how can i use it for my use case. about spatialattentiongan HOT 3 OPEN

Hi @alan-ai-learner, please try the celeba_128_eyeglasses pre-trained model. (1) Clone the repo. (2) Install all dependencies. (3) Download the pre-trained model and put it under ./results/. (4) Write an inference script like generate.py. You cannot use generate.py directly because it generates images with opposite attributes against the original images (labels).

from spatialattentiongan.

hey @elvisyjlin thanks for replying, but I tried the way you said, but the results are not that promising.
If you can suggest any different way that will be great.
Thanks in advance!

from spatialattentiongan.

Sorry that I'm not sure about your "not that promising". Is the generated image has low quality? looks the same as input? or total a garbage?

Please understand that the model learns the distribution translation between with attribute A and without attribute A. However, it is quite a naive model with limited capacity. The results are not guaranteed to be perfect.

If you're looking for a stabler model which generates amazing images, StyleGAN is a good choice.

from spatialattentiongan.