Skip to content

O
opencv
Commit 02fb3f0a authored by Vadim Pisarevsky's avatar Vadim Pisarevsky
Browse files
Options

it finally works!!!

parent ef509ace
Hide whitespace changes
Inline Side-by-side
Showing with 151 additions and 575 deletions
modules/objdetect/src/cascadedetect.cpp
View file @ 02fb3f0a
...@@ -44,6 +44,7 @@ ...@@ -44,6 +44,7 @@
#include "cascadedetect.hpp" #include "cascadedetect.hpp"
#include "opencv2/objdetect/objdetect_c.h" #include "opencv2/objdetect/objdetect_c.h"
#include "opencl_kernels.hpp"
#if defined (LOG_CASCADE_STATISTIC) #if defined (LOG_CASCADE_STATISTIC)
struct Logger struct Logger
...@@ -491,7 +492,7 @@ bool HaarEvaluator::read(const FileNode& node) ...@@ -491,7 +492,7 @@ bool HaarEvaluator::read(const FileNode& node)
features->resize(n); features->resize(n);
FileNodeIterator it = node.begin(); FileNodeIterator it = node.begin();
hasTiltedFeatures = false; hasTiltedFeatures = false;
std::vector<Feature> ff = *features; std::vector<Feature>& ff = *features;
sumSize0 = Size(); sumSize0 = Size();
ufbuf.release(); ufbuf.release();
...@@ -552,30 +553,37 @@ bool HaarEvaluator::setImage( InputArray _image, Size _origWinSize, Size _sumSiz ...@@ -552,30 +553,37 @@ bool HaarEvaluator::setImage( InputArray _image, Size _origWinSize, Size _sumSiz
tofs = (int)((utilted.offset - usum.offset)/sizeof(int)); tofs = (int)((utilted.offset - usum.offset)/sizeof(int));
} }
else else
{
integral(_image, usum, noArray(), noArray(), CV_32S); integral(_image, usum, noArray(), noArray(), CV_32S);
}
sqrBoxFilter(_image, usqsum, CV_32S, sqrBoxFilter(_image, usqsum, CV_32S,
Size(normrect.width, normrect.height), Size(normrect.width, normrect.height),
Point(0, 0), false); Point(0, 0), false);
/*sqrBoxFilter(_image.getMat(), sqsum, CV_32S,
Size(normrect.width, normrect.height),
Point(0, 0), false);
sqsum.copyTo(usqsum);*/
sumStep = (int)(usum.step/usum.elemSize()); sumStep = (int)(usum.step/usum.elemSize());
} }
else else
{ {
sum0.create(rn*rn_scale, cn, CV_32S); sum0.create(rn*rn_scale, cn, CV_32S);
sqsum0.create(rn, cn, CV_64F); sqsum0.create(rn, cn, CV_32S);
sum = sum0(Rect(0, 0, cols+1, rows+1)); sum = sum0(Rect(0, 0, cols+1, rows+1));
sqsum = sqsum0(Rect(0, 0, cols+1, rows+1)); sqsum = sqsum0(Rect(0, 0, cols, rows));
if( hasTiltedFeatures ) if( hasTiltedFeatures )
{ {
Mat tilted = sum0(Rect(0, _sumSize.height, cols+1, rows+1)); Mat tilted = sum0(Rect(0, _sumSize.height, cols+1, rows+1));
integral(_image, sum, sqsum, tilted, CV_32S); integral(_image, sum, noArray(), tilted, CV_32S);
tofs = (int)((tilted.data - sum.data)/sizeof(int)); tofs = (int)((tilted.data - sum.data)/sizeof(int));
} }
else else
integral(_image, sum, sqsum, noArray(), CV_32S); integral(_image, sum, noArray(), noArray(), CV_32S);
/*sqrBoxFilter(_image, sqsum, CV_32S, sqrBoxFilter(_image, sqsum, CV_32S,
Size(normrect.width, normrect.height), Size(normrect.width, normrect.height),
Point(0, 0), false);*/ Point(0, 0), false);
sumStep = (int)(sum.step/sum.elemSize()); sumStep = (int)(sum.step/sum.elemSize());
} }
...@@ -592,7 +600,7 @@ bool HaarEvaluator::setImage( InputArray _image, Size _origWinSize, Size _sumSiz ...@@ -592,7 +600,7 @@ bool HaarEvaluator::setImage( InputArray _image, Size _origWinSize, Size _sumSiz
optfeaturesPtr[fi].setOffsets( ff[fi], sumStep, tofs ); optfeaturesPtr[fi].setOffsets( ff[fi], sumStep, tofs );
} }
if( _image.isUMat() && (sumSize0 != _sumSize || ufbuf.empty()) ) if( _image.isUMat() && (sumSize0 != _sumSize || ufbuf.empty()) )
copyVectorToUMat(ff, ufbuf); copyVectorToUMat(*optfeatures, ufbuf);
sumSize0 = _sumSize; sumSize0 = _sumSize;
return true; return true;
...@@ -608,13 +616,7 @@ bool HaarEvaluator::setWindow( Point pt ) ...@@ -608,13 +616,7 @@ bool HaarEvaluator::setWindow( Point pt )
const int* p = &sum.at<int>(pt); const int* p = &sum.at<int>(pt);
int valsum = CALC_SUM_OFS(nofs, p); int valsum = CALC_SUM_OFS(nofs, p);
double valsqsum = sqsum.at<int>(pt.y + normrect.y, pt.x + normrect.x);
int nqofs[4];
CV_SUM_OFS( nqofs[0], nqofs[1], nqofs[2], nqofs[3], 0, normrect, (int)(sqsum.step/sizeof(double)) );
const double* pq = &sqsum.at<double>(pt);
double valsqsum = CALC_SUM_OFS(nqofs, pq);
//double valsqsum = sqsum.at<int>(pt.y + normrect.y, pt.x + normrect.x);
double nf = (double)normrect.area() * valsqsum - (double)valsum * valsum; double nf = (double)normrect.area() * valsqsum - (double)valsum * valsum;
if( nf > 0. ) if( nf > 0. )
...@@ -1131,8 +1133,6 @@ bool CascadeClassifierImpl::detectSingleScale( InputArray _image, Size processin ...@@ -1131,8 +1133,6 @@ bool CascadeClassifierImpl::detectSingleScale( InputArray _image, Size processin
bool CascadeClassifierImpl::ocl_detectSingleScale( InputArray _image, Size processingRectSize, bool CascadeClassifierImpl::ocl_detectSingleScale( InputArray _image, Size processingRectSize,
int yStep, double factor, Size sumSize0 ) int yStep, double factor, Size sumSize0 )
{ {
const int MAX_FACES = 10000;
Ptr<HaarEvaluator> haar = featureEvaluator.dynamicCast<HaarEvaluator>(); Ptr<HaarEvaluator> haar = featureEvaluator.dynamicCast<HaarEvaluator>();
if( haar.empty() ) if( haar.empty() )
return false; return false;
...@@ -1141,7 +1141,8 @@ bool CascadeClassifierImpl::ocl_detectSingleScale( InputArray _image, Size proce ...@@ -1141,7 +1141,8 @@ bool CascadeClassifierImpl::ocl_detectSingleScale( InputArray _image, Size proce
if( cascadeKernel.empty() ) if( cascadeKernel.empty() )
{ {
//cascadeKernel.create(") cascadeKernel.create("runHaarClassifierStump", ocl::objdetect::haarobjectdetect_oclsrc,
format("-D MAX_FACES=%d", MAX_FACES));
if( cascadeKernel.empty() ) if( cascadeKernel.empty() )
return false; return false;
} }
...@@ -1152,30 +1153,35 @@ bool CascadeClassifierImpl::ocl_detectSingleScale( InputArray _image, Size proce ...@@ -1152,30 +1153,35 @@ bool CascadeClassifierImpl::ocl_detectSingleScale( InputArray _image, Size proce
copyVectorToUMat(data.classifiers, uclassifiers); copyVectorToUMat(data.classifiers, uclassifiers);
copyVectorToUMat(data.nodes, unodes); copyVectorToUMat(data.nodes, unodes);
copyVectorToUMat(data.leaves, uleaves); copyVectorToUMat(data.leaves, uleaves);
ufacepos.create(1, MAX_FACES*4 + 1, CV_32S);
} }
std::vector<UMat> bufs; std::vector<UMat> bufs;
haar->getUMats(bufs); haar->getUMats(bufs);
CV_Assert(bufs.size() == 3); CV_Assert(bufs.size() == 3);
Rect normrect = haar->getNormRect();
//processingRectSize = Size(yStep, yStep);
size_t globalsize[] = { processingRectSize.width/yStep, processingRectSize.height/yStep }; size_t globalsize[] = { processingRectSize.width/yStep, processingRectSize.height/yStep };
return cascadeKernel.args(ocl::KernelArg::ReadOnly(bufs[0]), // sum cascadeKernel.args(ocl::KernelArg::ReadOnlyNoSize(bufs[0]), // sum
ocl::KernelArg::ReadOnly(bufs[1]), // sqsum ocl::KernelArg::ReadOnlyNoSize(bufs[1]), // sqsum
ocl::KernelArg::PtrReadOnly(bufs[2]), // optfeatures ocl::KernelArg::PtrReadOnly(bufs[2]), // optfeatures
// cascade classifier // cascade classifier
(int)data.stages.size(),
ocl::KernelArg::PtrReadOnly(ustages), ocl::KernelArg::PtrReadOnly(ustages),
ocl::KernelArg::PtrReadOnly(uclassifiers), ocl::KernelArg::PtrReadOnly(uclassifiers),
ocl::KernelArg::PtrReadOnly(unodes), ocl::KernelArg::PtrReadOnly(unodes),
ocl::KernelArg::PtrReadOnly(uleaves), ocl::KernelArg::PtrReadOnly(uleaves),
ocl::KernelArg::WriteOnly(ufacepos), // positions ocl::KernelArg::PtrWriteOnly(ufacepos), // positions
ocl::KernelArg::PtrReadOnly(uparams), processingRectSize,
processingRectSize.width, yStep, (float)factor,
processingRectSize.height, normrect, data.origWinSize);
yStep, (float)factor, MAX_FACES).run(2, globalsize, 0, false); bool ok = cascadeKernel.run(2, globalsize, 0, true);
//CV_Assert(ok);
return ok;
} }
bool CascadeClassifierImpl::isOldFormatCascade() const bool CascadeClassifierImpl::isOldFormatCascade() const
...@@ -1234,12 +1240,13 @@ void CascadeClassifierImpl::detectMultiScaleNoGrouping( InputArray _image, std:: ...@@ -1234,12 +1240,13 @@ void CascadeClassifierImpl::detectMultiScaleNoGrouping( InputArray _image, std::
if( maxObjectSize.height == 0 || maxObjectSize.width == 0 ) if( maxObjectSize.height == 0 || maxObjectSize.width == 0 )
maxObjectSize = imgsz; maxObjectSize = imgsz;
bool use_ocl = false;/*ocl::useOpenCL() && bool use_ocl = ocl::useOpenCL() &&
getFeatureType() == FeatureEvaluator::HAAR && getFeatureType() == FeatureEvaluator::HAAR &&
!isOldFormatCascade() && !isOldFormatCascade() &&
data.isStumpBased &&
maskGenerator.empty() && maskGenerator.empty() &&
!outputRejectLevels && !outputRejectLevels &&
tryOpenCL;*/ tryOpenCL;
if( !use_ocl ) if( !use_ocl )
{ {
...@@ -1268,13 +1275,20 @@ void CascadeClassifierImpl::detectMultiScaleNoGrouping( InputArray _image, std:: ...@@ -1268,13 +1275,20 @@ void CascadeClassifierImpl::detectMultiScaleNoGrouping( InputArray _image, std::
} }
Size sumSize0((imgsz.width + SUM_ALIGN) & -SUM_ALIGN, imgsz.height+1); Size sumSize0((imgsz.width + SUM_ALIGN) & -SUM_ALIGN, imgsz.height+1);
if( use_ocl )
{
ufacepos.create(1, MAX_FACES*4 + 1, CV_32S);
UMat ufacecount(ufacepos, Rect(0,0,1,1));
ufacecount.setTo(Scalar::all(0));
}
