void BaseQualityEvaluator::readAllDatasetsRunParams()
{
string filename = getRunParamsFilename();
FileStorage fs( filename, FileStorage::READ );
if( !fs.isOpened() )
{
isWriteParams = true;
setDefaultAllDatasetsRunParams();
printf("All runParams are default.\n");
}
else
{
isWriteParams = false;
FileNode topfn = fs.getFirstTopLevelNode();
FileNode fn = topfn[DEFAULT_PARAMS];
readDefaultRunParams(fn);
for( int i = 0; i < DATASETS_COUNT; i++ )
{
FileNode fn = topfn[DATASET_NAMES[i]];
if( fn.empty() )
{
printf( "%d-runParams is default.\n", i);
setDefaultDatasetRunParams(i);
}
else
readDatasetRunParams(fn, i);
}
}
}
inline void read(const FileNode& node, params &x, const params & default_value = params())
{
if(node.empty())
x = default_value;
else
x.read(node);
}
void CameraCalibration::read( const FileNode& node, Settings& x, const Settings& default_value )
{
if ( node.empty() )
x = default_value;
else {
x.read( node );
}
}
void read(const FileNode& node, ICFDetector& d,
const ICFDetector& default_value)
{
if( node.empty() )
d = default_value;
else
d.read(node);
}
void DetectorQualityEvaluator::readDefaultRunParams (FileNode &fn)
{
if (! fn.empty() )
{
isSaveKeypointsDefault = (int)fn[IS_SAVE_KEYPOINTS] != 0;
defaultDetector->read (fn);
}
}
/// static function that reads the settings file
/// @param[in] node FileNode to read from
/// @param[out] Settings Settings to be saved from FileNode
/// @param[in] default_value value by default if FileNode is empty
static void read(const FileNode& node, Settings& x, const Settings& default_value = Settings())
{
if (node.empty()){
x = default_value;
}
else{
x.read(node);
}
}
void DescriptorQualityEvaluator::readDefaultRunParams (FileNode &fn)
{
if (! fn.empty() )
{
commRunParamsDefault.projectKeypointsFrom1Image = (int)fn[PROJECT_KEYPOINTS_FROM_1IMAGE] != 0;
commRunParamsDefault.matchFilter = (int)fn[MATCH_FILTER];
defaultDescMatcher->read (fn);
}
}
bool CvFeatureParams::read( const FileNode &node )
{
if( node.empty() )
return false;
maxCatCount = node[CC_MAX_CAT_COUNT];
featSize = node[CC_FEATURE_SIZE];
numFeatures = node[CC_NUM_FEATURES];
return ( maxCatCount >= 0 && featSize >= 1 );
}
/// static function that reads the results file
/// @param[in] node FileNode to read from
/// @param[out] Results result to be saved from FileNode
/// @param[in] default_value value by default if FileNode is empty
static void read(const FileNode& node, Results& result,const Results& default_value = Results())
{
if (node.empty())
{
result = default_value;
} else
{
result.read(node);
}
}
void PCAread(const FileNode& fs, PCA& pca)
{
CV_Assert( !fs.empty() );
String name = (String)fs["name"];
CV_Assert( name == "PCA" );
cv::read(fs["vectors"], pca.eigenvectors);
cv::read(fs["values"], pca.eigenvalues);
cv::read(fs["mean"], pca.mean);
}
void BaseHoughLineTest::run_test(int type)
{
string filename = cvtest::TS::ptr()->get_data_path() + picture_name;
Mat src = imread(filename, IMREAD_GRAYSCALE);
EXPECT_FALSE(src.empty()) << "Invalid test image: " << filename;
string xml;
if (type == STANDART)
xml = string(cvtest::TS::ptr()->get_data_path()) + "imgproc/HoughLines.xml";
else if (type == PROBABILISTIC)
xml = string(cvtest::TS::ptr()->get_data_path()) + "imgproc/HoughLinesP.xml";
Mat dst;
