deep-photo-enhance-t2
deep-photo-enhance-t2 copied to clipboard
MrRobot2211
Working code
Hi. I greatly appreciate your work on upgrading the DPE model to TF2, and thanks for your effort. I'm wondering, have you managed to run the prediction code using any provided model? I cloned your repo but get big errors when trying to run it. Thanks!
Yes i managed to run it using python predict.py it just uses the pretrained 736
On Tue, Apr 7, 2020, 08:09 Andrei Pauliuc [email protected] wrote:
Hi. I greatly appreciate your work on upgrading the DPE model to TF2, and thanks for your effort. I'm wondering, have you managed to run the prediction code using any provided model? I cloned your repo but get big errors when trying to run it. Thanks!
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/MrRobot2211/deep-photo-enhance-t2/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFTMVKHFZAGET6GJY567LDDRLMCWFANCNFSM4MDAXCNQ .
Hi, how to run the training code? I want to train the model on my custom dataset.
Hi , you should run train.py and set the images in the directories as explained in the original dpe repo as well as setup the corresponding files . That is you need to have an input folder that conttains LPGAn and inside a folder called input with the original pictures and another one called labels with target pictures. Inside LPGAn you should add files train_input.txt, test_input.txt and test.txt (in there you should have the images that are on both folders , that's to say tthe ones you have the original and the translation) . THis works forr images that are of maximum size 512 in any direction, you can use the resize.py script for that. I am adding a clear readme for the whole prrocess today.
Hi , you should run train.py and set the images in the directories as explained in the original dpe repo as well as setup the corresponding files . That is you need to have an input folder that conttains LPGAn and inside a folder called input with the original pictures and another one called labels with target pictures. Inside LPGAn you should add files train_input.txt, test_input.txt and test.txt (in there you should have the images that are on both folders , that's to say tthe ones you have the original and the translation) . THis works forr images that are of maximum size 512 in any direction, you can use the resize.py script for that. I am adding a clear readme for the whole prrocess today.
Hi, I would also be interested to try your model in inference (on the pretrained model). Any news about a readme ? ;)
