Hallo everyone!
I have used one versus all multi-class SVM for image classification and I would like to compute probabilities of each class at each node/pixel from the decision function. How do I go about this? I have not seen an example computing probabilities from a multi-class trainer from dlib documentation (http://dlib.net/ml.html), kindly help/advice. Advice from different SVM C++ libraries preferable with demo code is also welcome.
Here is my code (https://sourceforge.net/p/dclib/discussion/442517/thread/c34fe256/):
Regards