Canny(src, dst, 100, 150, 3);
EXPECT_FALSE(dst.empty()) << "Failed Canny edge detector";
Mat lines;
if (type == STANDART)
HoughLines(dst, lines, rhoStep, thetaStep, threshold, 0, 0);
else if (type == PROBABILISTIC)
HoughLinesP(dst, lines, rhoStep, thetaStep, threshold, minLineLength, maxGap);
String test_case_name = format("lines_%s_%.0f_%.2f_%d_%d_%d", picture_name.c_str(), rhoStep, thetaStep,
threshold, minLineLength, maxGap);
test_case_name = getTestCaseName(test_case_name);
FileStorage fs(xml, FileStorage::READ);
FileNode node = fs[test_case_name];
if (node.empty())
{
fs.release();
fs.open(xml, FileStorage::APPEND);
EXPECT_TRUE(fs.isOpened()) << "Cannot open sanity data file: " << xml;
fs << test_case_name << lines;
fs.release();
fs.open(xml, FileStorage::READ);
EXPECT_TRUE(fs.isOpened()) << "Cannot open sanity data file: " << xml;
}
Mat exp_lines;
read( fs[test_case_name], exp_lines, Mat() );
fs.release();
int count = -1;
if (type == STANDART)
count = countMatIntersection<Vec2f>(exp_lines, lines, (float) thetaStep + FLT_EPSILON, (float) rhoStep + FLT_EPSILON);
else if (type == PROBABILISTIC)
count = countMatIntersection<Vec4i>(exp_lines, lines, 1e-4f, 0.f);
#if defined HAVE_IPP && IPP_VERSION_X100 >= 810 && IPP_DISABLE_BLOCK
EXPECT_GE( count, (int) (exp_lines.total() * 0.8) );
#else
EXPECT_EQ( count, (int) exp_lines.total());
#endif
}
void read ( const FileNode& node, MyData& x, const MyData& default_value = MyData ( ) )
{
if ( node.empty ( ) )
{
x = default_value;
}
else
{
x.read ( node );
}
}
bool CascadeClassifier::read(const FileNode& root)
{
if( !data.read(root) )
return false;
// load features
featureEvaluator = FeatureEvaluator::create(data.featureType);
FileNode fn = root[CC_FEATURES];
if( fn.empty() )
return false;
return featureEvaluator->read(fn);
}
bool CvCascadeClassifier::readStages( const FileNode &node)
{
FileNode rnode = node[CC_STAGES];
if (!rnode.empty() || !rnode.isSeq())
return false;
stageClassifiers.reserve(numStages);
FileNodeIterator it = rnode.begin();
for( int i = 0; i < min( (int)rnode.size(), numStages ); i++, it++ )
{
Ptr<CvCascadeBoost> tempStage = makePtr<CvCascadeBoost>();
if ( !tempStage->read( *it, featureEvaluator, *stageParams) )
return false;
stageClassifiers.push_back(tempStage);
}
return true;
}
void readVectorOfVector(FileStorage &fns, string name, vector<vector<KeyPoint> > &vov)
{
vov.clear();
FileNode fn = fns[name];
if (fn.empty()){
return;
}
FileNodeIterator current = fn.begin(), it_end = fn.end(); // Go through the node
for (; current != it_end; ++current)
{
vector<KeyPoint> tmp;
FileNode item = *current;
read(item, tmp);
vov.push_back(tmp);
}
}
bool f_lcc::load_rad_data()
{
char buf[1024];
snprintf(buf, 1024, "%s.yml", m_fmap);
int size;
FileStorage fs(buf, FileStorage::READ);
if(!fs.isOpened()){
cerr << "Failed to open " << buf << endl;
return false;
}else{
FileNode fn = fs["LCCR"];
if(!fn.empty()){
fn >> size;
}
fn = fs["LCCAMap"];
if(!fn.empty()){
fn >> m_amap;
}
bool CvCascadeClassifier::readStages( const FileNode &node)
{
FileNode rnode = node[CC_STAGES];
if (!rnode.empty() || !rnode.isSeq())
return false;
stageClassifiers.reserve(numStages);
FileNodeIterator it = rnode.begin();
for( int i = 0; i < min( (int)rnode.size(), numStages ); i++, it++ )
{
CvCascadeBoost* tempStage = new CvCascadeBoost;
if ( !tempStage->read( *it, (CvFeatureEvaluator *)featureEvaluator, *((CvCascadeBoostParams*)stageParams) ) )
{
delete tempStage;
return false;
}
stageClassifiers.push_back(tempStage);
}
return true;
}
int CV_MLBaseTest::prepare_test_case( int test_case_idx )
{
int trainSampleCount, respIdx;
string varTypes;
clear();
string dataPath = ts->get_data_path();
if ( dataPath.empty() )
{
ts->printf( cvtest::TS::LOG, "data path is empty" );
return cvtest::TS::FAIL_INVALID_TEST_DATA;
}
string dataName = dataSetNames[test_case_idx],
filename = dataPath + dataName + ".data";
if ( data.read_csv( filename.c_str() ) != 0)
{
char msg[100];
sprintf( msg, "file %s can not be read", filename.c_str() );
ts->printf( cvtest::TS::LOG, msg );
return cvtest::TS::FAIL_INVALID_TEST_DATA;
}
FileNode dataParamsNode = validationFS.getFirstTopLevelNode()["validation"][modelName][dataName]["data_params"];
CV_DbgAssert( !dataParamsNode.empty() );
CV_DbgAssert( !dataParamsNode["LS"].empty() );
dataParamsNode["LS"] >> trainSampleCount;
CvTrainTestSplit spl( trainSampleCount );
data.set_train_test_split( &spl );
CV_DbgAssert( !dataParamsNode["resp_idx"].empty() );
dataParamsNode["resp_idx"] >> respIdx;
data.set_response_idx( respIdx );
CV_DbgAssert( !dataParamsNode["types"].empty() );
dataParamsNode["types"] >> varTypes;
data.set_var_types( varTypes.c_str() );
return cvtest::TS::OK;
}
void init_nn( NeuralNet_MLP& mlp, const string& data )
{
assert( mlp.get_layer_count() == 0 );
try
{
FileStorage fs( data, cv::FileStorage::READ | cv::FileStorage::MEMORY );
FileNode fn = fs[ "mlp" ];
if ( !fn.empty() )
{
mlp.read( *fs, *fn );
}
else
{
throw runtime_error( "invalid data in file" );
}
}
catch ( const exception& e )
{
throw runtime_error( string( "Failed load train neural network state, reason: " ) + e.what() );
}
}
void SVMSGDImpl::readParams( const FileNode& fn )
{
String svmsgdTypeStr = (String)fn["svmsgdType"];
int svmsgdType =
svmsgdTypeStr == "SGD" ? SGD :
svmsgdTypeStr == "ASGD" ? ASGD : -1;
if( svmsgdType < 0 )
CV_Error( CV_StsParseError, "Missing or invalid SVMSGD type" );
params.svmsgdType = svmsgdType;
String marginTypeStr = (String)fn["marginType"];
int marginType =
marginTypeStr == "SOFT_MARGIN" ? SOFT_MARGIN :
marginTypeStr == "HARD_MARGIN" ? HARD_MARGIN : -1;
if( marginType < 0 )
CV_Error( CV_StsParseError, "Missing or invalid margin type" );
params.marginType = marginType;
CV_Assert ( fn["marginRegularization"].isReal() );
params.marginRegularization = (float)fn["marginRegularization"];
CV_Assert ( fn["initialStepSize"].isReal() );
params.initialStepSize = (float)fn["initialStepSize"];
CV_Assert ( fn["stepDecreasingPower"].isReal() );
params.stepDecreasingPower = (float)fn["stepDecreasingPower"];
FileNode tcnode = fn["term_criteria"];
CV_Assert(!tcnode.empty());
params.termCrit.epsilon = (double)tcnode["epsilon"];
params.termCrit.maxCount = (int)tcnode["iterations"];
params.termCrit.type = (params.termCrit.epsilon > 0 ? TermCriteria::EPS : 0) +
(params.termCrit.maxCount > 0 ? TermCriteria::COUNT : 0);
CV_Assert ((params.termCrit.type & TermCriteria::COUNT || params.termCrit.type & TermCriteria::EPS));
}
int CV_MLBaseTest::prepare_test_case( int test_case_idx )
{
clear();
string dataPath = ts->get_data_path();
if ( dataPath.empty() )
{
ts->printf( cvtest::TS::LOG, "data path is empty" );
return cvtest::TS::FAIL_INVALID_TEST_DATA;
}
string dataName = dataSetNames[test_case_idx],
filename = dataPath + dataName + ".data";
FileNode dataParamsNode = validationFS.getFirstTopLevelNode()["validation"][modelName][dataName]["data_params"];
CV_DbgAssert( !dataParamsNode.empty() );
CV_DbgAssert( !dataParamsNode["LS"].empty() );
int trainSampleCount = (int)dataParamsNode["LS"];
CV_DbgAssert( !dataParamsNode["resp_idx"].empty() );
int respIdx = (int)dataParamsNode["resp_idx"];
CV_DbgAssert( !dataParamsNode["types"].empty() );
String varTypes = (String)dataParamsNode["types"];
data = TrainData::loadFromCSV(filename, 0, respIdx, respIdx+1, varTypes);
if( data.empty() )
{
ts->printf( cvtest::TS::LOG, "file %s can not be read\n", filename.c_str() );
return cvtest::TS::FAIL_INVALID_TEST_DATA;
}
data->setTrainTestSplit(trainSampleCount);
return cvtest::TS::OK;
}
bool CascadeClassifier::Data::read(const FileNode &root)
{
static const float THRESHOLD_EPS = 1e-5f;
// load stage params
String stageTypeStr = (String)root[CC_STAGE_TYPE];
if( stageTypeStr == CC_BOOST )
stageType = BOOST;
else
return false;
String featureTypeStr = (String)root[CC_FEATURE_TYPE];
if( featureTypeStr == CC_HAAR )
featureType = FeatureEvaluator::HAAR;
else if( featureTypeStr == CC_LBP )
featureType = FeatureEvaluator::LBP;
else if( featureTypeStr == CC_HOG )
featureType = FeatureEvaluator::HOG;
else
return false;
origWinSize.width = (int)root[CC_WIDTH];
origWinSize.height = (int)root[CC_HEIGHT];
CV_Assert( origWinSize.height > 0 && origWinSize.width > 0 );
isStumpBased = (int)(root[CC_STAGE_PARAMS][CC_MAX_DEPTH]) == 1 ? true : false;
// load feature params
FileNode fn = root[CC_FEATURE_PARAMS];
if( fn.empty() )
return false;
ncategories = fn[CC_MAX_CAT_COUNT];
int subsetSize = (ncategories + 31)/32,
nodeStep = 3 + ( ncategories>0 ? subsetSize : 1 );
// load stages
fn = root[CC_STAGES];
if( fn.empty() )
return false;
stages.reserve(fn.size());
classifiers.clear();
nodes.clear();
FileNodeIterator it = fn.begin(), it_end = fn.end();
for( int si = 0; it != it_end; si++, ++it )
{
FileNode fns = *it;
Stage stage;
stage.threshold = (float)fns[CC_STAGE_THRESHOLD] - THRESHOLD_EPS;
fns = fns[CC_WEAK_CLASSIFIERS];
if(fns.empty())
return false;
stage.ntrees = (int)fns.size();
stage.first = (int)classifiers.size();
stages.push_back(stage);
classifiers.reserve(stages[si].first + stages[si].ntrees);
FileNodeIterator it1 = fns.begin(), it1_end = fns.end();
for( ; it1 != it1_end; ++it1 ) // weak trees
{
FileNode fnw = *it1;
FileNode internalNodes = fnw[CC_INTERNAL_NODES];
FileNode leafValues = fnw[CC_LEAF_VALUES];
if( internalNodes.empty() || leafValues.empty() )
return false;
DTree tree;
tree.nodeCount = (int)internalNodes.size()/nodeStep;
classifiers.push_back(tree);
nodes.reserve(nodes.size() + tree.nodeCount);
leaves.reserve(leaves.size() + leafValues.size());
if( subsetSize > 0 )
subsets.reserve(subsets.size() + tree.nodeCount*subsetSize);
FileNodeIterator internalNodesIter = internalNodes.begin(), internalNodesEnd = internalNodes.end();
for( ; internalNodesIter != internalNodesEnd; ) // nodes
{
DTreeNode node;
node.left = (int)*internalNodesIter; ++internalNodesIter;