for( double factor = 1; ; factor *= scaleFactor ) for( double factor = 1; ; factor *= scaleFactor )
{ {
Size originalWindowSize = getOriginalWindowSize(); Size originalWindowSize = getOriginalWindowSize();
Size windowSize( cvRound(originalWindowSize.width*factor), cvRound(originalWindowSize.height*factor) ); Size windowSize( cvRound(originalWindowSize.width*factor), cvRound(originalWindowSize.height*factor) );
Size scaledImageSize( cvRound( grayImage.cols/factor ), cvRound( grayImage.rows/factor ) ); Size scaledImageSize( cvRound( imgsz.width/factor ), cvRound( imgsz.height/factor ) );
Size processingRectSize( scaledImageSize.width - originalWindowSize.width, Size processingRectSize( scaledImageSize.width - originalWindowSize.width,
scaledImageSize.height - originalWindowSize.height ); scaledImageSize.height - originalWindowSize.height );
...@@ -1331,6 +1345,7 @@ void CascadeClassifierImpl::detectMultiScaleNoGrouping( InputArray _image, std:: ...@@ -1331,6 +1345,7 @@ void CascadeClassifierImpl::detectMultiScaleNoGrouping( InputArray _image, std::
Mat facepos = ufacepos.getMat(ACCESS_READ); Mat facepos = ufacepos.getMat(ACCESS_READ);
const int* fptr = facepos.ptr<int>(); const int* fptr = facepos.ptr<int>();
int i, nfaces = fptr[0]; int i, nfaces = fptr[0];
printf("nfaces = %d\n", nfaces);
for( i = 0; i < nfaces; i++ ) for( i = 0; i < nfaces; i++ )
{ {
candidates.push_back(Rect(fptr[i*4+1], fptr[i*4+2], fptr[i*4+3], fptr[i*4+4])); candidates.push_back(Rect(fptr[i*4+1], fptr[i*4+2], fptr[i*4+3], fptr[i*4+4]));
...@@ -1439,8 +1454,6 @@ bool CascadeClassifierImpl::Data::read(const FileNode &root) ...@@ -1439,8 +1454,6 @@ bool CascadeClassifierImpl::Data::read(const FileNode &root)
origWinSize.height = (int)root[CC_HEIGHT]; origWinSize.height = (int)root[CC_HEIGHT];
CV_Assert( origWinSize.height > 0 && origWinSize.width > 0 ); CV_Assert( origWinSize.height > 0 && origWinSize.width > 0 );
isStumpBased = (int)(root[CC_STAGE_PARAMS][CC_MAX_DEPTH]) == 1 ? true : false;
// load feature params // load feature params
FileNode fn = root[CC_FEATURE_PARAMS]; FileNode fn = root[CC_FEATURE_PARAMS];
if( fn.empty() ) if( fn.empty() )
...@@ -1460,6 +1473,7 @@ bool CascadeClassifierImpl::Data::read(const FileNode &root) ...@@ -1460,6 +1473,7 @@ bool CascadeClassifierImpl::Data::read(const FileNode &root)
nodes.clear(); nodes.clear();
FileNodeIterator it = fn.begin(), it_end = fn.end(); FileNodeIterator it = fn.begin(), it_end = fn.end();
isStumpBased = true;
for( int si = 0; it != it_end; si++, ++it ) for( int si = 0; it != it_end; si++, ++it )
{ {
...@@ -1485,6 +1499,9 @@ bool CascadeClassifierImpl::Data::read(const FileNode &root) ...@@ -1485,6 +1499,9 @@ bool CascadeClassifierImpl::Data::read(const FileNode &root)
DTree tree; DTree tree;
tree.nodeCount = (int)internalNodes.size()/nodeStep; tree.nodeCount = (int)internalNodes.size()/nodeStep;
if( tree.nodeCount > 1 )
isStumpBased = false;
classifiers.push_back(tree); classifiers.push_back(tree);
nodes.reserve(nodes.size() + tree.nodeCount); nodes.reserve(nodes.size() + tree.nodeCount);
......
modules/objdetect/src/cascadedetect.hpp
View file @ 02fb3f0a
...@@ -63,8 +63,8 @@ protected: ...@@ -63,8 +63,8 @@ protected:
double scaleFactor, Size minObjectSize, Size maxObjectSize, double scaleFactor, Size minObjectSize, Size maxObjectSize,
bool outputRejectLevels = false ); bool outputRejectLevels = false );
enum { BOOST = 0 enum { MAX_FACES = 10000 };
}; enum { BOOST = 0 };
enum { DO_CANNY_PRUNING = CASCADE_DO_CANNY_PRUNING, enum { DO_CANNY_PRUNING = CASCADE_DO_CANNY_PRUNING,
SCALE_IMAGE = CASCADE_SCALE_IMAGE, SCALE_IMAGE = CASCADE_SCALE_IMAGE,
FIND_BIGGEST_OBJECT = CASCADE_FIND_BIGGEST_OBJECT, FIND_BIGGEST_OBJECT = CASCADE_FIND_BIGGEST_OBJECT,
...@@ -132,7 +132,7 @@ protected: ...@@ -132,7 +132,7 @@ protected:
Ptr<MaskGenerator> maskGenerator; Ptr<MaskGenerator> maskGenerator;
UMat ugrayImage, uimageBuffer; UMat ugrayImage, uimageBuffer;
UMat ufacepos, ustages, uclassifiers, unodes, uleaves, usubsets, uparams; UMat ufacepos, ustages, uclassifiers, unodes, uleaves, usubsets;
ocl::Kernel cascadeKernel; ocl::Kernel cascadeKernel;
bool tryOpenCL; bool tryOpenCL;
...@@ -327,19 +327,19 @@ inline void HaarEvaluator::OptFeature :: setOffsets( const Feature& _f, int step ...@@ -327,19 +327,19 @@ inline void HaarEvaluator::OptFeature :: setOffsets( const Feature& _f, int step
weight[0] = _f.rect[0].weight; weight[0] = _f.rect[0].weight;
weight[1] = _f.rect[1].weight; weight[1] = _f.rect[1].weight;
weight[2] = _f.rect[2].weight; weight[2] = _f.rect[2].weight;
Rect r2 = weight[2] > 0 ? _f.rect[2].r : Rect(0,0,0,0);
if (_f.tilted) if (_f.tilted)
{ {
CV_TILTED_OFS( ofs[0][0], ofs[0][1], ofs[0][2], ofs[0][3], tofs, _f.rect[0].r, step ); CV_TILTED_OFS( ofs[0][0], ofs[0][1], ofs[0][2], ofs[0][3], tofs, _f.rect[0].r, step );
CV_TILTED_OFS( ofs[1][0], ofs[1][1], ofs[1][2], ofs[1][3], tofs, _f.rect[1].r, step ); CV_TILTED_OFS( ofs[1][0], ofs[1][1], ofs[1][2], ofs[1][3], tofs, _f.rect[1].r, step );
if (weight[2]) CV_TILTED_PTRS( ofs[2][0], ofs[2][1], ofs[2][2], ofs[2][3], tofs, r2, step );
CV_TILTED_PTRS( ofs[2][0], ofs[2][1], ofs[2][2], ofs[2][3], tofs, _f.rect[2].r, step );
} }
else else
{ {
CV_SUM_OFS( ofs[0][0], ofs[0][1], ofs[0][2], ofs[0][3], 0, _f.rect[0].r, step ); CV_SUM_OFS( ofs[0][0], ofs[0][1], ofs[0][2], ofs[0][3], 0, _f.rect[0].r, step );
CV_SUM_OFS( ofs[1][0], ofs[1][1], ofs[1][2], ofs[1][3], 0, _f.rect[1].r, step ); CV_SUM_OFS( ofs[1][0], ofs[1][1], ofs[1][2], ofs[1][3], 0, _f.rect[1].r, step );
if (weight[2]) CV_SUM_OFS( ofs[2][0], ofs[2][1], ofs[2][2], ofs[2][3], 0, r2, step );
CV_SUM_OFS( ofs[2][0], ofs[2][1], ofs[2][2], ofs[2][3], 0, _f.rect[2].r, step );
} }
} }
......
modules/objdetect/src/opencl/haarobjectdetect.cl
View file @ 02fb3f0a
...@@ -12,6 +12,7 @@ ...@@ -12,6 +12,7 @@
// Nathan, liujun@multicorewareinc.com // Nathan, liujun@multicorewareinc.com
// Peng Xiao, pengxiao@outlook.com // Peng Xiao, pengxiao@outlook.com
// Erping Pang, erping@multicorewareinc.com // Erping Pang, erping@multicorewareinc.com
// Vadim Pisarevsky, vadim.pisarevsky@itseez.com
// Redistribution and use in source and binary forms, with or without modification, // Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met: // are permitted provided that the following conditions are met:
// //
...@@ -38,559 +39,117 @@ ...@@ -38,559 +39,117 @@
// //
// //
#define CV_HAAR_FEATURE_MAX 3 typedef struct __attribute__((aligned(4))) OptFeature
{
#define calc_sum(rect,offset) (sum[(rect).p0+offset] - sum[(rect).p1+offset] - sum[(rect).p2+offset] + sum[(rect).p3+offset]) int4 ofs[3] __attribute__((aligned (4)));
#define calc_sum1(rect,offset,i) (sum[(rect).p0[i]+offset] - sum[(rect).p1[i]+offset] - sum[(rect).p2[i]+offset] + sum[(rect).p3[i]+offset]) float4 weight __attribute__((aligned (4)));
}
typedef int sumtype; OptFeature;
typedef float sqsumtype;
#ifndef STUMP_BASED
#define STUMP_BASED 1
#endif
typedef struct __attribute__((aligned (128) )) GpuHidHaarTreeNode typedef struct __attribute__((aligned(4))) DTreeNode
{ {
int p[CV_HAAR_FEATURE_MAX][4] __attribute__((aligned (64))); int featureIdx __attribute__((aligned (4)));
float weight[CV_HAAR_FEATURE_MAX]; float threshold __attribute__((aligned (4))); // for ordered features only
float threshold;
float alpha[3] __attribute__((aligned (16)));
int left __attribute__((aligned (4))); int left __attribute__((aligned (4)));
int right __attribute__((aligned (4))); int right __attribute__((aligned (4)));
} }
GpuHidHaarTreeNode; DTreeNode;
//typedef struct __attribute__((aligned (32))) GpuHidHaarClassifier typedef struct __attribute__((aligned (4))) DTree
//{
// int count __attribute__((aligned (4)));
// GpuHidHaarTreeNode* node __attribute__((aligned (8)));
// float* alpha __attribute__((aligned (8)));
//}
//GpuHidHaarClassifier;
typedef struct __attribute__((aligned (64))) GpuHidHaarStageClassifier
{ {
int count __attribute__((aligned (4))); int nodeCount __attribute__((aligned (4)));
float threshold __attribute__((aligned (4)));
int two_rects __attribute__((aligned (4)));
int reserved0 __attribute__((aligned (8)));
int reserved1 __attribute__((aligned (8)));
int reserved2 __attribute__((aligned (8)));
int reserved3 __attribute__((aligned (8)));
} }
GpuHidHaarStageClassifier; DTree;
//typedef struct __attribute__((aligned (64))) GpuHidHaarClassifierCascade
//{
// int count __attribute__((aligned (4)));
// int is_stump_based __attribute__((aligned (4)));
// int has_tilted_features __attribute__((aligned (4)));
// int is_tree __attribute__((aligned (4)));
// int pq0 __attribute__((aligned (4)));
// int pq1 __attribute__((aligned (4)));
// int pq2 __attribute__((aligned (4)));
// int pq3 __attribute__((aligned (4)));
// int p0 __attribute__((aligned (4)));
// int p1 __attribute__((aligned (4)));
// int p2 __attribute__((aligned (4)));
// int p3 __attribute__((aligned (4)));
// float inv_window_area __attribute__((aligned (4)));
//} GpuHidHaarClassifierCascade;
#ifdef PACKED_CLASSIFIER
// this code is scalar, one pixel -> one workitem
__kernel void gpuRunHaarClassifierCascadePacked(
global const GpuHidHaarStageClassifier * stagecascadeptr,
