python train.py --dataroot ../data/Local_Watermark/trainAB --semifineSize 0 --quartfineSize 0 --resize_or_crop resize_and_crop --loadSize 16 --fineSize 16 --nThreads 0 --checkpoints_dir checkpoint_UNET_L1_cGAN_PL_NNresize4 --which_model_netG unet_16_nnresize
python test.py --dataroot ../newopenset --resize_or_crop scale_width --nThreads 0 --dataset_mode single --model test --how_many 12000 --checkpoints_dir checkpoint_UNET_L1_cGAN_PL_NNresize4 --which_model_netG unet_16_nnresize --which_epoch 40
pix2pix: Image-to-image translation with conditional adversarial nets