR%\@OPENCV_LIB_RUNTIME_DIR_RELATIVE_CMAKECONFIG@;%PATH%"
 
-IF NOT DEFINED OPENCV_SKIP_PYTHON (
-  SET "PYTHONPATH_OPENCV=!SCRIPT_DIR!\@OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG@"
-  IF NOT DEFINED OPENCV_QUIET ( ECHO Append PYTHONPATH: !PYTHONPATH_OPENCV! )
-  SET "PYTHONPATH=!PYTHONPATH_OPENCV!;%PYTHONPATH%"
-)
+IF NOT DEFINED OPENCV_SKIP_PYTHON CALL :SET_PYTHON
 
-IF NOT [%1] == [] (
-  %*
-  EXIT /B !errorlevel!
+SET SCRIPT_DIR=
+
+IF NOT [%1] == [] GOTO :RUN_COMMAND
+
+GOTO :EOF
+
+:RUN_COMMAND
+SET RUN_INTERACTIVE=1
+echo %CMDCMDLINE% | find /i "%~0" >nul
+IF NOT errorlevel 1 set RUN_INTERACTIVE=0
+
+%*
+SET RESULT=%ERRORLEVEL%
+IF %RESULT% NEQ 0 (
+  IF _%RUN_INTERACTIVE%_==_0_ ( IF NOT DEFINED OPENCV_BATCH_MODE ( pause ) )
 )
+EXIT /B %RESULT%
+
+:SET_PYTHON
+SET "PYTHONPATH_OPENCV=%SCRIPT_DIR%\@OPENCV_PYTHON_DIR_RELATIVE_CMAKECONFIG@"
+IF NOT DEFINED OPENCV_QUIET ( ECHO Append PYTHONPATH: %PYTHONPATH_OPENCV% )
+SET "PYTHONPATH=%PYTHONPATH_OPENCV%;%PYTHONPATH%"
+SET PYTHONPATH_OPENCV=
+EXIT /B
+
+
+:EOF

doc/Doxyfile.in

@@ -252,6 +252,7 @@ PREDEFINED             = __cplusplus=1 \
                          CV_SSE2=1 \
                          CV__DEBUG_NS_BEGIN= \
                          CV__DEBUG_NS_END= \
+                         CV_DEPRECATED_EXTERNAL= \
                          CV_DEPRECATED=
 EXPAND_AS_DEFINED      =
 SKIP_FUNCTION_MACROS   = YES

doc/py_tutorials/py_imgproc/py_transforms/py_fourier_transform/py_fourier_transform.markdown

@@ -79,11 +79,11 @@ using **np.ifft2()** function. The result, again, will be a complex number. You
 absolute value.
 @code{.py}
 rows, cols = img.shape
-crow,ccol = rows/2 , cols/2
-fshift[crow-30:crow+30, ccol-30:ccol+30] = 0
+crow,ccol = rows//2 , cols//2
+fshift[crow-30:crow+31, ccol-30:ccol+31] = 0
 f_ishift = np.fft.ifftshift(fshift)
 img_back = np.fft.ifft2(f_ishift)
-img_back = np.abs(img_back)
+img_back = np.real(img_back)
 
 plt.subplot(131),plt.imshow(img, cmap = 'gray')
 plt.title('Input Image'), plt.xticks([]), plt.yticks([])

doc/py_tutorials/py_photo/py_hdr/py_hdr.markdown

@@ -85,10 +85,8 @@ we will later have to clip the data in order to avoid overflow.
 
 @code{.py}
 # Tonemap HDR image
-tonemap1 = cv.createTonemapDurand(gamma=2.2)
+tonemap1 = cv.createTonemap(gamma=2.2)
 res_debevec = tonemap1.process(hdr_debevec.copy())
-tonemap2 = cv.createTonemapDurand(gamma=1.3)
-res_robertson = tonemap2.process(hdr_robertson.copy())
 @endcode
 
 ### 4. Merge exposures using Mertens fusion
@@ -173,5 +171,5 @@ Additional Resources
 
 Exercises
 ---------
-1. Try all tonemap algorithms: cv::TonemapDrago, cv::TonemapDurand, cv::TonemapMantiuk and cv::TonemapReinhard
+1. Try all tonemap algorithms: cv::TonemapDrago, cv::TonemapMantiuk and cv::TonemapReinhard
 2. Try changing the parameters in the HDR calibration and tonemap methods.

doc/tutorials/dnn/dnn_android/dnn_android.markdown

@@ -12,7 +12,7 @@ Tutorial was written for the following versions of corresponding software:
 
 - Download and install Android Studio from https://developer.android.com/studio.
 
-- Get the latest pre-built OpenCV for Android release from https://github.com/opencv/opencv/releases and unpack it (for example, `opencv-3.4.3-android-sdk.zip`).
+- Get the latest pre-built OpenCV for Android release from https://github.com/opencv/opencv/releases and unpack it (for example, `opencv-3.4.5-android-sdk.zip`).
 
 - Download MobileNet object detection model from https://github.com/chuanqi305/MobileNet-SSD. We need a configuration file `MobileNetSSD_deploy.prototxt` and weights `MobileNetSSD_deploy.caffemodel`.
 

doc/tutorials/imgproc/histograms/template_matching/template_matching.markdown

@@ -65,7 +65,7 @@ that should be used to find the match.
     -#  **Mask image (M):** The mask, a grayscale image that masks the template
 
 
--   Only two matching methods currently accept a mask: CV_TM_SQDIFF and CV_TM_CCORR_NORMED (see
+-   Only two matching methods currently accept a mask: TM_SQDIFF and TM_CCORR_NORMED (see
     below for explanation of all the matching methods available in opencv).
 
 
@@ -86,23 +86,23 @@ that should be used to find the match.
 Good question. OpenCV implements Template matching in the function **matchTemplate()**. The
 available methods are 6:
 
--#  **method=CV_TM_SQDIFF**
+-#  **method=TM_SQDIFF**
 
     \f[R(x,y)= \sum _{x',y'} (T(x',y')-I(x+x',y+y'))^2\f]
 
--#  **method=CV_TM_SQDIFF_NORMED**
+-#  **method=TM_SQDIFF_NORMED**
 
     \f[R(x,y)= \frac{\sum_{x',y'} (T(x',y')-I(x+x',y+y'))^2}{\sqrt{\sum_{x',y'}T(x',y')^2 \cdot \sum_{x',y'} I(x+x',y+y')^2}}\f]
 
--#  **method=CV_TM_CCORR**
+-#  **method=TM_CCORR**
 
     \f[R(x,y)= \sum _{x',y'} (T(x',y')  \cdot I(x+x',y+y'))\f]
 
--#  **method=CV_TM_CCORR_NORMED**
+-#  **method=TM_CCORR_NORMED**
 
     \f[R(x,y)= \frac{\sum_{x',y'} (T(x',y') \cdot I(x+x',y+y'))}{\sqrt{\sum_{x',y'}T(x',y')^2 \cdot \sum_{x',y'} I(x+x',y+y')^2}}\f]
 
--#  **method=CV_TM_CCOEFF**
+-#  **method=TM_CCOEFF**
 
     \f[R(x,y)= \sum _{x',y'} (T'(x',y')  \cdot I'(x+x',y+y'))\f]
 
@@ -110,7 +110,7 @@ available methods are 6:
 
     \f[\begin{array}{l} T'(x',y')=T(x',y') - 1/(w  \cdot h)  \cdot \sum _{x'',y''} T(x'',y'') \\ I'(x+x',y+y')=I(x+x',y+y') - 1/(w  \cdot h)  \cdot \sum _{x'',y''} I(x+x'',y+y'') \end{array}\f]
 
--#  **method=CV_TM_CCOEFF_NORMED**
+-#  **method=TM_CCOEFF_NORMED**
 
     \f[R(x,y)= \frac{ \sum_{x',y'} (T'(x',y') \cdot I'(x+x',y+y')) }{ \sqrt{\sum_{x',y'}T'(x',y')^2 \cdot \sum_{x',y'} I'(x+x',y+y')^2} }\f]
 

doc/tutorials/introduction/cross_referencing/tutorial_cross_referencing.markdown

@@ -36,14 +36,14 @@ Open your Doxyfile using your favorite text editor and search for the key
 `TAGFILES`. Change it as follows:
 
 @code
-TAGFILES = ./docs/doxygen-tags/opencv.tag=http://docs.opencv.org/3.4.3
+TAGFILES = ./docs/doxygen-tags/opencv.tag=http://docs.opencv.org/3.4.5
 @endcode
 
 If you had other definitions already, you can append the line using a `\`:
 
 @code
 TAGFILES = ./docs/doxygen-tags/libstdc++.tag=https://gcc.gnu.org/onlinedocs/libstdc++/latest-doxygen \
-           ./docs/doxygen-tags/opencv.tag=http://docs.opencv.org/3.4.3
+           ./docs/doxygen-tags/opencv.tag=http://docs.opencv.org/3.4.5
 @endcode
 
 Doxygen can now use the information from the tag file to link to the OpenCV

doc/tutorials/photo/hdr_imaging/hdr_imaging.markdown

@@ -171,7 +171,7 @@ Now it's time to look at the results. Note that HDR image can't be stored in one
 formats, so we save it to Radiance image (.hdr). Also all HDR imaging functions return results in
 [0, 1] range so we should multiply result by 255.
 
-You can try other tonemap algorithms: cv::TonemapDrago, cv::TonemapDurand, cv::TonemapMantiuk and cv::TonemapReinhard
+You can try other tonemap algorithms: cv::TonemapDrago, cv::TonemapMantiuk and cv::TonemapReinhard
 You can also adjust the parameters in the HDR calibration and tonemap methods for your own photos.
 
 Results

doc/tutorials/videoio/intelperc.markdown

@@ -78,5 +78,5 @@ there are two flags that should be used to set/get property of the needed genera
     flag value is assumed by default if neither of the two possible values of the property is set.
 
 For more information please refer to the example of usage
-[intelperc_capture.cpp](https://github.com/opencv/opencv/tree/3.4/samples/cpp/intelperc_capture.cpp)
+[videocapture_intelperc.cpp](https://github.com/opencv/opencv/tree/3.4/samples/cpp/videocapture_intelperc.cpp)
 in opencv/samples/cpp folder.

doc/tutorials/videoio/kinect_openni.markdown

@@ -134,5 +134,5 @@ property. The following properties of cameras available through OpenNI interface
     -   CAP_OPENNI_DEPTH_GENERATOR_REGISTRATION = CAP_OPENNI_DEPTH_GENERATOR + CAP_PROP_OPENNI_REGISTRATION
 
 For more information please refer to the example of usage
-[openni_capture.cpp](https://github.com/opencv/opencv/tree/3.4/samples/cpp/openni_capture.cpp) in
+[videocapture_openni.cpp](https://github.com/opencv/opencv/tree/3.4/samples/cpp/videocapture_openni.cpp) in
 opencv/samples/cpp folder.

modules/core/CMakeLists.txt

@@ -2,7 +2,7 @@ set(the_description "The Core Functionality")
 
 ocv_add_dispatched_file(mathfuncs_core SSE2 AVX AVX2)
 ocv_add_dispatched_file(stat SSE4_2 AVX2)
-ocv_add_dispatched_file(arithm SSE2 SSE4_1 AVX2)
+ocv_add_dispatched_file(arithm SSE2 SSE4_1 AVX2 VSX3)
 
 # dispatching for accuracy tests
 ocv_add_dispatched_file_force_all(test_intrin128 TEST SSE2 SSE3 SSSE3 SSE4_1 SSE4_2 AVX FP16 AVX2)
@@ -82,3 +82,47 @@ ocv_add_accuracy_tests()
 ocv_add_perf_tests()
 
 ocv_install_3rdparty_licenses(SoftFloat "${CMAKE_CURRENT_SOURCE_DIR}/3rdparty/SoftFloat/COPYING.txt")
+
+
+# generate data (samples data) config file
+set(OPENCV_DATA_CONFIG_FILE "${CMAKE_BINARY_DIR}/opencv_data_config.hpp")
+set(OPENCV_DATA_CONFIG_STR "")
+
+if(CMAKE_INSTALL_PREFIX)
+  set(OPENCV_DATA_CONFIG_STR "${OPENCV_DATA_CONFIG_STR}
+#define OPENCV_INSTALL_PREFIX \"${CMAKE_INSTALL_PREFIX}\"
+")
+endif()
+if(OPENCV_OTHER_INSTALL_PATH)
+  set(OPENCV_DATA_CONFIG_STR "${OPENCV_DATA_CONFIG_STR}
+#define OPENCV_DATA_INSTALL_PATH \"${OPENCV_OTHER_INSTALL_PATH}\"
+")
+endif()
+
+set(OPENCV_DATA_CONFIG_STR "${OPENCV_DATA_CONFIG_STR}
+#define OPENCV_BUILD_DIR \"${CMAKE_BINARY_DIR}\"
+")
+
+file(RELATIVE_PATH SOURCE_DIR_RELATIVE ${CMAKE_BINARY_DIR} ${CMAKE_SOURCE_DIR})
+set(OPENCV_DATA_CONFIG_STR "${OPENCV_DATA_CONFIG_STR}
+#define OPENCV_DATA_BUILD_DIR_SEARCH_PATHS \\
+    \"${SOURCE_DIR_RELATIVE}/\"
+")
+
+if(WIN32)
+  file(RELATIVE_PATH INSTALL_DATA_DIR_RELATIVE "${CMAKE_INSTALL_PREFIX}/${OPENCV_BIN_INSTALL_PATH}" "${CMAKE_INSTALL_PREFIX}/${OPENCV_OTHER_INSTALL_PATH}")
+else()
+  file(RELATIVE_PATH INSTALL_DATA_DIR_RELATIVE "${CMAKE_INSTALL_PREFIX}/${OPENCV_LIB_INSTALL_PATH}" "${CMAKE_INSTALL_PREFIX}/${OPENCV_OTHER_INSTALL_PATH}")
+endif()
+list(APPEND OPENCV_INSTALL_DATA_DIR_RELATIVE "${INSTALL_DATA_DIR_RELATIVE}")
+string(REPLACE ";" "\",\\\n    \"" OPENCV_INSTALL_DATA_DIR_RELATIVE_STR "\"${OPENCV_INSTALL_DATA_DIR_RELATIVE}\"")
+set(OPENCV_DATA_CONFIG_STR "${OPENCV_DATA_CONFIG_STR}
+#define OPENCV_INSTALL_DATA_DIR_RELATIVE ${OPENCV_INSTALL_DATA_DIR_RELATIVE_STR}
+")
+
+if(EXISTS "${OPENCV_DATA_CONFIG_FILE}")
+  file(READ "${OPENCV_DATA_CONFIG_FILE}" __content)
+endif()
+if(NOT OPENCV_DATA_CONFIG_STR STREQUAL "${__content}")
+  file(WRITE "${OPENCV_DATA_CONFIG_FILE}" "${OPENCV_DATA_CONFIG_STR}")
+endif()

modules/core/include/opencv2/core.hpp

@@ -75,6 +75,7 @@
         @defgroup core_utils_sse SSE utilities
         @defgroup core_utils_neon NEON utilities
         @defgroup core_utils_softfloat Softfloat support
+        @defgroup core_utils_samples Utility functions for OpenCV samples
     @}
     @defgroup core_opengl OpenGL interoperability
     @defgroup core_ipp Intel IPP Asynchronous C/C++ Converters

modules/core/include/opencv2/core/cv_cpu_dispatch.h

@@ -107,7 +107,7 @@
 #  include <arm_neon.h>
 #endif
 
-#if defined(__VSX__) && defined(__PPC64__) && defined(__LITTLE_ENDIAN__)
+#ifdef CV_CPU_COMPILE_VSX
 #  include <altivec.h>
 #  undef vector
 #  undef pixel
@@ -115,6 +115,10 @@
 #  define CV_VSX 1
 #endif
 
+#ifdef CV_CPU_COMPILE_VSX3
+#  define CV_VSX3 1
+#endif
+
 #endif // CV_ENABLE_INTRINSICS && !CV_DISABLE_OPTIMIZATION && !__CUDACC__
 
 #if defined CV_CPU_COMPILE_AVX && !defined CV_CPU_BASELINE_COMPILE_AVX
@@ -237,3 +241,7 @@ struct VZeroUpperGuard {
 #ifndef CV_VSX
 #  define CV_VSX 0
 #endif
+
+#ifndef CV_VSX3
+#  define CV_VSX3 0
+#endif

modules/core/include/opencv2/core/cv_cpu_helper.h

@@ -315,5 +315,26 @@
 #endif
 #define __CV_CPU_DISPATCH_CHAIN_VSX(fn, args, mode, ...)  CV_CPU_CALL_VSX(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
 
+#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_VSX3
+#  define CV_TRY_VSX3 1
+#  define CV_CPU_FORCE_VSX3 1
+#  define CV_CPU_HAS_SUPPORT_VSX3 1
+#  define CV_CPU_CALL_VSX3(fn, args) return (cpu_baseline::fn args)
+#  define CV_CPU_CALL_VSX3_(fn, args) return (opt_VSX3::fn args)
+#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_VSX3
+#  define CV_TRY_VSX3 1
+#  define CV_CPU_FORCE_VSX3 0
+#  define CV_CPU_HAS_SUPPORT_VSX3 (cv::checkHardwareSupport(CV_CPU_VSX3))
+#  define CV_CPU_CALL_VSX3(fn, args) if (CV_CPU_HAS_SUPPORT_VSX3) return (opt_VSX3::fn args)
+#  define CV_CPU_CALL_VSX3_(fn, args) if (CV_CPU_HAS_SUPPORT_VSX3) return (opt_VSX3::fn args)
+#else
+#  define CV_TRY_VSX3 0
+#  define CV_CPU_FORCE_VSX3 0
+#  define CV_CPU_HAS_SUPPORT_VSX3 0
+#  define CV_CPU_CALL_VSX3(fn, args)
+#  define CV_CPU_CALL_VSX3_(fn, args)
+#endif
+#define __CV_CPU_DISPATCH_CHAIN_VSX3(fn, args, mode, ...)  CV_CPU_CALL_VSX3(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
+
 #define CV_CPU_CALL_BASELINE(fn, args) return (cpu_baseline::fn args)
 #define __CV_CPU_DISPATCH_CHAIN_BASELINE(fn, args, mode, ...)  CV_CPU_CALL_BASELINE(fn, args) /* last in sequence */

modules/core/include/opencv2/core/cvdef.h

@@ -240,9 +240,10 @@ namespace cv { namespace debug_build_guard { } using namespace debug_build_guard
 #define CV_CPU_AVX_512VBMI      20
 #define CV_CPU_AVX_512VL        21
 
-#define CV_CPU_NEON   100
+#define CV_CPU_NEON             100
 
-#define CV_CPU_VSX 200
+#define CV_CPU_VSX              200
+#define CV_CPU_VSX3             201
 
 // CPU features groups
 #define CV_CPU_AVX512_SKX       256
@@ -280,6 +281,7 @@ enum CpuFeatures {
     CPU_NEON            = 100,
 
     CPU_VSX             = 200,
+    CPU_VSX3            = 201,
 
     CPU_AVX512_SKX      = 256, //!< Skylake-X with AVX-512F/CD/BW/DQ/VL
 
@@ -363,6 +365,15 @@ Cv64suf;
 #  endif
 #endif
 
+#ifndef CV_DEPRECATED_EXTERNAL
+#  if defined(__OPENCV_BUILD)
+#    define CV_DEPRECATED_EXTERNAL /* nothing */
+#  else
+#    define CV_DEPRECATED_EXTERNAL CV_DEPRECATED
+#  endif
+#endif
+
+
 #ifndef CV_EXTERN_C
 #  ifdef __cplusplus
 #    define CV_EXTERN_C extern "C"

modules/core/include/opencv2/core/eigen.hpp

@@ -60,7 +60,7 @@ namespace cv
 //! @{
 
 template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
-void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& src, Mat& dst )
+void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& src, OutputArray dst )
 {
     if( !(src.Flags & Eigen::RowMajorBit) )
     {

modules/core/include/opencv2/core/hal/intrin_avx.hpp

@@ -905,6 +905,11 @@ OPENCV_HAL_IMPL_AVX_CMP_OP_64BIT(v_int64x4)
 OPENCV_HAL_IMPL_AVX_CMP_OP_FLT(v_float32x8, ps)
 OPENCV_HAL_IMPL_AVX_CMP_OP_FLT(v_float64x4, pd)
 
+inline v_float32x8 v_not_nan(const v_float32x8& a)
+{ return v_float32x8(_mm256_cmp_ps(a.val, a.val, _CMP_ORD_Q)); }
+inline v_float64x4 v_not_nan(const v_float64x4& a)
+{ return v_float64x4(_mm256_cmp_pd(a.val, a.val, _CMP_ORD_Q)); }
+
 /** min/max **/
 OPENCV_HAL_IMPL_AVX_BIN_FUNC(v_min, v_uint8x32,  _mm256_min_epu8)
 OPENCV_HAL_IMPL_AVX_BIN_FUNC(v_max, v_uint8x32,  _mm256_max_epu8)
@@ -1120,6 +1125,12 @@ inline float v_reduce_sum(const v_float32x8& a)
     return _mm_cvtss_f32(s1);
 }
 
+inline double v_reduce_sum(const v_float64x4& a)
+{
+    __m256d s0 = _mm256_hadd_pd(a.val, a.val);
+    return _mm_cvtsd_f64(_mm_add_pd(_v256_extract_low(s0), _v256_extract_high(s0)));
+}
+
 inline v_float32x8 v_reduce_sum4(const v_float32x8& a, const v_float32x8& b,
                                  const v_float32x8& c, const v_float32x8& d)
 {
@@ -1128,6 +1139,41 @@ inline v_float32x8 v_reduce_sum4(const v_float32x8& a, const v_float32x8& b,
     return v_float32x8(_mm256_hadd_ps(ab, cd));
 }
 
+inline unsigned v_reduce_sad(const v_uint8x32& a, const v_uint8x32& b)
+{
+    return (unsigned)_v_cvtsi256_si32(_mm256_sad_epu8(a.val, b.val));
+}
+inline unsigned v_reduce_sad(const v_int8x32& a, const v_int8x32& b)
+{
+    __m256i half = _mm256_set1_epi8(0x7f);
+    return (unsigned)_v_cvtsi256_si32(_mm256_sad_epu8(_mm256_add_epi8(a.val, half), _mm256_add_epi8(b.val, half)));
+}
+inline unsigned v_reduce_sad(const v_uint16x16& a, const v_uint16x16& b)
+{
+    v_uint32x8 l, h;
+    v_expand(v_add_wrap(a - b, b - a), l, h);
+    return v_reduce_sum(l + h);
+}
+inline unsigned v_reduce_sad(const v_int16x16& a, const v_int16x16& b)
+{
+    v_uint32x8 l, h;
+    v_expand(v_reinterpret_as_u16(v_sub_wrap(v_max(a, b), v_min(a, b))), l, h);
+    return v_reduce_sum(l + h);
+}
+inline unsigned v_reduce_sad(const v_uint32x8& a, const v_uint32x8& b)
+{
+    return v_reduce_sum(v_max(a, b) - v_min(a, b));
+}
+inline unsigned v_reduce_sad(const v_int32x8& a, const v_int32x8& b)
+{
+    v_int32x8 m = a < b;
+    return v_reduce_sum(v_reinterpret_as_u32(((a - b) ^ m) - m));
+}
+inline float v_reduce_sad(const v_float32x8& a, const v_float32x8& b)
+{
+    return v_reduce_sum((a - b) & v_float32x8(_mm256_castsi256_ps(_mm256_set1_epi32(0x7fffffff))));
+}
+
 /** Popcount **/
 #define OPENCV_HAL_IMPL_AVX_POPCOUNT(_Tpvec)                     \
     inline v_uint32x8 v_popcount(const _Tpvec& a)                \
@@ -1232,6 +1278,16 @@ OPENCV_HAL_IMPL_AVX_CHECK_FLT(v_float64x4, 15)
 OPENCV_HAL_IMPL_AVX_MULADD(v_float32x8, ps)
 OPENCV_HAL_IMPL_AVX_MULADD(v_float64x4, pd)
 