global const int4 * info,
global const GpuHidHaarTreeNode * nodeptr,
global const int * restrict sum,
global const float * restrict sqsum,
volatile global int4 * candidate,
const int pixelstep,
const int loopcount,
const int start_stage,
const int split_stage,
const int end_stage,
const int startnode,
const int splitnode,
const int4 p,
const int4 pq,
const float correction,
global const int* pNodesPK,
global const int4* pWGInfo
)
typedef struct __attribute__((aligned (4))) Stage
{ {
// this version used information provided for each workgroup int first __attribute__((aligned (4)));
// no empty WG int ntrees __attribute__((aligned (4)));
int gid = (int)get_group_id(0); float threshold __attribute__((aligned (4)));
int lid_x = (int)get_local_id(0); }
int lid_y = (int)get_local_id(1); Stage;
int lid = lid_y*LSx+lid_x;
int4 WGInfo = pWGInfo[gid]; __kernel void runHaarClassifierStump(
int GroupX = (WGInfo.y >> 16)&0xFFFF; __global const int* sum,
int GroupY = (WGInfo.y >> 0 )& 0xFFFF; int sumstep, int sumoffset,
int Width = (WGInfo.x >> 16)&0xFFFF; __global const int* sqsum,
int Height = (WGInfo.x >> 0 )& 0xFFFF; int sqsumstep, int sqsumoffset,
int ImgOffset = WGInfo.z; __global const OptFeature* optfeatures,
float ScaleFactor = as_float(WGInfo.w);
int nstages,
#define DATA_SIZE_X (LSx+WND_SIZE_X) __global const Stage* stages,
#define DATA_SIZE_Y (LSy+WND_SIZE_Y) __global const DTree* trees,
#define DATA_SIZE (DATA_SIZE_X*DATA_SIZE_Y) __global const DTreeNode* nodes,
__global const float* leaves,
local int SumL[DATA_SIZE];
volatile __global int* facepos,
// read input data window into local mem int2 imgsize, int xyscale, float factor,
for(int i = 0; i<DATA_SIZE; i+=(LSx*LSy)) int4 normrect, int2 windowsize)
{
int index = i+lid; // index in shared local memory
if(index<DATA_SIZE)
{// calc global x,y coordinat and read data from there
int x = min(GroupX + (index % (DATA_SIZE_X)),Width-1);
int y = min(GroupY + (index / (DATA_SIZE_X)),Height-1);
SumL[index] = sum[ImgOffset+y*pixelstep+x];
}
}
barrier(CLK_LOCAL_MEM_FENCE);
// calc variance_norm_factor for all stages
float variance_norm_factor;
int nodecounter= startnode;
int4 info1 = p;
int4 info2 = pq;
{
int xl = lid_x;
int yl = lid_y;
int OffsetLocal = yl * DATA_SIZE_X + xl;
int OffsetGlobal = (GroupY+yl)* pixelstep + (GroupX+xl);
// add shift to get position on scaled image
OffsetGlobal += ImgOffset;
float mean =
SumL[info1.y*DATA_SIZE_X+info1.x+OffsetLocal] -
SumL[info1.y*DATA_SIZE_X+info1.z+OffsetLocal] -
SumL[info1.w*DATA_SIZE_X+info1.x+OffsetLocal] +
SumL[info1.w*DATA_SIZE_X+info1.z+OffsetLocal];
float sq =
sqsum[info2.y*pixelstep+info2.x+OffsetGlobal] -
sqsum[info2.y*pixelstep+info2.z+OffsetGlobal] -
sqsum[info2.w*pixelstep+info2.x+OffsetGlobal] +
sqsum[info2.w*pixelstep+info2.z+OffsetGlobal];
mean *= correction;
sq *= correction;
variance_norm_factor = sq - mean * mean;
variance_norm_factor = (variance_norm_factor >=0.f) ? sqrt(variance_norm_factor) : 1.f;
}// end calc variance_norm_factor for all stages
int result = (1.0f>0.0f);
for(int stageloop = start_stage; (stageloop < end_stage) && result; stageloop++ )
{// iterate until candidate is exist
float stage_sum = 0.0f;
__global GpuHidHaarStageClassifier* stageinfo = (__global GpuHidHaarStageClassifier*)
((__global uchar*)stagecascadeptr+stageloop*sizeof(GpuHidHaarStageClassifier));
int stagecount = stageinfo->count;
float stagethreshold = stageinfo->threshold;
int lcl_off = (lid_y*DATA_SIZE_X)+(lid_x);
for(int nodeloop = 0; nodeloop < stagecount; nodecounter++,nodeloop++ )
{
// simple macro to extract shorts from int
#define M0(_t) ((_t)&0xFFFF)
#define M1(_t) (((_t)>>16)&0xFFFF)
// load packed node data from global memory (L3) into registers
global const int4* pN = (__global int4*)(pNodesPK+nodecounter*NODE_SIZE);
int4 n0 = pN[0];
int4 n1 = pN[1];
int4 n2 = pN[2];
float nodethreshold = as_float(n2.y) * variance_norm_factor;
// calc sum of intensity pixels according to node information
float classsum =
(SumL[M0(n0.x)+lcl_off] - SumL[M1(n0.x)+lcl_off] - SumL[M0(n0.y)+lcl_off] + SumL[M1(n0.y)+lcl_off]) * as_float(n1.z) +
(SumL[M0(n0.z)+lcl_off] - SumL[M1(n0.z)+lcl_off] - SumL[M0(n0.w)+lcl_off] + SumL[M1(n0.w)+lcl_off]) * as_float(n1.w) +
(SumL[M0(n1.x)+lcl_off] - SumL[M1(n1.x)+lcl_off] - SumL[M0(n1.y)+lcl_off] + SumL[M1(n1.y)+lcl_off]) * as_float(n2.x);
//accumulate stage responce
stage_sum += (classsum >= nodethreshold) ? as_float(n2.w) : as_float(n2.z);
}
result = (stage_sum >= stagethreshold);
}// next stage if needed
if(result)
{// all stages will be passed and there is a detected face on the tested position
int index = 1+atomic_inc((volatile global int*)candidate); //get index to write global data with face info
if(index<OUTPUTSZ)
{
int x = GroupX+lid_x;
int y = GroupY+lid_y;
int4 candidate_result;
candidate_result.x = convert_int_rtn(x*ScaleFactor);
candidate_result.y = convert_int_rtn(y*ScaleFactor);
candidate_result.z = convert_int_rtn(ScaleFactor*WND_SIZE_X);
candidate_result.w = convert_int_rtn(ScaleFactor*WND_SIZE_Y);
candidate[index] = candidate_result;
}
}
}//end gpuRunHaarClassifierCascade
#else
__kernel void __attribute__((reqd_work_group_size(8,8,1)))gpuRunHaarClassifierCascade(
global GpuHidHaarStageClassifier * stagecascadeptr,
global int4 * info,
global GpuHidHaarTreeNode * nodeptr,
global const int * restrict sum1,
global const float * restrict sqsum1,
global int4 * candidate,
const int pixelstep,
const int loopcount,
const int start_stage,
const int split_stage,
const int end_stage,
const int startnode,
const int splitnode,
const int4 p,
const int4 pq,
const float correction)
{ {
int grpszx = get_local_size(0); int ix = get_global_id(0)*xyscale;
int grpszy = get_local_size(1); int iy = get_global_id(1)*xyscale;
int grpnumx = get_num_groups(0); sumstep /= sizeof(int);
int grpidx = get_group_id(0); sqsumstep /= sizeof(int);
int lclidx = get_local_id(0);
int lclidy = get_local_id(1); if( ix < imgsize.x && iy < imgsize.y )
int lcl_sz = mul24(grpszx,grpszy);
int lcl_id = mad24(lclidy,grpszx,lclidx);
__local int lclshare[1024];
__local int* lcldata = lclshare;//for save win data
__local int* glboutindex = lcldata + 28*28;//for save global out index
__local int* lclcount = glboutindex + 1;//for save the numuber of temp pass pixel
__local int* lcloutindex = lclcount + 1;//for save info of temp pass pixel
__local float* partialsum = (__local float*)(lcloutindex + (lcl_sz<<1));
glboutindex[0]=0;
int outputoff = mul24(grpidx,256);
//assume window size is 20X20
#define WINDOWSIZE 20+1
//make sure readwidth is the multiple of 4
//ystep =1, from host code
int readwidth = ((grpszx-1 + WINDOWSIZE+3)>>2)<<2;
int readheight = grpszy-1+WINDOWSIZE;
int read_horiz_cnt = readwidth >> 2;//each read int4
int total_read = mul24(read_horiz_cnt,readheight);
int read_loop = (total_read + lcl_sz - 1) >> 6;
candidate[outputoff+(lcl_id<<2)] = (int4)0;
candidate[outputoff+(lcl_id<<2)+1] = (int4)0;
candidate[outputoff+(lcl_id<<2)+2] = (int4)0;
candidate[outputoff+(lcl_id<<2)+3] = (int4)0;
for(int scalei = 0; scalei <loopcount; scalei++)
{ {
int4 scaleinfo1= info[scalei]; int ntrees, nodeOfs = 0, leafOfs = 0;
int height = scaleinfo1.x & 0xffff; int stageIdx, i;
int grpnumperline =(scaleinfo1.y & 0xffff0000) >> 16; float s = 0.f;
int totalgrp = scaleinfo1.y & 0xffff; __global const DTreeNode* node;
int imgoff = scaleinfo1.z; __global const OptFeature* f;
float factor = as_float(scaleinfo1.w);
__global const int* psum = sum + mad24(iy, sumstep, ix);
__global const int * sum = sum1 + imgoff; __global const int* pnsum = psum + mad24(normrect.y, sumstep, normrect.x);
__global const float * sqsum = sqsum1 + imgoff; int normarea = normrect.z * normrect.w;
for(int grploop=grpidx; grploop<totalgrp; grploop+=grpnumx) float invarea = 1.f/normarea;
float sval = (pnsum[0] - pnsum[normrect.z] - pnsum[mul24(normrect.w, sumstep)] +
pnsum[mad24(normrect.w, sumstep, normrect.z)])*invarea;
float sqval = (sqsum[mad24(iy + normrect.y, sqsumstep, ix + normrect.x)])*invarea;
float nf = (float)normarea * sqrt(max(sqval - sval * sval, 0.f));
float4 weight;
int4 ofs;
nf = nf > 0 ? nf : 1.f;
for( stageIdx = 0; stageIdx < nstages; stageIdx++ )
{ {
int grpidy = grploop / grpnumperline; ntrees = stages[stageIdx].ntrees;
int grpidx = grploop - mul24(grpidy, grpnumperline); s = 0.f;
int x = mad24(grpidx,grpszx,lclidx); for( i = 0; i < ntrees; i++, nodeOfs++, leafOfs += 2 )
int y = mad24(grpidy,grpszy,lclidy);
int grpoffx = x-lclidx;
int grpoffy = y-lclidy;
for(int i=0; i<read_loop; i++)
{ {
int pos_id = mad24(i,lcl_sz,lcl_id); node = nodes + nodeOfs;
pos_id = pos_id < total_read ? pos_id : 0; f = optfeatures + node->featureIdx;
int lcl_y = pos_id / read_horiz_cnt; weight = f->weight;
int lcl_x = pos_id - mul24(lcl_y, read_horiz_cnt);
ofs = f->ofs[0];
int glb_x = grpoffx + (lcl_x<<2); sval = (psum[ofs.x] - psum[ofs.y] - psum[ofs.z] + psum[ofs.w])*weight.x;
int glb_y = grpoffy + lcl_y; ofs = f->ofs[1];
sval += (psum[ofs.x] - psum[ofs.y] - psum[ofs.z] + psum[ofs.w])*weight.y;
int glb_off = mad24(min(glb_y, height + WINDOWSIZE - 1),pixelstep,glb_x); if( weight.z > 0 )
