Emgu SVM classifier, predicts incorrect with poly and RBF kernels

Question

I used EmguCV in C# for a face recognition project, but I found Emgu's support vector machine (SVM) predicts wrong classes when I use poly and RBF (redial basis function) kernels.

I compared Emgu's SVM answers with Matlab's svmclassify and found Matlab classifies all test cases right but Emgu predicts wrong. may be there is something wrong with my code, but I don't this think so.

I reported the bug I found ----> here <----, but I got no answers.

can anyone help me to find out what is wrong with Emgu's SVM or to find a library for a good multi-class support vector machine that uses different kernels like RBF (redial basis function)

Answer 1

Given your emgu bug has not been investigated, I can suggest you to try out some Libsvm c# wrapper.

In the past I personally made use of the excellent porting SVM.NET (it wraps libsvm 2.89 functionality), but if you want to stay more up to date you should also consider user another C# libsvm wrapper called LibSvmWrapper (it wraps 3.1 functionality).