reproducing results using pretrained Deep Fashion Model -- quality seems not as good as expected about xinggan HOT 3 CLOSED

The results are worse than what I expected, it may be caused by a different environment.

In addition, you can also try BiGraphGAN.

from xinggan.

Thanks for clarifying that.
It is not so easy to reconstruct exactly the code environment you had when running XingGAN and getting good results.
But do you have an archive folder of test results, the 8570 generate images supposed to be in the DeepFashion test set?
If you have this data, I can still run a clean comparison and cite your paper.

thanks a lot for the help

from xinggan.

This project was done last year, and the generated results have not been preserved. You can absolutely use the results you get to compare in your paper.

Good luck with your NeurIPS paper!

from xinggan.