danieltan07 / spatialaffinitynetwork

PyTorch Implementation of the paper Learning Affinity via Spatial Propagation Networks
61 stars 12 forks source link

zero_pad = 256 * 3 - len(self.palette) #3

Open liuyi1989 opened 6 years ago

liuyi1989 commented 6 years ago

What does this command do? Besides, if I solve a binary segmentation problem, i.e., only 2 classes, can I directly modify palette = [0, 0, 0, 255, 255, 255]?