opencv
diff --git a/‎cmake/OpenCVCompilerOptimizations.cmake
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion b/‎cmake/OpenCVCompilerOptimizations.cmake
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion
diff --git a/‎cmake/checks/cpu_fp16.cpp
Copy file name to clipboard
+5-5Lines changed: 5 additions & 5 deletions b/‎cmake/checks/cpu_fp16.cpp
Copy file name to clipboard
+5-5Lines changed: 5 additions & 5 deletions
diff --git a/‎modules/core/include/opencv2/core/cv_cpu_helper.h
Copy file name to clipboard
+42Lines changed: 42 additions & 0 deletions b/‎modules/core/include/opencv2/core/cv_cpu_helper.h
Copy file name to clipboard
+42Lines changed: 42 additions & 0 deletions
diff --git a/‎modules/core/test/ocl/test_image2d.cpp
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion b/‎modules/core/test/ocl/test_image2d.cpp
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion
diff --git a/‎modules/dnn/CMakeLists.txt
Copy file name to clipboard
+2-2Lines changed: 2 additions & 2 deletions b/‎modules/dnn/CMakeLists.txt
Copy file name to clipboard
+2-2Lines changed: 2 additions & 2 deletions
diff --git a/‎modules/dnn/include/opencv2/dnn/all_layers.hpp
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion b/‎modules/dnn/include/opencv2/dnn/all_layers.hpp
Copy file name to clipboard
+1-1Lines changed: 1 addition & 1 deletion
diff --git a/‎modules/dnn/include/opencv2/dnn/dnn.hpp
Copy file name to clipboard
+3Lines changed: 3 additions & 0 deletions b/‎modules/dnn/include/opencv2/dnn/dnn.hpp
Copy file name to clipboard
+3Lines changed: 3 additions & 0 deletions
@@ -49,7 +49,7 @@
 
 set(CPU_ALL_OPTIMIZATIONS "SSE;SSE2;SSE3;SSSE3;SSE4_1;SSE4_2;POPCNT;AVX;FP16;AVX2;FMA3;AVX_512F")
 list(APPEND CPU_ALL_OPTIMIZATIONS "AVX512_COMMON;AVX512_KNL;AVX512_KNM;AVX512_SKX;AVX512_CNL;AVX512_CLX;AVX512_ICL")
-list(APPEND CPU_ALL_OPTIMIZATIONS NEON VFPV3 FP16 NEON_DOTPROD)
+list(APPEND CPU_ALL_OPTIMIZATIONS NEON VFPV3 FP16 NEON_DOTPROD NEON_FP16 NEON_BF16)
 list(APPEND CPU_ALL_OPTIMIZATIONS MSA)
 list(APPEND CPU_ALL_OPTIMIZATIONS VSX VSX3)
 list(APPEND CPU_ALL_OPTIMIZATIONS RVV)
 
@@ -15,12 +15,12 @@ int test()
 #include "arm_neon.h"
 int test()
 {
-    const float src[] = { 0.0f, 0.0f, 0.0f, 0.0f };
-    short dst[8];
-    float32x4_t v_src = *(float32x4_t*)src;
+    const float src[] = { 0.0f, 1.0f, 2.0f, 3.0f };
+    short dst[4];
+    float32x4_t v_src = vld1q_f32(src);
     float16x4_t v_dst = vcvt_f16_f32(v_src);
-    *(float16x4_t*)dst = v_dst;
-    return (int)dst[0];
+    vst1_f16((__fp16*)dst, v_dst);
+    return dst[0] + dst[1] + dst[2] + dst[3];
 }
 #else
 #error "FP16 is not supported"
 
