potok/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2026-07-31 08:13:04 +04:00
Code Packages Releases Wiki Activity

Merge remote-tracking branch 'upstream/3.4' into merge-3.4

Browse Source
This commit is contained in:
Alexander Alekhin
2018-11-17 13:50:04 +00:00
parent d7540c9a3c 183bc5c281
commit 22dbcf98c5
237 changed files with 2990 additions and 1660 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/dnn/CMakeLists.txt
+5 -10
View File
@@ -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
+2 -2
View File
@@ -236,7 +236,7 @@ CV__DNN_INLINE_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 +578,7 @@ CV__DNN_INLINE_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
+3 -2
View File
@@ -60,12 +60,13 @@ CV__DNN_INLINE_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
+6 -3
View File
@@ -181,7 +181,8 @@ CV__DNN_INLINE_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
@@ -198,7 +199,8 @@ CV__DNN_INLINE_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.
@@ -218,7 +220,8 @@ CV__DNN_INLINE_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
modules/dnn/perf/perf_net.cpp
+6 -7
View File
@@ -175,8 +175,7 @@ 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 +184,7 @@ 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 +193,7 @@ 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 +202,7 @@ 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 +229,7 @@ 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 +240,7 @@ 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
+6 -6
View File
@@ -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
+21 -4
View File
@@ -353,7 +353,7 @@ 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
@@ -428,7 +428,7 @@ 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
@@ -1665,6 +1665,23 @@ 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;
}
@@ -1787,8 +1804,8 @@ 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();
modules/dnn/src/layers/batch_norm_layer.cpp
+2 -2
View File
@@ -151,8 +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
modules/dnn/src/layers/blank_layer.cpp
+9 -2
View File
@@ -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", input->dims.size() - 1);
ieLayer->params["out_sizes"] = format("%d", input->dims[0]);
#endif
return Ptr<BackendNode>(new InfEngineBackendNode(ieLayer));
#endif // HAVE_INF_ENGINE
return Ptr<BackendNode>();
modules/dnn/src/layers/concat_layer.cpp
+3 -3
View File
@@ -105,9 +105,9 @@ 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_VKCOM && haveVulkan() && !padding;
(backendId == DNN_BACKEND_HALIDE && haveHalide() && axis == 1 && !padding) || // By channels
(backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && !padding) ||
(backendId == DNN_BACKEND_VKCOM && haveVulkan() && !padding);
}
class ChannelConcatInvoker : public ParallelLoopBody
modules/dnn/src/layers/convolution_layer.cpp
+7 -1
View File
@@ -225,7 +225,7 @@ public:
else
return backendId == DNN_BACKEND_OPENCV ||
backendId == DNN_BACKEND_HALIDE ||
backendId == DNN_BACKEND_VKCOM && haveVulkan();
(backendId == DNN_BACKEND_VKCOM && haveVulkan());
}
bool getMemoryShapes(const std::vector<MatShape> &inputs,
@@ -530,6 +530,12 @@ 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
+10 -2
View File
@@ -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
+1 -1
View File
@@ -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/eltwise_layer.cpp
+1 -1
View File
@@ -98,7 +98,7 @@ public:
{
return backendId == DNN_BACKEND_OPENCV ||
backendId == DNN_BACKEND_HALIDE ||
backendId == DNN_BACKEND_INFERENCE_ENGINE && (op != SUM || coeffs.empty());
(backendId == DNN_BACKEND_INFERENCE_ENGINE && (op != SUM || coeffs.empty()));
}
bool getMemoryShapes(const std::vector<MatShape> &inputs,
modules/dnn/src/layers/flatten_layer.cpp
+1 -1
View File
@@ -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
+5 -2
View File
@@ -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
+2 -2
View File
@@ -93,8 +93,8 @@ 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_VKCOM && haveVulkan() && (size % 2 == 1) && (type == CHANNEL_NRM);
(backendId == DNN_BACKEND_INFERENCE_ENGINE && (preferableTarget != DNN_TARGET_MYRIAD || type == CHANNEL_NRM)) ||
(backendId == DNN_BACKEND_VKCOM && haveVulkan() && (size % 2 == 1) && (type == CHANNEL_NRM));
}
#ifdef HAVE_OPENCL
modules/dnn/src/layers/max_unpooling_layer.cpp
+1 -2
View File
@@ -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/padding_layer.cpp
+1 -1
View File
@@ -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
+2 -2
View File
@@ -106,8 +106,8 @@ public:
virtual bool supportBackend(int backendId) CV_OVERRIDE
{
return backendId == DNN_BACKEND_OPENCV ||
backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() ||
backendId == DNN_BACKEND_VKCOM && haveVulkan();
(backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine()) ||
(backendId == DNN_BACKEND_VKCOM && haveVulkan());
}
bool getMemoryShapes(const std::vector<MatShape> &inputs,
modules/dnn/src/layers/pooling_layer.cpp
+10 -6
View File
@@ -155,10 +155,10 @@ 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_VKCOM && haveVulkan() &&
(type == MAX || type == AVE);
(backendId == DNN_BACKEND_HALIDE && haveHalide() &&
(type == MAX || (type == AVE && !pad_t && !pad_l && !pad_b && !pad_r))) ||
(backendId == DNN_BACKEND_VKCOM && haveVulkan() &&
(type == MAX || type == AVE));
}
#ifdef HAVE_OPENCL
@@ -313,6 +313,10 @@ 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;
@@ -380,8 +384,8 @@ 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
+2 -2
View File
@@ -272,8 +272,8 @@ public:
virtual bool supportBackend(int backendId) CV_OVERRIDE
{
return backendId == DNN_BACKEND_OPENCV ||
backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() ||
backendId == DNN_BACKEND_VKCOM && haveVulkan();
(backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine()) ||
(backendId == DNN_BACKEND_VKCOM && haveVulkan());
}
bool getMemoryShapes(const std::vector<MatShape> &inputs,
modules/dnn/src/layers/proposal_layer.cpp
+1 -1
View File
@@ -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
+2 -2
View File
@@ -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
+1 -1
View File
@@ -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
+24 -7
View File
@@ -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
+2 -2
View File
@@ -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
+1 -1
View File
@@ -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
+3 -3
View File
@@ -90,9 +90,9 @@ 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_VKCOM && haveVulkan();
(backendId == DNN_BACKEND_HALIDE && haveHalide() && axisRaw == 1) ||
(backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine() && !logSoftMax) ||
(backendId == DNN_BACKEND_VKCOM && haveVulkan());
}
#ifdef HAVE_OPENCL
modules/dnn/src/ocl4dnn/src/ocl4dnn_conv_spatial.cpp
+1 -1
View File
@@ -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
+10
View File
@@ -508,6 +508,16 @@ 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";
modules/dnn/src/op_inf_engine.cpp
+23 -2
View File
@@ -309,7 +309,7 @@ void InfEngineBackendNet::setTargetDevice(InferenceEngine::TargetDevice device)
InferenceEngine::TargetDevice InfEngineBackendNet::getTargetDevice() CV_NOEXCEPT
{
return targetDevice;
return const_cast<const InfEngineBackendNet*>(this)->getTargetDevice();
}
InferenceEngine::TargetDevice InfEngineBackendNet::getTargetDevice() const CV_NOEXCEPT
@@ -387,6 +387,27 @@ 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())
@@ -559,7 +580,7 @@ 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
+3 -2
View File
@@ -25,10 +25,11 @@
#define INF_ENGINE_RELEASE_2018R1 2018010000
#define INF_ENGINE_RELEASE_2018R2 2018020000
#define INF_ENGINE_RELEASE_2018R3 2018030000
#define INF_ENGINE_RELEASE_2018R4 2018040000
#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 2018R4 by default")
#define INF_ENGINE_RELEASE INF_ENGINE_RELEASE_2018R4
#endif
#define INF_ENGINE_VER_MAJOR_GT(ver) (((INF_ENGINE_RELEASE) / 10000) > ((ver) / 10000))
modules/dnn/src/tensorflow/tf_importer.cpp
+2
View File
@@ -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)
{
modules/dnn/src/tensorflow/tf_io.cpp
