vector<float> HOGDetectorGPU::getHOGdescriptors(cv::HOGDescriptor& hog, Mat& image){
resize(image, image, Size(WIN_X,WIN_Y) );
Mat img;
cvtColor(image, img, CV_RGB2GRAY);
hog.winSize.height = WIN_X;
hog.winSize.width = WIN_Y;
// Size(128,64), //winSize
// Size(16,16), //blocksize
// Size(8,8), //blockStride,
// Size(8,8), //cellSize,
// 9, //nbins,
// 0, //derivAper,
// -1, //winSigma,
// 0, //histogramNormType,
// 0.2, //L2HysThresh,
// 0 //gammal correction,
// //nlevels=64
//);
vector<float> descriptorsValues;
vector<Point> locations;
hog.compute( img, descriptorsValues, Size(0,0), Size(0,0), locations);
cout << "HOG descriptor size is " << hog.getDescriptorSize() << endl;
cout << "img dimensions: " << img.cols << " width x " << img.rows << "height" << endl;
cout << "Found " << descriptorsValues.size() << " descriptor values" << endl;
cout << "Nr of locations specified : " << locations.size() << endl;
return descriptorsValues;
}
static void calculateFeaturesFromInput( cv::Mat imageData, std::vector< float >& featureVector, cv::HOGDescriptor& hog )
{
std::vector< cv::Point > locations;
hog.compute( imageData, featureVector, winStride, trainingPadding, locations );
imageData.release();
}
cv::Mat computeHOG(const cv::Mat &img_gray, cv::HOGDescriptor &d) {
int correct_width = img_gray.cols - (img_gray.cols - d.blockSize.width) % d.blockStride.width;
int correct_height = img_gray.rows - (img_gray.rows - d.blockSize.height) % d.blockStride.height;
d.winSize = cv::Size(correct_width, correct_height);
//Scale input image to the correct size
cv::Mat resize_img;
cv::resize(img_gray, resize_img, d.winSize);
std::vector<float> descriptorsValues;
std::vector<cv::Point> locations;
d.compute(resize_img, descriptorsValues, cv::Size(0,0), cv::Size(0,0), locations);
cv::Mat matDescriptorsValues(descriptorsValues, true);
return matDescriptorsValues;
}