+inline v_int32x8 v_fma(const v_int32x8& a, const v_int32x8& b, const v_int32x8& c)
+{
+    return a * b + c;
+}
+
+inline v_int32x8 v_muladd(const v_int32x8& a, const v_int32x8& b, const v_int32x8& c)
+{
+    return v_fma(a, b, c);
+}
+
 inline v_float32x8 v_invsqrt(const v_float32x8& x)
 {
     v_float32x8 half = x * v256_setall_f32(0.5);
@@ -1363,25 +1419,22 @@ inline v_float64x4 v_cvt_f64_high(const v_float32x8& a)
 
 inline v_int32x8 v_lut(const int* tab, const v_int32x8& idxvec)
 {
-    int CV_DECL_ALIGNED(32) idx[8];
-    v_store_aligned(idx, idxvec);
-    return v_int32x8(_mm256_setr_epi32(tab[idx[0]], tab[idx[1]], tab[idx[2]], tab[idx[3]],
-                                       tab[idx[4]], tab[idx[5]], tab[idx[6]], tab[idx[7]]));
+    return v_int32x8(_mm256_i32gather_epi32(tab, idxvec.val, 4));
+}
+
+inline v_uint32x8 v_lut(const unsigned* tab, const v_int32x8& idxvec)
+{
+    return v_reinterpret_as_u32(v_lut((const int *)tab, idxvec));
 }
 
 inline v_float32x8 v_lut(const float* tab, const v_int32x8& idxvec)
 {
-    int CV_DECL_ALIGNED(32) idx[8];
-    v_store_aligned(idx, idxvec);
-    return v_float32x8(_mm256_setr_ps(tab[idx[0]], tab[idx[1]], tab[idx[2]], tab[idx[3]],
-                                      tab[idx[4]], tab[idx[5]], tab[idx[6]], tab[idx[7]]));
+    return v_float32x8(_mm256_i32gather_ps(tab, idxvec.val, 4));
 }
 
 inline v_float64x4 v_lut(const double* tab, const v_int32x8& idxvec)
 {
-    int CV_DECL_ALIGNED(32) idx[8];
-    v_store_aligned(idx, idxvec);
-    return v_float64x4(_mm256_setr_pd(tab[idx[0]], tab[idx[1]], tab[idx[2]], tab[idx[3]]));
+    return v_float64x4(_mm256_i32gather_pd(tab, _mm256_castsi256_si128(idxvec.val), 8));
 }
 
 inline void v_lut_deinterleave(const float* tab, const v_int32x8& idxvec, v_float32x8& x, v_float32x8& y)

modules/core/include/opencv2/core/hal/intrin_cpp.hpp

@@ -683,6 +683,25 @@ OPENCV_HAL_IMPL_CMP_OP(==)
 For all types except 64-bit integer values. */
 OPENCV_HAL_IMPL_CMP_OP(!=)
 
+template<int n>
+inline v_reg<float, n> v_not_nan(const v_reg<float, n>& a)
+{
+    typedef typename V_TypeTraits<float>::int_type itype;
+    v_reg<float, n> c;
+    for (int i = 0; i < n; i++)
+        c.s[i] = V_TypeTraits<float>::reinterpret_from_int((itype)-(int)(a.s[i] == a.s[i]));
+    return c;
+}
+template<int n>
+inline v_reg<double, n> v_not_nan(const v_reg<double, n>& a)
+{
+    typedef typename V_TypeTraits<double>::int_type itype;
+    v_reg<double, n> c;
+    for (int i = 0; i < n; i++)
+        c.s[i] = V_TypeTraits<double>::reinterpret_from_int((itype)-(int)(a.s[i] == a.s[i]));
+    return c;
+}
+
 //! @brief Helper macro
 //! @ingroup core_hal_intrin_impl
 #define OPENCV_HAL_IMPL_ARITHM_OP(func, bin_op, cast_op, _Tp2) \
@@ -1044,6 +1063,21 @@ inline v_float32x4 v_reduce_sum4(const v_float32x4& a, const v_float32x4& b,
     return r;
 }
 
+/** @brief Sum absolute differences of values
+
+Scheme:
+@code
+{A1 A2 A3 ...} {B1 B2 B3 ...} => sum{ABS(A1-B1),abs(A2-B2),abs(A3-B3),...}
+@endcode
+For all types except 64-bit types.*/
+template<typename _Tp, int n> inline typename V_TypeTraits< typename V_TypeTraits<_Tp>::abs_type >::sum_type v_reduce_sad(const v_reg<_Tp, n>& a, const v_reg<_Tp, n>& b)
+{
+    typename V_TypeTraits< typename V_TypeTraits<_Tp>::abs_type >::sum_type c = _absdiff(a.s[0], b.s[0]);
+    for (int i = 1; i < n; i++)
+        c += _absdiff(a.s[i], b.s[i]);
+    return c;
+}
+
 /** @brief Get negative values mask
 
 Returned value is a bit mask with bits set to 1 on places corresponding to negative packed values indexes.

modules/core/include/opencv2/core/hal/intrin_neon.hpp

@@ -764,6 +764,13 @@ OPENCV_HAL_IMPL_NEON_INT_CMP_OP(v_int64x2, vreinterpretq_s64_u64, s64, u64)
 OPENCV_HAL_IMPL_NEON_INT_CMP_OP(v_float64x2, vreinterpretq_f64_u64, f64, u64)
 #endif
 
+inline v_float32x4 v_not_nan(const v_float32x4& a)
+{ return v_float32x4(vreinterpretq_f32_u32(vceqq_f32(a.val, a.val))); }
+#if CV_SIMD128_64F
+inline v_float64x2 v_not_nan(const v_float64x2& a)
+{ return v_float64x2(vreinterpretq_f64_u64(vceqq_f64(a.val, a.val))); }
+#endif
+
 OPENCV_HAL_IMPL_NEON_BIN_FUNC(v_uint8x16, v_add_wrap, vaddq_u8)
 OPENCV_HAL_IMPL_NEON_BIN_FUNC(v_int8x16, v_add_wrap, vaddq_s8)
 OPENCV_HAL_IMPL_NEON_BIN_FUNC(v_uint16x8, v_add_wrap, vaddq_u16)
@@ -977,6 +984,13 @@ OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(v_float32x4, float32x2, float, sum, add, f32)
 OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(v_float32x4, float32x2, float, max, max, f32)
 OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(v_float32x4, float32x2, float, min, min, f32)
 
+#if CV_SIMD128_64F
+inline double v_reduce_sum(const v_float64x2& a)
+{
+    return vgetq_lane_f64(a.val, 0) + vgetq_lane_f64(a.val, 1);
+}
+#endif
+
 inline v_float32x4 v_reduce_sum4(const v_float32x4& a, const v_float32x4& b,
                                  const v_float32x4& c, const v_float32x4& d)
 {
@@ -992,6 +1006,49 @@ inline v_float32x4 v_reduce_sum4(const v_float32x4& a, const v_float32x4& b,
     return v_float32x4(vaddq_f32(v0, v1));
 }
 
+inline unsigned v_reduce_sad(const v_uint8x16& a, const v_uint8x16& b)
+{
+    uint32x4_t t0 = vpaddlq_u16(vpaddlq_u8(vabdq_u8(a.val, b.val)));
+    uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
+    return vget_lane_u32(vpadd_u32(t1, t1), 0);
+}
+inline unsigned v_reduce_sad(const v_int8x16& a, const v_int8x16& b)
+{
+    uint32x4_t t0 = vpaddlq_u16(vpaddlq_u8(vreinterpretq_u8_s8(vabdq_s8(a.val, b.val))));
+    uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
+    return vget_lane_u32(vpadd_u32(t1, t1), 0);
+}
+inline unsigned v_reduce_sad(const v_uint16x8& a, const v_uint16x8& b)
+{
+    uint32x4_t t0 = vpaddlq_u16(vabdq_u16(a.val, b.val));
+    uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
+    return vget_lane_u32(vpadd_u32(t1, t1), 0);
+}
+inline unsigned v_reduce_sad(const v_int16x8& a, const v_int16x8& b)
+{
+    uint32x4_t t0 = vpaddlq_u16(vreinterpretq_u16_s16(vabdq_s16(a.val, b.val)));
+    uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
+    return vget_lane_u32(vpadd_u32(t1, t1), 0);
+}
+inline unsigned v_reduce_sad(const v_uint32x4& a, const v_uint32x4& b)
+{
+    uint32x4_t t0 = vabdq_u32(a.val, b.val);
+    uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
+    return vget_lane_u32(vpadd_u32(t1, t1), 0);
+}
+inline unsigned v_reduce_sad(const v_int32x4& a, const v_int32x4& b)
+{
+    uint32x4_t t0 = vreinterpretq_u32_s32(vabdq_s32(a.val, b.val));
+    uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
+    return vget_lane_u32(vpadd_u32(t1, t1), 0);
+}
+inline float v_reduce_sad(const v_float32x4& a, const v_float32x4& b)
+{
+    float32x4_t t0 = vabdq_f32(a.val, b.val);
+    float32x2_t t1 = vpadd_f32(vget_low_f32(t0), vget_high_f32(t0));
+    return vget_lane_f32(vpadd_f32(t1, t1), 0);
+}
+
 #define OPENCV_HAL_IMPL_NEON_POPCOUNT(_Tpvec, cast) \
 inline v_uint32x4 v_popcount(const _Tpvec& a) \
 { \

modules/core/include/opencv2/core/hal/intrin_sse.hpp

@@ -1041,6 +1041,11 @@ inline _Tpvec operator != (const _Tpvec& a, const _Tpvec& b) \
 OPENCV_HAL_IMPL_SSE_64BIT_CMP_OP(v_uint64x2, v_reinterpret_as_u64)
 OPENCV_HAL_IMPL_SSE_64BIT_CMP_OP(v_int64x2, v_reinterpret_as_s64)
 
+inline v_float32x4 v_not_nan(const v_float32x4& a)
+{ return v_float32x4(_mm_cmpord_ps(a.val, a.val)); }
+inline v_float64x2 v_not_nan(const v_float64x2& a)
+{ return v_float64x2(_mm_cmpord_pd(a.val, a.val)); }
+
 OPENCV_HAL_IMPL_SSE_BIN_FUNC(v_uint8x16, v_add_wrap, _mm_add_epi8)
 OPENCV_HAL_IMPL_SSE_BIN_FUNC(v_int8x16, v_add_wrap, _mm_add_epi8)
 OPENCV_HAL_IMPL_SSE_BIN_FUNC(v_uint16x8, v_add_wrap, _mm_add_epi16)
@@ -1451,6 +1456,13 @@ OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(v_uint32x4, unsigned, __m128i, epi32, OPENCV
 OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(v_int32x4, int, __m128i, epi32, OPENCV_HAL_NOP, OPENCV_HAL_NOP, si128_si32)
 OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(v_float32x4, float, __m128, ps, _mm_castps_si128, _mm_castsi128_ps, ss_f32)
 
+inline double v_reduce_sum(const v_float64x2& a)
+{
+    double CV_DECL_ALIGNED(32) idx[2];
+    v_store_aligned(idx, a);
+    return idx[0] + idx[1];
+}
+
 inline v_float32x4 v_reduce_sum4(const v_float32x4& a, const v_float32x4& b,
                                  const v_float32x4& c, const v_float32x4& d)
 {
@@ -1472,6 +1484,41 @@ OPENCV_HAL_IMPL_SSE_REDUCE_OP_4(v_int32x4, int, min, std::min)
 OPENCV_HAL_IMPL_SSE_REDUCE_OP_4(v_float32x4, float, max, std::max)
 OPENCV_HAL_IMPL_SSE_REDUCE_OP_4(v_float32x4, float, min, std::min)
 
+inline unsigned v_reduce_sad(const v_uint8x16& a, const v_uint8x16& b)
+{
+    return (unsigned)_mm_cvtsi128_si32(_mm_sad_epu8(a.val, b.val));
+}
+inline unsigned v_reduce_sad(const v_int8x16& a, const v_int8x16& b)
+{
+    __m128i half = _mm_set1_epi8(0x7f);
+    return (unsigned)_mm_cvtsi128_si32(_mm_sad_epu8(_mm_add_epi8(a.val, half),
+                                                    _mm_add_epi8(b.val, half)));
+}
+inline unsigned v_reduce_sad(const v_uint16x8& a, const v_uint16x8& b)
+{
+    v_uint32x4 l, h;
+    v_expand(v_absdiff(a, b), l, h);
+    return v_reduce_sum(l + h);
+}
+inline unsigned v_reduce_sad(const v_int16x8& a, const v_int16x8& b)
+{
+    v_uint32x4 l, h;
+    v_expand(v_absdiff(a, b), l, h);
+    return v_reduce_sum(l + h);
+}
+inline unsigned v_reduce_sad(const v_uint32x4& a, const v_uint32x4& b)
+{
+    return v_reduce_sum(v_absdiff(a, b));
+}
+inline unsigned v_reduce_sad(const v_int32x4& a, const v_int32x4& b)
+{
+    return v_reduce_sum(v_absdiff(a, b));
+}
+inline float v_reduce_sad(const v_float32x4& a, const v_float32x4& b)
+{
+    return v_reduce_sum(v_absdiff(a, b));
+}
+
 #define OPENCV_HAL_IMPL_SSE_POPCOUNT(_Tpvec) \
 inline v_uint32x4 v_popcount(const _Tpvec& a) \
 { \
@@ -1925,13 +1972,11 @@ inline void v_load_deinterleave(const unsigned* ptr, v_uint32x4& a, v_uint32x4&
 
 inline void v_load_deinterleave(const float* ptr, v_float32x4& a, v_float32x4& b)
 {
-    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
-
     __m128 u0 = _mm_loadu_ps(ptr);       // a0 b0 a1 b1
     __m128 u1 = _mm_loadu_ps((ptr + 4)); // a2 b2 a3 b3
 
-    a.val = _mm_shuffle_ps(u0, u1, mask_lo); // a0 a1 a2 a3
-    b.val = _mm_shuffle_ps(u0, u1, mask_hi); // b0 b1 ab b3
+    a.val = _mm_shuffle_ps(u0, u1, _MM_SHUFFLE(2, 0, 2, 0)); // a0 a1 a2 a3
+    b.val = _mm_shuffle_ps(u0, u1, _MM_SHUFFLE(3, 1, 3, 1)); // b0 b1 ab b3
 }
 
 inline void v_load_deinterleave(const float* ptr, v_float32x4& a, v_float32x4& b, v_float32x4& c)

modules/core/include/opencv2/core/hal/intrin_vsx.hpp

@@ -607,6 +607,11 @@ OPENCV_HAL_IMPL_VSX_INT_CMP_OP(v_float64x2)
 OPENCV_HAL_IMPL_VSX_INT_CMP_OP(v_uint64x2)
 OPENCV_HAL_IMPL_VSX_INT_CMP_OP(v_int64x2)
 
+inline v_float32x4 v_not_nan(const v_float32x4& a)
+{ return v_float32x4(vec_cmpeq(a.val, a.val)); }
+inline v_float64x2 v_not_nan(const v_float64x2& a)
+{ return v_float64x2(vec_cmpeq(a.val, a.val)); }
+
 /** min/max **/
 OPENCV_HAL_IMPL_VSX_BIN_FUNC(v_min, vec_min)
 OPENCV_HAL_IMPL_VSX_BIN_FUNC(v_max, vec_max)
@@ -711,6 +716,11 @@ OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(v_float32x4, vec_float4, float, sum, vec_add)
 OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(v_float32x4, vec_float4, float, max, vec_max)
 OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(v_float32x4, vec_float4, float, min, vec_min)
 
+inline double v_reduce_sum(const v_float64x2& a)
+{
+    return vec_extract(vec_add(a.val, vec_permi(a.val, a.val, 3)), 0);
+}
+
 #define OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(_Tpvec, _Tpvec2, scalartype, suffix, func) \
 inline scalartype v_reduce_##suffix(const _Tpvec& a)                               \
 {                                                                                  \
@@ -734,6 +744,50 @@ inline v_float32x4 v_reduce_sum4(const v_float32x4& a, const v_float32x4& b,
     return v_float32x4(vec_mergeh(ac, bd));
 }
 
+inline unsigned v_reduce_sad(const v_uint8x16& a, const v_uint8x16& b)
+{
+    const vec_uint4 zero4 = vec_uint4_z;
+    vec_uint4 sum4 = vec_sum4s(vec_absd(a.val, b.val), zero4);
+    return (unsigned)vec_extract(vec_sums(vec_int4_c(sum4), vec_int4_c(zero4)), 3);
+}
+inline unsigned v_reduce_sad(const v_int8x16& a, const v_int8x16& b)
+{
+    const vec_int4 zero4 = vec_int4_z;
+    vec_char16 ad = vec_abss(vec_subs(a.val, b.val));
+    vec_int4 sum4 = vec_sum4s(ad, zero4);
+    return (unsigned)vec_extract(vec_sums(sum4, zero4), 3);
+}
+inline unsigned v_reduce_sad(const v_uint16x8& a, const v_uint16x8& b)
+{
+    vec_ushort8 ad = vec_absd(a.val, b.val);
+    VSX_UNUSED(vec_int4) sum = vec_sums(vec_int4_c(vec_unpackhu(ad)), vec_int4_c(vec_unpacklu(ad)));
+    return (unsigned)vec_extract(sum, 3);
+}
+inline unsigned v_reduce_sad(const v_int16x8& a, const v_int16x8& b)
+{
+    const vec_int4 zero4 = vec_int4_z;
+    vec_short8 ad = vec_abss(vec_subs(a.val, b.val));
+    vec_int4 sum4 = vec_sum4s(ad, zero4);
+    return (unsigned)vec_extract(vec_sums(sum4, zero4), 3);
+}
+inline unsigned v_reduce_sad(const v_uint32x4& a, const v_uint32x4& b)
+{
+    const vec_uint4 ad = vec_absd(a.val, b.val);
+    const vec_uint4 rd = vec_add(ad, vec_sld(ad, ad, 8));
+    return vec_extract(vec_add(rd, vec_sld(rd, rd, 4)), 0);
+}
+inline unsigned v_reduce_sad(const v_int32x4& a, const v_int32x4& b)
+{
+    vec_int4 ad = vec_abss(vec_sub(a.val, b.val));
+    return (unsigned)vec_extract(vec_sums(ad, vec_int4_z), 3);
+}
+inline float v_reduce_sad(const v_float32x4& a, const v_float32x4& b)
+{
+    const vec_float4 ad = vec_abs(vec_sub(a.val, b.val));
+    const vec_float4 rd = vec_add(ad, vec_sld(ad, ad, 8));
+    return vec_extract(vec_add(rd, vec_sld(rd, rd, 4)), 0);
+}
+
 /** Popcount **/
 template<typename _Tpvec>
 inline v_uint32x4 v_popcount(const _Tpvec& a)

modules/core/include/opencv2/core/private.hpp

@@ -801,6 +801,82 @@ CV_EXPORTS InstrNode*   getCurrentNode();
 #define CV_INSTRUMENT_REGION(); CV_INSTRUMENT_REGION_();
 #endif
 
+namespace cv {
+
+namespace utils {
+
+//! @addtogroup core_utils
+//! @{
+
+/** @brief Try to find requested data file
+
+Search directories:
+
+1. Directories passed via `addDataSearchPath()`
+2. Check path specified by configuration parameter with "_HINT" suffix (name of environment variable).
+3. Check path specified by configuration parameter (name of environment variable).
+   If parameter value is not empty and nothing is found then stop searching.
+4. Detects build/install path based on:
+   a. current working directory (CWD)
+   b. and/or binary module location (opencv_core/opencv_world, doesn't work with static linkage)
+5. Scan `<source>/{,data}` directories if build directory is detected or the current directory is in source tree.
+6. Scan `<install>/share/OpenCV` directory if install directory is detected.
+
+@param relative_path Relative path to data file
+@param required Specify "file not found" handling.
+       If true, function prints information message and raises cv::Exception.
+       If false, function returns empty result
+@param configuration_parameter specify configuration parameter name. Default NULL value means "OPENCV_DATA_PATH".
+@return Returns path (absolute or relative to the current directory) or empty string if file is not found
+
+@note Implementation is not thread-safe.
+*/
+CV_EXPORTS
+cv::String findDataFile(const cv::String& relative_path, bool required = true,
+                        const char* configuration_parameter = NULL);
+
+/** @overload
+@param relative_path Relative path to data file
+@param configuration_parameter specify configuration parameter name. Default NULL value means "OPENCV_DATA_PATH".
+@param search_paths override addDataSearchPath() settings.
+@param subdir_paths override addDataSearchSubDirectory() settings.
+@return Returns path (absolute or relative to the current directory) or empty string if file is not found
+
+@note Implementation is not thread-safe.
+*/
+CV_EXPORTS
+cv::String findDataFile(const cv::String& relative_path,
+                        const char* configuration_parameter,
+                        const std::vector<String>* search_paths,
+                        const std::vector<String>* subdir_paths);
+
+/** @brief Override default search data path by adding new search location
+
+Use this only to override default behavior
+Passed paths are used in LIFO order.
+
+@param path Path to used samples data
+
+@note Implementation is not thread-safe.
+*/
+CV_EXPORTS void addDataSearchPath(const cv::String& path);
+
+/** @brief Append default search data sub directory
+
+General usage is to add OpenCV modules name (`<opencv_contrib>/modules/<name>/data` -> `modules/<name>/data` + `<name>/data`).
+Passed subdirectories are used in LIFO order.
+
+@param subdir samples data sub directory
+
+@note Implementation is not thread-safe.
+*/
+CV_EXPORTS void addDataSearchSubDirectory(const cv::String& subdir);
+
+//! @}
+
+} // namespace utils
+} // namespace cv
+
 //! @endcond
 
 #endif // OPENCV_CORE_PRIVATE_HPP

modules/core/include/opencv2/core/sse_utils.hpp

@@ -567,7 +567,7 @@ inline void _mm_deinterleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m
 
 inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m128 & v_g1)
 {
-    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
+    enum { mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1) };
 