-2
View File
@@ -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/test/test_backends.cpp
+5 -5
View File
@@ -174,7 +174,7 @@ 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 +184,7 @@ 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",
@@ -194,7 +194,7 @@ 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)
(backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
throw SkipTestException("");
processNet("dnn/openpose_pose_coco.caffemodel", "dnn/openpose_pose_coco.prototxt",
Size(368, 368));
@@ -203,7 +203,7 @@ TEST_P(DNNTestNetwork, OpenPose_pose_coco)
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",
Size(368, 368));
@@ -212,7 +212,7 @@ TEST_P(DNNTestNetwork, OpenPose_pose_mpi)
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
modules/dnn/test/test_caffe_importer.cpp
+5 -1
View File
@@ -512,7 +512,11 @@ 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,
modules/dnn/test/test_common.hpp
+103 -3
View File
@@ -57,7 +57,7 @@ static inline void PrintTo(const cv::dnn::Backend& v, std::ostream* os)
case DNN_BACKEND_OPENCV: *os << "OCV"; return;
case DNN_BACKEND_VKCOM: *os << "VKCOM"; return;
} // don't use "default:" to emit compiler warnings
*os << "DNN_BACKEND_UNKNOWN(" << v << ")";
*os << "DNN_BACKEND_UNKNOWN(" << (int)v << ")";
}
static inline void PrintTo(const cv::dnn::Target& v, std::ostream* os)
@@ -69,7 +69,7 @@ static inline void PrintTo(const cv::dnn::Target& v, std::ostream* os)
case DNN_TARGET_MYRIAD: *os << "MYRIAD"; return;
case DNN_TARGET_VULKAN: *os << "VULKAN"; return;
} // don't use "default:" to emit compiler warnings
*os << "DNN_TARGET_UNKNOWN(" << v << ")";
*os << "DNN_TARGET_UNKNOWN(" << (int)v << ")";
}
using opencv_test::tuple;
@@ -237,7 +237,8 @@ namespace opencv_test {
using namespace cv::dnn;
static testing::internal::ParamGenerator<tuple<Backend, Target> > dnnBackendsAndTargets(
static inline
testing::internal::ParamGenerator<tuple<Backend, Target> > dnnBackendsAndTargets(
bool withInferenceEngine = true,
bool withHalide = false,
bool withCpuOCV = true,
@@ -290,4 +291,103 @@ static testing::internal::ParamGenerator<tuple<Backend, Target> > dnnBackendsAnd
} // namespace
namespace opencv_test {
using namespace cv::dnn;
static inline
testing::internal::ParamGenerator<Target> availableDnnTargets()
{
static std::vector<Target> targets;
if (targets.empty())
{
targets.push_back(DNN_TARGET_CPU);
#ifdef HAVE_OPENCL
if (cv::ocl::useOpenCL())
targets.push_back(DNN_TARGET_OPENCL);
#endif
}
return testing::ValuesIn(targets);
}
class DNNTestLayer : public TestWithParam<tuple<Backend, Target> >
{
public:
dnn::Backend backend;
dnn::Target target;
double default_l1, default_lInf;
DNNTestLayer()
{
backend = (dnn::Backend)(int)get<0>(GetParam());
target = (dnn::Target)(int)get<1>(GetParam());
getDefaultThresholds(backend, target, &default_l1, &default_lInf);
}
static void getDefaultThresholds(int backend, int target, double* l1, double* lInf)
{
if (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD)
{
*l1 = 4e-3;
*lInf = 2e-2;
}
else
{
*l1 = 1e-5;
*lInf = 1e-4;
}
}
static void checkBackend(int backend, int target, Mat* inp = 0, Mat* ref = 0)
{
if (backend == DNN_BACKEND_OPENCV && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16))
{
#ifdef HAVE_OPENCL
if (!cv::ocl::useOpenCL())
#endif
{
throw 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");
}
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE < 2018030000
if (inp && ref && inp->size[0] != 1)
{
// Myriad plugin supports only batch size 1. Slice a single sample.
if (inp->size[0] == ref->size[0])
{
std::vector<cv::Range> range(inp->dims, Range::all());
range[0] = Range(0, 1);
*inp = inp->operator()(range);
range = std::vector<cv::Range>(ref->dims, Range::all());
range[0] = Range(0, 1);
*ref = ref->operator()(range);
}
else
throw SkipTestException("Myriad plugin supports only batch size 1");
}
#else
if (inp && ref && inp->dims == 4 && ref->dims == 4 &&
inp->size[0] != 1 && inp->size[0] != ref->size[0])
throw SkipTestException("Inconsistent batch size of input and output blobs for Myriad plugin");
#endif
}
}
protected:
void checkBackend(Mat* inp = 0, Mat* ref = 0)
{
checkBackend(backend, target, inp, ref);
}
};
} // namespace
#endif
modules/dnn/test/test_darknet_importer.cpp
+3
View File
@@ -306,6 +306,9 @@ TEST_P(Test_Darknet_nets, TinyYoloVoc)
// batch size 1
testDarknetModel(config_file, weights_file, ref.rowRange(0, 2), scoreDiff, iouDiff);
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE == 2018040000
if (backend == DNN_BACKEND_INFERENCE_ENGINE && target != DNN_TARGET_MYRIAD)
#endif
// batch size 2
testDarknetModel(config_file, weights_file, ref, scoreDiff, iouDiff);
}
modules/dnn/test/test_halide_layers.cpp
+5
View File
@@ -166,6 +166,11 @@ TEST_P(Deconvolution, Accuracy)
if (backendId == DNN_BACKEND_INFERENCE_ENGINE && targetId == DNN_TARGET_CPU &&
dilation.width == 2 && dilation.height == 2)
throw SkipTestException("");
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE == 2018040000
if (backendId == DNN_BACKEND_INFERENCE_ENGINE && targetId == DNN_TARGET_CPU &&
hasBias && group != 1)
throw SkipTestException("Test is disabled for OpenVINO 2018R4");
#endif
int sz[] = {inChannels, outChannels / group, kernel.height, kernel.width};
Mat weights(4, &sz[0], CV_32F);
modules/dnn/test/test_ie_models.cpp
+10
View File
@@ -177,10 +177,20 @@ TEST_P(DNNTestOpenVINO, models)
Target target = (dnn::Target)(int)get<0>(GetParam());
std::string modelName = get<1>(GetParam());
#ifdef INF_ENGINE_RELEASE
#if INF_ENGINE_RELEASE <= 2018030000
if (target == DNN_TARGET_MYRIAD && (modelName == "landmarks-regression-retail-0001" ||
modelName == "semantic-segmentation-adas-0001" ||
modelName == "face-reidentification-retail-0001"))
throw SkipTestException("");
#elif INF_ENGINE_RELEASE == 2018040000
if (modelName == "single-image-super-resolution-0034" ||
(target == DNN_TARGET_MYRIAD && (modelName == "license-plate-recognition-barrier-0001" ||
modelName == "landmarks-regression-retail-0009" ||
modelName == "semantic-segmentation-adas-0001")))
throw SkipTestException("");
#endif
#endif
std::string precision = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? "FP16" : "FP32";
std::string prefix = utils::fs::join("intel_models",
modules/dnn/test/test_layers.cpp
+6
View File
@@ -137,6 +137,10 @@ TEST_P(Test_Caffe_layers, Convolution)
TEST_P(Test_Caffe_layers, DeConvolution)
{
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE == 2018040000
if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_CPU)
throw SkipTestException("Test is disabled for OpenVINO 2018R4");
#endif
testLayerUsingCaffeModels("layer_deconvolution", true, false);
}
@@ -558,7 +562,9 @@ TEST_P(Test_Caffe_layers, FasterRCNN_Proposal)
normAssert(outs[i].rowRange(0, numDets), ref);
if (numDets < outs[i].size[0])
{
EXPECT_EQ(countNonZero(outs[i].rowRange(numDets, outs[i].size[0])), 0);
}
}
}
modules/dnn/test/test_misc.cpp
+3 -3
View File
@@ -140,9 +140,9 @@ TEST(LayerFactory, custom_layers)
net.setPreferableBackend(DNN_BACKEND_OPENCV);
Mat output = net.forward();
if (i == 0) EXPECT_EQ(output.at<float>(0), 1);
else if (i == 1) EXPECT_EQ(output.at<float>(0), 2);
else if (i == 2) EXPECT_EQ(output.at<float>(0), 1);
if (i == 0) { EXPECT_EQ(output.at<float>(0), 1); }
else if (i == 1) { EXPECT_EQ(output.at<float>(0), 2); }
else if (i == 2) { EXPECT_EQ(output.at<float>(0), 1); }
}
LayerFactory::unregisterLayer("CustomType");
}
modules/dnn/test/test_onnx_importer.cpp
+9 -3
View File
@@ -68,6 +68,12 @@ TEST_P(Test_ONNX_layers, Convolution)
testONNXModels("two_convolution");
}
TEST_P(Test_ONNX_layers, Deconvolution)
{
testONNXModels("deconvolution");
testONNXModels("two_deconvolution");
}
TEST_P(Test_ONNX_layers, Dropout)
{
testONNXModels("dropout");
@@ -118,8 +124,8 @@ TEST_P(Test_ONNX_layers, Transpose)
TEST_P(Test_ONNX_layers, Multiplication)
{
if (backend == DNN_BACKEND_OPENCV && target == DNN_TARGET_OPENCL_FP16 ||
backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
if ((backend == DNN_BACKEND_OPENCV && target == DNN_TARGET_OPENCL_FP16) ||
(backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD))
throw SkipTestException("");
testONNXModels("mul");
}
@@ -296,7 +302,7 @@ TEST_P(Test_ONNX_nets, ResNet101_DUC_HDC)
TEST_P(Test_ONNX_nets, TinyYolov2)
{
if (cvtest::skipUnstableTests ||
backend == DNN_BACKEND_INFERENCE_ENGINE && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16)) {
(backend == DNN_BACKEND_INFERENCE_ENGINE && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16))) {
throw SkipTestException("");
}
// output range: [-11; 8]
modules/dnn/test/test_precomp.hpp
-96
View File
@@ -49,100 +49,4 @@
#include "opencv2/dnn.hpp"
#include "test_common.hpp"
namespace opencv_test {
using namespace cv::dnn;
static testing::internal::ParamGenerator<Target> availableDnnTargets()
{
static std::vector<Target> targets;
if (targets.empty())
{
targets.push_back(DNN_TARGET_CPU);
#ifdef HAVE_OPENCL
if (cv::ocl::useOpenCL())
targets.push_back(DNN_TARGET_OPENCL);
#endif
}
return testing::ValuesIn(targets);
}
class DNNTestLayer : public TestWithParam<tuple<Backend, Target> >
{
public:
dnn::Backend backend;
dnn::Target target;
double default_l1, default_lInf;
DNNTestLayer()
{
backend = (dnn::Backend)(int)get<0>(GetParam());
target = (dnn::Target)(int)get<1>(GetParam());
getDefaultThresholds(backend, target, &default_l1, &default_lInf);
}
static void getDefaultThresholds(int backend, int target, double* l1, double* lInf)
{
if (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD)
{
*l1 = 4e-3;
*lInf = 2e-2;
}
else
{
*l1 = 1e-5;
*lInf = 1e-4;
}
}
static void checkBackend(int backend, int target, Mat* inp = 0, Mat* ref = 0)
{
if (backend == DNN_BACKEND_OPENCV && (target == DNN_TARGET_OPENCL || target == DNN_TARGET_OPENCL_FP16))
{
#ifdef HAVE_OPENCL
if (!cv::ocl::useOpenCL())
#endif
{
throw 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");
}
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE < 2018030000
if (inp && ref && inp->size[0] != 1)
{
// Myriad plugin supports only batch size 1. Slice a single sample.
if (inp->size[0] == ref->size[0])
{
std::vector<cv::Range> range(inp->dims, Range::all());
range[0] = Range(0, 1);
*inp = inp->operator()(range);
range = std::vector<cv::Range>(ref->dims, Range::all());
range[0] = Range(0, 1);
*ref = ref->operator()(range);
}
else
throw SkipTestException("Myriad plugin supports only batch size 1");
}
#else
if (inp && ref && inp->dims == 4 && ref->dims == 4 &&
inp->size[0] != 1 && inp->size[0] != ref->size[0])
throw SkipTestException("Inconsistent batch size of input and output blobs for Myriad plugin");
#endif
}
}
protected:
void checkBackend(Mat* inp = 0, Mat* ref = 0)
{
checkBackend(backend, target, inp, ref);
}
};
} // namespace
#endif
modules/dnn/test/test_tf_importer.cpp
+3 -1
View File
@@ -101,7 +101,9 @@ public:
string dataConfig;
if (hasText)
{
ASSERT_TRUE(readFileInMemory(netConfig, dataConfig));
}
net = readNetFromTensorflow(dataModel.c_str(), dataModel.size(),
dataConfig.c_str(), dataConfig.size());
@@ -473,7 +475,7 @@ TEST_P(Test_TensorFlow_nets, EAST_text_detection)
double l1_geometry = default_l1, lInf_geometry = default_lInf;
if (target == DNN_TARGET_OPENCL_FP16)
{
lInf_scores = 0.11;
lInf_scores = backend == DNN_BACKEND_INFERENCE_ENGINE ? 0.16 : 0.11;
l1_geometry = 0.28; lInf_geometry = 5.94;
}
else if (target == DNN_TARGET_MYRIAD)
modules/dnn/test/test_torch_importer.cpp
+8
View File
@@ -136,6 +136,10 @@ TEST_P(Test_Torch_layers, run_reshape_change_batch_size)
TEST_P(Test_Torch_layers, run_reshape)
{
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE == 2018040000
if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
throw SkipTestException("Test is disabled for OpenVINO 2018R4");
#endif
runTorchNet("net_reshape_batch");
runTorchNet("net_reshape_channels", "", false, true);
}
@@ -168,6 +172,10 @@ TEST_P(Test_Torch_layers, run_depth_concat)
TEST_P(Test_Torch_layers, run_deconv)
{
#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_RELEASE == 2018040000
if (backend == DNN_BACKEND_INFERENCE_ENGINE && target == DNN_TARGET_MYRIAD)
throw SkipTestException("Test is disabled for OpenVINO 2018R4");
#endif
runTorchNet("net_deconv");
}