@@ -441,6 +441,48 @@
 #endif
 #define __CV_CPU_DISPATCH_CHAIN_NEON_DOTPROD(fn, args, mode, ...)  CV_CPU_CALL_NEON_DOTPROD(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_NEON_FP16
+#  define CV_TRY_NEON_FP16 1
+#  define CV_CPU_FORCE_NEON_FP16 1
+#  define CV_CPU_HAS_SUPPORT_NEON_FP16 1
+#  define CV_CPU_CALL_NEON_FP16(fn, args) return (cpu_baseline::fn args)
+#  define CV_CPU_CALL_NEON_FP16_(fn, args) return (opt_NEON_FP16::fn args)
+#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_NEON_FP16
+#  define CV_TRY_NEON_FP16 1
+#  define CV_CPU_FORCE_NEON_FP16 0
+#  define CV_CPU_HAS_SUPPORT_NEON_FP16 (cv::checkHardwareSupport(CV_CPU_NEON_FP16))
+#  define CV_CPU_CALL_NEON_FP16(fn, args) if (CV_CPU_HAS_SUPPORT_NEON_FP16) return (opt_NEON_FP16::fn args)
+#  define CV_CPU_CALL_NEON_FP16_(fn, args) if (CV_CPU_HAS_SUPPORT_NEON_FP16) return (opt_NEON_FP16::fn args)
+#else
+#  define CV_TRY_NEON_FP16 0
+#  define CV_CPU_FORCE_NEON_FP16 0
+#  define CV_CPU_HAS_SUPPORT_NEON_FP16 0
+#  define CV_CPU_CALL_NEON_FP16(fn, args)
+#  define CV_CPU_CALL_NEON_FP16_(fn, args)
+#endif
+#define __CV_CPU_DISPATCH_CHAIN_NEON_FP16(fn, args, mode, ...)  CV_CPU_CALL_NEON_FP16(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_NEON_BF16
+#  define CV_TRY_NEON_BF16 1
+#  define CV_CPU_FORCE_NEON_BF16 1
+#  define CV_CPU_HAS_SUPPORT_NEON_BF16 1
+#  define CV_CPU_CALL_NEON_BF16(fn, args) return (cpu_baseline::fn args)
+#  define CV_CPU_CALL_NEON_BF16_(fn, args) return (opt_NEON_BF16::fn args)
+#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_NEON_BF16
+#  define CV_TRY_NEON_BF16 1
+#  define CV_CPU_FORCE_NEON_BF16 0
+#  define CV_CPU_HAS_SUPPORT_NEON_BF16 (cv::checkHardwareSupport(CV_CPU_NEON_BF16))
+#  define CV_CPU_CALL_NEON_BF16(fn, args) if (CV_CPU_HAS_SUPPORT_NEON_BF16) return (opt_NEON_BF16::fn args)
+#  define CV_CPU_CALL_NEON_BF16_(fn, args) if (CV_CPU_HAS_SUPPORT_NEON_BF16) return (opt_NEON_BF16::fn args)
+#else
+#  define CV_TRY_NEON_BF16 0
+#  define CV_CPU_FORCE_NEON_BF16 0
+#  define CV_CPU_HAS_SUPPORT_NEON_BF16 0
+#  define CV_CPU_CALL_NEON_BF16(fn, args)
+#  define CV_CPU_CALL_NEON_BF16_(fn, args)
+#endif
+#define __CV_CPU_DISPATCH_CHAIN_NEON_BF16(fn, args, mode, ...)  CV_CPU_CALL_NEON_BF16(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_MSA
 #  define CV_TRY_MSA 1
 #  define CV_CPU_FORCE_MSA 1
 
@@ -83,7 +83,7 @@ TEST(Image2D, turnOffOpenCL)
         }
         else
             std::cout << "CV_8UC1 is not supported for OpenCL images. Test skipped." << std::endl;
-    
+
         // reset state to the previous one
         cv::ocl::setUseOpenCL(useOCL);
     }
 
@@ -6,9 +6,9 @@ set(the_description "Deep neural network module. It allows to load models from d
 
 ocv_add_dispatched_file_force_all("layers/layers_common" AVX AVX2 AVX512_SKX RVV LASX)
 ocv_add_dispatched_file_force_all("int8layers/layers_common" AVX2 AVX512_SKX LASX)
-ocv_add_dispatched_file_force_all("layers/cpu_kernels/conv_block" AVX AVX2)
+ocv_add_dispatched_file_force_all("layers/cpu_kernels/conv_block" AVX AVX2 NEON NEON_FP16)
 ocv_add_dispatched_file_force_all("layers/cpu_kernels/conv_depthwise" AVX AVX2 RVV LASX)
-ocv_add_dispatched_file_force_all("layers/cpu_kernels/conv_winograd_f63" AVX AVX2)
+ocv_add_dispatched_file_force_all("layers/cpu_kernels/conv_winograd_f63" AVX AVX2 NEON_FP16)
 ocv_add_dispatched_file_force_all("layers/cpu_kernels/fast_gemm_kernels" AVX AVX2 NEON LASX)
 
 ocv_add_module(dnn opencv_core opencv_imgproc WRAP python java objc js)
 
@@ -303,7 +303,7 @@ CV__DNN_INLINE_NS_BEGIN
         // quantization type flag. The perChannel default is true, that means it contains the parameters
         // of per-Channel quantization. Otherwise, that means this layer contains per-Tensor quantized parameters.
         bool per_channel;
-        bool useWinograd = true; // Flag whether to use Winograd to speed up 3x3 convolution.
+        bool useWinograd = false; // Flag whether to use Winograd to speed up 3x3 convolution.
         static Ptr<BaseConvolutionLayer> create(const LayerParams& params);
     };
 
 
@@ -1458,6 +1458,9 @@ CV__DNN_INLINE_NS_BEGIN
          /// @sa Net::setPreferableTarget
          CV_WRAP Model& setPreferableTarget(dnn::Target targetId);
 
+         /// @sa Net::enableWinograd
+         CV_WRAP Model& enableWinograd(bool useWinograd);
+
          CV_DEPRECATED_EXTERNAL
          operator Net&() const { return getNetwork_(); }
Original file line number	Diff line number	Diff line change
`@@ -83,7 +83,7 @@ TEST(Image2D, turnOffOpenCL)`
`83`	`83`	`}`
`84`	`84`	`else`
`85`	`85`	`std::cout << "CV_8UC1 is not supported for OpenCL images. Test skipped." << std::endl;`
`86`		`-`
	`86`	`+`
`87`	`87`	`// reset state to the previous one`
`88`	`88`	`cv::ocl::setUseOpenCL(useOCL);`
`89`	`89`	`}`