     __m128 layer2_chunk0 = _mm_shuffle_ps(v_r0, v_r1, mask_lo);
     __m128 layer2_chunk2 = _mm_shuffle_ps(v_r0, v_r1, mask_hi);
@@ -588,7 +588,7 @@ inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m12
 inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0,
                               __m128 & v_g1, __m128 & v_b0, __m128 & v_b1)
 {
-    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
+    enum { mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1) };
 
     __m128 layer2_chunk0 = _mm_shuffle_ps(v_r0, v_r1, mask_lo);
     __m128 layer2_chunk3 = _mm_shuffle_ps(v_r0, v_r1, mask_hi);
@@ -615,7 +615,7 @@ inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0,
 inline void _mm_interleave_ps(__m128 & v_r0, __m128 & v_r1, __m128 & v_g0, __m128 & v_g1,
                               __m128 & v_b0, __m128 & v_b1, __m128 & v_a0, __m128 & v_a1)
 {
-    const int mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1);
+    enum { mask_lo = _MM_SHUFFLE(2, 0, 2, 0), mask_hi = _MM_SHUFFLE(3, 1, 3, 1) };
 
     __m128 layer2_chunk0 = _mm_shuffle_ps(v_r0, v_r1, mask_lo);
     __m128 layer2_chunk4 = _mm_shuffle_ps(v_r0, v_r1, mask_hi);

modules/core/include/opencv2/core/types.hpp

@@ -1884,8 +1884,11 @@ Rect_<_Tp>& operator += ( Rect_<_Tp>& a, const Size_<_Tp>& b )
 template<typename _Tp> static inline
 Rect_<_Tp>& operator -= ( Rect_<_Tp>& a, const Size_<_Tp>& b )
 {
-    a.width -= b.width;
-    a.height -= b.height;
+    const _Tp width = a.width - b.width;
+    const _Tp height = a.height - b.height;
+    CV_DbgAssert(width >= 0 && height >= 0);
+    a.width = width;
+    a.height = height;
     return a;
 }
 
@@ -1950,6 +1953,15 @@ Rect_<_Tp> operator + (const Rect_<_Tp>& a, const Size_<_Tp>& b)
     return Rect_<_Tp>( a.x, a.y, a.width + b.width, a.height + b.height );
 }
 
+template<typename _Tp> static inline
+Rect_<_Tp> operator - (const Rect_<_Tp>& a, const Size_<_Tp>& b)
+{
+    const _Tp width = a.width - b.width;
+    const _Tp height = a.height - b.height;
+    CV_DbgAssert(width >= 0 && height >= 0);
+    return Rect_<_Tp>( a.x, a.y, width, height );
+}
+
 template<typename _Tp> static inline
 Rect_<_Tp> operator & (const Rect_<_Tp>& a, const Rect_<_Tp>& b)
 {

modules/core/include/opencv2/core/types_c.h

@@ -364,7 +364,7 @@ IplImage;
 
 CV_INLINE IplImage cvIplImage()
 {
-#if !defined(CV__ENABLE_C_API_CTORS)
+#if !(defined(CV__ENABLE_C_API_CTORS) && defined(__cplusplus))
     IplImage self = CV_STRUCT_INITIALIZER; self.nSize = sizeof(IplImage); return self;
 #else
     return _IplImage();

modules/core/include/opencv2/core/utility.hpp

@@ -1274,8 +1274,75 @@ enum FLAGS
 CV_EXPORTS void       setFlags(FLAGS modeFlags);
 static inline void    setFlags(int modeFlags) { setFlags((FLAGS)modeFlags); }
 CV_EXPORTS FLAGS      getFlags();
+
+} // namespace instr
+
+
+namespace samples {
+
+//! @addtogroup core_utils_samples
+// This section describes utility functions for OpenCV samples.
+//
+// @note Implementation of these utilities is not thread-safe.
+//
+//! @{
+
+/** @brief Try to find requested data file
+
+Search directories:
+
+1. Directories passed via `addSamplesDataSearchPath()`
+2. OPENCV_SAMPLES_DATA_PATH_HINT environment variable
+3. OPENCV_SAMPLES_DATA_PATH environment variable
+   If parameter value is not empty and nothing is found then stop searching.
+4. Detects build/install path based on:
+   a. current working directory (CWD)
+   b. and/or binary module location (opencv_core/opencv_world, doesn't work with static linkage)
+5. Scan `<source>/{,data,samples/data}` directories if build directory is detected or the current directory is in source tree.
+6. Scan `<install>/share/OpenCV` directory if install directory is detected.
+
+@see cv::utils::findDataFile
+
+@param relative_path Relative path to data file
+@param required Specify "file not found" handling.
+       If true, function prints information message and raises cv::Exception.
+       If false, function returns empty result
+@param silentMode Disables messages
+@return Returns path (absolute or relative to the current directory) or empty string if file is not found
+*/
+CV_EXPORTS_W cv::String findFile(const cv::String& relative_path, bool required = true, bool silentMode = false);
+
+CV_EXPORTS_W cv::String findFileOrKeep(const cv::String& relative_path, bool silentMode = false);
+
+inline cv::String findFileOrKeep(const cv::String& relative_path, bool silentMode)
+{
+    cv::String res = findFile(relative_path, false, silentMode);
+    if (res.empty())
+        return relative_path;
+    return res;
 }
 
+/** @brief Override search data path by adding new search location
+
+Use this only to override default behavior
+Passed paths are used in LIFO order.
+
+@param path Path to used samples data
+*/
+CV_EXPORTS_W void addSamplesDataSearchPath(const cv::String& path);
+
+/** @brief Append samples search data sub directory
+
+General usage is to add OpenCV modules name (`<opencv_contrib>/modules/<name>/samples/data` -> `<name>/samples/data` + `modules/<name>/samples/data`).
+Passed subdirectories are used in LIFO order.
+
+@param subdir samples data sub directory
+*/
+CV_EXPORTS_W void addSamplesDataSearchSubDirectory(const cv::String& subdir);
+
+//! @}
+} // namespace samples
+
 namespace utils {
 
 CV_EXPORTS int getThreadID();

modules/core/include/opencv2/core/utils/filesystem.hpp

@@ -16,6 +16,13 @@ CV_EXPORTS void remove_all(const cv::String& path);
 
 CV_EXPORTS cv::String getcwd();
 
+/** @brief Converts path p to a canonical absolute path
+ * Symlinks are processed if there is support for them on running platform.
+ *
+ * @param path input path. Target file/directory should exist.
+ */
+CV_EXPORTS cv::String canonical(const cv::String& path);
+
 /** Join path components */
 CV_EXPORTS cv::String join(const cv::String& base, const cv::String& path);
 

modules/core/include/opencv2/core/version.hpp

@@ -7,8 +7,8 @@
 
 #define CV_VERSION_MAJOR    3
 #define CV_VERSION_MINOR    4
-#define CV_VERSION_REVISION 3
-#define CV_VERSION_STATUS   "-dev"
+#define CV_VERSION_REVISION 5
+#define CV_VERSION_STATUS   ""
 
 #define CVAUX_STR_EXP(__A)  #__A
 #define CVAUX_STR(__A)      CVAUX_STR_EXP(__A)

modules/core/perf/perf_norm.cpp

@@ -253,4 +253,53 @@ PERF_TEST_P( Size_MatType, normalize_minmax, TYPICAL_MATS )
     SANITY_CHECK(dst, 1e-6, ERROR_RELATIVE);
 }
 
+typedef TestBaseWithParam< int > test_len;
+PERF_TEST_P(test_len, hal_normL1_u8,
+            testing::Values(300000, 2000000)
+           )
+{
+    int len = GetParam();
+
+    Mat src1(1, len, CV_8UC1);
+    Mat src2(1, len, CV_8UC1);
+
+    declare.in(src1, src2, WARMUP_RNG);
+    double n;
+    TEST_CYCLE() n = hal::normL1_(src1.ptr<uchar>(0), src2.ptr<uchar>(0), len);
+    CV_UNUSED(n);
+    SANITY_CHECK_NOTHING();
+}
+
+PERF_TEST_P(test_len, hal_normL1_f32,
+            testing::Values(300000, 2000000)
+           )
+{
+    int len = GetParam();
+
+    Mat src1(1, len, CV_32FC1);
+    Mat src2(1, len, CV_32FC1);
+
+    declare.in(src1, src2, WARMUP_RNG);
+    double n;
+    TEST_CYCLE() n = hal::normL1_(src1.ptr<float>(0), src2.ptr<float>(0), len);
+    CV_UNUSED(n);
+    SANITY_CHECK_NOTHING();
+}
+
+PERF_TEST_P(test_len, hal_normL2Sqr,
+            testing::Values(300000, 2000000)
+           )
+{
+    int len = GetParam();
+
+    Mat src1(1, len, CV_32FC1);
+    Mat src2(1, len, CV_32FC1);
+
+    declare.in(src1, src2, WARMUP_RNG);
+    double n;
+    TEST_CYCLE() n = hal::normL2Sqr_(src1.ptr<float>(0), src2.ptr<float>(0), len);
+    CV_UNUSED(n);
+    SANITY_CHECK_NOTHING();
+}
+
 } // namespace

modules/core/src/arithm.cpp

@@ -1379,7 +1379,7 @@ struct InRange_SIMD
     }
 };
 
-#if CV_SIMD128
+#if CV_SIMD
 
 template <>
 struct InRange_SIMD<uchar>
@@ -1388,16 +1388,17 @@ struct InRange_SIMD<uchar>
         uchar * dst, int len) const
     {
         int x = 0;
-        const int width = v_uint8x16::nlanes;
+        const int width = v_uint8::nlanes;
 
         for (; x <= len - width; x += width)
         {
-            v_uint8x16 values = v_load(src1 + x);
-            v_uint8x16 low = v_load(src2 + x);
-            v_uint8x16 high = v_load(src3 + x);
+            v_uint8 values = vx_load(src1 + x);
+            v_uint8 low = vx_load(src2 + x);
+            v_uint8 high = vx_load(src3 + x);
 
             v_store(dst + x, (values >= low) & (high >= values));
         }
+        vx_cleanup();
         return x;
     }
 };
@@ -1409,16 +1410,17 @@ struct InRange_SIMD<schar>
         uchar * dst, int len) const
     {
         int x = 0;
-        const int width = v_int8x16::nlanes;
+        const int width = v_int8::nlanes;
 
         for (; x <= len - width; x += width)
         {
-            v_int8x16 values = v_load(src1 + x);
-            v_int8x16 low = v_load(src2 + x);
-            v_int8x16 high = v_load(src3 + x);
+            v_int8 values = vx_load(src1 + x);
+            v_int8 low = vx_load(src2 + x);
+            v_int8 high = vx_load(src3 + x);
 
             v_store((schar*)(dst + x), (values >= low) & (high >= values));
         }
+        vx_cleanup();
         return x;
     }
 };
@@ -1430,20 +1432,21 @@ struct InRange_SIMD<ushort>
         uchar * dst, int len) const
     {
         int x = 0;
-        const int width = v_uint16x8::nlanes * 2;
+        const int width = v_uint16::nlanes * 2;
 
         for (; x <= len - width; x += width)
         {
-            v_uint16x8 values1 = v_load(src1 + x);
-            v_uint16x8 low1 = v_load(src2 + x);
-            v_uint16x8 high1 = v_load(src3 + x);
+            v_uint16 values1 = vx_load(src1 + x);
+            v_uint16 low1 = vx_load(src2 + x);
+            v_uint16 high1 = vx_load(src3 + x);
 
-            v_uint16x8 values2 = v_load(src1 + x + v_uint16x8::nlanes);
-            v_uint16x8 low2 = v_load(src2 + x + v_uint16x8::nlanes);
-            v_uint16x8 high2 = v_load(src3 + x + v_uint16x8::nlanes);
+            v_uint16 values2 = vx_load(src1 + x + v_uint16::nlanes);
+            v_uint16 low2 = vx_load(src2 + x + v_uint16::nlanes);
+            v_uint16 high2 = vx_load(src3 + x + v_uint16::nlanes);
 
             v_store(dst + x, v_pack((values1 >= low1) & (high1 >= values1), (values2 >= low2) & (high2 >= values2)));
         }
+        vx_cleanup();
         return x;
     }
 };
@@ -1455,20 +1458,21 @@ struct InRange_SIMD<short>
         uchar * dst, int len) const
     {
         int x = 0;
-        const int width = (int)v_int16x8::nlanes * 2;
+        const int width = (int)v_int16::nlanes * 2;
 
         for (; x <= len - width; x += width)
         {
-            v_int16x8 values1 = v_load(src1 + x);
-            v_int16x8 low1 = v_load(src2 + x);
-            v_int16x8 high1 = v_load(src3 + x);
+            v_int16 values1 = vx_load(src1 + x);
+            v_int16 low1 = vx_load(src2 + x);
+            v_int16 high1 = vx_load(src3 + x);
 
-            v_int16x8 values2 = v_load(src1 + x + v_int16x8::nlanes);
-            v_int16x8 low2 = v_load(src2 + x + v_int16x8::nlanes);
-            v_int16x8 high2 = v_load(src3 + x + v_int16x8::nlanes);
+            v_int16 values2 = vx_load(src1 + x + v_int16::nlanes);
+            v_int16 low2 = vx_load(src2 + x + v_int16::nlanes);
+            v_int16 high2 = vx_load(src3 + x + v_int16::nlanes);
 
             v_store((schar*)(dst + x), v_pack((values1 >= low1) & (high1 >= values1), (values2 >= low2) & (high2 >= values2)));
         }
+        vx_cleanup();
         return x;
     }
 };
@@ -1480,20 +1484,21 @@ struct InRange_SIMD<int>
         uchar * dst, int len) const
     {
         int x = 0;
-        const int width = (int)v_int32x4::nlanes * 2;
+        const int width = (int)v_int32::nlanes * 2;
 
         for (; x <= len - width; x += width)
         {
-            v_int32x4 values1 = v_load(src1 + x);
-            v_int32x4 low1 = v_load(src2 + x);
-            v_int32x4 high1 = v_load(src3 + x);
+            v_int32 values1 = vx_load(src1 + x);
+            v_int32 low1 = vx_load(src2 + x);
+            v_int32 high1 = vx_load(src3 + x);
 
-            v_int32x4 values2 = v_load(src1 + x + v_int32x4::nlanes);
-            v_int32x4 low2 = v_load(src2 + x + v_int32x4::nlanes);
-            v_int32x4 high2 = v_load(src3 + x + v_int32x4::nlanes);
+            v_int32 values2 = vx_load(src1 + x + v_int32::nlanes);
+            v_int32 low2 = vx_load(src2 + x + v_int32::nlanes);
+            v_int32 high2 = vx_load(src3 + x + v_int32::nlanes);
 
             v_pack_store(dst + x, v_reinterpret_as_u16(v_pack((values1 >= low1) & (high1 >= values1), (values2 >= low2) & (high2 >= values2))));
         }
+        vx_cleanup();
         return x;
     }
 };
@@ -1505,20 +1510,21 @@ struct InRange_SIMD<float>
         uchar * dst, int len) const
     {
         int x = 0;
-        const int width = (int)v_float32x4::nlanes * 2;
+        const int width = (int)v_float32::nlanes * 2;
 
         for (; x <= len - width; x += width)
         {
-            v_float32x4 values1 = v_load(src1 + x);
-            v_float32x4 low1 = v_load(src2 + x);
-            v_float32x4 high1 = v_load(src3 + x);
+            v_float32 values1 = vx_load(src1 + x);
+            v_float32 low1 = vx_load(src2 + x);
+            v_float32 high1 = vx_load(src3 + x);
 
-            v_float32x4 values2 = v_load(src1 + x + v_float32x4::nlanes);
-            v_float32x4 low2 = v_load(src2 + x + v_float32x4::nlanes);
-            v_float32x4 high2 = v_load(src3 + x + v_float32x4::nlanes);
+            v_float32 values2 = vx_load(src1 + x + v_float32::nlanes);
+            v_float32 low2 = vx_load(src2 + x + v_float32::nlanes);
+            v_float32 high2 = vx_load(src3 + x + v_float32::nlanes);
 
             v_pack_store(dst + x, v_pack(v_reinterpret_as_u32((values1 >= low1) & (high1 >= values1)), v_reinterpret_as_u32((values2 >= low2) & (high2 >= values2))));
         }
+        vx_cleanup();
         return x;
     }
 };

modules/core/src/batch_distance.cpp

@@ -297,19 +297,21 @@ void cv::batchDistance( InputArray _src1, InputArray _src2,
         nidx = Scalar::all(-1);
     }
 
+
     if( crosscheck )
     {
         CV_Assert( K == 1 && update == 0 && mask.empty() );
         CV_Assert(!nidx.empty());
-        Mat tdist, tidx;
+        Mat tdist, tidx, sdist, sidx;
         batchDistance(src2, src1, tdist, dtype, tidx, normType, K, mask, 0, false);
+        batchDistance(src1, src2, sdist, dtype, sidx, normType, K, mask, 0, false);
 
         // if an idx-th element from src1 appeared to be the nearest to i-th element of src2,
         // we update the minimum mutual distance between idx-th element of src1 and the whole src2 set.
         // As a result, if nidx[idx] = i*, it means that idx-th element of src1 is the nearest
         // to i*-th element of src2 and i*-th element of src2 is the closest to idx-th element of src1.
         // If nidx[idx] = -1, it means that there is no such ideal couple for it in src2.
-        // This O(N) procedure is called cross-check and it helps to eliminate some false matches.
+        // This O(2N) procedure is called cross-check and it helps to eliminate some false matches.
         if( dtype == CV_32S )
         {
             for( int i = 0; i < tdist.rows; i++ )
@@ -336,6 +338,13 @@ void cv::batchDistance( InputArray _src1, InputArray _src2,
                 }
             }
         }
+        for( int i = 0; i < sdist.rows; i++ )
+        {
+            if( tidx.at<int>(sidx.at<int>(i)) != i )
+            {
+                nidx.at<int>(i) = -1;
+            }
+        }
         return;
     }
 

modules/core/src/copy.cpp

@@ -1183,9 +1183,9 @@ void cv::copyMakeBorder( InputArray _src, OutputArray _dst, int top, int bottom,
 {
     CV_INSTRUMENT_REGION();
 
-    CV_Assert( top >= 0 && bottom >= 0 && left >= 0 && right >= 0 );
+    CV_Assert( top >= 0 && bottom >= 0 && left >= 0 && right >= 0 && _src.dims() <= 2);
 
-    CV_OCL_RUN(_dst.isUMat() && _src.dims() <= 2,
+    CV_OCL_RUN(_dst.isUMat(),
                ocl_copyMakeBorder(_src, _dst, top, bottom, left, right, borderType, value))
 
     Mat src = _src.getMat();

modules/core/src/matmul.cpp

@@ -1699,7 +1699,7 @@ transform_( const T* src, T* dst, const WT* m, int len, int scn, int dcn )
     }
 }
 
-#if CV_SIMD128
+#if CV_SIMD128 && !defined(__aarch64__)
 static inline void
 load3x3Matrix(const float* m, v_float32x4& m0, v_float32x4& m1, v_float32x4& m2, v_float32x4& m3)
 {
@@ -1708,7 +1708,9 @@ load3x3Matrix(const float* m, v_float32x4& m0, v_float32x4& m1, v_float32x4& m2,
     m2 = v_float32x4(m[2], m[6], m[10], 0);
     m3 = v_float32x4(m[3], m[7], m[11], 0);
 }
+#endif
 
+#if CV_SIMD128
 static inline v_int16x8
 v_matmulvec(const v_int16x8 &v0, const v_int16x8 &m0, const v_int16x8 &m1, const v_int16x8 &m2, const v_int32x4 &m3, const int BITS)
 {

