thias15 / context-aware-cf-tracking Goto Github PK

Thank you very much for your paper. The implementation of the code is only in the primal domain. There is no in dual domain code. How can I change it to a dual domain? Thank you.
kf = conj(xf) .* xf;

    kfn = zeros([size(xf) length(offset)]);
    for j=1:length(offset)
        %obtain a subwindow close to target for regression to 0
        patch = get_subwindow(im, pos+offset(j,:), window_sz);
        xfn = fft2(get_features(patch, features, cell_size, cos_window));
        kfn(:,:,:,j) = conj(xfn) .*xfn;
    end

    %Assumption: Features are independent
    num = bsxfun(@times, conj(xf),yf); 
    den = kf + lambda1 + lambda2.*sum(kfn,4);
    wf = num ./ den;
    
	if frame == 1,  %first frame, train with a single image
		model_wf = wf;
	else
		%subsequent frames, interpolate model
		model_wf = (1 - interp_factor) * model_wf + interp_factor * wf;
	end

Thanks a lot for posting the different tracking algorithms. I was just wondering do you have any plans to push CACF code for tracking?

Hello,

Could you please also include the steps to build and run (with the expected pre-requisites) in the Readme .. or point me to where to find these?

Thank you.

Thank you for the code.
I tried to run the code for on mac system but there is no compiled mex function for the mexResize.cpp
is it possible for you to provide the compiled mex function for the mac OS?
I tried to install opencv and compile the mex files but never succeed