Pre-trained weights and Inference

KupynOrest / DeblurGAN

Image Deblurring using Generative Adversarial Networks

Other

2.52k stars 521 forks source link

Pre-trained weights and Inference #197

Closed innat closed 2 years ago

innat commented 4 years ago

Even in ReadMe, it says the weight files are stored in google drive, surely it's not there. Files are located now simply in the following directories: DeblurGAN/checkpoints/experiment_name/, namely: latest_net_D.pth and latest_net_G.pth.

In inference time, what I have found that the model randomly crops the image, practically which is not desirable. Is there any workaround to get results with the pre-trained weights on the entire input (blurred) sample?

SuJiaKuan commented 3 years ago

Please try to run test.py with --resize_or_crop scale_width parameter.

Example:

python test.py --dataroot /.path_to_your_data --model test --dataset_mode single --resize_or_crop scale_width --learn_residual