modules/core/src/norm.cpp

@@ -98,43 +98,15 @@ int normHamming(const uchar* a, const uchar* b, int n, int cellSize)
 float normL2Sqr_(const float* a, const float* b, int n)
 {
     int j = 0; float d = 0.f;
-#if CV_AVX2
-    float CV_DECL_ALIGNED(32) buf[8];
-    __m256 d0 = _mm256_setzero_ps();
-
-    for( ; j <= n - 8; j += 8 )
+#if CV_SIMD
+    v_float32 v_d = vx_setzero_f32();
+    for (; j <= n - v_float32::nlanes; j += v_float32::nlanes)
     {
-        __m256 t0 = _mm256_sub_ps(_mm256_loadu_ps(a + j), _mm256_loadu_ps(b + j));
-#if CV_FMA3
-        d0 = _mm256_fmadd_ps(t0, t0, d0);
-#else
-        d0 = _mm256_add_ps(d0, _mm256_mul_ps(t0, t0));
+        v_float32 t = vx_load(a + j) - vx_load(b + j);
+        v_d = v_muladd(t, t, v_d);
+    }
+    d = v_reduce_sum(v_d);
 #endif
-    }
-    _mm256_store_ps(buf, d0);
-    d = buf[0] + buf[1] + buf[2] + buf[3] + buf[4] + buf[5] + buf[6] + buf[7];
-#elif CV_SSE
-    float CV_DECL_ALIGNED(16) buf[4];
-    __m128 d0 = _mm_setzero_ps(), d1 = _mm_setzero_ps();
-
-    for( ; j <= n - 8; j += 8 )
-    {
-        __m128 t0 = _mm_sub_ps(_mm_loadu_ps(a + j), _mm_loadu_ps(b + j));
-        __m128 t1 = _mm_sub_ps(_mm_loadu_ps(a + j + 4), _mm_loadu_ps(b + j + 4));
-        d0 = _mm_add_ps(d0, _mm_mul_ps(t0, t0));
-        d1 = _mm_add_ps(d1, _mm_mul_ps(t1, t1));
-    }
-    _mm_store_ps(buf, _mm_add_ps(d0, d1));
-    d = buf[0] + buf[1] + buf[2] + buf[3];
-#endif
-    {
-        for( ; j <= n - 4; j += 4 )
-        {
-            float t0 = a[j] - b[j], t1 = a[j+1] - b[j+1], t2 = a[j+2] - b[j+2], t3 = a[j+3] - b[j+3];
-            d += t0*t0 + t1*t1 + t2*t2 + t3*t3;
-        }
-    }
-
     for( ; j < n; j++ )
     {
         float t = a[j] - b[j];
@@ -147,38 +119,12 @@ float normL2Sqr_(const float* a, const float* b, int n)
 float normL1_(const float* a, const float* b, int n)
 {
     int j = 0; float d = 0.f;
-#if CV_SSE
-    float CV_DECL_ALIGNED(16) buf[4];
-    static const int CV_DECL_ALIGNED(16) absbuf[4] = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff};
-    __m128 d0 = _mm_setzero_ps(), d1 = _mm_setzero_ps();
-    __m128 absmask = _mm_load_ps((const float*)absbuf);
-
-    for( ; j <= n - 8; j += 8 )
-    {
-        __m128 t0 = _mm_sub_ps(_mm_loadu_ps(a + j), _mm_loadu_ps(b + j));
-        __m128 t1 = _mm_sub_ps(_mm_loadu_ps(a + j + 4), _mm_loadu_ps(b + j + 4));
-        d0 = _mm_add_ps(d0, _mm_and_ps(t0, absmask));
-        d1 = _mm_add_ps(d1, _mm_and_ps(t1, absmask));
-    }
-    _mm_store_ps(buf, _mm_add_ps(d0, d1));
-    d = buf[0] + buf[1] + buf[2] + buf[3];
-#elif CV_NEON
-    float32x4_t v_sum = vdupq_n_f32(0.0f);
-    for ( ; j <= n - 4; j += 4)
-        v_sum = vaddq_f32(v_sum, vabdq_f32(vld1q_f32(a + j), vld1q_f32(b + j)));
-
-    float CV_DECL_ALIGNED(16) buf[4];
-    vst1q_f32(buf, v_sum);
-    d = buf[0] + buf[1] + buf[2] + buf[3];
+#if CV_SIMD
+    v_float32 v_d = vx_setzero_f32();
+    for (; j <= n - v_float32::nlanes; j += v_float32::nlanes)
+        v_d += v_absdiff(vx_load(a + j), vx_load(b + j));
+    d = v_reduce_sum(v_d);
 #endif
-    {
-        for( ; j <= n - 4; j += 4 )
-        {
-            d += std::abs(a[j] - b[j]) + std::abs(a[j+1] - b[j+1]) +
-            std::abs(a[j+2] - b[j+2]) + std::abs(a[j+3] - b[j+3]);
-        }
-    }
-
     for( ; j < n; j++ )
         d += std::abs(a[j] - b[j]);
     return d;
@@ -187,46 +133,10 @@ float normL1_(const float* a, const float* b, int n)
 int normL1_(const uchar* a, const uchar* b, int n)
 {
     int j = 0, d = 0;
-#if CV_SSE
-    __m128i d0 = _mm_setzero_si128();
-
-    for( ; j <= n - 16; j += 16 )
-    {
-        __m128i t0 = _mm_loadu_si128((const __m128i*)(a + j));
-        __m128i t1 = _mm_loadu_si128((const __m128i*)(b + j));
-
-        d0 = _mm_add_epi32(d0, _mm_sad_epu8(t0, t1));
-    }
-
-    for( ; j <= n - 4; j += 4 )
-    {
-        __m128i t0 = _mm_cvtsi32_si128(*(const int*)(a + j));
-        __m128i t1 = _mm_cvtsi32_si128(*(const int*)(b + j));
-
-        d0 = _mm_add_epi32(d0, _mm_sad_epu8(t0, t1));
-    }
-    d = _mm_cvtsi128_si32(_mm_add_epi32(d0, _mm_unpackhi_epi64(d0, d0)));
-#elif CV_NEON
-    uint32x4_t v_sum = vdupq_n_u32(0.0f);
-    for ( ; j <= n - 16; j += 16)
-    {
-        uint8x16_t v_dst = vabdq_u8(vld1q_u8(a + j), vld1q_u8(b + j));
-        uint16x8_t v_low = vmovl_u8(vget_low_u8(v_dst)), v_high = vmovl_u8(vget_high_u8(v_dst));
-        v_sum = vaddq_u32(v_sum, vaddl_u16(vget_low_u16(v_low), vget_low_u16(v_high)));
-        v_sum = vaddq_u32(v_sum, vaddl_u16(vget_high_u16(v_low), vget_high_u16(v_high)));
-    }
-
-    uint CV_DECL_ALIGNED(16) buf[4];
-    vst1q_u32(buf, v_sum);
-    d = buf[0] + buf[1] + buf[2] + buf[3];
+#if CV_SIMD
+    for (; j <= n - v_uint8::nlanes; j += v_uint8::nlanes)
+        d += v_reduce_sad(vx_load(a + j), vx_load(b + j));
 #endif
-    {
-        for( ; j <= n - 4; j += 4 )
-        {
-            d += std::abs(a[j] - b[j]) + std::abs(a[j+1] - b[j+1]) +
-            std::abs(a[j+2] - b[j+2]) + std::abs(a[j+3] - b[j+3]);
-        }
-    }
     for( ; j < n; j++ )
         d += std::abs(a[j] - b[j]);
     return d;

modules/core/src/ocl.cpp

@@ -1257,6 +1257,14 @@ struct Device::Impl
         else
             vendorID_ = UNKNOWN_VENDOR;
 
+        const size_t CV_OPENCL_DEVICE_MAX_WORK_GROUP_SIZE = utils::getConfigurationParameterSizeT("OPENCV_OPENCL_DEVICE_MAX_WORK_GROUP_SIZE", 0);
+        if (CV_OPENCL_DEVICE_MAX_WORK_GROUP_SIZE > 0)
+        {
+            const size_t new_maxWorkGroupSize = std::min(maxWorkGroupSize_, CV_OPENCL_DEVICE_MAX_WORK_GROUP_SIZE);
+            if (new_maxWorkGroupSize != maxWorkGroupSize_)
+                CV_LOG_WARNING(NULL, "OpenCL: using workgroup size: " << new_maxWorkGroupSize << " (was " << maxWorkGroupSize_ << ")");
+            maxWorkGroupSize_ = new_maxWorkGroupSize;
+        }
 #if 0
         if (isExtensionSupported("cl_khr_spir"))
         {
@@ -2777,6 +2785,7 @@ struct Kernel::Impl
         for( int i = 0; i < MAX_ARRS; i++ )
             u[i] = 0;
         haveTempDstUMats = false;
+        haveTempSrcUMats = false;
     }
 
     void cleanupUMats()
@@ -2793,6 +2802,7 @@ struct Kernel::Impl
             }
         nu = 0;
         haveTempDstUMats = false;
+        haveTempSrcUMats = false;
     }
 
     void addUMat(const UMat& m, bool dst)
@@ -2803,6 +2813,8 @@ struct Kernel::Impl
         nu++;
         if(dst && m.u->tempUMat())
             haveTempDstUMats = true;
+        if(m.u->originalUMatData == NULL && m.u->tempUMat())
+            haveTempSrcUMats = true;  // UMat is created on RAW memory (without proper lifetime management, even from Mat)
     }
 
     void addImage(const Image2D& image)
@@ -2840,6 +2852,7 @@ struct Kernel::Impl
     int nu;
     std::list<Image2D> images;
     bool haveTempDstUMats;
+    bool haveTempSrcUMats;
 };
 
 }} // namespace cv::ocl
@@ -3113,6 +3126,8 @@ bool Kernel::Impl::run(int dims, size_t globalsize[], size_t localsize[],
     cl_command_queue qq = getQueue(q);
     if (haveTempDstUMats)
         sync = true;
+    if (haveTempSrcUMats)
+        sync = true;
     if (timeNS)
         sync = true;
     cl_event asyncEvent = 0;

modules/core/src/parallel.cpp

@@ -427,7 +427,7 @@ static inline int _initMaxThreads()
     {
         omp_set_dynamic(maxThreads);
     }
-    return numThreads;
+    return maxThreads;
 }
 static int numThreadsMax = _initMaxThreads();
 #elif defined HAVE_GCD

modules/core/src/system.cpp

@@ -107,15 +107,14 @@ void* allocSingletonBuffer(size_t size) { return fastMalloc(size); }
 #  include <cpu-features.h>
 #endif
 
-#ifndef __VSX__
-# if defined __PPC64__ && defined __linux__
-#   include "sys/auxv.h"
-#   ifndef AT_HWCAP2
-#     define AT_HWCAP2 26
-#   endif
-#   ifndef PPC_FEATURE2_ARCH_2_07
-#     define PPC_FEATURE2_ARCH_2_07 0x80000000
-#   endif
+
+#if CV_VSX && defined __linux__
+# include "sys/auxv.h"
+# ifndef AT_HWCAP2
+#   define AT_HWCAP2 26
+# endif
+# ifndef PPC_FEATURE2_ARCH_3_00
+#   define PPC_FEATURE2_ARCH_3_00 0x00800000
 # endif
 #endif
 
@@ -359,6 +358,7 @@ struct HWFeatures
         g_hwFeatureNames[CPU_NEON] = "NEON";
 
         g_hwFeatureNames[CPU_VSX] = "VSX";
+        g_hwFeatureNames[CPU_VSX3] = "VSX3";
 
         g_hwFeatureNames[CPU_AVX512_SKX] = "AVX512-SKX";
     }
@@ -513,14 +513,14 @@ struct HWFeatures
     #endif
     #endif
 
-    #ifdef __VSX__
-        have[CV_CPU_VSX] = true;
-    #elif (defined __PPC64__ && defined __linux__)
-        uint64 hwcaps = getauxval(AT_HWCAP);
+    // there's no need to check VSX availability in runtime since it's always available on ppc64le CPUs
+    have[CV_CPU_VSX] = (CV_VSX);
+    // TODO: Check VSX3 availability in runtime for other platforms
+    #if CV_VSX && defined __linux__
         uint64 hwcap2 = getauxval(AT_HWCAP2);
-        have[CV_CPU_VSX] = (hwcaps & PPC_FEATURE_PPC_LE && hwcaps & PPC_FEATURE_HAS_VSX && hwcap2 & PPC_FEATURE2_ARCH_2_07);
+        have[CV_CPU_VSX3] = (hwcap2 & PPC_FEATURE2_ARCH_3_00);
     #else
-        have[CV_CPU_VSX] = false;
+        have[CV_CPU_VSX3] = (CV_VSX3);
     #endif
 
         int baseline_features[] = { CV_CPU_BASELINE_FEATURES };

modules/core/src/utils/datafile.cpp

@@ -0,0 +1,383 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+
+#include "../precomp.hpp"
+
+#include "opencv_data_config.hpp"
+
+#include <vector>
+#include <fstream>
+
+#include <opencv2/core/utils/logger.defines.hpp>
+#undef CV_LOG_STRIP_LEVEL
+#define CV_LOG_STRIP_LEVEL CV_LOG_LEVEL_VERBOSE + 1
+#include "opencv2/core/utils/logger.hpp"
+#include "opencv2/core/utils/filesystem.hpp"
+
+#include <opencv2/core/utils/configuration.private.hpp>
+
+#ifdef _WIN32
+#define WIN32_LEAN_AND_MEAN
+#include <windows.h>
+#undef small
+#undef min
+#undef max
+#undef abs
+#elif defined(__linux__)
+#include <dlfcn.h>  // requires -ldl
+#elif defined(__APPLE__)
+#include <TargetConditionals.h>
+#if TARGET_OS_MAC
+#include <dlfcn.h>
+#endif
+#endif
+
+namespace cv { namespace utils {
+
+static cv::Ptr< std::vector<cv::String> > g_data_search_path;
+static cv::Ptr< std::vector<cv::String> > g_data_search_subdir;
+
+static std::vector<cv::String>& _getDataSearchPath()
+{
+    if (g_data_search_path.empty())
+        g_data_search_path.reset(new std::vector<cv::String>());
+    return *(g_data_search_path.get());
+}
+
+static std::vector<cv::String>& _getDataSearchSubDirectory()
+{
+    if (g_data_search_subdir.empty())
+    {
+        g_data_search_subdir.reset(new std::vector<cv::String>());
+        g_data_search_subdir->push_back("data");
+        g_data_search_subdir->push_back("");
+    }
+    return *(g_data_search_subdir.get());
+}
+
+
+CV_EXPORTS void addDataSearchPath(const cv::String& path)
+{
+    if (utils::fs::isDirectory(path))
+        _getDataSearchPath().push_back(path);
+}
+CV_EXPORTS void addDataSearchSubDirectory(const cv::String& subdir)
+{
+    _getDataSearchSubDirectory().push_back(subdir);
+}
+
+static bool isPathSep(char c)
+{
+    return c == '/' || c == '\\';
+}
+static bool isSubDirectory_(const cv::String& base_path, const cv::String& path)
+{
+    size_t N = base_path.size();
+    if (N == 0)
+        return false;
+    if (isPathSep(base_path[N - 1]))
+        N--;
+    if (path.size() < N)
+        return false;
+    for (size_t i = 0; i < N; i++)
+    {
+        if (path[i] == base_path[i])
+            continue;
+        if (isPathSep(path[i]) && isPathSep(base_path[i]))
+            continue;
+        return false;
+    }
+    size_t M = path.size();
+    if (M > N)
+    {
+        if (!isPathSep(path[N]))
+            return false;
+    }
+    return true;
+}
+static bool isSubDirectory(const cv::String& base_path, const cv::String& path)
+{
+    bool res = isSubDirectory_(base_path, path);
+    CV_LOG_VERBOSE(NULL, 0, "isSubDirectory(): base: " << base_path << "  path: " << path << "  => result: " << (res ? "TRUE" : "FALSE"));
+    return res;
+}
+
+static cv::String getModuleLocation(const void* addr)
+{
+    CV_UNUSED(addr);
+#ifdef _WIN32
+    HMODULE m = 0;
+#if _WIN32_WINNT >= 0x0501
+    ::GetModuleHandleEx(GET_MODULE_HANDLE_EX_FLAG_FROM_ADDRESS | GET_MODULE_HANDLE_EX_FLAG_UNCHANGED_REFCOUNT,
+        reinterpret_cast<LPCTSTR>(addr),
+        &m);
+#endif
+    if (m)
+    {
+        char path[MAX_PATH];
+        const size_t path_size = sizeof(path)/sizeof(*path);
+        size_t sz = GetModuleFileNameA(m, path, path_size); // no unicode support
+        if (sz > 0 && sz < path_size)
+        {
+            path[sz] = '\0';
+            return cv::String(path);
+        }
+    }
+#elif defined(__linux__)
+    Dl_info info;
+    if (0 != dladdr(addr, &info))
+    {
+        return cv::String(info.dli_fname);
+    }
+#elif defined(__APPLE__)
+# if TARGET_OS_MAC
+    Dl_info info;
+    if (0 != dladdr(addr, &info))
+    {
+        return cv::String(info.dli_fname);
+    }
+# endif
+#else
+    // not supported, skip
+#endif
+    return cv::String();
+}
+
+cv::String findDataFile(const cv::String& relative_path,
+                        const char* configuration_parameter,
+                        const std::vector<String>* search_paths,
+                        const std::vector<String>* subdir_paths)
+{
+    configuration_parameter = configuration_parameter ? configuration_parameter : "OPENCV_DATA_PATH";
+    CV_LOG_DEBUG(NULL, cv::format("utils::findDataFile('%s', %s)", relative_path.c_str(), configuration_parameter));
+
+#define TRY_FILE_WITH_PREFIX(prefix) \
+{ \
+    cv::String path = utils::fs::join(prefix, relative_path); \
+    CV_LOG_DEBUG(NULL, cv::format("... Line %d: trying open '%s'", __LINE__, path.c_str())); \
+    FILE* f = fopen(path.c_str(), "rb"); \
+    if(f) { \
+        fclose(f); \
+        return path; \
+    } \
+}
+
+
+    // Step 0: check current directory or absolute path at first
+    TRY_FILE_WITH_PREFIX("");
+
+
+    // Step 1
+    const std::vector<cv::String>& search_path = search_paths ? *search_paths : _getDataSearchPath();
+    for(size_t i = search_path.size(); i > 0; i--)
+    {
+        const cv::String& prefix = search_path[i - 1];
+        TRY_FILE_WITH_PREFIX(prefix);
+    }
+
+    const std::vector<cv::String>& search_subdir = subdir_paths ? *subdir_paths : _getDataSearchSubDirectory();
+
+
+    // Step 2
+    const cv::String configuration_parameter_s(configuration_parameter ? configuration_parameter : "");
+    const cv::utils::Paths& search_hint = configuration_parameter_s.empty() ? cv::utils::Paths()
+                                          : getConfigurationParameterPaths((configuration_parameter_s + "_HINT").c_str());
+    for (size_t k = 0; k < search_hint.size(); k++)
+    {
+        cv::String datapath = search_hint[k];
+        if (datapath.empty())
+            continue;
+        if (utils::fs::isDirectory(datapath))
+        {
+            CV_LOG_DEBUG(NULL, "utils::findDataFile(): trying " << configuration_parameter << "_HINT=" << datapath);
+            for(size_t i = search_subdir.size(); i > 0; i--)
+            {
+                const cv::String& subdir = search_subdir[i - 1];
+                cv::String prefix = utils::fs::join(datapath, subdir);
+                TRY_FILE_WITH_PREFIX(prefix);
+            }
+        }
+        else
+        {
+            CV_LOG_WARNING(NULL, configuration_parameter << "_HINT is specified but it is not a directory: " << datapath);
+        }
+    }
+
+
+    // Step 3
+    const cv::utils::Paths& override_paths = configuration_parameter_s.empty() ? cv::utils::Paths()
+                                           : getConfigurationParameterPaths(configuration_parameter);
+    for (size_t k = 0; k < override_paths.size(); k++)
+    {
+        cv::String datapath = override_paths[k];
+        if (datapath.empty())
+            continue;
+        if (utils::fs::isDirectory(datapath))
+        {
+            CV_LOG_DEBUG(NULL, "utils::findDataFile(): trying " << configuration_parameter << "=" << datapath);
+            for(size_t i = search_subdir.size(); i > 0; i--)
+            {
+                const cv::String& subdir = search_subdir[i - 1];
+                cv::String prefix = utils::fs::join(datapath, subdir);
+                TRY_FILE_WITH_PREFIX(prefix);
+            }
+        }
+        else
+        {
+            CV_LOG_WARNING(NULL, configuration_parameter << " is specified but it is not a directory: " << datapath);
+        }
+    }
+    if (!override_paths.empty())
+    {
+        CV_LOG_INFO(NULL, "utils::findDataFile(): can't find data file via " << configuration_parameter << " configuration override: " << relative_path);
+        return cv::String();
+    }
+
+
+    // Steps: 4, 5, 6
+    cv::String cwd = utils::fs::getcwd();
+    cv::String build_dir(OPENCV_BUILD_DIR);
+    bool has_tested_build_directory = false;
+    if (isSubDirectory(build_dir, cwd) || isSubDirectory(utils::fs::canonical(build_dir), utils::fs::canonical(cwd)))
+    {
+        CV_LOG_DEBUG(NULL, "utils::findDataFile(): the current directory is build sub-directory: " << cwd);
+        const char* build_subdirs[] = { OPENCV_DATA_BUILD_DIR_SEARCH_PATHS };
+        for (size_t k = 0; k < sizeof(build_subdirs)/sizeof(build_subdirs[0]); k++)
+        {
+            CV_LOG_DEBUG(NULL, "utils::findDataFile(): <build>/" << build_subdirs[k]);
+            cv::String datapath = utils::fs::join(build_dir, build_subdirs[k]);
+            if (utils::fs::isDirectory(datapath))
+            {
+                for(size_t i = search_subdir.size(); i > 0; i--)
+                {
+                    const cv::String& subdir = search_subdir[i - 1];
+                    cv::String prefix = utils::fs::join(datapath, subdir);
+                    TRY_FILE_WITH_PREFIX(prefix);
+                }
+            }
+        }
+        has_tested_build_directory = true;
+    }
+
+    cv::String source_dir;
+    cv::String try_source_dir = cwd;
+    for (int levels = 0; levels < 3; ++levels)
+    {
+        if (utils::fs::exists(utils::fs::join(try_source_dir, "modules/core/include/opencv2/core/version.hpp")))
+        {
+            source_dir = try_source_dir;
+            break;
+        }
+        try_source_dir = utils::fs::join(try_source_dir, "/..");
+    }
+    if (!source_dir.empty())
+    {
+        CV_LOG_DEBUG(NULL, "utils::findDataFile(): the current directory is source sub-directory: " << source_dir);
+        CV_LOG_DEBUG(NULL, "utils::findDataFile(): <source>" << source_dir);
+        cv::String datapath = source_dir;
+        if (utils::fs::isDirectory(datapath))
+        {
+            for(size_t i = search_subdir.size(); i > 0; i--)
+            {
+                const cv::String& subdir = search_subdir[i - 1];
+                cv::String prefix = utils::fs::join(datapath, subdir);
+                TRY_FILE_WITH_PREFIX(prefix);
+            }
+        }
+    }
+
+    cv::String module_path = getModuleLocation((void*)getModuleLocation);  // use code addr, doesn't work with static linkage!
+    CV_LOG_DEBUG(NULL, "Detected module path: '" << module_path << '\'');
+
+    if (!has_tested_build_directory &&
+        (isSubDirectory(build_dir, module_path) || isSubDirectory(utils::fs::canonical(build_dir), utils::fs::canonical(module_path)))
+    )
+    {
+        CV_LOG_DEBUG(NULL, "utils::findDataFile(): the binary module directory is build sub-directory: " << module_path);
+        const char* build_subdirs[] = { OPENCV_DATA_BUILD_DIR_SEARCH_PATHS };
+        for (size_t k = 0; k < sizeof(build_subdirs)/sizeof(build_subdirs[0]); k++)
+        {
+            CV_LOG_DEBUG(NULL, "utils::findDataFile(): <build>/" << build_subdirs[k]);
+            cv::String datapath = utils::fs::join(build_dir, build_subdirs[k]);
+            if (utils::fs::isDirectory(datapath))
+            {
+                for(size_t i = search_subdir.size(); i > 0; i--)
+                {
+                    const cv::String& subdir = search_subdir[i - 1];
+                    cv::String prefix = utils::fs::join(datapath, subdir);
+                    TRY_FILE_WITH_PREFIX(prefix);
+                }
+            }
+        }
+    }
+
+#if defined OPENCV_INSTALL_DATA_DIR_RELATIVE
+    if (!module_path.empty())  // require module path
+    {
+        size_t pos = module_path.rfind('/');
+        if (pos == cv::String::npos)
+            pos = module_path.rfind('\\');
+        cv::String module_dir = (pos == cv::String::npos) ? module_path : module_path.substr(0, pos);
+        const char* install_subdirs[] = { OPENCV_INSTALL_DATA_DIR_RELATIVE };
+        for (size_t k = 0; k < sizeof(install_subdirs)/sizeof(install_subdirs[0]); k++)
+        {
+            cv::String datapath = utils::fs::join(module_dir, install_subdirs[k]);
+            CV_LOG_DEBUG(NULL, "utils::findDataFile(): trying install path (from binary path): " << datapath);
+            if (utils::fs::isDirectory(datapath))
+            {
+                for(size_t i = search_subdir.size(); i > 0; i--)
+                {
+                    const cv::String& subdir = search_subdir[i - 1];
+                    cv::String prefix = utils::fs::join(datapath, subdir);
+                    TRY_FILE_WITH_PREFIX(prefix);
+                }
+            }
+            else
+            {
+                CV_LOG_DEBUG(NULL, "utils::findDataFile(): ... skip, not a valid directory: " << datapath);
+            }
+        }
+    }
+#endif
+
+#if defined OPENCV_INSTALL_PREFIX && defined OPENCV_DATA_INSTALL_PATH
+    cv::String install_dir(OPENCV_INSTALL_PREFIX);
+    // use core/world module path and verify that library is running from installation directory
+    // It is neccessary to avoid touching of unrelated common /usr/local path
+    if (module_path.empty()) // can't determine
+        module_path = install_dir;
+    if (isSubDirectory(install_dir, module_path) || isSubDirectory(utils::fs::canonical(install_dir), utils::fs::canonical(module_path)))
+    {
+        cv::String datapath = utils::fs::join(install_dir, OPENCV_DATA_INSTALL_PATH);
+        if (utils::fs::isDirectory(datapath))
+        {
+            CV_LOG_DEBUG(NULL, "utils::findDataFile(): trying install path: " << datapath);
+            for(size_t i = search_subdir.size(); i > 0; i--)
+            {
+                const cv::String& subdir = search_subdir[i - 1];
+                cv::String prefix = utils::fs::join(datapath, subdir);
+                TRY_FILE_WITH_PREFIX(prefix);
+            }
+        }
+    }
+#endif
+
+    return cv::String();  // not found
+}
+
+cv::String findDataFile(const cv::String& relative_path, bool required, const char* configuration_parameter)
+{
+    CV_LOG_DEBUG(NULL, cv::format("cv::utils::findDataFile('%s', %s, %s)",
+                                  relative_path.c_str(), required ? "true" : "false",
+                                  configuration_parameter ? configuration_parameter : "NULL"));
+    cv::String result = cv::utils::findDataFile(relative_path,
+                                                configuration_parameter,
+                                                NULL,
+                                                NULL);
+    if (result.empty() && required)
+        CV_Error(cv::Error::StsError, cv::format("OpenCV: Can't find required data file: %s", relative_path.c_str()));
+    return result;
+}
+
+}} // namespace

modules/core/src/utils/filesystem.cpp

@@ -34,7 +34,7 @@
 #include <errno.h>
 #include <io.h>
 #include <stdio.h>
-#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__
+#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__ || defined __FreeBSD__
 #include <sys/types.h>
 #include <sys/stat.h>
 #include <fcntl.h>
@@ -85,6 +85,23 @@ cv::String join(const cv::String& base, const cv::String& path)
 
 #if OPENCV_HAVE_FILESYSTEM_SUPPORT
 
+cv::String canonical(const cv::String& path)
+{
+    cv::String result;
+#ifdef _WIN32
+    const char* result_str = _fullpath(NULL, path.c_str(), 0);
+#else
+    const char* result_str = realpath(path.c_str(), NULL);
+#endif
+    if (result_str)
+    {
+        result = cv::String(result_str);
+        free((void*)result_str);
+    }
+    return result.empty() ? path : result;
+}
+
+
 bool exists(const cv::String& path)
 {
     CV_INSTRUMENT_REGION();
@@ -161,7 +178,7 @@ cv::String getcwd()
     sz = GetCurrentDirectoryA((DWORD)buf.size(), buf.data());
     return cv::String(buf.data(), (size_t)sz);
 #endif
-#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__
+#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__ || defined __FreeBSD__
     for(;;)
     {
         char* p = ::getcwd(buf.data(), buf.size());
@@ -195,7 +212,7 @@ bool createDirectory(const cv::String& path)
 #else
     int result = _mkdir(path.c_str());
 #endif
-#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__
+#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__ || defined __FreeBSD__
     int result = mkdir(path.c_str(), 0777);
 #else
     int result = -1;
@@ -310,7 +327,7 @@ private:
     Impl& operator=(const Impl&); // disabled
 };
 
-#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__
+#elif defined __linux__ || defined __APPLE__ || defined __HAIKU__ || defined __FreeBSD__
 
 struct FileLock::Impl
 {
@@ -424,7 +441,7 @@ cv::String getCacheDirectory(const char* sub_directory_name, const char* configu
             default_cache_path = "/tmp/";
             CV_LOG_WARNING(NULL, "Using world accessible cache directory. This may be not secure: " << default_cache_path);
         }
-#elif defined __linux__ || defined __HAIKU__
+#elif defined __linux__ || defined __HAIKU__ || defined __FreeBSD__
         // https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html
         if (default_cache_path.empty())
         {
@@ -543,11 +560,13 @@ cv::String getCacheDirectory(const char* sub_directory_name, const char* configu
 
 #else
 #define NOT_IMPLEMENTED CV_Error(Error::StsNotImplemented, "");
-CV_EXPORTS bool exists(const cv::String& /*path*/) { NOT_IMPLEMENTED }
-CV_EXPORTS void remove_all(const cv::String& /*path*/) { NOT_IMPLEMENTED }
-CV_EXPORTS bool createDirectory(const cv::String& /*path*/) { NOT_IMPLEMENTED }
-CV_EXPORTS bool createDirectories(const cv::String& /*path*/) { NOT_IMPLEMENTED }
-CV_EXPORTS cv::String getCacheDirectory(const char* /*sub_directory_name*/, const char* /*configuration_name = NULL*/) { NOT_IMPLEMENTED }
+cv::String canonical(const cv::String& /*path*/) { NOT_IMPLEMENTED }
+bool exists(const cv::String& /*path*/) { NOT_IMPLEMENTED }
+void remove_all(const cv::String& /*path*/) { NOT_IMPLEMENTED }
+cv::String getcwd() { NOT_IMPLEMENTED }
+bool createDirectory(const cv::String& /*path*/) { NOT_IMPLEMENTED }
+bool createDirectories(const cv::String& /*path*/) { NOT_IMPLEMENTED }
+cv::String getCacheDirectory(const char* /*sub_directory_name*/, const char* /*configuration_name = NULL*/) { NOT_IMPLEMENTED }
 #undef NOT_IMPLEMENTED
 #endif // OPENCV_HAVE_FILESYSTEM_SUPPORT
 

modules/core/src/utils/samples.cpp

@@ -0,0 +1,67 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+
+#include "../precomp.hpp"
+
+#include <vector>
+
+#include <opencv2/core/utils/logger.defines.hpp>
+#undef CV_LOG_STRIP_LEVEL
+#define CV_LOG_STRIP_LEVEL CV_LOG_LEVEL_VERBOSE + 1
+#include "opencv2/core/utils/logger.hpp"
+#include "opencv2/core/utils/filesystem.hpp"
+
+namespace cv { namespace samples {
+
+static cv::Ptr< std::vector<cv::String> > g_data_search_path;
+static cv::Ptr< std::vector<cv::String> > g_data_search_subdir;
+
+static std::vector<cv::String>& _getDataSearchPath()
+{
+    if (g_data_search_path.empty())
+        g_data_search_path.reset(new std::vector<cv::String>());
+    return *(g_data_search_path.get());
+}
+
+static std::vector<cv::String>& _getDataSearchSubDirectory()
+{
+    if (g_data_search_subdir.empty())
+    {
+        g_data_search_subdir.reset(new std::vector<cv::String>());
+        g_data_search_subdir->push_back("samples/data");
+        g_data_search_subdir->push_back("data");
+        g_data_search_subdir->push_back("");
+    }
+    return *(g_data_search_subdir.get());
+}
+
+
+CV_EXPORTS void addSamplesDataSearchPath(const cv::String& path)
+{
+    if (utils::fs::isDirectory(path))
+        _getDataSearchPath().push_back(path);
+}
+CV_EXPORTS void addSamplesDataSearchSubDirectory(const cv::String& subdir)
+{
+    _getDataSearchSubDirectory().push_back(subdir);
+}
+
+cv::String findFile(const cv::String& relative_path, bool required, bool silentMode)
+{
+    CV_LOG_DEBUG(NULL, cv::format("cv::samples::findFile('%s', %s)", relative_path.c_str(), required ? "true" : "false"));
+    cv::String result = cv::utils::findDataFile(relative_path,
+                                                "OPENCV_SAMPLES_DATA_PATH",
+                                                &_getDataSearchPath(),
+                                                &_getDataSearchSubDirectory());
+    if (result != relative_path && !silentMode)
+    {
+        CV_LOG_WARNING(NULL, "cv::samples::findFile('" << relative_path << "') => '" << result << "'");
+    }
+    if (result.empty() && required)
+        CV_Error(cv::Error::StsError, cv::format("OpenCV samples: Can't find required data file: %s", relative_path.c_str()));
+    return result;
+}
+
+
+}} // namespace

modules/core/src/va_intel.cpp

@@ -340,8 +340,8 @@ static void copy_convert_yv12_to_bgr(const VAImage& image, const unsigned char*
             1.5959997177f
         };
 
-    CV_CheckEQ(image.format.fourcc, VA_FOURCC_YV12, "Unexpected image format");
-    CV_CheckEQ(image.num_planes, 3, "");
+    CV_CheckEQ((size_t)image.format.fourcc, (size_t)VA_FOURCC_YV12, "Unexpected image format");
+    CV_CheckEQ((size_t)image.num_planes, (size_t)3, "");
 
     const size_t srcOffsetY = image.offsets[0];
     const size_t srcOffsetV = image.offsets[1];
@@ -417,8 +417,8 @@ static void copy_convert_bgr_to_yv12(const VAImage& image, const Mat& bgr, unsig
             -0.2909994125f, 0.438999176f, -0.3679990768f, -0.0709991455f
         };
 
-    CV_CheckEQ(image.format.fourcc, VA_FOURCC_YV12, "Unexpected image format");
-    CV_CheckEQ(image.num_planes, 3, "");
+    CV_CheckEQ((size_t)image.format.fourcc, (size_t)VA_FOURCC_YV12, "Unexpected image format");
+    CV_CheckEQ((size_t)image.num_planes, (size_t)3, "");
 
     const size_t dstOffsetY = image.offsets[0];
     const size_t dstOffsetV = image.offsets[1];

modules/core/test/test_mat.cpp

@@ -3,6 +3,12 @@
 // of this distribution and at http://opencv.org/license.html.
 #include "test_precomp.hpp"
 
+#ifdef HAVE_EIGEN
+#include <Eigen/Core>
+#include <Eigen/Dense>
+#include "opencv2/core/eigen.hpp"
+#endif
+
 namespace opencv_test { namespace {
 
 class Core_ReduceTest : public cvtest::BaseTest
@@ -1972,4 +1978,22 @@ TEST(Core_Vectors, issue_13078_workaround)
     ASSERT_EQ(7, ints[3]);
 }
 
+
+#ifdef HAVE_EIGEN
+TEST(Core_Eigen, eigen2cv_check_Mat_type)
+{
+    Mat A(4, 4, CV_32FC1, Scalar::all(0));
+    Eigen::MatrixXf eigen_A;
+    cv2eigen(A, eigen_A);
+
+    Mat_<float> f_mat;
+    EXPECT_NO_THROW(eigen2cv(eigen_A, f_mat));
+    EXPECT_EQ(CV_32FC1, f_mat.type());
+
+    Mat_<double> d_mat;
+    EXPECT_ANY_THROW(eigen2cv(eigen_A, d_mat));
+    //EXPECT_EQ(CV_64FC1, d_mat.type());
+}
+#endif // HAVE_EIGEN
+
 }} // namespace

modules/core/test/test_operations.cpp

@@ -972,6 +972,13 @@ bool CV_OperationsTest::operations1()
         if (sz.width != 10 || sz.height != 20) throw test_excep();
         if (cvSize(sz).width != 10 || cvSize(sz).height != 20) throw test_excep();
 
+        Rect r1(0, 0, 10, 20);
+        Size sz1(5, 10);
+        r1 -= sz1;
+        if (r1.size().width != 5 || r1.size().height != 10) throw test_excep();
+        Rect r2 = r1 - sz1;
+        if (r2.size().width != 0 || r2.size().height != 0) throw test_excep();
+
         Vec<double, 5> v5d(1, 1, 1, 1, 1);
         Vec<double, 6> v6d(1, 1, 1, 1, 1, 1);
         Vec<double, 7> v7d(1, 1, 1, 1, 1, 1, 1);

modules/core/test/test_utils.cpp

@@ -2,6 +2,7 @@
 // It is subject to the license terms in the LICENSE file found in the top-level directory
 // of this distribution and at http://opencv.org/license.html.
 #include "test_precomp.hpp"
+#include "opencv2/core/utils/logger.hpp"
 
 namespace opencv_test { namespace {
 
@@ -283,4 +284,21 @@ TEST(CommandLineParser, testScalar)
     EXPECT_EQ(parser.get<Scalar>("s5"), Scalar(5, -4, 3, 2));
 }
 
+TEST(Samples, findFile)
+{
+    cv::utils::logging::LogLevel prev = cv::utils::logging::setLogLevel(cv::utils::logging::LOG_LEVEL_VERBOSE);
+    cv::String path;
+    ASSERT_NO_THROW(path = samples::findFile("lena.jpg", false));
+    EXPECT_NE(std::string(), path.c_str());
+    cv::utils::logging::setLogLevel(prev);
+}
+
+TEST(Samples, findFile_missing)
+{
+    cv::utils::logging::LogLevel prev = cv::utils::logging::setLogLevel(cv::utils::logging::LOG_LEVEL_VERBOSE);
+    cv::String path;
+    ASSERT_ANY_THROW(path = samples::findFile("non-existed.file", true));
+    cv::utils::logging::setLogLevel(prev);
+}
+
 }} // namespace

modules/dnn/CMakeLists.txt

@@ -20,11 +20,6 @@ else()
   ocv_cmake_hook_append(INIT_MODULE_SOURCES_opencv_dnn "${CMAKE_CURRENT_LIST_DIR}/cmake/hooks/INIT_MODULE_SOURCES_opencv_dnn.cmake")
 endif()
 
-ocv_warnings_disable(CMAKE_CXX_FLAGS -Wno-shadow -Wno-parentheses -Wmaybe-uninitialized -Wsign-promo
-                                     -Wmissing-declarations -Wmissing-prototypes
-)
-ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4701 /wd4100)
-
 if(MSVC)
   add_definitions( -D_CRT_SECURE_NO_WARNINGS=1 )
   ocv_warnings_disable(CMAKE_CXX_FLAGS /wd4244 /wd4267 /wd4018 /wd4355 /wd4800 /wd4251 /wd4996 /wd4146
@@ -33,12 +28,14 @@ if(MSVC)
   )
 else()
   ocv_warnings_disable(CMAKE_CXX_FLAGS -Wno-deprecated -Wmissing-prototypes -Wmissing-declarations -Wshadow
-                                       -Wunused-parameter -Wunused-local-typedefs -Wsign-compare -Wsign-promo
-                                       -Wundef -Wtautological-undefined-compare -Wignored-qualifiers -Wextra
-                                       -Wunused-function -Wunused-const-variable -Wdeprecated-declarations
+                                       -Wunused-parameter -Wsign-compare
   )
 endif()
 
+if(NOT HAVE_CXX11)
+  ocv_warnings_disable(CMAKE_CXX_FLAGS -Wno-undef)  # LANG_CXX11 from protobuf files
+endif()
+
 if(APPLE_FRAMEWORK)
   ocv_warnings_disable(CMAKE_CXX_FLAGS -Wshorten-64-to-32)
 endif()
@@ -55,8 +52,6 @@ add_definitions(-DHAVE_PROTOBUF=1)
 
 #suppress warnings in autogenerated caffe.pb.* files
 ocv_warnings_disable(CMAKE_CXX_FLAGS
-    -Wunused-parameter -Wundef -Wignored-qualifiers -Wno-enum-compare
-    -Wdeprecated-declarations
     /wd4125 /wd4267 /wd4127 /wd4244 /wd4512 /wd4702
     /wd4456 /wd4510 /wd4610 /wd4800
     /wd4701 /wd4703                    # potentially uninitialized local/pointer variable 'value' used

modules/dnn/include/opencv2/dnn/all_layers.hpp

@@ -77,6 +77,15 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
         static Ptr<Layer> create(const LayerParams &params);
     };
 
+    /**
+     * Constant layer produces the same data blob at an every forward pass.
+     */
+    class CV_EXPORTS ConstLayer : public Layer
+    {
+    public:
+        static Ptr<Layer> create(const LayerParams &params);
+    };
+
     //! LSTM recurrent layer
     class CV_EXPORTS LSTMLayer : public Layer
     {
@@ -236,7 +245,7 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
         int type;
         Size kernel, stride;
         int pad_l, pad_t, pad_r, pad_b;
-        CV_DEPRECATED Size pad;
+        CV_DEPRECATED_EXTERNAL Size pad;
         bool globalPooling;
         bool computeMaxIdx;
         String padMode;
@@ -578,7 +587,7 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
     {
     public:
         float pnorm, epsilon;
-        CV_DEPRECATED bool acrossSpatial;
+        CV_DEPRECATED_EXTERNAL bool acrossSpatial;
 
         static Ptr<NormalizeBBoxLayer> create(const LayerParams& params);
     };

modules/dnn/include/opencv2/dnn/dict.hpp

@@ -60,12 +60,13 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
 struct CV_EXPORTS_W DictValue
 {
     DictValue(const DictValue &r);
+    DictValue(bool i)           : type(Param::INT), pi(new AutoBuffer<int64,1>) { (*pi)[0] = i ? 1 : 0; }       //!< Constructs integer scalar
     DictValue(int64 i = 0)      : type(Param::INT), pi(new AutoBuffer<int64,1>) { (*pi)[0] = i; }       //!< Constructs integer scalar
-    CV_WRAP DictValue(int i)            : type(Param::INT), pi(new AutoBuffer<int64,1>) { (*pi)[0] = i; }       //!< Constructs integer scalar
+    CV_WRAP DictValue(int i)    : type(Param::INT), pi(new AutoBuffer<int64,1>) { (*pi)[0] = i; }       //!< Constructs integer scalar
     DictValue(unsigned p)       : type(Param::INT), pi(new AutoBuffer<int64,1>) { (*pi)[0] = p; }       //!< Constructs integer scalar
     CV_WRAP DictValue(double p)         : type(Param::REAL), pd(new AutoBuffer<double,1>) { (*pd)[0] = p; }     //!< Constructs floating point scalar
     CV_WRAP DictValue(const String &s)  : type(Param::STRING), ps(new AutoBuffer<String,1>) { (*ps)[0] = s; }   //!< Constructs string scalar
-    DictValue(const char *s)    : type(Param::STRING), ps(new AutoBuffer<String,1>) { (*ps)[0] = s; }   //!< @overload
+    DictValue(const char *s)            : type(Param::STRING), ps(new AutoBuffer<String,1>) { (*ps)[0] = s; }   //!< @overload
 
     template<typename TypeIter>
     static DictValue arrayInt(TypeIter begin, int size);    //!< Constructs integer array

modules/dnn/include/opencv2/dnn/dnn.hpp

@@ -46,9 +46,9 @@
 #include <opencv2/core.hpp>
 
 #if !defined CV_DOXYGEN && !defined CV_DNN_DONT_ADD_EXPERIMENTAL_NS
-#define CV__DNN_EXPERIMENTAL_NS_BEGIN namespace experimental_dnn_34_v9 {
+#define CV__DNN_EXPERIMENTAL_NS_BEGIN namespace experimental_dnn_34_v11 {
 #define CV__DNN_EXPERIMENTAL_NS_END }
-namespace cv { namespace dnn { namespace experimental_dnn_34_v9 { } using namespace experimental_dnn_34_v9; }}
+namespace cv { namespace dnn { namespace experimental_dnn_34_v11 { } using namespace experimental_dnn_34_v11; }}
 #else
 #define CV__DNN_EXPERIMENTAL_NS_BEGIN
 #define CV__DNN_EXPERIMENTAL_NS_END
@@ -88,9 +88,14 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
         DNN_TARGET_CPU,
         DNN_TARGET_OPENCL,
         DNN_TARGET_OPENCL_FP16,
-        DNN_TARGET_MYRIAD
+        DNN_TARGET_MYRIAD,
+        //! FPGA device with CPU fallbacks using Inference Engine's Heterogeneous plugin.
+        DNN_TARGET_FPGA
     };
 
+    CV_EXPORTS std::vector< std::pair<Backend, Target> > getAvailableBackends();
+    CV_EXPORTS std::vector<Target> getAvailableTargets(Backend be);
+
     /** @brief This class provides all data needed to initialize layer.
      *
      * It includes dictionary with scalar params (which can be read by using Dict interface),
@@ -186,7 +191,8 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
          * If this method is called after network has allocated all memory for input and output blobs
          * and before inferencing.
          */
-        CV_DEPRECATED virtual void finalize(const std::vector<Mat*> &input, std::vector<Mat> &output);
+        CV_DEPRECATED_EXTERNAL
+        virtual void finalize(const std::vector<Mat*> &input, std::vector<Mat> &output);
 
         /** @brief Computes and sets internal parameters according to inputs, outputs and blobs.
          *  @param[in]  inputs  vector of already allocated input blobs
@@ -203,7 +209,8 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
          *  @param[out] output allocated output blobs, which will store results of the computation.
          *  @param[out] internals allocated internal blobs
          */
-        CV_DEPRECATED virtual void forward(std::vector<Mat*> &input, std::vector<Mat> &output, std::vector<Mat> &internals);
+        CV_DEPRECATED_EXTERNAL
+        virtual void forward(std::vector<Mat*> &input, std::vector<Mat> &output, std::vector<Mat> &internals);
 
         /** @brief Given the @p input blobs, computes the output @p blobs.
          *  @param[in]  inputs  the input blobs.
@@ -223,7 +230,8 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
          * @overload
          * @deprecated Use Layer::finalize(InputArrayOfArrays, OutputArrayOfArrays) instead
          */
-        CV_DEPRECATED void finalize(const std::vector<Mat> &inputs, CV_OUT std::vector<Mat> &outputs);
+        CV_DEPRECATED_EXTERNAL
+        void finalize(const std::vector<Mat> &inputs, CV_OUT std::vector<Mat> &outputs);
 
         /** @brief
          * @overload
@@ -498,6 +506,7 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
          * | DNN_TARGET_OPENCL      |                  + |                            + |                  + |
          * | DNN_TARGET_OPENCL_FP16 |                  + |                            + |                    |
          * | DNN_TARGET_MYRIAD      |                    |                            + |                    |
+         * | DNN_TARGET_FPGA        |                    |                            + |                    |
          */
         CV_WRAP void setPreferableTarget(int targetId);
 
@@ -745,6 +754,7 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
      *  @brief Reads a network model stored in <a href="http://torch.ch">Torch7</a> framework's format.
      *  @param model    path to the file, dumped from Torch by using torch.save() function.
      *  @param isBinary specifies whether the network was serialized in ascii mode or binary.
+     *  @param evaluate specifies testing phase of network. If true, it's similar to evaluate() method in Torch.
      *  @returns Net object.
      *
      *  @note Ascii mode of Torch serializer is more preferable, because binary mode extensively use `long` type of C language,
@@ -766,7 +776,7 @@ CV__DNN_EXPERIMENTAL_NS_BEGIN
      *
      * Also some equivalents of these classes from cunn, cudnn, and fbcunn may be successfully imported.
      */
-     CV_EXPORTS_W Net readNetFromTorch(const String &model, bool isBinary = true);
+     CV_EXPORTS_W Net readNetFromTorch(const String &model, bool isBinary = true, bool evaluate = true);
 
      /**
       * @brief Read deep learning network represented in one of the supported formats.

