-
Notifications
You must be signed in to change notification settings - Fork 533
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Convergency of edge model #177
Comments
你好,我想知道你跑这个模型,跑了多长时间,我把它放在3060上跑,max_iters=140000,跑了两天了,还没结束 |
方便的话,可以加一下qq:2635505974 |
hello,your ID shows cannot be found |
Hi, have you solved this problem yet? |
Hello, I also encountered the same problem as you when training the edge model, the repaired edge is very bad. Have you solved the problem? If you solved the problem can you tell me how you did it? It means a lot to me. Thank you! |
Hello, I have been reproducing this code recently, and the edge repair results were poor when training MODEL=1. Have you solved the problem? |
Hello, I also encountered this problem, did you solve it? |
Did you solve the issue? |
I think it can not be solved. |
This work has trained connect work on a custom dataset and they have a model working fine. Its trained in 2022 https://github.com/amiretefaghi/E2F-GAN |
Hello @knazeri Thanks for your opening source code. I have used your implementation to train from scratch. I stay all your settings unchanged, but only use another remote sensing dataset, which contains only 800 training images with size 256x256.
At the beginning, I trained a edge model with "MODEL=1". It is observed that after 1000 epoches training, the edge model does learn something, but very slow and far from convergency.
The above images are the samples during training after 1 and 67000 iterations respectively. It shows that the edge model was learning throughout the whole training procedure, but could you tell me why does the estimated results so bad?
Beyond that, I have also trained a inpainting model alone with "MODEL=2", but I get a very different results from the results above, as I list below. It is clear that the inpainting model convergenced successfully, which proves the correctness of the training strategy. So Why I get a results above?
Looking for your replys.
The text was updated successfully, but these errors were encountered: