search

vickyFox / Region-wise-Inpainting

Region-wise GAN, image inpainting
MIT License
53 stars 17 forks source link

train on places2 #4

Open caffeAndtf opened 4 years ago

caffeAndtf commented 4 years ago

did you adopt the whole places2 dataset to train the model? how many hours does it take for each stage on places? thx~

vickyFox commented 4 years ago

Thanks for your attenton. We did use the whole places2 dataset to train the model. We first trained our model using six blocks of 1080ti in almost six days， and then added the discriminator and trained it in almost three days.