junyanz / pytorch-CycleGAN-and-pix2pix

Image-to-Image Translation in PyTorch
Other
22.8k stars 6.29k forks source link

Help with Inference #1675

Closed roshan2024nar closed 5 days ago

roshan2024nar commented 1 week ago

Hello,

I have trained a pix2pix model using

python train.py --dataroot datasets/shadow --name shadow --model pix2pix --direction AtoB --load_size 2048 --crop_size 2048 --checkpoints_dir checkpoints

how can i test on a images without B paths or on a single image by providing path of it

catyoyo commented 5 days ago

I have read your question. Is it the case of using a single image for testing when it is not an image pair? The author has already provided a method. Using the single dataset mode is fine.Really, on GitHub, this is the first time I have seen an author who writes such detailed comments and makes module operations so simple. It is really admirable. See a list of currently available models at ./scripts/download_pix2pix_model.sh python test.py --dataroot ./datasets/facades/testB/ --name facades_pix2pix --model test --netG unet_256 --direction AtoB --dataset_mode single --norm batch

roshan2024nar commented 5 days ago

Thank you for your response i know that author has done a awesome work i might missed some details what i want is something where i can provide path of a single image without going through creation of data loaders etc. i tried to mimic the transformation used in the original code but it did not work well