search

odegeasslbc / Self-Supervised-Sketch-to-Image-Synthesis-PyTorch

The pytorch implementation for the paper "Self-Supervised Sketch-to-Image Synthesis" in AAAI-2021
GNU General Public License v3.0
65 stars 6 forks source link

Pretrained models #2

Closed lastapple closed 3 years ago

lastapple commented 3 years ago

Are pretrained models available? Can you upload them to Google drive？Thanks!

odegeasslbc commented 3 years ago

Sorry, we can't release a pre-trained model due to commercial constraints. However, the model shall be trained with the provided code and data, a single GPU is able to reproduce similar results.

lastapple commented 3 years ago

Ok, thanks! I‘ll train the model.

lyhh123 commented 3 years ago

Do all models need to be trained ? Including VGG?

odegeasslbc commented 3 years ago

Do all models need to be trained ? Including VGG?

No, if I understand your question correctly, VGG for the LPIPS loss does not need to be trained. The pre-trained model will be automatically downloaded and used.

lyhh123 commented 3 years ago

Do all models need to be trained ? Including VGG?

No, if I understand your question correctly, VGG for the LPIPS loss does not need to be trained. The pre-trained model will be automatically downloaded and used. Thanks for your reply. The problem has been solved.

lyhh123 commented 3 years ago

Do all models need to be trained ? Including VGG?

No, if I understand your question correctly, VGG for the LPIPS loss does not need to be trained. The pre-trained model will be automatically downloaded and used. Thanks for your reply. The problem has been solved.

ZhouCX117 commented 1 year ago

@lyhh123 @odegeasslbc Hi, how to automatically download the vgg-feature-weights.pth used in sketch_styletransfer/train.py?