From c16c9a2e8e239b2b50144046928549778038c1c3 Mon Sep 17 00:00:00 2001 From: Ilya Lavrenov Date: Sun, 8 Dec 2013 14:45:25 +0400 Subject: [PATCH] added cv::split to T-API --- modules/core/src/convert.cpp | 48 ++++++++- modules/core/src/ocl.cpp | 23 ++--- modules/core/src/opencl/split_merge.cl | 13 ++- modules/core/test/ocl/test_arithm.cpp | 4 + modules/core/test/ocl/test_split_merge.cpp | 150 +++++++++++++++++------------ 5 files changed, 150 insertions(+), 88 deletions(-) diff --git a/modules/core/src/convert.cpp b/modules/core/src/convert.cpp index c6cc0fd..1b1ceac 100644 --- a/modules/core/src/convert.cpp +++ b/modules/core/src/convert.cpp @@ -264,8 +264,50 @@ void cv::split(const Mat& src, Mat* mv) } } +namespace cv { + +static bool ocl_split( InputArray _m, OutputArrayOfArrays _mv ) +{ + int type = _m.type(), depth = CV_MAT_DEPTH(type), cn = CV_MAT_CN(type); + + String dstargs, dstdecl, processelem; + for (int i = 0; i < cn; ++i) + { + dstargs += format("DECLARE_DST_PARAM(%d)", i); + dstdecl += format("DECLARE_DATA(%d)", i); + processelem += format("PROCESS_ELEM(%d)", i); + } + + ocl::Kernel k("split", ocl::core::split_merge_oclsrc, + format("-D T=%s -D OP_SPLIT -D cn=%d -D DECLARE_DST_PARAMS=%s " + "-D DECLARE_DATA_N=%s -D PROCESS_ELEMS_N=%s", + ocl::memopTypeToStr(depth), cn, dstargs.c_str(), + dstdecl.c_str(), processelem.c_str())); + if (k.empty()) + return false; + + Size size = _m.size(); + std::vector & dst = *(std::vector *)_mv.getObj(); + dst.resize(cn); + for (int i = 0; i < cn; ++i) + dst[i].create(size, depth); + + int argidx = k.set(0, ocl::KernelArg::ReadOnly(_m.getUMat())); + for (int i = 0; i < cn; ++i) + argidx = k.set(argidx, ocl::KernelArg::WriteOnlyNoSize(dst[i])); + + size_t globalsize[2] = { size.width, size.height }; + return k.run(2, globalsize, NULL, false); +} + +} + void cv::split(InputArray _m, OutputArrayOfArrays _mv) { + if (ocl::useOpenCL() && _m.dims() <= 2 && _mv.isUMatVector() && + ocl_split(_m, _mv)) + return; + Mat m = _m.getMat(); if( m.empty() ) { @@ -362,10 +404,6 @@ static bool ocl_merge( InputArrayOfArrays _mv, OutputArray _dst ) int type = src[0].type(), depth = CV_MAT_DEPTH(type); Size size = src[0].size(); - bool doubleSupport = ocl::Device::getDefault().doubleFPConfig() > 0; - - if (doubleSupport && depth == CV_64F) - return false; size_t srcsize = src.size(); for (size_t i = 0; i < srcsize; ++i) @@ -390,7 +428,7 @@ static bool ocl_merge( InputArrayOfArrays _mv, OutputArray _dst ) if (k.empty()) return false; - _dst.create(size, CV_MAKE_TYPE(depth, srcsize)); + _dst.create(size, CV_MAKE_TYPE(depth, (int)srcsize)); UMat dst = _dst.getUMat(); int argidx = 0; diff --git a/modules/core/src/ocl.cpp b/modules/core/src/ocl.cpp index 5fb0d35..f733dd1 100644 --- a/modules/core/src/ocl.cpp +++ b/modules/core/src/ocl.cpp @@ -1893,7 +1893,7 @@ Context2& Context2::getDefault() // First, try to retrieve existing context of the same type. // In its turn, Platform::getContext() may call Context2::create() // if there is no such context. - ctx.create(Device::TYPE_CPU); + ctx.create(Device::TYPE_ACCELERATOR); if(!ctx.p) ctx.create(Device::TYPE_DGPU); if(!ctx.p) @@ -2189,13 +2189,8 @@ int Kernel::set(int i, const void* value, size_t sz) CV_Assert(i >= 0); if( i == 0 ) p->cleanupUMats(); - cl_int retval; - if( !p || !p->handle || (retval = clSetKernelArg(p->handle, (cl_uint)i, sz, value)) < 0 ) - { - printf("%d\n", retval); + if( !p || !p->handle || clSetKernelArg(p->handle, (cl_uint)i, sz, value) < 0 ) return -1; - } - printf("%d\n", retval); return i+1; } @@ -2206,7 +2201,6 @@ int Kernel::set(int i, const UMat& m) int Kernel::set(int i, const KernelArg& arg) { - printf("Setting to index %d\n", i); CV_Assert( i >= 0 ); if( !p || !p->handle ) return -1; @@ -2220,21 +2214,20 @@ int Kernel::set(int i, const KernelArg& arg) cl_mem h = (cl_mem)arg.m->handle(accessFlags); if (ptronly) - printf("%d\n", clSetKernelArg(p->handle, (cl_uint)i++, sizeof(h), &h)); + clSetKernelArg(p->handle, (cl_uint)i++, sizeof(h), &h); else if( arg.m->dims <= 2 ) { UMat2D u2d(*arg.m); - printf("setting ... \n"); - printf("%d\n", clSetKernelArg(p->handle, (cl_uint)i, sizeof(h), &h)); - printf("%d\n", clSetKernelArg(p->handle, (cl_uint)(i+1), sizeof(u2d.step), &u2d.step)); - printf("%d\n", clSetKernelArg(p->handle, (cl_uint)(i+2), sizeof(u2d.offset), &u2d.offset)); + clSetKernelArg(p->handle, (cl_uint)i, sizeof(h), &h); + clSetKernelArg(p->handle, (cl_uint)(i+1), sizeof(u2d.step), &u2d.step); + clSetKernelArg(p->handle, (cl_uint)(i+2), sizeof(u2d.offset), &u2d.offset); i += 3; if( !(arg.flags & KernelArg::NO_SIZE) ) { int cols = u2d.cols*arg.wscale; - printf("%d\n", clSetKernelArg(p->handle, (cl_uint)i, sizeof(u2d.rows), &u2d.rows)); - printf("%d\n", clSetKernelArg(p->handle, (cl_uint)(i+1), sizeof(cols), &cols)); + clSetKernelArg(p->handle, (cl_uint)i, sizeof(u2d.rows), &u2d.rows); + clSetKernelArg(p->handle, (cl_uint)(i+1), sizeof(cols), &cols); i += 2; } } diff --git a/modules/core/src/opencl/split_merge.cl b/modules/core/src/opencl/split_merge.cl index 2fd7b51..d246275 100644 --- a/modules/core/src/opencl/split_merge.cl +++ b/modules/core/src/opencl/split_merge.cl @@ -65,16 +65,21 @@ __kernel void merge(DECLARE_SRC_PARAMS_N #elif defined OP_SPLIT -__kernel void set(__global uchar* dstptr, int dststep, int dstoffset, - int rows, int cols, dstT value ) +#define DECLARE_DST_PARAM(index) , __global uchar * dst##index##ptr, int dst##index##_step, int dst##index##_offset +#define DECLARE_DATA(index) __global T * dst##index = \ + (__global T *)(dst##index##ptr + mad24(y, dst##index##_step, x * (int)sizeof(T) + dst##index##_offset)); +#define PROCESS_ELEM(index) dst##index[0] = src[index]; + +__kernel void split(__global uchar* srcptr, int src_step, int src_offset, int rows, int cols DECLARE_DST_PARAMS) { int x = get_global_id(0); int y = get_global_id(1); if (x < cols && y < rows) { - int dst_index = mad24(y, dststep, x*(int)sizeof(dstT) + dstoffset); - *(__global