int4 data = *(__global int4*)&sum[glb_off];
int lcl_off = mad24(lcl_y, readwidth, lcl_x<<2);
vstore4(data, 0, &lcldata[lcl_off]);
}
lcloutindex[lcl_id] = 0;
lclcount[0] = 0;
int result = 1;
int nodecounter= startnode;
float mean, variance_norm_factor;
barrier(CLK_LOCAL_MEM_FENCE);
int lcl_off = mad24(lclidy,readwidth,lclidx);
int4 cascadeinfo1, cascadeinfo2;
cascadeinfo1 = p;
cascadeinfo2 = pq;
cascadeinfo1.x +=lcl_off;
cascadeinfo1.z +=lcl_off;
mean = (lcldata[mad24(cascadeinfo1.y,readwidth,cascadeinfo1.x)] - lcldata[mad24(cascadeinfo1.y,readwidth,cascadeinfo1.z)] -
lcldata[mad24(cascadeinfo1.w,readwidth,cascadeinfo1.x)] + lcldata[mad24(cascadeinfo1.w,readwidth,cascadeinfo1.z)])
*correction;
int p_offset = mad24(y, pixelstep, x);
cascadeinfo2.x +=p_offset;
cascadeinfo2.z +=p_offset;
variance_norm_factor =sqsum[mad24(cascadeinfo2.y, pixelstep, cascadeinfo2.x)] - sqsum[mad24(cascadeinfo2.y, pixelstep, cascadeinfo2.z)] -
sqsum[mad24(cascadeinfo2.w, pixelstep, cascadeinfo2.x)] + sqsum[mad24(cascadeinfo2.w, pixelstep, cascadeinfo2.z)];
variance_norm_factor = variance_norm_factor * correction - mean * mean;
variance_norm_factor = variance_norm_factor >=0.f ? sqrt(variance_norm_factor) : 1.f;
for(int stageloop = start_stage; (stageloop < split_stage) && result; stageloop++ )
{
float stage_sum = 0.f;
__global GpuHidHaarStageClassifier* stageinfo = (__global GpuHidHaarStageClassifier*)
((__global uchar*)stagecascadeptr+stageloop*sizeof(GpuHidHaarStageClassifier));
int stagecount = stageinfo->count;
float stagethreshold = stageinfo->threshold;
for(int nodeloop = 0; nodeloop < stagecount; )
{
__global GpuHidHaarTreeNode* currentnodeptr = (__global GpuHidHaarTreeNode*)
(((__global uchar*)nodeptr) + nodecounter * sizeof(GpuHidHaarTreeNode));
int4 info1 = *(__global int4*)(&(currentnodeptr->p[0][0]));
int4 info2 = *(__global int4*)(&(currentnodeptr->p[1][0]));
int4 info3 = *(__global int4*)(&(currentnodeptr->p[2][0]));
float4 w = *(__global float4*)(&(currentnodeptr->weight[0]));
float3 alpha3 = *(__global float3*)(&(currentnodeptr->alpha[0]));
float nodethreshold = w.w * variance_norm_factor;
info1.x +=lcl_off;
info1.z +=lcl_off;
info2.x +=lcl_off;
info2.z +=lcl_off;
float classsum = (lcldata[mad24(info1.y,readwidth,info1.x)] - lcldata[mad24(info1.y,readwidth,info1.z)] -
lcldata[mad24(info1.w,readwidth,info1.x)] + lcldata[mad24(info1.w,readwidth,info1.z)]) * w.x;
classsum += (lcldata[mad24(info2.y,readwidth,info2.x)] - lcldata[mad24(info2.y,readwidth,info2.z)] -
lcldata[mad24(info2.w,readwidth,info2.x)] + lcldata[mad24(info2.w,readwidth,info2.z)]) * w.y;
info3.x +=lcl_off;
info3.z +=lcl_off;
classsum += (lcldata[mad24(info3.y,readwidth,info3.x)] - lcldata[mad24(info3.y,readwidth,info3.z)] -
lcldata[mad24(info3.w,readwidth,info3.x)] + lcldata[mad24(info3.w,readwidth,info3.z)]) * w.z;
bool passThres = classsum >= nodethreshold;
#if STUMP_BASED
stage_sum += passThres ? alpha3.y : alpha3.x;
nodecounter++;
nodeloop++;
#else
bool isRootNode = (nodecounter & 1) == 0;
if(isRootNode)
{
if( (passThres && currentnodeptr->right) ||
(!passThres && currentnodeptr->left))
{
nodecounter ++;
}
else
{
stage_sum += alpha3.x;
nodecounter += 2;
nodeloop ++;
}
}
else
{
stage_sum += passThres ? alpha3.z : alpha3.y;
nodecounter ++;
nodeloop ++;
}
#endif
}
result = (stage_sum >= stagethreshold) ? 1 : 0;
}
if(factor < 2)
{
if(result && lclidx %2 ==0 && lclidy %2 ==0 )
{
int queueindex = atomic_inc(lclcount);
lcloutindex[queueindex<<1] = (lclidy << 16) | lclidx;
lcloutindex[(queueindex<<1)+1] = as_int((float)variance_norm_factor);
}
}
else
{
if(result)
{ {
int queueindex = atomic_inc(lclcount); ofs = f->ofs[2];
lcloutindex[queueindex<<1] = (lclidy << 16) | lclidx; sval += (psum[ofs.x] - psum[ofs.y] - psum[ofs.z] + psum[ofs.w])*weight.z;
lcloutindex[(queueindex<<1)+1] = as_int((float)variance_norm_factor);
} }
s += leaves[ sval < node->threshold*nf ? leafOfs : leafOfs + 1 ];
} }
barrier(CLK_LOCAL_MEM_FENCE);
int queuecount = lclcount[0]; if( s < stages[stageIdx].threshold )
barrier(CLK_LOCAL_MEM_FENCE); break;
nodecounter = splitnode; }
for(int stageloop = split_stage; stageloop< end_stage && queuecount>0; stageloop++)
{ if( stageIdx == nstages )
lclcount[0]=0; {
barrier(CLK_LOCAL_MEM_FENCE); int nfaces = atomic_inc(facepos);
//printf("detected face #d!!!!\n", nfaces);
//int2 stageinfo = *(global int2*)(stagecascadeptr+stageloop); if( nfaces < MAX_FACES )
__global GpuHidHaarStageClassifier* stageinfo = (__global GpuHidHaarStageClassifier*)
((__global uchar*)stagecascadeptr+stageloop*sizeof(GpuHidHaarStageClassifier));
