Original dataset is available : https://www.kaggle.com/datasets/andrewmvd/ct-low-dose-reconstruction?resource=download
Here We choose the preprocessed images at : https://share.weiyun.com/tWTsIAde passwd: dry8pe
all in environment.yaml
CUDA_VISIBLE_DEVICES=3 python main.py \
--data_root=./CT_Reconstruction_256x256_1m \
--print_freq=10 --save_freq=100 --img_size=256 \
--epochs=100 --lr=1e-4 --num_workers=32 --device=cuda \
--model_name=unet4 --batch=16 --run_name=noflip_L2_0427