ContourGAN implement using Pytorch
Continuous updating...
paper: https://www.sciencedirect.com/science/article/abs/pii/S0950705118304842
tensorflow Implement(Official ): https://github.com/sxdxedu/ContourGAN
-
try get as close as possible to the results of the paper...
- fix the content loss for G (Crucial)
- fix the Loss for adversarial loss (e,g, using wgan-gp etc..)
- add Non-Maximum Suppression for image view
- decoder conv bilinear interpolation init ... have not added now but refered in paper
- normalize the input of Discriminator? now Original binary image
-
Enhancing Engineering