Implement Multi class SVM classifier with HOG descriptor

edit

Hello,

Please I need your help with some code or the algorithm to follow to give HOG (Histogram of oriented gradient) to an SVM for identification of a valve in a video stream.

I want to implement Multi class SVM classifier with HOG, I have done some research but I don't think I understood well how the training works. In my problem I have two classes, let's labilize the first 1 and the second -1.

Then in my case I have to give the SVM if I want the SVM to detect the first class:

• Images of the first classes as positive samples, and images from the second class as negative samples.

And if I want the SVM to detect the second class I give it:

• Images of the second classes as positive samples, and images from the second class as negative samples.

please tell me if I am wrong or not about this, and if you can help me with a code with C++ and Opencv to start with I will be very grateful.

And thank you.