int stagecount = stageinfo->count;
float stagethreshold = stageinfo->threshold;
int perfscale = queuecount > 4 ? 3 : 2;
int queuecount_loop = (queuecount + (1<<perfscale)-1) >> perfscale;
int lcl_compute_win = lcl_sz >> perfscale;
int lcl_compute_win_id = (lcl_id >>(6-perfscale));
int lcl_loops = (stagecount + lcl_compute_win -1) >> (6-perfscale);
int lcl_compute_id = lcl_id - (lcl_compute_win_id << (6-perfscale));
for(int queueloop=0; queueloop<queuecount_loop; queueloop++)
{
float stage_sum = 0.f;
int temp_coord = lcloutindex[lcl_compute_win_id<<1];
float variance_norm_factor = as_float(lcloutindex[(lcl_compute_win_id<<1)+1]);
int queue_pixel = mad24(((temp_coord & (int)0xffff0000)>>16),readwidth,temp_coord & 0xffff);
if(lcl_compute_win_id < queuecount)
{
int tempnodecounter = lcl_compute_id;
float part_sum = 0.f;
const int stump_factor = STUMP_BASED ? 1 : 2;
int root_offset = 0;
for(int lcl_loop=0; lcl_loop<lcl_loops && tempnodecounter<stagecount;)
{
__global GpuHidHaarTreeNode* currentnodeptr = (__global GpuHidHaarTreeNode*)
(((__global uchar*)nodeptr) + sizeof(GpuHidHaarTreeNode) * ((nodecounter + tempnodecounter) * stump_factor + root_offset));
int4 info1 = *(__global int4*)(&(currentnodeptr->p[0][0]));
int4 info2 = *(__global int4*)(&(currentnodeptr->p[1][0]));
int4 info3 = *(__global int4*)(&(currentnodeptr->p[2][0]));
float4 w = *(__global float4*)(&(currentnodeptr->weight[0]));
float3 alpha3 = *(__global float3*)(&(currentnodeptr->alpha[0]));
float nodethreshold = w.w * variance_norm_factor;
info1.x +=queue_pixel;
info1.z +=queue_pixel;
info2.x +=queue_pixel;
info2.z +=queue_pixel;
float classsum = (lcldata[mad24(info1.y,readwidth,info1.x)] - lcldata[mad24(info1.y,readwidth,info1.z)] -
lcldata[mad24(info1.w,readwidth,info1.x)] + lcldata[mad24(info1.w,readwidth,info1.z)]) * w.x;
classsum += (lcldata[mad24(info2.y,readwidth,info2.x)] - lcldata[mad24(info2.y,readwidth,info2.z)] -
lcldata[mad24(info2.w,readwidth,info2.x)] + lcldata[mad24(info2.w,readwidth,info2.z)]) * w.y;
info3.x +=queue_pixel;
info3.z +=queue_pixel;
classsum += (lcldata[mad24(info3.y,readwidth,info3.x)] - lcldata[mad24(info3.y,readwidth,info3.z)] -
lcldata[mad24(info3.w,readwidth,info3.x)] + lcldata[mad24(info3.w,readwidth,info3.z)]) * w.z;
bool passThres = classsum >= nodethreshold;
#if STUMP_BASED
part_sum += passThres ? alpha3.y : alpha3.x;
tempnodecounter += lcl_compute_win;
lcl_loop++;
#else
if(root_offset == 0)
{
if( (passThres && currentnodeptr->right) ||
(!passThres && currentnodeptr->left))
{
root_offset = 1;
}
else
{
part_sum += alpha3.x;
tempnodecounter += lcl_compute_win;
lcl_loop++;
}
}
else
{
part_sum += passThres ? alpha3.z : alpha3.y;
tempnodecounter += lcl_compute_win;
lcl_loop++;
root_offset = 0;
}
#endif
}//end for(int lcl_loop=0;lcl_loop<lcl_loops;lcl_loop++)
partialsum[lcl_id]=part_sum;
}
barrier(CLK_LOCAL_MEM_FENCE);
if(lcl_compute_win_id < queuecount)
{
for(int i=0; i<lcl_compute_win && (lcl_compute_id==0); i++)
{
stage_sum += partialsum[lcl_id+i];
}
if(stage_sum >= stagethreshold && (lcl_compute_id==0))
{
int queueindex = atomic_inc(lclcount);
lcloutindex[queueindex<<1] = temp_coord;
lcloutindex[(queueindex<<1)+1] = as_int(variance_norm_factor);
}
lcl_compute_win_id +=(1<<perfscale);
}
barrier(CLK_LOCAL_MEM_FENCE);
}//end for(int queueloop=0;queueloop<queuecount_loop;queueloop++)
queuecount = lclcount[0];
barrier(CLK_LOCAL_MEM_FENCE);
nodecounter += stagecount;
}//end for(int stageloop = splitstage; stageloop< endstage && queuecount>0;stageloop++)
if(lcl_id<queuecount)
{ {
int temp = lcloutindex[lcl_id<<1]; volatile __global int* face = facepos + 1 + nfaces*4;
int x = mad24(grpidx,grpszx,temp & 0xffff); face[0] = convert_int_rte(ix*factor);
int y = mad24(grpidy,grpszy,((temp & (int)0xffff0000) >> 16)); face[1] = convert_int_rte(iy*factor);
temp = glboutindex[0]; face[2] = convert_int_rte(windowsize.x*factor);
int4 candidate_result; face[3] = convert_int_rte(windowsize.y*factor);
candidate_result.zw = (int2)convert_int_rte(factor*20.f);
candidate_result.x = convert_int_rte(x*factor);
candidate_result.y = convert_int_rte(y*factor);
atomic_inc(glboutindex);
int i = outputoff+temp+lcl_id;
if(candidate[i].z == 0)
{
candidate[i] = candidate_result;
}
else
{
for(i=i+1;;i++)
{
if(candidate[i].z == 0)
{
candidate[i] = candidate_result;
break;
}
}
}
} }
barrier(CLK_LOCAL_MEM_FENCE); }
}//end for(int grploop=grpidx;grploop<totalgrp;grploop+=grpnumx) }
}//end for(int scalei = 0; scalei <loopcount; scalei++)
} }
#endif
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment