code for "DedustGAN: Unpaired learning for image dedusting based on Retinex with GANs"(Expert Systems with Applications).
The indoor synthetic dusty dataset we used is derived from Jiayan Huang's SIDNet: A single image dedusting network with color cast correction, you can download it by the link(c0kq).
We provide pre-trained weights on synthetic dusty dataset SID and unpaired real dusty dataset.link(54uh).We also provide some samples testd with real pre-trained weight.link(mr9f)
python train.py --dataroot [dataset_path] --dataset_mode unpaired --model Dedust --name [experiment_name] --niter 30 --niter_decay 60 --lr_decay_iters 10 --preprocess scale_min_and_crop --load_size 300 --crop_size 256 --nu
m_threads 0 --save_epoch_freq 3
python test.py --dataroot ./datasets/test --dataset_mode single --name [experiment_name] --model Dedust --phase test --preprocess none --save_image --method_name [method_name] --epoch 60