dstT*)(dstptr + dst_index) = value; + DECLARE_DATA_N + __global const T * src = (__global const T *)(srcptr + mad24(y, src_step, x * cn * (int)sizeof(T) + src_offset)); + PROCESS_ELEMS_N } } diff --git a/modules/core/test/ocl/test_arithm.cpp b/modules/core/test/ocl/test_arithm.cpp index 9ef0d21..844be7b 100644 --- a/modules/core/test/ocl/test_arithm.cpp +++ b/modules/core/test/ocl/test_arithm.cpp @@ -42,6 +42,8 @@ #include "test_precomp.hpp" #include "opencv2/ts/ocl_test.hpp" +#ifdef HAVE_OPENCL + namespace cvtest { namespace ocl { @@ -1034,3 +1036,5 @@ OCL_INSTANTIATE_TEST_CASE_P(Arithm, Magnitude, Combine(::testing::Values(CV_32F, OCL_INSTANTIATE_TEST_CASE_P(Arithm, Flip, Combine(Values(CV_8U, CV_8S, CV_16U, CV_16S, CV_32S, CV_32F, CV_64F), Values(1, 2, 3, 4), Bool())); } } // namespace cvtest::ocl + +#endif // HAVE_OPENCL diff --git a/modules/core/test/ocl/test_split_merge.cpp b/modules/core/test/ocl/test_split_merge.cpp index 70ba2d5..224963c 100644 --- a/modules/core/test/ocl/test_split_merge.cpp +++ b/modules/core/test/ocl/test_split_merge.cpp @@ -47,6 +47,8 @@ #include "test_precomp.hpp" #include "opencv2/ts/ocl_test.hpp" +#ifdef HAVE_OPENCL + namespace cvtest { namespace ocl { @@ -69,11 +71,12 @@ PARAM_TEST_CASE(MergeTestBase, MatDepth, Channels, bool) depth = GET_PARAM(0); cn = GET_PARAM(1); use_roi = GET_PARAM(2); + + CV_Assert(cn >= 1 && cn <= 4); } - virtual void random_roi() + void random_roi() { - CV_Assert(cn >= 1 && cn <= 4); Size roiSize = randomSize(1, MAX_VALUE); { @@ -130,72 +133,91 @@ OCL_TEST_P(Merge, Accuracy) } } -//PARAM_TEST_CASE(SplitTestBase, MatType, int, bool) -//{ -// int type; -// int channels; -// bool use_roi; - -// cv::Mat src, src_roi; -// cv::Mat dst[MAX_CHANNELS], dst_roi[MAX_CHANNELS]; - -// cv::ocl::oclMat gsrc_whole, gsrc_roi; -// cv::ocl::oclMat gdst_whole[MAX_CHANNELS], gdst_roi[MAX_CHANNELS]; - -// virtual void SetUp() -// { -// type = GET_PARAM(0); -// channels = GET_PARAM(1); -// use_roi = GET_PARAM(2); -// } - -// void random_roi() -// { -// Size roiSize = randomSize(1, MAX_VALUE); -// Border srcBorder = randomBorder(0, use_roi ? MAX_VALUE : 0); -// randomSubMat(src, src_roi, roiSize, srcBorder, CV_MAKETYPE(type, channels), 0, 256); -// generateOclMat(gsrc_whole, gsrc_roi, src, roiSize, srcBorder); - -// for (int i = 0; i < channels; ++i) -// { -// Border dstBorder = randomBorder(0, use_roi ? MAX_VALUE : 0); -// randomSubMat(dst[i], dst_roi[i], roiSize, dstBorder, CV_MAKETYPE(type, 1), 5, 16); -// generateOclMat(gdst_whole[i], gdst_roi[i], dst[i], roiSize, dstBorder); -// } -// } -//}; - -//struct Split : SplitTestBase {}; - -//#ifdef ANDROID -//// NOTE: The test fail on Android is the top of the iceberg only -//// The real fail reason is memory access vialation somewhere else -//OCL_TEST_P(Split, DISABLED_Accuracy) -//#else -//OCL_TEST_P(Split, Accuracy) -//#endif -//{ -// for(int j = 0; j < LOOP_TIMES; j++) -// { -// random_roi(); - -// cv::split(src_roi, dst_roi); -// cv::ocl::split(gsrc_roi, gdst_roi); - -// for (int i = 0; i < channels; ++i) -// { -// EXPECT_MAT_NEAR(dst[i], gdst_whole[i], 0.0); -// EXPECT_MAT_NEAR(dst_roi[i], gdst_roi[i], 0.0); -// } -// } -//} +PARAM_TEST_CASE(SplitTestBase, MatType, Channels, bool) +{ + int depth, cn; + bool use_roi; + TEST_DECLARE_INPUT_PARAMETER(src) + TEST_DECLARE_OUTPUT_PARAMETER(dst1) + TEST_DECLARE_OUTPUT_PARAMETER(dst2) + TEST_DECLARE_OUTPUT_PARAMETER(dst3) + TEST_DECLARE_OUTPUT_PARAMETER(dst4) -OCL_INSTANTIATE_TEST_CASE_P(SplitMerge, Merge, Combine(OCL_ALL_DEPTHS, OCL_ALL_CHANNELS, Bool())); + std::vector dst_roi, dst; + std::vector udst_roi, udst; + + virtual void SetUp() + { + depth = GET_PARAM(0); + cn = GET_PARAM(1); + use_roi = GET_PARAM(2); + + CV_Assert(cn >= 1 && cn <= 4); + } + + void random_roi() + { + Size roiSize = randomSize(1, MAX_VALUE); + Border srcBorder = randomBorder(0, use_roi ? MAX_VALUE : 0); + randomSubMat(src, src_roi, roiSize, srcBorder, CV_MAKE_TYPE(depth, cn), 5, 16); + + { + Border dst1Border = randomBorder(0, use_roi ? MAX_VALUE : 0); + randomSubMat(dst1, dst1_roi, roiSize, dst1Border, depth, 2, 11); + Border dst2Border = randomBorder(0, use_roi ? MAX_VALUE : 0); + randomSubMat(dst2, dst2_roi, roiSize, dst2Border, depth, -1540, 1740); -//INSTANTIATE_TEST_CASE_P(SplitMerge, Split , Combine( -// Values(CV_8U, CV_8S, CV_16U, CV_16S, CV_32S, CV_32F), Values(1, 2, 3, 4), Bool())); + Border dst3Border = randomBorder(0, use_roi ? MAX_VALUE : 0); + randomSubMat(dst3, dst3_roi, roiSize, dst3Border, depth, -1540, 1740); + Border dst4Border = randomBorder(0, use_roi ? MAX_VALUE : 0); + randomSubMat(dst4, dst4_roi, roiSize, dst4Border, depth, -1540, 1740); + } + + UMAT_UPLOAD_INPUT_PARAMETER(src) + UMAT_UPLOAD_OUTPUT_PARAMETER(dst1) + UMAT_UPLOAD_OUTPUT_PARAMETER(dst2) + UMAT_UPLOAD_OUTPUT_PARAMETER(dst3) + UMAT_UPLOAD_OUTPUT_PARAMETER(dst4) + + dst_roi.push_back(dst1_roi), udst_roi.push_back(udst1_roi), + dst.push_back(dst1), udst.push_back(udst1); + if (cn >= 2) + dst_roi.push_back(dst2_roi), udst_roi.push_back(udst2_roi), + dst.push_back(dst2), udst.push_back(udst2); + if (cn >= 3) + dst_roi.push_back(dst3_roi), udst_roi.push_back(udst3_roi), + dst.push_back(dst3), udst.push_back(udst3); + if (cn >= 4) + dst_roi.push_back(dst4_roi), udst_roi.push_back(udst4_roi), + dst.push_back(dst4), udst.push_back(udst4); + } +}; + +typedef SplitTestBase Split; + +OCL_TEST_P(Split, Accuracy) +{ + for (int j = 0; j < test_loop_times; j++) + { + random_roi(); + + OCL_OFF(cv::split(src_roi, dst_roi)); + OCL_ON(cv::split(usrc_roi, udst_roi)); + + for (int i = 0; i < cn; ++i) + { + EXPECT_MAT_NEAR(dst[i], udst[i], 0.0); + EXPECT_MAT_NEAR(dst_roi[i], udst_roi[i], 0.0); + } + } +} + +OCL_INSTANTIATE_TEST_CASE_P(SplitMerge, Merge, Combine(OCL_ALL_DEPTHS, OCL_ALL_CHANNELS, Bool())); +OCL_INSTANTIATE_TEST_CASE_P(SplitMerge, Split, Combine(OCL_ALL_DEPTHS, OCL_ALL_CHANNELS, Bool())); } } // namespace cvtest::ocl + +#endif // HAVE_OPENCL -- 2.7.4