Relative to a parameter-tuned bandpass filter, siamese convolutional networks significantly reduce false matches. The improvement is quantified using patches of brain images from serial section electron microscopy. Unlike fully supervised approaches to metric learning, our method can improve upon vanilla NCC without being given locations of true matches during training. Our main technical contribution is a weakly supervised learning algorithm for training siamese networks. We improve the robustness of this algorithm by transforming image features with "siamese" convolutional networks trained to maximize the contrast between NCC values of true and false matches. Template matching by normalized cross correlation (NCC) is widely used for finding image correspondences.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |