generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

classifier loading has been improved

Browse Source
This commit is contained in:
Marina Kolpakova 2012-07-02 08:07:46 +00:00
parent 83ccdbd0ce
commit e7f5978768
1 changed files with 24 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

61
modules/gpu/src/cascadeclassifier.cpp
View File

@ -53,7 +53,6 @@ struct Stage
int first; int first;
int ntrees; int ntrees;
float threshold; float threshold;
Stage(int f = 0, int n = 0, float t = 0.f) : first(f), ntrees(n), threshold(t) {}
}; };
struct DTreeNode struct DTreeNode
@ -61,7 +60,6 @@ struct DTreeNode
int featureIdx; int featureIdx;
int left; int left;
int right; int right;
DTreeNode(int f = 0, int l = 0, int r = 0) : featureIdx(f), left(l), right(r) {}
}; };
#if !defined (HAVE_CUDA) #if !defined (HAVE_CUDA)
@ -70,9 +68,9 @@ cv::gpu::CascadeClassifier_GPU::CascadeClassifier_GPU() { throw_no
cv::gpu::CascadeClassifier_GPU::CascadeClassifier_GPU(const string&) { throw_nogpu(); } cv::gpu::CascadeClassifier_GPU::CascadeClassifier_GPU(const string&) { throw_nogpu(); }
cv::gpu::CascadeClassifier_GPU::~CascadeClassifier_GPU() { throw_nogpu(); } cv::gpu::CascadeClassifier_GPU::~CascadeClassifier_GPU() { throw_nogpu(); }
bool cv::gpu::CascadeClassifier_GPU::empty() const { throw_nogpu(); return true; } bool cv::gpu::CascadeClassifier_GPU::empty() const { throw_nogpu(); return true; }
bool cv::gpu::CascadeClassifier_GPU::load(const string&) { throw_nogpu(); return true; } bool cv::gpu::CascadeClassifier_GPU::load(const string&) { throw_nogpu(); return true; }
Size cv::gpu::CascadeClassifier_GPU::getClassifierSize() const { throw_nogpu(); return Size(); } Size cv::gpu::CascadeClassifier_GPU::getClassifierSize() const { throw_nogpu(); return Size(); }
int cv::gpu::CascadeClassifier_GPU::detectMultiScale( const GpuMat& , GpuMat& , double , int , Size) { throw_nogpu(); return 0; } int cv::gpu::CascadeClassifier_GPU::detectMultiScale( const GpuMat& , GpuMat& , double , int , Size) { throw_nogpu(); return 0; }
@ -89,13 +87,9 @@ int cv::gpu::CascadeClassifier_GPU_LBP::detectMultiScale(const cv::gpu::GpuMat&,
#else #else
cv::gpu::CascadeClassifier_GPU_LBP::CascadeClassifier_GPU_LBP() cv::gpu::CascadeClassifier_GPU_LBP::CascadeClassifier_GPU_LBP(){}
{
}
cv::gpu::CascadeClassifier_GPU_LBP::~CascadeClassifier_GPU_LBP() cv::gpu::CascadeClassifier_GPU_LBP::~CascadeClassifier_GPU_LBP(){}
{
}
void cv::gpu::CascadeClassifier_GPU_LBP::preallocateIntegralBuffer(cv::Size desired) void cv::gpu::CascadeClassifier_GPU_LBP::preallocateIntegralBuffer(cv::Size desired)
{ {
@ -134,6 +128,7 @@ bool cv::gpu::CascadeClassifier_GPU_LBP::load(const string& classifierAsXml)
#define GPU_CC_FEATURES "features" #define GPU_CC_FEATURES "features"
#define GPU_CC_RECT "rect" #define GPU_CC_RECT "rect"
// currently only stump based boost classifiers are supported
bool CascadeClassifier_GPU_LBP::read(const FileNode &root) bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
{ {
std::string stageTypeStr = (string)root[GPU_CC_STAGE_TYPE]; std::string stageTypeStr = (string)root[GPU_CC_STAGE_TYPE];
@ -149,14 +144,14 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
isStumps = ((int)(root[GPU_CC_STAGE_PARAMS][GPU_CC_MAX_DEPTH]) == 1) ? true : false; isStumps = ((int)(root[GPU_CC_STAGE_PARAMS][GPU_CC_MAX_DEPTH]) == 1) ? true : false;
CV_Assert(isStumps); CV_Assert(isStumps);
// features
FileNode fn = root[GPU_CC_FEATURE_PARAMS]; FileNode fn = root[GPU_CC_FEATURE_PARAMS];
if (fn.empty()) if (fn.empty())
return false; return false;
ncategories = fn[GPU_CC_MAX_CAT_COUNT]; ncategories = fn[GPU_CC_MAX_CAT_COUNT];
subsetSize = (ncategories + 31) / 32, nodeStep = 3 + ( ncategories > 0 ? subsetSize : 1 ); subsetSize = (ncategories + 31) / 32;
nodeStep = 3 + ( ncategories > 0 ? subsetSize : 1 );
fn = root[GPU_CC_STAGES]; fn = root[GPU_CC_STAGES];
if (fn.empty()) if (fn.empty())
@ -166,7 +161,7 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
stages.reserve(fn.size()); stages.reserve(fn.size());
std::vector<int> cl_trees; std::vector<int> cl_trees;
std::vector<DTreeNode> cl_nodes; std::vector<int> cl_nodes;
std::vector<float> cl_leaves; std::vector<float> cl_leaves;
std::vector<int> subsets; std::vector<int> subsets;
@ -184,7 +179,7 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
st.ntrees = (int)fns.size(); st.ntrees = (int)fns.size();
st.first = (int)cl_trees.size(); st.first = (int)cl_trees.size();
stages.push_back(st); stages.push_back(st);// (int, int, float)
cl_trees.reserve(stages[si].first + stages[si].ntrees); cl_trees.reserve(stages[si].first + stages[si].ntrees);
@ -198,10 +193,11 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
FileNode leafValues = fnw[GPU_CC_LEAF_VALUES]; FileNode leafValues = fnw[GPU_CC_LEAF_VALUES];
if ( internalNodes.empty() || leafValues.empty() ) if ( internalNodes.empty() || leafValues.empty() )
return false; return false;
int nodeCount = (int)internalNodes.size()/nodeStep; int nodeCount = (int)internalNodes.size()/nodeStep;
cl_trees.push_back(nodeCount); cl_trees.push_back(nodeCount);
cl_nodes.reserve(cl_nodes.size() + nodeCount); cl_nodes.reserve((cl_nodes.size() + nodeCount) * 3);
cl_leaves.reserve(cl_leaves.size() + leafValues.size()); cl_leaves.reserve(cl_leaves.size() + leafValues.size());
if( subsetSize > 0 ) if( subsetSize > 0 )
@ -212,8 +208,9 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
for( ; iIt != iEnd; ) for( ; iIt != iEnd; )
{ {
DTreeNode node((int)*(iIt++), (int)*(iIt++), (int)*(iIt++)); cl_nodes.push_back((int)*(iIt++));
cl_nodes.push_back(node); cl_nodes.push_back((int)*(iIt++));
cl_nodes.push_back((int)*(iIt++));
if( subsetSize > 0 ) if( subsetSize > 0 )
for( int j = 0; j < subsetSize; j++, ++iIt ) for( int j = 0; j < subsetSize; j++, ++iIt )
@ -226,15 +223,16 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
cl_leaves.push_back((float)*iIt); cl_leaves.push_back((float)*iIt);
} }
} }
fn = root[GPU_CC_FEATURES]; fn = root[GPU_CC_FEATURES];
if( fn.empty() ) if( fn.empty() )
return false; return false;
std::vector<char> features; std::vector<uchar> features;
features.reserve(fn.size() * 4); features.reserve(fn.size() * 4);
FileNodeIterator f_it = fn.begin(), f_end = fn.end(); FileNodeIterator f_it = fn.begin(), f_end = fn.end();
for (; f_it != f_end; ++f_it) for (; f_it != f_end; ++f_it)
{ {
FileNode rect = (*it)[GPU_CC_RECT]; FileNode rect = (*f_it)[GPU_CC_RECT];
FileNodeIterator r_it = rect.begin(); FileNodeIterator r_it = rect.begin();
features.push_back(saturate_cast<uchar>((int)*(r_it++))); features.push_back(saturate_cast<uchar>((int)*(r_it++)));
features.push_back(saturate_cast<uchar>((int)*(r_it++))); features.push_back(saturate_cast<uchar>((int)*(r_it++)));
@ -243,23 +241,12 @@ bool CascadeClassifier_GPU_LBP::read(const FileNode &root)
} }
// copy data structures on gpu // copy data structures on gpu
stage_mat = cv::gpu::GpuMat(1, (int)stages.size() * sizeof(Stage), CV_8UC1); stage_mat.upload(cv::Mat(1, stages.size() * sizeof(Stage), CV_8UC1, (uchar*)&(stages[0]) ));
stage_mat.upload(cv::Mat(1, stages.size() * sizeof(Stage), CV_8UC1, &(stages[0]) )); trees_mat.upload(cv::Mat(cl_trees).reshape(1,1));
nodes_mat.upload(cv::Mat(cl_nodes).reshape(1,1));
trees_mat = cv::gpu::GpuMat(1, (int)cl_trees.size(), CV_32SC1); leaves_mat.upload(cv::Mat(cl_leaves).reshape(1,1));
stage_mat.upload(cv::Mat(cl_trees)); subsets_mat.upload(cv::Mat(subsets).reshape(1,1));
features_mat.upload(cv::Mat(features).reshape(4,1));
nodes_mat = cv::gpu::GpuMat(1, (int)cl_nodes.size() * sizeof(DTreeNode), CV_8UC1);
stage_mat.upload(cv::Mat(1, cl_nodes.size() * sizeof(DTreeNode), CV_8UC1, &(cl_nodes[0]) ));
leaves_mat = cv::gpu::GpuMat(1, (int)cl_leaves.size(), CV_32FC1);
stage_mat.upload(cv::Mat(cl_leaves));
subsets_mat = cv::gpu::GpuMat(1, (int)subsets.size(), CV_32SC1);
stage_mat.upload(cv::Mat(subsets));
features_mat = cv::gpu::GpuMat(1, (int)features.size(), CV_8UC1);
features_mat.upload(cv::Mat(features));
return true; return true;
} }