node.right = (int)*internalNodesIter; ++internalNodesIter;
node.featureIdx = (int)*internalNodesIter; ++internalNodesIter;
if( subsetSize > 0 )
{
for( int j = 0; j < subsetSize; j++, ++internalNodesIter )
subsets.push_back((int)*internalNodesIter);
node.threshold = 0.f;
}
else
{
node.threshold = (float)*internalNodesIter; ++internalNodesIter;
}
nodes.push_back(node);
}
internalNodesIter = leafValues.begin(), internalNodesEnd = leafValues.end();
for( ; internalNodesIter != internalNodesEnd; ++internalNodesIter ) // leaves
leaves.push_back((float)*internalNodesIter);
}
}
return true;
}
static void read(const FileNode& node, MyData& x, const MyData& default_value = MyData()){
if(node.empty())
x = default_value;
else
x.read(node);
}
static void read(const FileNode& node, TestHashEntry& x, const TestHashEntry& default_value = TestHashEntry()){
if(node.empty())
x = default_value;
else
x.read(node);
}
int main(int argc, char** argv)
{
//vector of points to store the feature points calculated from part 3
vector<Point> points1;
vector<Point> points2;
//loading the first .yml file
//should have the same name as image file + _Points.yml
string filename = string(argv[1]);
filename = filename.substr(0, filename.find(".pgm"));
string pointfilename = filename+"_Points.yml";
FileStorage fs(pointfilename, FileStorage::READ);
FileNode ptFileNode;
int num1 = 0;
while(true)
{
string pointName = "point";
Point tempPoint;
stringstream ss;
ss << num1;
string temp;
ss >> temp;
pointName+=temp;
ptFileNode = fs[pointName];
if(ptFileNode.empty() == true)
{
break; //break when finding an empty filenode, this means we've reached the end of our list
}
ptFileNode >> tempPoint;
points1.push_back(tempPoint);
num1++;
}
//loading the second file
string filename2 = string(argv[2]);
filename2 = filename2.substr(0, filename2.find(".pgm"));
string pointfilename2 = filename2+"_Points.yml";
FileStorage fs2(pointfilename2, FileStorage::READ);
FileNode ptFileNode2;
int num2 = 0;
while(true)
{
string pointName = "point";
Point tempPoint;
stringstream ss;
ss << num2;
string temp;
ss >> temp;
pointName+=temp;
ptFileNode2 = fs2[pointName];
if(ptFileNode2.empty() == true)
{
break;
}
ptFileNode2 >> tempPoint;
points2.push_back(tempPoint);
num2++;
}
Mat image;
Mat image2;
//reads image name from command line
image = imread(argv[1], CV_LOAD_IMAGE_GRAYSCALE);
image2 = imread(argv[2], CV_LOAD_IMAGE_GRAYSCALE);
Mat imageColor;
imageColor = imread(argv[2], 1);
int sum;
int difference;
Scalar im1;
Scalar im2;
vector<feature> potentialMatches;
for(int j=0; j<points2.size(); j++)
{
Point currPoint2 = points2[j]; //iterate through all features in the reference frame
for(int k=0; k<points1.size(); k++)
{
Point currPoint1 = points1[k]; //iterate through all features in the other image
sum = 0;
for(int r = -windHalf; r <= windHalf; r++)
{
for(int c = -windHalf; c <= windHalf; c++)
{
//check a window x window sized square of pixels to compute ssd
// do thie for every feature of image 1 with respect to image 2 (reference
Point addPoint = Point(r,c);
Point squarePoint2 = currPoint2+addPoint;
Point squarePoint1 = currPoint1+addPoint;
if(squarePoint2.x < 0 || squarePoint1.x < 0 || squarePoint2.x >= image.cols
|| squarePoint1.x >= image.cols)
{
continue; //make sure the square doesn't go out of bounds