modules/dnn/perf/perf_net.cpp

@@ -31,23 +31,6 @@ public:
     void processNet(std::string weights, std::string proto, std::string halide_scheduler,
                     const Mat& input, const std::string& outputLayer = "")
     {
-        if (backend == DNN_BACKEND_OPENCV && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16))
-        {
-#if defined(HAVE_OPENCL)
-            if (!cv::ocl::useOpenCL())
-#endif
-            {
-                throw cvtest::SkipTestException("OpenCL is not available/disabled in OpenCV");
-            }
-        }
-        if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
-        {
-            if (!checkMyriadTarget())
-            {
-                throw SkipTestException("Myriad is not available/disabled in OpenCV");
-            }
-        }
-
         randu(input, 0.0f, 1.0f);
 
         weights = findDataFile(weights, false);
@@ -175,8 +158,7 @@ PERF_TEST_P_(DNNTestNetwork, MobileNet_SSD_v2_TensorFlow)
 PERF_TEST_P_(DNNTestNetwork, DenseNet_121)
 {
     if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && (target == DNN_TARGET_OPENCL_FP16 ||
-                                                    target == DNN_TARGET_MYRIAD))
+        (backend == DNN_BACKEND_INFERENCE_ENGINE && (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD)))
         throw SkipTestException("");
     processNet("dnn/DenseNet_121.caffemodel", "dnn/DenseNet_121.prototxt", "",
                Mat(cv::Size(224, 224), CV_32FC3));
@@ -185,7 +167,7 @@ PERF_TEST_P_(DNNTestNetwork, DenseNet_121)
 PERF_TEST_P_(DNNTestNetwork, OpenPose_pose_coco)
 {
     if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+        (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
         throw SkipTestException("");
     processNet("dnn/openpose_pose_coco.caffemodel", "dnn/openpose_pose_coco.prototxt", "",
                Mat(cv::Size(368, 368), CV_32FC3));
@@ -194,7 +176,7 @@ PERF_TEST_P_(DNNTestNetwork, OpenPose_pose_coco)
 PERF_TEST_P_(DNNTestNetwork, OpenPose_pose_mpi)
 {
     if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+        (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
         throw SkipTestException("");
     processNet("dnn/openpose_pose_mpi.caffemodel", "dnn/openpose_pose_mpi.prototxt", "",
                Mat(cv::Size(368, 368), CV_32FC3));
@@ -203,7 +185,7 @@ PERF_TEST_P_(DNNTestNetwork, OpenPose_pose_mpi)
 PERF_TEST_P_(DNNTestNetwork, OpenPose_pose_mpi_faster_4_stages)
 {
     if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+        (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
         throw SkipTestException("");
     // The same .caffemodel but modified .prototxt
     // See https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/src/openpose/pose/poseParameters.cpp
@@ -230,7 +212,7 @@ PERF_TEST_P_(DNNTestNetwork, Inception_v2_SSD_TensorFlow)
 PERF_TEST_P_(DNNTestNetwork, YOLOv3)
 {
     if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+        (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
         throw SkipTestException("");
     Mat sample = imread(findDataFile("dnn/dog416.png", false));
     Mat inp;
@@ -241,7 +223,7 @@ PERF_TEST_P_(DNNTestNetwork, YOLOv3)
 PERF_TEST_P_(DNNTestNetwork, EAST_text_detection)
 {
     if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+        (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
         throw SkipTestException("");
     processNet("dnn/frozen_east_text_detection.pb", "", "", Mat(cv::Size(320, 320), CV_32FC3));
 }

modules/dnn/src/caffe/caffe_io.cpp

@@ -404,7 +404,7 @@ bool UpgradeV0LayerParameter(V1LayerParameter* v0_layer_connection_,
               PoolingParameter_PoolMethod_STOCHASTIC);
           break;
         default:
-          LOG(ERROR) << "Unknown pool method " << pool;
+          LOG(ERROR) << "Unknown pool method " << (int)pool;
           is_fully_compatible = false;
         }
       } else {
@@ -863,7 +863,7 @@ bool UpgradeV1LayerParameter(V1LayerParameter* v1_layer_param_,
     while (layer_param->param_size() <= i) { layer_param->add_param(); }
     layer_param->mutable_param(i)->set_name(v1_layer_param.param(i));
   }
-  ParamSpec_DimCheckMode mode;
+  ParamSpec_DimCheckMode mode = ParamSpec_DimCheckMode_STRICT;
   for (int i = 0; i < v1_layer_param.blob_share_mode_size(); ++i) {
     while (layer_param->param_size() <= i) { layer_param->add_param(); }
     switch (v1_layer_param.blob_share_mode(i)) {
@@ -875,8 +875,8 @@ bool UpgradeV1LayerParameter(V1LayerParameter* v1_layer_param_,
       break;
     default:
       LOG(FATAL) << "Unknown blob_share_mode: "
-                 << v1_layer_param.blob_share_mode(i);
-      break;
+                 << (int)v1_layer_param.blob_share_mode(i);
+      CV_Error_(Error::StsError, ("Unknown blob_share_mode: %d", (int)v1_layer_param.blob_share_mode(i)));
     }
     layer_param->mutable_param(i)->set_share_mode(mode);
   }
@@ -1102,12 +1102,12 @@ const char* UpgradeV1LayerType(const V1LayerParameter_LayerType type) {
   case V1LayerParameter_LayerType_THRESHOLD:
     return "Threshold";
   default:
-    LOG(FATAL) << "Unknown V1LayerParameter layer type: " << type;
+    LOG(FATAL) << "Unknown V1LayerParameter layer type: " << (int)type;
     return "";
   }
 }
 
-const int kProtoReadBytesLimit = INT_MAX;  // Max size of 2 GB minus 1 byte.
+static const int kProtoReadBytesLimit = INT_MAX;  // Max size of 2 GB minus 1 byte.
 
 bool ReadProtoFromBinary(ZeroCopyInputStream* input, Message *proto) {
     CodedInputStream coded_input(input);

modules/dnn/src/dnn.cpp

@@ -84,6 +84,107 @@ using std::map;
 using std::make_pair;
 using std::set;
 
+//==================================================================================================
+
+class BackendRegistry
+{
+public:
+    typedef std::vector< std::pair<Backend, Target> > BackendsList;
+    const BackendsList & getBackends() const { return backends; }
+    static BackendRegistry & getRegistry()
+    {
+        static BackendRegistry impl;
+        return impl;
+    }
+private:
+    BackendRegistry()
+    {
+#ifdef HAVE_HALIDE
+        backends.push_back(std::make_pair(DNN_BACKEND_HALIDE, DNN_TARGET_CPU));
+#  ifdef HAVE_OPENCL
+        if (cv::ocl::useOpenCL())
+            backends.push_back(std::make_pair(DNN_BACKEND_HALIDE, DNN_TARGET_OPENCL));
+#  endif
+#endif // HAVE_HALIDE
+
+#ifdef HAVE_INF_ENGINE
+        if (checkIETarget(DNN_TARGET_CPU))
+            backends.push_back(std::make_pair(DNN_BACKEND_INFERENCE_ENGINE, DNN_TARGET_CPU));
+        if (checkIETarget(DNN_TARGET_MYRIAD))
+            backends.push_back(std::make_pair(DNN_BACKEND_INFERENCE_ENGINE, DNN_TARGET_MYRIAD));
+        if (checkIETarget(DNN_TARGET_FPGA))
+            backends.push_back(std::make_pair(DNN_BACKEND_INFERENCE_ENGINE, DNN_TARGET_FPGA));
+#  ifdef HAVE_OPENCL
+        if (cv::ocl::useOpenCL() && ocl::Device::getDefault().isIntel())
+        {
+            if (checkIETarget(DNN_TARGET_OPENCL))
+                backends.push_back(std::make_pair(DNN_BACKEND_INFERENCE_ENGINE, DNN_TARGET_OPENCL));
+            if (checkIETarget(DNN_TARGET_OPENCL_FP16))
+                backends.push_back(std::make_pair(DNN_BACKEND_INFERENCE_ENGINE, DNN_TARGET_OPENCL_FP16));
+        }
+#  endif
+#endif // HAVE_INF_ENGINE
+
+#ifdef HAVE_OPENCL
+        if (cv::ocl::useOpenCL())
+        {
+            backends.push_back(std::make_pair(DNN_BACKEND_OPENCV, DNN_TARGET_OPENCL));
+            backends.push_back(std::make_pair(DNN_BACKEND_OPENCV, DNN_TARGET_OPENCL_FP16));
+        }
+#endif
+
+        backends.push_back(std::make_pair(DNN_BACKEND_OPENCV, DNN_TARGET_CPU));
+    }
+    static inline bool checkIETarget(int target)
+    {
+#ifndef HAVE_INF_ENGINE
+        return false;
+#else
+        cv::dnn::Net net;
+        cv::dnn::LayerParams lp;
+        net.addLayerToPrev("testLayer", "Identity", lp);
+        net.setPreferableBackend(cv::dnn::DNN_BACKEND_INFERENCE_ENGINE);
+        net.setPreferableTarget(target);
+        static int inpDims[] = {1, 2, 3, 4};
+        net.setInput(cv::Mat(4, &inpDims[0], CV_32FC1, cv::Scalar(0)));
+        try
+        {
+            net.forward();
+        }
+        catch(...)
+        {
+            return false;
+        }
+        return true;
+#endif
+    }
+
+    BackendsList backends;
+};
+
+
+std::vector< std::pair<Backend, Target> > getAvailableBackends()
+{
+    return BackendRegistry::getRegistry().getBackends();
+}
+
+std::vector<Target> getAvailableTargets(Backend be)
+{
+    if (be == DNN_BACKEND_DEFAULT)
+        be = (Backend)PARAM_DNN_BACKEND_DEFAULT;
+
+    std::vector<Target> result;
+    const BackendRegistry::BackendsList all_backends = getAvailableBackends();
+    for(BackendRegistry::BackendsList::const_iterator i = all_backends.begin(); i != all_backends.end(); ++i )
+    {
+        if (i->first == be)
+            result.push_back(i->second);
+    }
+    return result;
+}
+
+//==================================================================================================
+
 namespace
 {
     typedef std::vector<MatShape> ShapesVec;
@@ -352,7 +453,7 @@ struct LayerPin
 
     bool operator<(const LayerPin &r) const
     {
-        return lid < r.lid || lid == r.lid && oid < r.oid;
+        return lid < r.lid || (lid == r.lid && oid < r.oid);
     }
 
     bool operator ==(const LayerPin &r) const
@@ -427,7 +528,7 @@ struct DataLayer : public Layer
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && inputsData.size() == 1;
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && inputsData.size() == 1);
     }
 
     void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE
@@ -1077,7 +1178,8 @@ struct Net::Impl
                   preferableTarget == DNN_TARGET_CPU ||
                   preferableTarget == DNN_TARGET_OPENCL ||
                   preferableTarget == DNN_TARGET_OPENCL_FP16 ||
-                  preferableTarget == DNN_TARGET_MYRIAD);
+                  preferableTarget == DNN_TARGET_MYRIAD ||
+                  preferableTarget == DNN_TARGET_FPGA);
         if (!netWasAllocated || this->blobsToKeep != blobsToKeep_)
         {
             if (preferableBackend == DNN_BACKEND_OPENCV && IS_DNN_OPENCL_TARGET(preferableTarget))
@@ -1512,7 +1614,9 @@ struct Net::Impl
             ieNode->net = net;
 
             auto weightableLayer = std::dynamic_pointer_cast<InferenceEngine::WeightableLayer>(ieNode->layer);
-            if ((preferableTarget == DNN_TARGET_OPENCL_FP16 || preferableTarget == DNN_TARGET_MYRIAD) && !fused)
+            if ((preferableTarget == DNN_TARGET_OPENCL_FP16 ||
+                 preferableTarget == DNN_TARGET_MYRIAD ||
+                 preferableTarget == DNN_TARGET_FPGA) && !fused)
             {
                 ieNode->layer->precision = InferenceEngine::Precision::FP16;
                 if (weightableLayer)
@@ -1568,6 +1672,23 @@ struct Net::Impl
 
             if (!ieNode->net->isInitialized())
             {
+#if INF_ENGINE_VER_MAJOR_GT(INF_ENGINE_RELEASE_2018R3)
+                // For networks which is built in runtime we need to specify a
+                // version of it's hyperparameters.
+                std::string versionTrigger = "<net name=\"TestInput\" version=\"3\" batch=\"1\">"
+                                               "<layers>"
+                                                 "<layer name=\"data\" type=\"Input\" precision=\"FP32\" id=\"0\">"
+                                                   "<output>"
+                                                     "<port id=\"0\">"
+                                                       "<dim>1</dim>"
+                                                     "</port>"
+                                                   "</output>"
+                                                 "</layer>"
+                                               "</layers>"
+                                             "</net>";
+                InferenceEngine::CNNNetReader reader;
+                reader.ReadNetwork(versionTrigger.data(), versionTrigger.size());
+#endif
                 ieNode->net->init(preferableTarget);
                 ld.skip = false;
             }
@@ -1690,8 +1811,8 @@ struct Net::Impl
 
     void fuseLayers(const std::vector<LayerPin>& blobsToKeep_)
     {
-        if( !fusion || preferableBackend != DNN_BACKEND_OPENCV &&
-                       preferableBackend != DNN_BACKEND_INFERENCE_ENGINE)
+        if( !fusion || (preferableBackend != DNN_BACKEND_OPENCV &&
+                        preferableBackend != DNN_BACKEND_INFERENCE_ENGINE))
             return;
 
         CV_TRACE_FUNCTION();
@@ -1899,6 +2020,9 @@ struct Net::Impl
                 }
             }
 
+            if (preferableBackend != DNN_BACKEND_OPENCV)
+                continue;  // Go to the next layer.
+
             // the optimization #2. if there is no layer that takes max pooling layer's computed
             // max indices (and only some semantical segmentation networks might need this;
             // many others only take the maximum values), then we switch the max pooling

modules/dnn/src/init.cpp

@@ -112,6 +112,7 @@ void initializeLayerFactory()
     CV_DNN_REGISTER_LAYER_CLASS(Dropout,        BlankLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Identity,       BlankLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Silence,        BlankLayer);
+    CV_DNN_REGISTER_LAYER_CLASS(Const,          ConstLayer);
 
     CV_DNN_REGISTER_LAYER_CLASS(Crop,           CropLayer);
     CV_DNN_REGISTER_LAYER_CLASS(Eltwise,        EltwiseLayer);

modules/dnn/src/layers/batch_norm_layer.cpp

@@ -10,6 +10,7 @@ Implementation of Batch Normalization layer.
 */
 
 #include "../precomp.hpp"
+#include "layers_common.hpp"
 #include "../op_halide.hpp"
 #include "../op_inf_engine.hpp"
 #include <opencv2/dnn/shape_utils.hpp>
@@ -150,8 +151,8 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_HALIDE && haveHalide() ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+               (backendId == DNN_BACKEND_HALIDE && haveHalide()) ||
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
     }
 
 #ifdef HAVE_OPENCL
@@ -284,10 +285,10 @@ public:
                 v_float32x4 x1 = v_load(srcptr + i + 4);
                 v_float32x4 x2 = v_load(srcptr + i + 8);
                 v_float32x4 x3 = v_load(srcptr + i + 12);
-                x0 = v_muladd(x0, w, b);
-                x1 = v_muladd(x1, w, b);
-                x2 = v_muladd(x2, w, b);
-                x3 = v_muladd(x3, w, b);
+                x0 = v_muladd(x0, wV, bV);
+                x1 = v_muladd(x1, wV, bV);
+                x2 = v_muladd(x2, wV, bV);
+                x3 = v_muladd(x3, wV, bV);
                 v_store(dstptr + i, x0);
                 v_store(dstptr + i + 4, x1);
                 v_store(dstptr + i + 8, x2);

modules/dnn/src/layers/blank_layer.cpp

@@ -57,7 +57,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,
@@ -107,14 +107,21 @@ public:
                 inputs[i].copyTo(outputs[i]);
     }
 
-    virtual Ptr<BackendNode> initInfEngine(const std::vector<Ptr<BackendWrapper> >&) CV_OVERRIDE
+    virtual Ptr<BackendNode> initInfEngine(const std::vector<Ptr<BackendWrapper> >& inputs) CV_OVERRIDE
     {
 #ifdef HAVE_INF_ENGINE
+        InferenceEngine::DataPtr input = infEngineDataNode(inputs[0]);
+        CV_Assert(!input->dims.empty());
+
         InferenceEngine::LayerParams lp;
         lp.name = name;
         lp.type = "Split";
         lp.precision = InferenceEngine::Precision::FP32;
         std::shared_ptr<InferenceEngine::SplitLayer> ieLayer(new InferenceEngine::SplitLayer(lp));
+#if INF_ENGINE_VER_MAJOR_GT(INF_ENGINE_RELEASE_2018R3)
+        ieLayer->params["axis"] = format("%d", (int)input->dims.size() - 1);
+        ieLayer->params["out_sizes"] = format("%d", (int)input->dims[0]);
+#endif
         return Ptr<BackendNode>(new InfEngineBackendNode(ieLayer));
 #endif  // HAVE_INF_ENGINE
         return Ptr<BackendNode>();

modules/dnn/src/layers/concat_layer.cpp

@@ -104,8 +104,8 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_HALIDE && haveHalide() && axis == 1 && !padding ||  // By channels
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && !padding;
+               (backendId == DNN_BACKEND_HALIDE && haveHalide() && axis == 1 && !padding) ||  // By channels
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && !padding);
     }
 
     class ChannelConcatInvoker : public ParallelLoopBody

modules/dnn/src/layers/const_layer.cpp

@@ -0,0 +1,68 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+
+// Copyright (C) 2018, Intel Corporation, all rights reserved.
+// Third party copyrights are property of their respective owners.
+
+#include "../precomp.hpp"
+#include "layers_common.hpp"
+
+#ifdef HAVE_OPENCL
+#include "opencl_kernels_dnn.hpp"
+#endif
+
+namespace cv { namespace dnn {
+
+class ConstLayerImpl CV_FINAL : public ConstLayer
+{
+public:
+    ConstLayerImpl(const LayerParams& params)
+    {
+        setParamsFrom(params);
+        CV_Assert(blobs.size() == 1);
+    }
+
+    virtual bool getMemoryShapes(const std::vector<MatShape> &inputs,
+                                 const int requiredOutputs,
+                                 std::vector<MatShape> &outputs,
+                                 std::vector<MatShape> &internals) const CV_OVERRIDE
+    {
+        CV_Assert(inputs.empty());
+        outputs.assign(1, shape(blobs[0]));
+        return false;
+    }
+
+#ifdef HAVE_OPENCL
+    bool forward_ocl(InputArrayOfArrays inps, OutputArrayOfArrays outs, OutputArrayOfArrays internals)
+    {
+        std::vector<UMat> outputs;
+        outs.getUMatVector(outputs);
+        if (outs.depth() == CV_16S)
+            convertFp16(blobs[0], outputs[0]);
+        else
+            blobs[0].copyTo(outputs[0]);
+        return true;
+    }
+#endif
+
+    void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE
+    {
+        CV_TRACE_FUNCTION();
+        CV_TRACE_ARG_VALUE(name, "name", name.c_str());
+
+        CV_OCL_RUN(IS_DNN_OPENCL_TARGET(preferableTarget),
+                   forward_ocl(inputs_arr, outputs_arr, internals_arr))
+
+        std::vector<Mat> outputs;
+        outputs_arr.getMatVector(outputs);
+        blobs[0].copyTo(outputs[0]);
+    }
+};
+
+Ptr<Layer> ConstLayer::create(const LayerParams& params)
+{
+    return Ptr<Layer>(new ConstLayerImpl(params));
+}
+
+}}  // namespace cv::dnn

modules/dnn/src/layers/convolution_layer.cpp

@@ -219,9 +219,14 @@ public:
 
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
+#ifdef HAVE_INF_ENGINE
         if (backendId == DNN_BACKEND_INFERENCE_ENGINE)
-            return preferableTarget != DNN_TARGET_MYRIAD || dilation.width == dilation.height;
+        {
+            return INF_ENGINE_VER_MAJOR_GE(INF_ENGINE_RELEASE_2018R4) ||
+                   (preferableTarget != DNN_TARGET_MYRIAD || dilation.width == dilation.height);
+        }
         else
+#endif
             return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE;
     }
 
@@ -460,6 +465,12 @@ public:
         ieLayer->_pads_end.insert(InferenceEngine::Y_AXIS, pad.height);
         ieLayer->_dilation.insert(InferenceEngine::X_AXIS, dilation.width);
         ieLayer->_dilation.insert(InferenceEngine::Y_AXIS, dilation.height);
+        ieLayer->params["output"] = format("%d", outCn);
+        ieLayer->params["kernel"] = format("%d,%d,%d,%d", outCn, inpGroupCn, kernel.height, kernel.width);
+        ieLayer->params["pads_begin"] = format("%d,%d", pad.height, pad.width);
+        ieLayer->params["pads_end"] = format("%d,%d", pad.height, pad.width);
+        ieLayer->params["strides"] = format("%d,%d", stride.height, stride.width);
+        ieLayer->params["dilations"] = format("%d,%d", dilation.height, dilation.width);
 #else
         ieLayer->_kernel_x = kernel.width;
         ieLayer->_kernel_y = kernel.height;

modules/dnn/src/layers/crop_layer.cpp

@@ -68,7 +68,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && crop_ranges.size() == 4;
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && crop_ranges.size() == 4);
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,
@@ -156,6 +156,14 @@ public:
 
         CV_Assert(crop_ranges.size() == 4);
 
+#if INF_ENGINE_VER_MAJOR_GT(INF_ENGINE_RELEASE_2018R3)
+        for (int i = 0; i < 4; ++i)
+        {
+            ieLayer->axis.push_back(i);
+            ieLayer->offset.push_back(crop_ranges[i].start);
+            ieLayer->dim.push_back(crop_ranges[i].end - crop_ranges[i].start);
+        }
+#else
         ieLayer->axis.push_back(0);  // batch
         ieLayer->offset.push_back(crop_ranges[0].start);
         ieLayer->dim.push_back(crop_ranges[0].end - crop_ranges[0].start);
@@ -171,7 +179,7 @@ public:
         ieLayer->axis.push_back(2);  // width
         ieLayer->offset.push_back(crop_ranges[3].start);
         ieLayer->dim.push_back(crop_ranges[3].end - crop_ranges[3].start);
-
+#endif
         return Ptr<BackendNode>(new InfEngineBackendNode(ieLayer));
 #endif  // HAVE_INF_ENGINE
         return Ptr<BackendNode>();

modules/dnn/src/layers/detection_output_layer.cpp

@@ -198,7 +198,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && !_locPredTransposed && _bboxesNormalized && !_clip;
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && !_locPredTransposed && _bboxesNormalized && !_clip);
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/elementwise_layers.cpp

@@ -44,7 +44,6 @@
 #include "layers_common.hpp"
 #include "../op_halide.hpp"
 #include "../op_inf_engine.hpp"
-#include "opencv2/imgproc.hpp"
 #include <opencv2/dnn/shape_utils.hpp>
 #include <iostream>
 
@@ -701,7 +700,8 @@ struct AbsValFunctor
 
     bool supportBackend(int backendId, int)
     {
-        return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE;
+        return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE ||
+               backendId == DNN_BACKEND_INFERENCE_ENGINE;
     }
 
     void apply(const float* srcptr, float* dstptr, int len, size_t planeSize, int cn0, int cn1) const
@@ -755,8 +755,11 @@ struct AbsValFunctor
 #ifdef HAVE_INF_ENGINE
     InferenceEngine::CNNLayerPtr initInfEngine(InferenceEngine::LayerParams& lp)
     {
-        CV_Error(Error::StsNotImplemented, "Abs");
-        return InferenceEngine::CNNLayerPtr();
+        lp.type = "ReLU";
+        std::shared_ptr<InferenceEngine::ReLULayer> ieLayer(new InferenceEngine::ReLULayer(lp));
+        ieLayer->negative_slope = -1;
+        ieLayer->params["negative_slope"] = "-1.0";
+        return ieLayer;
     }
 #endif  // HAVE_INF_ENGINE
 
@@ -833,7 +836,7 @@ struct PowerFunctor
     bool supportBackend(int backendId, int targetId)
     {
         if (backendId == DNN_BACKEND_INFERENCE_ENGINE)
-            return (targetId != DNN_TARGET_OPENCL && targetId != DNN_TARGET_OPENCL_FP16) || power == 1.0;
+            return (targetId != DNN_TARGET_OPENCL && targetId != DNN_TARGET_OPENCL_FP16) || power == 1.0 || power == 0.5;
         else
             return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE;
     }
@@ -979,7 +982,8 @@ struct ChannelsPReLUFunctor
 
     bool supportBackend(int backendId, int)
     {
-        return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE;
+        return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE ||
+               backendId == DNN_BACKEND_INFERENCE_ENGINE;
     }
 
     void apply(const float* srcptr, float* dstptr, int len, size_t planeSize, int cn0, int cn1) const
@@ -1065,8 +1069,11 @@ struct ChannelsPReLUFunctor
 #ifdef HAVE_INF_ENGINE
     InferenceEngine::CNNLayerPtr initInfEngine(InferenceEngine::LayerParams& lp)
     {
-        CV_Error(Error::StsNotImplemented, "PReLU");
-        return InferenceEngine::CNNLayerPtr();
+        lp.type = "PReLU";
+        std::shared_ptr<InferenceEngine::PReLULayer> ieLayer(new InferenceEngine::PReLULayer(lp));
+        const size_t numChannels = scale.total();
+        ieLayer->_weights = wrapToInfEngineBlob(scale, {numChannels}, InferenceEngine::Layout::C);
+        return ieLayer;
     }
 #endif  // HAVE_INF_ENGINE
 

modules/dnn/src/layers/eltwise_layer.cpp

@@ -98,7 +98,8 @@ public:
     {
         return backendId == DNN_BACKEND_OPENCV ||
                backendId == DNN_BACKEND_HALIDE ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && (op != SUM || coeffs.empty());
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE &&
+                (preferableTarget != DNN_TARGET_MYRIAD || coeffs.empty()));
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,
@@ -427,6 +428,7 @@ public:
         lp.type = "Eltwise";
         lp.precision = InferenceEngine::Precision::FP32;
         std::shared_ptr<InferenceEngine::EltwiseLayer> ieLayer(new InferenceEngine::EltwiseLayer(lp));
+        ieLayer->coeff = coeffs;
         if (op == SUM)
             ieLayer->_operation = InferenceEngine::EltwiseLayer::Sum;
         else if (op == PROD)

modules/dnn/src/layers/flatten_layer.cpp

@@ -65,7 +65,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/fully_connected_layer.cpp

@@ -123,8 +123,8 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_HALIDE && haveHalide() && axis == 1 ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && axis == 1;
+               (backendId == DNN_BACKEND_HALIDE && haveHalide() && axis == 1) ||
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && axis == 1);
     }
 
     virtual bool setActivation(const Ptr<ActivationLayer>& layer) CV_OVERRIDE
@@ -449,6 +449,9 @@ public:
         std::shared_ptr<InferenceEngine::FullyConnectedLayer> ieLayer(new InferenceEngine::FullyConnectedLayer(lp));
 
         ieLayer->_out_num = blobs[0].size[0];
+#if INF_ENGINE_VER_MAJOR_GT(INF_ENGINE_RELEASE_2018R3)
+        ieLayer->params["out-size"] = format("%d", blobs[0].size[0]);
+#endif
         ieLayer->_weights = wrapToInfEngineBlob(blobs[0], {(size_t)blobs[0].size[0], (size_t)blobs[0].size[1], 1, 1}, InferenceEngine::Layout::OIHW);
         if (blobs.size() > 1)
             ieLayer->_biases = wrapToInfEngineBlob(blobs[1], {(size_t)ieLayer->_out_num}, InferenceEngine::Layout::C);

modules/dnn/src/layers/lrn_layer.cpp

@@ -92,7 +92,7 @@ public:
     {
         return backendId == DNN_BACKEND_OPENCV ||
                backendId == DNN_BACKEND_HALIDE ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && (preferableTarget != DNN_TARGET_MYRIAD || type == CHANNEL_NRM);
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && (preferableTarget != DNN_TARGET_MYRIAD || type == CHANNEL_NRM));
     }
 
 #ifdef HAVE_OPENCL

modules/dnn/src/layers/max_unpooling_layer.cpp

@@ -35,8 +35,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_HALIDE && haveHalide() &&
-               !poolPad.width && !poolPad.height;
+               (backendId == DNN_BACKEND_HALIDE && haveHalide() && !poolPad.width && !poolPad.height);
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/mvn_layer.cpp

@@ -116,9 +116,15 @@ public:
 
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
+#ifdef HAVE_INF_ENGINE
         if (backendId == DNN_BACKEND_INFERENCE_ENGINE)
+#if INF_ENGINE_VER_MAJOR_GE(INF_ENGINE_RELEASE_2018R5)
+            return !zeroDev && eps <= 1e-7f;
+#else
             return !zeroDev && (preferableTarget == DNN_TARGET_CPU || eps <= 1e-7f);
+#endif
         else
+#endif  // HAVE_INF_ENGINE
             return backendId == DNN_BACKEND_OPENCV;
     }
 

modules/dnn/src/layers/padding_layer.cpp

@@ -91,7 +91,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_HALIDE && haveHalide() && dstRanges.size() == 4;
+               (backendId == DNN_BACKEND_HALIDE && haveHalide() && dstRanges.size() == 4);
     }
 
     void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE

modules/dnn/src/layers/permute_layer.cpp

@@ -105,7 +105,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/pooling_layer.cpp

@@ -154,8 +154,8 @@ public:
         }
         else
             return backendId == DNN_BACKEND_OPENCV ||
-                   backendId == DNN_BACKEND_HALIDE && haveHalide() &&
-                   (type == MAX || type == AVE && !pad_t && !pad_l && !pad_b && !pad_r);
+                   (backendId == DNN_BACKEND_HALIDE && haveHalide() &&
+                   (type == MAX || (type == AVE && !pad_t && !pad_l && !pad_b && !pad_r)));
     }
 
 #ifdef HAVE_OPENCL
@@ -275,6 +275,10 @@ public:
             poolLayer->_padding.insert(InferenceEngine::Y_AXIS, pad_t);
             poolLayer->_pads_end.insert(InferenceEngine::X_AXIS, pad_r);
             poolLayer->_pads_end.insert(InferenceEngine::Y_AXIS, pad_b);
+            poolLayer->params["kernel"] = format("%d,%d", kernel.height, kernel.width);
+            poolLayer->params["pads_begin"] = format("%d,%d", pad_t, pad_l);
+            poolLayer->params["pads_end"] = format("%d,%d", pad_b, pad_r);
+            poolLayer->params["strides"] = format("%d,%d", stride.height, stride.width);
 #else
             poolLayer->_kernel_x = kernel.width;
             poolLayer->_kernel_y = kernel.height;
@@ -341,8 +345,8 @@ public:
                       src.isContinuous(), dst.isContinuous(),
                       src.type() == CV_32F, src.type() == dst.type(),
                       src.dims == 4, dst.dims == 4,
-                      ((poolingType == ROI || poolingType == PSROI) && dst.size[0] ==rois.size[0] || src.size[0] == dst.size[0]),
-                       poolingType == PSROI || src.size[1] == dst.size[1],
+                      (((poolingType == ROI || poolingType == PSROI) && dst.size[0] == rois.size[0]) || src.size[0] == dst.size[0]),
+                      poolingType == PSROI || src.size[1] == dst.size[1],
                       (mask.empty() || (mask.type() == src.type() && mask.size == dst.size)));
 
             PoolingInvoker p;

modules/dnn/src/layers/prior_box_layer.cpp

@@ -271,7 +271,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/proposal_layer.cpp

@@ -87,7 +87,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && preferableTarget != DNN_TARGET_MYRIAD;
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && preferableTarget != DNN_TARGET_MYRIAD);
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/recurrent_layers.cpp

@@ -175,7 +175,7 @@ public:
                          std::vector<MatShape> &outputs,
                          std::vector<MatShape> &internals) const CV_OVERRIDE
     {
-        CV_Assert(!usePeephole && blobs.size() == 3 || usePeephole && blobs.size() == 6);
+        CV_Assert((!usePeephole && blobs.size() == 3) || (usePeephole && blobs.size() == 6));
         CV_Assert(inputs.size() == 1);
         const MatShape& inp0 = inputs[0];
 
@@ -221,7 +221,7 @@ public:
         std::vector<Mat> input;
         inputs_arr.getMatVector(input);
 
-        CV_Assert(!usePeephole && blobs.size() == 3 || usePeephole && blobs.size() == 6);
+        CV_Assert((!usePeephole && blobs.size() == 3) || (usePeephole && blobs.size() == 6));
         CV_Assert(input.size() == 1);
         const Mat& inp0 = input[0];
 

modules/dnn/src/layers/reshape_layer.cpp

@@ -178,7 +178,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/resize_layer.cpp

@@ -51,9 +51,14 @@ public:
 
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
+#ifdef HAVE_INF_ENGINE
         if (backendId == DNN_BACKEND_INFERENCE_ENGINE)
-            return interpolation == "nearest" && preferableTarget != DNN_TARGET_MYRIAD;
+        {
+            return (interpolation == "nearest" && preferableTarget != DNN_TARGET_MYRIAD) ||
+                   (interpolation == "bilinear" && INF_ENGINE_VER_MAJOR_GE(INF_ENGINE_RELEASE_2018R4));
+        }
         else
+#endif
             return backendId == DNN_BACKEND_OPENCV;
     }
 
@@ -160,15 +165,27 @@ public:
 #ifdef HAVE_INF_ENGINE
         InferenceEngine::LayerParams lp;
         lp.name = name;
-        lp.type = "Resample";
         lp.precision = InferenceEngine::Precision::FP32;
-
-        std::shared_ptr<InferenceEngine::CNNLayer> ieLayer(new InferenceEngine::CNNLayer(lp));
-        ieLayer->params["type"] = "caffe.ResampleParameter.NEAREST";
-        ieLayer->params["antialias"] = "0";
+        std::shared_ptr<InferenceEngine::CNNLayer> ieLayer;
+        if (interpolation == "nearest")
+        {
+            lp.type = "Resample";
+            ieLayer = std::shared_ptr<InferenceEngine::CNNLayer>(new InferenceEngine::CNNLayer(lp));
+            ieLayer->params["type"] = "caffe.ResampleParameter.NEAREST";
+            ieLayer->params["antialias"] = "0";
+        }
+        else if (interpolation == "bilinear")
+        {
+            lp.type = "Interp";
+            ieLayer = std::shared_ptr<InferenceEngine::CNNLayer>(new InferenceEngine::CNNLayer(lp));
+            ieLayer->params["pad_beg"] = "0";
+            ieLayer->params["pad_end"] = "0";
+            ieLayer->params["align_corners"] = "0";
+        }
+        else
+            CV_Error(Error::StsNotImplemented, "Unsupported interpolation: " + interpolation);
         ieLayer->params["width"] = cv::format("%d", outWidth);
         ieLayer->params["height"] = cv::format("%d", outHeight);
-
         return Ptr<BackendNode>(new InfEngineBackendNode(ieLayer));
 #endif  // HAVE_INF_ENGINE
         return Ptr<BackendNode>();

modules/dnn/src/layers/scale_layer.cpp

@@ -45,13 +45,13 @@ public:
         std::vector<Mat> inputs;
         inputs_arr.getMatVector(inputs);
         hasWeights = blobs.size() == 2 || (blobs.size() == 1 && !hasBias);
-        CV_Assert(inputs.size() == 2 && blobs.empty() || blobs.size() == (int)hasWeights + (int)hasBias);
+        CV_Assert((inputs.size() == 2 && blobs.empty()) || blobs.size() == (int)hasWeights + (int)hasBias);
     }
 
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV || backendId == DNN_BACKEND_HALIDE ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && axis == 1;
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && axis == 1);
     }
 
     void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE

modules/dnn/src/layers/slice_layer.cpp

@@ -111,7 +111,7 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && sliceRanges.size() == 1 && sliceRanges[0].size() == 4;
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && sliceRanges.size() == 1 && sliceRanges[0].size() == 4);
     }
 
     bool getMemoryShapes(const std::vector<MatShape> &inputs,

modules/dnn/src/layers/softmax_layer.cpp

@@ -89,8 +89,8 @@ public:
     virtual bool supportBackend(int backendId) CV_OVERRIDE
     {
         return backendId == DNN_BACKEND_OPENCV ||
-               backendId == DNN_BACKEND_HALIDE && haveHalide() && axisRaw == 1 ||
-               backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && !logSoftMax;
+               (backendId == DNN_BACKEND_HALIDE && haveHalide() && axisRaw == 1) ||
+               (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && !logSoftMax);
     }
 
 #ifdef HAVE_OPENCL

modules/dnn/src/ocl4dnn/src/ocl4dnn_conv_spatial.cpp

@@ -638,7 +638,7 @@ void OCL4DNNConvSpatial<Dtype>::generateKey()
                << "p" << pad_w_ << "x" << pad_h_ << "_"
                << "num" << num_ << "_"
                << "M" << M_ << "_"
-               << "activ" << fused_activ_ << "_"
+               << "activ" << (int)fused_activ_ << "_"
                << "eltwise" << fused_eltwise_ << "_"
                << precision;
 

modules/dnn/src/onnx/onnx_importer.cpp

@@ -6,6 +6,7 @@
 // Third party copyrights are property of their respective owners.
 
 #include "../precomp.hpp"
+#include <opencv2/dnn/shape_utils.hpp>
 
 #ifdef HAVE_PROTOBUF
 
@@ -134,9 +135,38 @@ Mat getMatFromTensor(opencv_onnx::TensorProto& tensor_proto)
     else
         CV_Error(Error::StsUnsupportedFormat, "Unsupported data type: " +
                         opencv_onnx::TensorProto_DataType_Name(datatype));
+    if (tensor_proto.dims_size() == 0)
+        blob.dims = 1;  // To force 1-dimensional cv::Mat for scalars.
     return blob;
 }
 
+void runLayer(Ptr<Layer> layer, const std::vector<Mat>& inputs,
+              std::vector<Mat>& outputs)
+{
+    std::vector<MatShape> inpShapes(inputs.size());
+    int ddepth = CV_32F;
+    for (size_t i = 0; i < inputs.size(); ++i)
+    {
+        inpShapes[i] = shape(inputs[i]);
+        if (i > 0 && ddepth != inputs[i].depth())
+            CV_Error(Error::StsNotImplemented, "Mixed input data types.");
+        ddepth = inputs[i].depth();
+    }
+
+    std::vector<MatShape> outShapes, internalShapes;
+    layer->getMemoryShapes(inpShapes, 0, outShapes, internalShapes);
+
+    std::vector<Mat> internals(internalShapes.size());
+    outputs.resize(outShapes.size());
+    for (size_t i = 0; i < outShapes.size(); ++i)
+        outputs[i].create(outShapes[i], ddepth);
+    for (size_t i = 0; i < internalShapes.size(); ++i)
+        internals[i].create(internalShapes[i], ddepth);
+
+    layer->finalize(inputs, outputs);
+    layer->forward(inputs, outputs, internals);
+}
+
 std::map<std::string, Mat> ONNXImporter::getGraphTensors(
                                         const opencv_onnx::GraphProto& graph_proto)
 {
@@ -292,6 +322,26 @@ void ONNXImporter::populateNet(Net dstNet)
     CV_Assert(model_proto.has_graph());
     opencv_onnx::GraphProto graph_proto = model_proto.graph();
     std::map<std::string, Mat> constBlobs = getGraphTensors(graph_proto);
+    // List of internal blobs shapes.
+    std::map<std::string, MatShape> outShapes;
+    // Add all the inputs shapes. It includes as constant blobs as network's inputs shapes.
+    for (int i = 0; i < graph_proto.input_size(); ++i)
+    {
+        opencv_onnx::ValueInfoProto valueInfoProto = graph_proto.input(i);
+        CV_Assert(valueInfoProto.has_type());
+        opencv_onnx::TypeProto typeProto = valueInfoProto.type();
+        CV_Assert(typeProto.has_tensor_type());
+        opencv_onnx::TypeProto::Tensor tensor = typeProto.tensor_type();
+        CV_Assert(tensor.has_shape());
+        opencv_onnx::TensorShapeProto tensorShape = tensor.shape();
+
+        MatShape inpShape(tensorShape.dim_size());
+        for (int j = 0; j < inpShape.size(); ++j)
+        {
+            inpShape[j] = tensorShape.dim(j).dim_value();
+        }
+        outShapes[valueInfoProto.name()] = inpShape;
+    }
 
     std::string framework_name;
     if (model_proto.has_producer_name()) {
@@ -301,6 +351,7 @@ void ONNXImporter::populateNet(Net dstNet)
     // create map with network inputs (without const blobs)
     std::map<std::string, LayerInfo> layer_id;
     std::map<std::string, LayerInfo>::iterator layerId;
+    std::map<std::string, MatShape>::iterator shapeIt;
     // fill map: push layer name, layer id and output id
     std::vector<String> netInputs;
     for (int j = 0; j < graph_proto.input_size(); j++)
@@ -317,9 +368,9 @@ void ONNXImporter::populateNet(Net dstNet)
     LayerParams layerParams;
     opencv_onnx::NodeProto node_proto;
 
-    for(int i = 0; i < layersSize; i++)
+    for(int li = 0; li < layersSize; li++)
     {
-        node_proto = graph_proto.node(i);
+        node_proto = graph_proto.node(li);
         layerParams = getLayerParams(node_proto);
         CV_Assert(node_proto.output_size() >= 1);
         layerParams.name = node_proto.output(0);
@@ -358,7 +409,8 @@ void ONNXImporter::populateNet(Net dstNet)
                     layerParams.set("shift", blob.at<float>(0));
                 }
                 else {
-                    layerParams.type = "Shift";
+                    layerParams.type = "Scale";
+                    layerParams.set("bias_term", true);
                     layerParams.blobs.push_back(blob);
                 }
             }
@@ -368,15 +420,32 @@ void ONNXImporter::populateNet(Net dstNet)
         }
         else if (layer_type == "Sub")
         {
-            Mat blob = (-1.0f) * getBlob(node_proto, constBlobs, 1);
-            blob = blob.reshape(1, 1);
+            Mat blob = getBlob(node_proto, constBlobs, 1);
             if (blob.total() == 1) {
                 layerParams.type = "Power";
-                layerParams.set("shift", blob.at<float>(0));
+                layerParams.set("shift", -blob.at<float>(0));
             }
             else {
-                layerParams.type = "Shift";
+                layerParams.type = "Scale";
+                layerParams.set("has_bias", true);
+                layerParams.blobs.push_back(-1.0f * blob.reshape(1, 1));
+            }
+        }
+        else if (layer_type == "Div")
+        {
+            Mat blob = getBlob(node_proto, constBlobs, 1);
+            CV_Assert_N(blob.type() == CV_32F, blob.total());
+            if (blob.total() == 1)
+            {
+                layerParams.set("scale", 1.0f / blob.at<float>(0));
+                layerParams.type = "Power";
+            }
+            else
+            {
+                layerParams.type = "Scale";
+                divide(1.0, blob, blob);
                 layerParams.blobs.push_back(blob);
+                layerParams.set("bias_term", false);
             }
         }
         else if (layer_type == "Constant")
@@ -508,6 +577,16 @@ void ONNXImporter::populateNet(Net dstNet)
             layerParams.set("num_output", layerParams.blobs[0].size[0]);
             layerParams.set("bias_term", node_proto.input_size() == 3);
         }
+        else if (layer_type == "ConvTranspose")
+        {
+            CV_Assert(node_proto.input_size() >= 2);
+            layerParams.type = "Deconvolution";
+            for (int j = 1; j < node_proto.input_size(); j++) {
+                layerParams.blobs.push_back(getBlob(node_proto, constBlobs, j));
+            }
+            layerParams.set("num_output", layerParams.blobs[0].size[1]);
+            layerParams.set("bias_term", node_proto.input_size() == 3);
+        }
         else if (layer_type == "Transpose")
         {
             layerParams.type = "Permute";
@@ -569,6 +648,65 @@ void ONNXImporter::populateNet(Net dstNet)
         {
             layerParams.type = "Padding";
         }
+        else if (layer_type == "Shape")
+        {
+            CV_Assert(node_proto.input_size() == 1);
+            shapeIt = outShapes.find(node_proto.input(0));
+            CV_Assert(shapeIt != outShapes.end());
+            MatShape inpShape = shapeIt->second;
+
+            Mat shapeMat(inpShape.size(), 1, CV_32S);
+            for (int j = 0; j < inpShape.size(); ++j)
+                shapeMat.at<int>(j) = inpShape[j];
+            shapeMat.dims = 1;
+
+            constBlobs.insert(std::make_pair(layerParams.name, shapeMat));
+            continue;
+        }
+        else if (layer_type == "Gather")
+        {
+            CV_Assert(node_proto.input_size() == 2);
+            CV_Assert(layerParams.has("axis"));
+            Mat input = getBlob(node_proto, constBlobs, 0);
+            Mat indexMat = getBlob(node_proto, constBlobs, 1);
+            CV_Assert_N(indexMat.type() == CV_32S, indexMat.total() == 1);
+            int index = indexMat.at<int>(0);
+            int axis = layerParams.get<int>("axis");
+
+            std::vector<cv::Range> ranges(input.dims, Range::all());
+            ranges[axis] = Range(index, index + 1);
+
+            Mat out = input(ranges);
+            constBlobs.insert(std::make_pair(layerParams.name, out));
+            continue;
+        }
+        else if (layer_type == "Concat")
+        {
+            bool hasVariableInps = false;
+            for (int i = 0; i < node_proto.input_size(); ++i)
+            {
+                if (layer_id.find(node_proto.input(i)) != layer_id.end())
+                {
+                    hasVariableInps = true;
+                    break;
+                }
+            }
+
+            if (!hasVariableInps)
+            {
+                std::vector<Mat> inputs(node_proto.input_size()), concatenated;
+                for (size_t i = 0; i < inputs.size(); ++i)
+                {
+                    inputs[i] = getBlob(node_proto, constBlobs, i);
+                }
+                Ptr<Layer> concat = ConcatLayer::create(layerParams);
+                runLayer(concat, inputs, concatenated);
+
+                CV_Assert(concatenated.size() == 1);
+                constBlobs.insert(std::make_pair(layerParams.name, concatenated[0]));
+                continue;
+            }
+        }
         else
         {
             for (int j = 0; j < node_proto.input_size(); j++) {
@@ -580,12 +718,24 @@ void ONNXImporter::populateNet(Net dstNet)
          int id = dstNet.addLayer(layerParams.name, layerParams.type, layerParams);
          layer_id.insert(std::make_pair(layerParams.name, LayerInfo(id, 0)));
 
+
+         std::vector<MatShape> layerInpShapes, layerOutShapes, layerInternalShapes;
          for (int j = 0; j < node_proto.input_size(); j++) {
              layerId = layer_id.find(node_proto.input(j));
              if (layerId != layer_id.end()) {
                  dstNet.connect(layerId->second.layerId, layerId->second.outputId, id, j);
+                 // Collect input shapes.
+                 shapeIt = outShapes.find(node_proto.input(j));
+                 CV_Assert(shapeIt != outShapes.end());
+                 layerInpShapes.push_back(shapeIt->second);
              }
          }
+
+         // Compute shape of output blob for this layer.
+         Ptr<Layer> layer = dstNet.getLayer(id);
+         layer->getMemoryShapes(layerInpShapes, 0, layerOutShapes, layerInternalShapes);
+         CV_Assert(!layerOutShapes.empty());
+         outShapes[layerParams.name] = layerOutShapes[0];
      }
  }
 

modules/dnn/src/op_inf_engine.cpp

@@ -152,6 +152,7 @@ InfEngineBackendNet::InfEngineBackendNet()
 {
     targetDevice = InferenceEngine::TargetDevice::eCPU;
     precision = InferenceEngine::Precision::FP32;
+    hasNetOwner = false;
 }
 
 InfEngineBackendNet::InfEngineBackendNet(InferenceEngine::CNNNetwork& net)
@@ -162,6 +163,7 @@ InfEngineBackendNet::InfEngineBackendNet(InferenceEngine::CNNNetwork& net)
     outputs = net.getOutputsInfo();
     layers.resize(net.layerCount());  // A hack to execute InfEngineBackendNet::layerCount correctly.
     netOwner = net;
+    hasNetOwner = true;
 }
 
 void InfEngineBackendNet::Release() noexcept
@@ -178,12 +180,12 @@ void InfEngineBackendNet::setPrecision(InferenceEngine::Precision p) noexcept
 
 InferenceEngine::Precision InfEngineBackendNet::getPrecision() noexcept
 {
-    return precision;
+    return hasNetOwner ? netOwner.getPrecision() : precision;
 }
 
 InferenceEngine::Precision InfEngineBackendNet::getPrecision() const noexcept
 {
-    return precision;
+    return hasNetOwner ? netOwner.getPrecision() : precision;
 }
 
 // Assume that outputs of network is unconnected blobs.
@@ -233,6 +235,12 @@ const std::string& InfEngineBackendNet::getName() const noexcept
     return name;
 }
 
+InferenceEngine::StatusCode InfEngineBackendNet::serialize(const std::string&, const std::string&, InferenceEngine::ResponseDesc*) const noexcept
+{
+    CV_Error(Error::StsNotImplemented, "");
+    return InferenceEngine::StatusCode::OK;
+}
+
 size_t InfEngineBackendNet::layerCount() noexcept
 {
     return const_cast<const InfEngineBackendNet*>(this)->layerCount();
@@ -302,19 +310,21 @@ void InfEngineBackendNet::setTargetDevice(InferenceEngine::TargetDevice device)
 {
     if (device != InferenceEngine::TargetDevice::eCPU &&
         device != InferenceEngine::TargetDevice::eGPU &&
-        device != InferenceEngine::TargetDevice::eMYRIAD)
+        device != InferenceEngine::TargetDevice::eMYRIAD &&
+        device != InferenceEngine::TargetDevice::eFPGA)
         CV_Error(Error::StsNotImplemented, "");
     targetDevice = device;
 }
 
 InferenceEngine::TargetDevice InfEngineBackendNet::getTargetDevice() noexcept
 {
-    return targetDevice;
+    return const_cast<const InfEngineBackendNet*>(this)->getTargetDevice();
 }
 
 InferenceEngine::TargetDevice InfEngineBackendNet::getTargetDevice() const noexcept
 {
-    return targetDevice;
+    return targetDevice == InferenceEngine::TargetDevice::eFPGA ?
+           InferenceEngine::TargetDevice::eHETERO : targetDevice;
 }
 
 InferenceEngine::StatusCode InfEngineBackendNet::setBatchSize(const size_t) noexcept
@@ -387,6 +397,27 @@ void InfEngineBackendNet::init(int targetId)
             }
         }
         CV_Assert(!inputs.empty());
+
+#if INF_ENGINE_VER_MAJOR_GT(INF_ENGINE_RELEASE_2018R3)
+        for (const auto& inp : inputs)
+        {
+            InferenceEngine::LayerParams lp;
+            lp.name = inp.first;
+            lp.type = "Input";
+            lp.precision = InferenceEngine::Precision::FP32;
+            std::shared_ptr<InferenceEngine::CNNLayer> inpLayer(new InferenceEngine::CNNLayer(lp));
+
+            layers.push_back(inpLayer);
+
+            InferenceEngine::DataPtr dataPtr = inp.second->getInputData();
+            // TODO: remove precision dependency (see setInput.normalization tests)
+            if (dataPtr->precision == InferenceEngine::Precision::FP32)
+            {
+                inpLayer->outData.assign(1, dataPtr);
+                dataPtr->creatorLayer = InferenceEngine::CNNLayerWeakPtr(inpLayer);
+            }
+        }
+#endif
     }
 
     if (outputs.empty())
@@ -395,6 +426,8 @@ void InfEngineBackendNet::init(int targetId)
         InferenceEngine::OutputsDataMap unconnectedOuts;
         for (const auto& l : layers)
         {
+            if (l->type == "Input")
+                continue;
             // Add all outputs.
             for (const InferenceEngine::DataPtr& out : l->outData)
             {
@@ -445,6 +478,11 @@ void InfEngineBackendNet::init(int targetId)
         setPrecision(InferenceEngine::Precision::FP16);
         setTargetDevice(InferenceEngine::TargetDevice::eMYRIAD); break;
     }
+    case DNN_TARGET_FPGA:
+    {
+        setPrecision(InferenceEngine::Precision::FP16);
+        setTargetDevice(InferenceEngine::TargetDevice::eFPGA); break;
+    }
     default:
         CV_Error(Error::StsError, format("Unknown target identifier: %d", targetId));
     }
@@ -468,10 +506,15 @@ void InfEngineBackendNet::initPlugin(InferenceEngine::ICNNNetwork& net)
         }
         else
         {
-            enginePtr = InferenceEngine::PluginDispatcher({""}).getSuitablePlugin(targetDevice);
+            auto dispatcher = InferenceEngine::PluginDispatcher({""});
+            if (targetDevice == InferenceEngine::TargetDevice::eFPGA)
+                enginePtr = dispatcher.getPluginByDevice("HETERO:FPGA,CPU");
+            else
+                enginePtr = dispatcher.getSuitablePlugin(targetDevice);
             sharedPlugins[targetDevice] = enginePtr;
 
-            if (targetDevice == InferenceEngine::TargetDevice::eCPU)
+            if (targetDevice == InferenceEngine::TargetDevice::eCPU ||
+                targetDevice == InferenceEngine::TargetDevice::eFPGA)
             {
                 std::string suffixes[] = {"_avx2", "_sse4", ""};
                 bool haveFeature[] = {
@@ -559,7 +602,7 @@ bool InfEngineBackendLayer::getMemoryShapes(const std::vector<MatShape> &inputs,
 bool InfEngineBackendLayer::supportBackend(int backendId)
 {
     return backendId == DNN_BACKEND_DEFAULT ||
-           backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine();
+           (backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
 }
 
 void InfEngineBackendLayer::forward(InputArrayOfArrays inputs, OutputArrayOfArrays outputs,

modules/dnn/src/op_inf_engine.hpp

@@ -25,10 +25,12 @@
 #define INF_ENGINE_RELEASE_2018R1 2018010000
 #define INF_ENGINE_RELEASE_2018R2 2018020000
 #define INF_ENGINE_RELEASE_2018R3 2018030000
+#define INF_ENGINE_RELEASE_2018R4 2018040000
+#define INF_ENGINE_RELEASE_2018R5 2018050000
 
 #ifndef INF_ENGINE_RELEASE
-#warning("IE version have not been provided via command-line. Using 2018R2 by default")
-#define INF_ENGINE_RELEASE INF_ENGINE_RELEASE_2018R2
+#warning("IE version have not been provided via command-line. Using 2018R5 by default")
+#define INF_ENGINE_RELEASE INF_ENGINE_RELEASE_2018R5
 #endif
 
 #define INF_ENGINE_VER_MAJOR_GT(ver) (((INF_ENGINE_RELEASE) / 10000) > ((ver) / 10000))
@@ -67,6 +69,8 @@ public:
 
     virtual InferenceEngine::InputInfo::Ptr getInput(const std::string &inputName) const noexcept;
 
+    virtual InferenceEngine::StatusCode serialize(const std::string &xmlPath, const std::string &binPath, InferenceEngine::ResponseDesc* resp) const noexcept;
+
     virtual void getName(char *pName, size_t len) noexcept;
 
     virtual void getName(char *pName, size_t len) const noexcept;
@@ -133,6 +137,9 @@ private:
     InferenceEngine::InferRequest infRequest;
     // In case of models from Model Optimizer we need to manage their lifetime.
     InferenceEngine::CNNNetwork netOwner;
+    // There is no way to check if netOwner is initialized or not so we use
+    // a separate flag to determine if the model has been loaded from IR.
+    bool hasNetOwner;
 
     std::string name;
 

modules/dnn/src/tensorflow/tf_importer.cpp

@@ -156,6 +156,7 @@ void blobFromTensor(const tensorflow::TensorProto &tensor, Mat &dstBlob)
     }
 }
 
+#if 0
 void printList(const tensorflow::AttrValue::ListValue &val)
 {
     std::cout << "(";
@@ -235,6 +236,7 @@ void printLayerAttr(const tensorflow::NodeDef &layer)
         std::cout << std::endl;
     }
 }
+#endif
 
 bool hasLayerAttr(const tensorflow::NodeDef &layer, const std::string &name)
 {
@@ -937,7 +939,7 @@ void TFImporter::populateNet(Net dstNet)
             if (getDataLayout(name, data_layouts) == DATA_LAYOUT_UNKNOWN)
                 data_layouts[name] = DATA_LAYOUT_NHWC;
         }
-        else if (type == "BiasAdd" || type == "Add")
+        else if (type == "BiasAdd" || type == "Add" || type == "Sub")
         {
             bool haveConst = false;
             for(int ii = 0; !haveConst && ii < layer.input_size(); ++ii)
@@ -951,6 +953,8 @@ void TFImporter::populateNet(Net dstNet)
             {
                 Mat values = getTensorContent(getConstBlob(layer, value_id));
                 CV_Assert(values.type() == CV_32FC1);
+                if (type == "Sub")
+                    values *= -1.0f;
 
                 int id;
                 if (values.total() == 1)  // is a scalar.
@@ -971,6 +975,12 @@ void TFImporter::populateNet(Net dstNet)
             else
             {
                 layerParams.set("operation", "sum");
+                if (type == "Sub")
+                {
+                    static float subCoeffs[] = {1.f, -1.f};
+                    layerParams.set("coeff", DictValue::arrayReal<float*>(subCoeffs, 2));
+                }
+
                 int id = dstNet.addLayer(name, "Eltwise", layerParams);
                 layer_id[name] = id;
 
@@ -983,36 +993,6 @@ void TFImporter::populateNet(Net dstNet)
                 }
             }
         }
-        else if (type == "Sub")
-        {
-            bool haveConst = false;
-            for(int ii = 0; !haveConst && ii < layer.input_size(); ++ii)
-            {
-                Pin input = parsePin(layer.input(ii));
-                haveConst = value_id.find(input.name) != value_id.end();
-            }
-            CV_Assert(haveConst);
-
-            Mat values = getTensorContent(getConstBlob(layer, value_id));
-            CV_Assert(values.type() == CV_32FC1);
-            values *= -1.0f;
-
-            int id;
-            if (values.total() == 1)  // is a scalar.
-            {
-                layerParams.set("shift", values.at<float>(0));
-                id = dstNet.addLayer(name, "Power", layerParams);
-            }
-            else  // is a vector
-            {
-                layerParams.blobs.resize(1, values);
-                id = dstNet.addLayer(name, "Shift", layerParams);
-            }
-            layer_id[name] = id;
-
-            // one input only
-            connect(layer_id, dstNet, parsePin(layer.input(0)), id, 0);
-        }
         else if (type == "MatMul")
         {
             CV_Assert(layer.input_size() == 2);
@@ -1264,14 +1244,31 @@ void TFImporter::populateNet(Net dstNet)
                 axis = toNCHW(axis);
             layerParams.set("axis", axis);
 
-            int id = dstNet.addLayer(name, "Concat", layerParams);
-            layer_id[name] = id;
-
-
+            // input(0) or input(n-1) is concat_dim
             int from = (type == "Concat" ? 1 : 0);
             int to = (type == "Concat" ? layer.input_size() : layer.input_size() - 1);
 
-            // input(0) or input(n-1) is concat_dim
+            for (int ii = from; ii < to; ii++)
+            {
+                Pin inp = parsePin(layer.input(ii));
+                if (layer_id.find(inp.name) == layer_id.end())
+                {
+                    // There are constant inputs.
+                    LayerParams lp;
+                    lp.name = inp.name;
+                    lp.type = "Const";
+                    lp.blobs.resize(1);
+                    blobFromTensor(getConstBlob(layer, value_id, ii), lp.blobs.back());
+                    CV_Assert_N(!lp.blobs[0].empty(), lp.blobs[0].type() == CV_32F);
+
+                    int constInpId = dstNet.addLayer(lp.name, lp.type, lp);
+                    layer_id[lp.name] = constInpId;
+                }
+            }
+
+            int id = dstNet.addLayer(name, "Concat", layerParams);
+            layer_id[name] = id;
+
             for (int ii = from; ii < to; ii++)
             {
                 Pin inp = parsePin(layer.input(ii));

modules/dnn/src/tensorflow/tf_io.cpp

@@ -37,8 +37,6 @@ using namespace tensorflow;
 using namespace ::google::protobuf;
 using namespace ::google::protobuf::io;
 
-const int kProtoReadBytesLimit = INT_MAX;  // Max size of 2 GB minus 1 byte.
-
 void ReadTFNetParamsFromBinaryFileOrDie(const char* param_file,
                                         tensorflow::GraphDef* param) {
     CHECK(ReadProtoFromBinaryFile(param_file, param))

modules/dnn/src/torch/torch_importer.cpp

@@ -129,13 +129,15 @@ struct TorchImporter
     Module *rootModule;
     Module *curModule;
     int moduleCounter;
+    bool testPhase;
 
-    TorchImporter(String filename, bool isBinary)
+    TorchImporter(String filename, bool isBinary, bool evaluate)
     {
         CV_TRACE_FUNCTION();
 
         rootModule = curModule = NULL;
         moduleCounter = 0;
+        testPhase = evaluate;
 
         file = cv::Ptr<THFile>(THDiskFile_new(filename, "r", 0), THFile_free);
         CV_Assert(file && THFile_isOpened(file));
@@ -680,7 +682,8 @@ struct TorchImporter
                     layerParams.blobs.push_back(tensorParams["bias"].second);
                 }
 
-                if (nnName == "InstanceNormalization")
+                bool trainPhase = scalarParams.get<bool>("train", false);
+                if (nnName == "InstanceNormalization" || (trainPhase && !testPhase))
                 {
                     cv::Ptr<Module> mvnModule(new Module(nnName));
                     mvnModule->apiType = "MVN";
@@ -1243,18 +1246,18 @@ struct TorchImporter
 
 Mat readTorchBlob(const String &filename, bool isBinary)
 {
-    TorchImporter importer(filename, isBinary);
+    TorchImporter importer(filename, isBinary, true);
     importer.readObject();
     CV_Assert(importer.tensors.size() == 1);
 
     return importer.tensors.begin()->second;
 }
 
-Net readNetFromTorch(const String &model, bool isBinary)
+Net readNetFromTorch(const String &model, bool isBinary, bool evaluate)
 {
     CV_TRACE_FUNCTION();
 
-    TorchImporter importer(model, isBinary);
+    TorchImporter importer(model, isBinary, evaluate);
     Net net;
     importer.populateNet(net);
     return net;

modules/dnn/test/test_backends.cpp

@@ -128,10 +128,16 @@ TEST_P(DNNTestNetwork, GoogLeNet)
 
 TEST_P(DNNTestNetwork, Inception_5h)
 {
-    if (backend == DNN_BACKEND_INFERENCE_ENGINE) throw SkipTestException("");
+    double l1 = default_l1, lInf = default_lInf;
+    if (backend == DNN_BACKEND_INFERENCE_ENGINE && (target == DNN_TARGET_CPU || target == DNN_TARGET_OPENCL))
+    {
+        l1 = 1.72e-5;
+        lInf = 8e-4;
+    }
     processNet("dnn/tensorflow_inception_graph.pb", "", Size(224, 224), "softmax2",
                target == DNN_TARGET_OPENCL ? "dnn/halide_scheduler_opencl_inception_5h.yml" :
-                                             "dnn/halide_scheduler_inception_5h.yml");
+                                             "dnn/halide_scheduler_inception_5h.yml",
+               l1, lInf);
 }
 
 TEST_P(DNNTestNetwork, ENet)
@@ -174,7 +180,7 @@ TEST_P(DNNTestNetwork, MobileNet_SSD_v2_TensorFlow)
         throw SkipTestException("");
     Mat sample = imread(findDataFile("dnn/street.png", false));
     Mat inp = blobFromImage(sample, 1.0f, Size(300, 300), Scalar(), false);
-    float l1 = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.011 : 0.0;
+    float l1 = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.013 : 0.0;
     float lInf = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.062 : 0.0;
     processNet("dnn/ssd_mobilenet_v2_coco_2018_03_29.pb", "dnn/ssd_mobilenet_v2_coco_2018_03_29.pbtxt",
                inp, "detection_out", "", l1, lInf, 0.25);
@@ -184,7 +190,7 @@ TEST_P(DNNTestNetwork, SSD_VGG16)
 {
     if (backend == DNN_BACKEND_HALIDE && target == DNN_TARGET_CPU)
         throw SkipTestException("");
-    double scoreThreshold = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.0252 : 0.0;
+    double scoreThreshold = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 0.0325 : 0.0;
     Mat sample = imread(findDataFile("dnn/street.png", false));
     Mat inp = blobFromImage(sample, 1.0f, Size(300, 300), Scalar(), false);
     processNet("dnn/VGG_ILSVRC2016_SSD_300x300_iter_440000.caffemodel",
@@ -193,41 +199,42 @@ TEST_P(DNNTestNetwork, SSD_VGG16)
 
 TEST_P(DNNTestNetwork, OpenPose_pose_coco)
 {
-    if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+    if (backend == DNN_BACKEND_HALIDE)
         throw SkipTestException("");
     processNet("dnn/openpose_pose_coco.caffemodel", "dnn/openpose_pose_coco.prototxt",
-               Size(368, 368));
+               Size(46, 46));
 }
 
 TEST_P(DNNTestNetwork, OpenPose_pose_mpi)
 {
-    if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+    if (backend == DNN_BACKEND_HALIDE)
         throw SkipTestException("");
     processNet("dnn/openpose_pose_mpi.caffemodel", "dnn/openpose_pose_mpi.prototxt",
-               Size(368, 368));
+               Size(46, 46));
 }
 
 TEST_P(DNNTestNetwork, OpenPose_pose_mpi_faster_4_stages)
 {
-    if (backend == DNN_BACKEND_HALIDE ||
-        backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+    if (backend == DNN_BACKEND_HALIDE)
         throw SkipTestException("");
     // The same .caffemodel but modified .prototxt
     // See https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/src/openpose/pose/poseParameters.cpp
     processNet("dnn/openpose_pose_mpi.caffemodel", "dnn/openpose_pose_mpi_faster_4_stages.prototxt",
-               Size(368, 368));
+               Size(46, 46));
 }
 
 TEST_P(DNNTestNetwork, OpenFace)
 {
-#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE < 2018030000
+#if defined(INF_ENGINE_RELEASE)
+#if (INF_ENGINE_RELEASE < 2018030000 || INF_ENGINE_RELEASE == 2018050000)
     if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
-        throw SkipTestException("Test is enabled starts from OpenVINO 2018R3");
+        throw SkipTestException("");
+#elif INF_ENGINE_RELEASE < 2018040000
+    if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_OPENCL_FP16)
+        throw SkipTestException("Test is enabled starts from OpenVINO 2018R4");
 #endif
-    if (backend == DNN_BACKEND_HALIDE ||
-        (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_OPENCL_FP16))
+#endif
+    if (backend == DNN_BACKEND_HALIDE)
         throw SkipTestException("");
     processNet("dnn/openface_nn4.small2.v1.t7", "", Size(96, 96), "");
 }

modules/dnn/test/test_caffe_importer.cpp

@@ -300,10 +300,11 @@ INSTANTIATE_TEST_CASE_P(/**/, Reproducibility_ResNet50,
 typedef testing::TestWithParam<Target> Reproducibility_SqueezeNet_v1_1;
 TEST_P(Reproducibility_SqueezeNet_v1_1, Accuracy)
 {
+    int targetId = GetParam();
+    if(targetId == DNN_TARGET_OPENCL_FP16)
+        throw SkipTestException("This test does not support FP16");
     Net net = readNetFromCaffe(findDataFile("dnn/squeezenet_v1.1.prototxt", false),
                                findDataFile("dnn/squeezenet_v1.1.caffemodel", false));
-
-    int targetId = GetParam();
     net.setPreferableBackend(DNN_BACKEND_OPENCV);
     net.setPreferableTarget(targetId);
 
@@ -324,7 +325,8 @@ TEST_P(Reproducibility_SqueezeNet_v1_1, Accuracy)
     Mat ref = blobFromNPY(_tf("squeezenet_v1.1_prob.npy"));
     normAssert(ref, out);
 }
-INSTANTIATE_TEST_CASE_P(/**/, Reproducibility_SqueezeNet_v1_1, availableDnnTargets());
+INSTANTIATE_TEST_CASE_P(/**/, Reproducibility_SqueezeNet_v1_1,
+    testing::ValuesIn(getAvailableTargets(DNN_BACKEND_OPENCV)));
 
 TEST(Reproducibility_AlexNet_fp16, Accuracy)
 {
@@ -469,6 +471,7 @@ TEST(Test_Caffe, shared_weights)
 
   net.setInput(blob_1, "input_1");
   net.setInput(blob_2, "input_2");
+  net.setPreferableBackend(DNN_BACKEND_OPENCV);
 
   Mat sum = net.forward();
 
@@ -512,7 +515,11 @@ INSTANTIATE_TEST_CASE_P(Test_Caffe, opencv_face_detector,
 
 TEST_P(Test_Caffe_nets, FasterRCNN_vgg16)
 {
-    if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+    if ((backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
+#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE > 2018030000
+     || (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_OPENCL_FP16)
+#endif
+    )
         throw SkipTestException("");
     static Mat ref = (Mat_<float>(3, 7) << 0, 2, 0.949398, 99.2454, 210.141, 601.205, 462.849,
                                            0, 7, 0.997022, 481.841, 92.3218, 722.685, 175.953,