Trying to add Face Recognizer

[kobush]

Hi,
I'm trying to add support for the FaceRecognizer API. I was able to configure and build the project from source but now I'm stuck on how to pass the parameters for train method:
http://docs.opencv.org/modules/contrib/doc/facerec/facerec_api.html#facerecognizer-train

It has signature

void FaceRecognizer::train(InputArrayOfArrays src, InputArray labels) 

But first argument is expected as vector and second is vector

So I came up with following code in OpenCvSharpExtern

CVAPI(void) objdetect_FaceRecognizer_train(cv::FaceRecognizer *obj, cv::Mat** src, int srcLength, int* labels, int labelsLength)
{
    std::vector<cv::Mat> srcVec(srcLength);
    for (int i = 0; i < srcLength; i++)
        srcVec[i] = *src[i];

    std::vector<int> labelsVec = std::vector<int>(labels, labels + labelsLength);
    obj->train(srcVec, labelsVec);
}

Then on C# side I have declared this in native methods:

        [DllImport(DllExtern, CallingConvention = CallingConvention.Cdecl)]
        public static extern void objdetect_FaceRecognizer_train(IntPtr obj, IntPtr[] src, int srcLength, 
            [MarshalAs(UnmanagedType.LPArray)] int[] labels, int labelsLength);

Finally this is the wrapper code:

        public void Train(Mat[] src, int[] labels)
        {
            IntPtr[] srcPtrs = EnumerableEx.SelectPtrs(src);

            NativeMethods.objdetect_FaceRecognizer_train(ptr, srcPtrs, src.Length, labels, labels.Length);
        }

I get AccessViolationException so clearly I must have done something stupid.

[shimat]

I added the FaceRecognizer wrapper to the latest branch (CPlusPlus-Improvement-feature2d).
I tested the following simple code, then it ran successfully. Please fetch the branch and try it.

FaceRecognizer fr = FaceRecognizer.CreateEigenFaceRecognizer();
fr.Train(new []{img1, img2}, new []{0, 1});

[kobush]

It works like a charm. Thank you sir for quick response!