}
if(squarePoint2.y < 0 || squarePoint1.y < 0 || squarePoint2.y >= image.rows
|| squarePoint1.y >= image.rows)
{
continue;
}
im2 = image2.at<unsigned char>(squarePoint2);
im1 = image.at<unsigned char>(squarePoint1);
difference = im1.val[0] - im2.val[0];
sum += difference*difference;
}
}
feature tempF;
tempF.ssd = sum;
tempF.p = currPoint1;
potentialMatches.push_back(tempF);
}
sort(potentialMatches.begin(), potentialMatches.end()); //sort matches by ssd value from low to high
if((double)potentialMatches[0].ssd/potentialMatches[1].ssd < threshold_val)
{
//if below threshold, it is a significant match, so draw a line between the points
line(imageColor, currPoint2, potentialMatches[0].p, Scalar(255, 0, 0));
}
potentialMatches.clear();
}
for(int i=0; i<points2.size(); i++)
{
circle(imageColor, points2[i], 3, Scalar(0,0,255));
}
imshow("Matched", imageColor);
//save image to file
string outputName = argv[2];
outputName = outputName.substr(0, outputName.find(".pgm"));
outputName += "_matched.pgm";
imwrite(outputName, imageColor);
waitKey(0);
return(0);
}
void HistogramsIO::read(const FileNode & node, Ptr<Histogram> & x) {
if (!node.empty())
x->read(node);
}
bool fill(const FileNode &root)
{
// cascade properties
static const char *const SC_STAGE_TYPE = "stageType";
static const char *const SC_BOOST = "BOOST";
static const char *const SC_FEATURE_TYPE = "featureType";
static const char *const SC_ICF = "ICF";
static const char *const SC_ORIG_W = "width";
static const char *const SC_ORIG_H = "height";
static const char *const SC_OCTAVES = "octaves";
static const char *const SC_TREES = "trees";
static const char *const SC_FEATURES = "features";
static const char *const SC_INTERNAL = "internalNodes";
static const char *const SC_LEAF = "leafValues";
static const char *const SC_SHRINKAGE = "shrinkage";
static const char *const FEATURE_FORMAT = "featureFormat";
// only Ada Boost supported
std::string stageTypeStr = (string)root[SC_STAGE_TYPE];
CV_Assert(stageTypeStr == SC_BOOST);
std::string fformat = (string)root[FEATURE_FORMAT];
bool useBoxes = (fformat == "BOX");
// only HOG-like integral channel features supported
string featureTypeStr = (string)root[SC_FEATURE_TYPE];
CV_Assert(featureTypeStr == SC_ICF);
origObjWidth = (int)root[SC_ORIG_W];
origObjHeight = (int)root[SC_ORIG_H];
shrinkage = (int)root[SC_SHRINKAGE];
FileNode fn = root[SC_OCTAVES];
if (fn.empty()) return false;
// for each octave
FileNodeIterator it = fn.begin(), it_end = fn.end();
for (int octIndex = 0; it != it_end; ++it, ++octIndex)
{
FileNode fns = *it;
Octave octave(octIndex, cv::Size(origObjWidth, origObjHeight), fns);
CV_Assert(octave.weaks > 0);
octaves.push_back(octave);
FileNode ffs = fns[SC_FEATURES];
if (ffs.empty()) return false;
fns = fns[SC_TREES];
if (fn.empty()) return false;
FileNodeIterator st = fns.begin(), st_end = fns.end();
for (; st != st_end; ++st )
{
weaks.push_back(Weak(*st));
fns = (*st)[SC_INTERNAL];
FileNodeIterator inIt = fns.begin(), inIt_end = fns.end();
for (; inIt != inIt_end;)
nodes.push_back(Node(features.size(), inIt));
fns = (*st)[SC_LEAF];
inIt = fns.begin(), inIt_end = fns.end();
for (; inIt != inIt_end; ++inIt)
leaves.push_back((float)(*inIt));
}
st = ffs.begin(), st_end = ffs.end();
for (; st != st_end; ++st )
features.push_back(Feature(*st, useBoxes));
}
return true;
}
static void read(const FileNode& node, Cluster& x, const Cluster& default_value = Cluster()){
if (node.empty())
x = default_value;
else
x.read(node);
}
static Fields* parseCascade(const FileNode &root, const float mins, const float maxs, const int totals, const int method)
{
static const char *const SC_STAGE_TYPE = "stageType";
static const char *const SC_BOOST = "BOOST";
static const char *const SC_FEATURE_TYPE = "featureType";
static const char *const SC_ICF = "ICF";
static const char *const SC_ORIG_W = "width";
static const char *const SC_ORIG_H = "height";
static const char *const SC_FEATURE_FORMAT = "featureFormat";
static const char *const SC_SHRINKAGE = "shrinkage";
static const char *const SC_OCTAVES = "octaves";
static const char *const SC_OCT_SCALE = "scale";
static const char *const SC_OCT_WEAKS = "weaks";
static const char *const SC_TREES = "trees";
static const char *const SC_WEAK_THRESHOLD = "treeThreshold";
static const char *const SC_FEATURES = "features";
static const char *const SC_INTERNAL = "internalNodes";
static const char *const SC_LEAF = "leafValues";
static const char *const SC_F_CHANNEL = "channel";
static const char *const SC_F_RECT = "rect";
// only Ada Boost supported
std::string stageTypeStr = (std::string)root[SC_STAGE_TYPE];
CV_Assert(stageTypeStr == SC_BOOST);
// only HOG-like integral channel features supported
std::string featureTypeStr = (std::string)root[SC_FEATURE_TYPE];
CV_Assert(featureTypeStr == SC_ICF);
int origWidth = (int)root[SC_ORIG_W];
int origHeight = (int)root[SC_ORIG_H];
std::string fformat = (std::string)root[SC_FEATURE_FORMAT];
bool useBoxes = (fformat == "BOX");
ushort shrinkage = cv::saturate_cast<ushort>((int)root[SC_SHRINKAGE]);
FileNode fn = root[SC_OCTAVES];
if (fn.empty()) return 0;
std::vector<device::Octave> voctaves;
std::vector<float> vstages;
std::vector<device::Node> vnodes;
std::vector<float> vleaves;
FileNodeIterator it = fn.begin(), it_end = fn.end();
for (ushort octIndex = 0; it != it_end; ++it, ++octIndex)
{
FileNode fns = *it;
float scale = powf(2.f,saturate_cast<float>((int)fns[SC_OCT_SCALE]));
bool isUPOctave = scale >= 1;
ushort nweaks = saturate_cast<ushort>((int)fns[SC_OCT_WEAKS]);
ushort2 size;
size.x = cvRound(origWidth * scale);
size.y = cvRound(origHeight * scale);
device::Octave octave(octIndex, nweaks, shrinkage, size, scale);
CV_Assert(octave.stages > 0);
voctaves.push_back(octave);
FileNode ffs = fns[SC_FEATURES];
if (ffs.empty()) return 0;
std::vector<cv::Rect> feature_rects;
std::vector<int> feature_channels;
FileNodeIterator ftrs = ffs.begin(), ftrs_end = ffs.end();
int feature_offset = 0;
for (; ftrs != ftrs_end; ++ftrs, ++feature_offset )
{
cv::FileNode ftn = (*ftrs)[SC_F_RECT];
cv::FileNodeIterator r_it = ftn.begin();
int x = (int)*(r_it++);
int y = (int)*(r_it++);
int w = (int)*(r_it++);
int h = (int)*(r_it++);
if (useBoxes)
{
if (isUPOctave)
{
w -= x;
h -= y;
}
}
else
{
if (!isUPOctave)
{
w += x;
h += y;
}
}
feature_rects.push_back(cv::Rect(x, y, w, h));
feature_channels.push_back((int)(*ftrs)[SC_F_CHANNEL]);
}
fns = fns[SC_TREES];
if (fn.empty()) return false;
// for each stage (~ decision tree with H = 2)
FileNodeIterator st = fns.begin(), st_end = fns.end();
for (; st != st_end; ++st )
{
FileNode octfn = *st;
float threshold = (float)octfn[SC_WEAK_THRESHOLD];
vstages.push_back(threshold);
FileNode intfns = octfn[SC_INTERNAL];
FileNodeIterator inIt = intfns.begin(), inIt_end = intfns.end();
for (; inIt != inIt_end;)
{
inIt +=2;
int featureIdx = (int)(*(inIt++));
float orig_threshold = (float)(*(inIt++));
unsigned int th = saturate_cast<unsigned int>((int)orig_threshold);
cv::Rect& r = feature_rects[featureIdx];
uchar4 rect;
rect.x = saturate_cast<uchar>(r.x);
rect.y = saturate_cast<uchar>(r.y);
rect.z = saturate_cast<uchar>(r.width);
rect.w = saturate_cast<uchar>(r.height);
unsigned int channel = saturate_cast<unsigned int>(feature_channels[featureIdx]);
vnodes.push_back(device::Node(rect, channel, th));
}
intfns = octfn[SC_LEAF];
inIt = intfns.begin(), inIt_end = intfns.end();
for (; inIt != inIt_end; ++inIt)
{
vleaves.push_back((float)(*inIt));
}
}
}
cv::Mat hoctaves(1, (int) (voctaves.size() * sizeof(device::Octave)), CV_8UC1, (uchar*)&(voctaves[0]));
CV_Assert(!hoctaves.empty());
cv::Mat hstages(cv::Mat(vstages).reshape(1,1));
CV_Assert(!hstages.empty());
cv::Mat hnodes(1, (int) (vnodes.size() * sizeof(device::Node)), CV_8UC1, (uchar*)&(vnodes[0]) );
CV_Assert(!hnodes.empty());
cv::Mat hleaves(cv::Mat(vleaves).reshape(1,1));
CV_Assert(!hleaves.empty());
Fields* fields = new Fields(mins, maxs, totals, origWidth, origHeight, shrinkage, 0,
hoctaves, hstages, hnodes, hleaves, method);
fields->voctaves = voctaves;
fields->createLevels(DEFAULT_FRAME_HEIGHT, DEFAULT_FRAME_WIDTH);
return fields;
}
void read_params( const FileNode& fn )
{
String activ_func_name = (String)fn["activation_function"];
if( !activ_func_name.empty() )
{
activ_func = activ_func_name == "SIGMOID_SYM" ? SIGMOID_SYM :
activ_func_name == "IDENTITY" ? IDENTITY :
activ_func_name == "GAUSSIAN" ? GAUSSIAN : -1;
CV_Assert( activ_func >= 0 );
}
else
activ_func = (int)fn["activation_function_id"];
f_param1 = (double)fn["f_param1"];
f_param2 = (double)fn["f_param2"];
setActivationFunction( activ_func, f_param1, f_param2 );
min_val = (double)fn["min_val"];
max_val = (double)fn["max_val"];
min_val1 = (double)fn["min_val1"];
max_val1 = (double)fn["max_val1"];
FileNode tpn = fn["training_params"];
params = AnnParams();
if( !tpn.empty() )
{
String tmethod_name = (String)tpn["train_method"];
if( tmethod_name == "BACKPROP" )
{
params.trainMethod = ANN_MLP::BACKPROP;
params.bpDWScale = (double)tpn["dw_scale"];
params.bpMomentScale = (double)tpn["moment_scale"];
}
else if( tmethod_name == "RPROP" )
{
params.trainMethod = ANN_MLP::RPROP;
params.rpDW0 = (double)tpn["dw0"];
params.rpDWPlus = (double)tpn["dw_plus"];
params.rpDWMinus = (double)tpn["dw_minus"];
params.rpDWMin = (double)tpn["dw_min"];
params.rpDWMax = (double)tpn["dw_max"];
}
else
CV_Error(CV_StsParseError, "Unknown training method (should be BACKPROP or RPROP)");
FileNode tcn = tpn["term_criteria"];
if( !tcn.empty() )
{
FileNode tcn_e = tcn["epsilon"];
FileNode tcn_i = tcn["iterations"];
params.termCrit.type = 0;
if( !tcn_e.empty() )
{
params.termCrit.type |= TermCriteria::EPS;
params.termCrit.epsilon = (double)tcn_e;
}
if( !tcn_i.empty() )
{
params.termCrit.type |= TermCriteria::COUNT;
params.termCrit.maxCount = (int)tcn_i;
}
}
}
}