CODEWITHSUNDEEP
opencv
SomethingSomething
SomethingSomething

Reputation: 12178

OpenCV 3.0 - HOGDescriptor::compute() code example

These are the signatures of HOGDescriptor C'tor and method compute() in OpenCV3.0:

HOGDescriptor (Size _winSize, Size _blockSize, Size _blockStride, Size _cellSize, 
                  int _nbins, int _derivAperture=1, double _winSigma=-1, 
                  int _histogramNormType=HOGDescriptor::L2Hys, double _L2HysThreshold=0.2, 
                  bool _gammaCorrection=false, int _nlevels=HOGDescriptor::DEFAULT_NLEVELS, 
                  bool _signedGradient=false)


virtual void compute (InputArray img, std::vector< float > &descriptors,
                          Size winStride=Size(), Size padding=Size(), 
                          const std::vector< Point > &locations=std::vector< Point >()) const

Does anyone have an example code or an idea of how to compute one vector for THE WHOLE image?

Upvotes: 2

Views: 7258

Answers (1)

Tobias Senst
Tobias Senst

Reputation: 2830

The following small piece of could should work for your problem:

// create descriptor object and set the size of the hog descriptor to image size
cv::HOGDescriptor hog;        
hog.winSize = grayImg.size(); 
// set the descriptor position to the middle of the image
std::vector<cv::Point> positions; 
positions.push_back(cv::Point(grayImg.cols / 2, grayImg.rows / 2));
std::vector<float> descriptor;
hog.compute(grayImg,descriptor,cv::Size(),cv::Size(),positions);

but please note that this descriptor will be larger than the descriptor used proposed to be optimal by Dalal and Triggs for detecting pedestrians.

Upvotes: 4

Related Questions