jonghwa-yim
commented
4 years ago Hi,
firstly, you need to prepare the dataset. You can apply pilgram filters to the COCO dataset. Then pair the filtered image with the original COCO version.
If you train the model without any uncertainty, it will converge after 6 to 9 epochs with train loss 0.05~0.08. Note that the train loss does not have to be lower, and it cannot exactly output the original image. The purpose of the DL model is to output mean-colored image.
2020년 10월 12일 (월) 21:10, YXXX notifications@github.com님이 작성:
hi, thanks for ur code. I want to use it, and I download the code, and train it just with only one pic,but the train loss converge to about 0.3. I want to know whether u could put a traind pth or give me some guide on how to train the model,thanks a lot.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jonghwa-yim/FilterStyleTransfer/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/AD54GUVRERFQ5TXHQLVE3WTSKLW2PANCNFSM4SMXBCNQ .
hi, thanks for ur code. I want to use it, and I download the code, and train it just with only one pic,but the train loss converge to about 0.3. I want to know whether u could put a traind pth or give me some guide on how to train the model,thanks a lot.