potok/opencv
1
0
Fork 0
mirror of https://github.com/opencv/opencv.git synced 2026-07-21 19:33:03 +04:00
Code Packages Releases Wiki Activity

Merge pull request #27776 from pratham-mcw:win-arm64-agast-detect-optimization

Browse Source 
features2d: performance optimization of detect function on Windows-ARM64 #27776

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch

- This PR improves the performance of the detect function on Windows ARM64 targets.
- Added the defined(_M_ARM64) macro in agast.cpp and agast_score.cpp files, aligning ARM64 behavior with how x64 selects internal functions for computation.
- As a result, ARM64 now executes the same internal functions as x64 where applicable, leading to measurable performance improvements in detect function.
- This changes is limited to Windows ARM64 and does not affect other architectures

**Performance impact:**
- Detect function shows improved runtime on ARM64 targets due to reuse of existing efficient computation paths.
<img width="1419" height="408" alt="image" src="https://github.com/user-attachments/assets/feab411a-d256-4bff-bec2-22b2583f63d1" />
This commit is contained in:
pratham-mcw
2025-09-12 19:58:54 +05:30
committed by GitHub
parent b7bc18670b
commit a7943cef60
2 changed files with 6 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/features2d/src/agast.cpp
+3 -3
View File
@@ -48,7 +48,7 @@ The references are:
namespace cv
{
#if (defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64))
#if (defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64) || defined(_M_ARM64) || defined(__aarch64__) || defined(__arm__))
static void AGAST_5_8(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold)
{
@@ -7444,7 +7444,7 @@ static void OAST_9_16(InputArray _img, std::vector<KeyPoint>& keypoints, int thr
#else // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64))
#else // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64) || defined(_M_ARM64) || defined(__aarch64__) || defined(__arm__))
static void AGAST_ALL(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold, AgastFeatureDetector::DetectorType agasttype)
{
@@ -7932,7 +7932,7 @@ static void OAST_9_16(InputArray _img, std::vector<KeyPoint>& keypoints, int thr
AGAST_ALL(_img, keypoints, threshold, AgastFeatureDetector::OAST_9_16);
}
#endif // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64))
#endif // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64) || defined(_M_ARM64) || defined(__aarch64__) || defined(__arm__))
void AGAST(InputArray _img, std::vector<KeyPoint>& keypoints, int threshold, bool nonmax_suppression)
{
modules/features2d/src/agast_score.cpp
+3 -3
View File
@@ -90,7 +90,7 @@ void makeAgastOffsets(int pixel[16], int rowStride, AgastFeatureDetector::Detect
pixel[k] = offsets[k][0] + offsets[k][1] * rowStride;
}
#if (defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64))
#if (defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64) || defined(_M_ARM64) || defined(__aarch64__) || defined(__arm__))
// 16 pixel mask
template<>
int agast_cornerScore<AgastFeatureDetector::OAST_9_16>(const uchar* ptr, const int pixel[], int threshold)
@@ -9373,7 +9373,7 @@ int agast_cornerScore<AgastFeatureDetector::AGAST_5_8>(const uchar* ptr, const i
b_test = (bmin + bmax) / 2;
}
}
#else // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64))
#else // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64) || defined(_M_ARM64) || defined(__aarch64__) || defined(__arm__))
int agast_tree_search(const uint32_t table_struct32[], int pixel_[], const unsigned char* const ptr, int threshold)
@@ -9858,6 +9858,6 @@ int agast_cornerScore<AgastFeatureDetector::OAST_9_16>(const uchar* ptr, const i
return AGAST_ALL_SCORE(ptr, pixel, threshold, AgastFeatureDetector::OAST_9_16);
}
#endif // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64))
#endif // !(defined __i386__ || defined(_M_IX86) || defined __x86_64__ || defined(_M_X64) || defined(_M_ARM64) || defined(__aarch64__) || defined(__arm__))
} // namespace cv