Hello, I print the pytorch model output, named "pts" and "desc" and find than the size of these tensors are pts=[4327, 3], desc=[4327, 32]. It seems that the model output has been changed to directly generate the feature point and descriptor. In the paper, the model output shold be pts=[320,240] and desc=[256, 320, 240]?
Hello, I print the pytorch model output, named "pts" and "desc" and find than the size of these tensors are pts=[4327, 3], desc=[4327, 32]. It seems that the model output has been changed to directly generate the feature point and descriptor. In the paper, the model output shold be pts=[320,240] and desc=[256, 320, 240]?