- It is important to note that this work was conducted using the source code of CycleGAN (Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks).
- The code of CycleGAN was shared by Jun-yan Zhu, et al. （http://people.csail.mit.edu/junyanz/）
- More details about this code can be found at https://github.com/junyanz/CycleGAN.
=======================
-
train a model python train.py --dataroot ./datasets/real_lrct2hrct_9to5 --name real_lrct2hrct_9to5_cyclegan --model cycle_gan --load_size 256 --crop_size 256 --batch_size 1 --niter 25 --niter_decay 25 --input_nc 1 --output_nc 1 --gpu_ids 0
-
test the model python test.py --dataroot datasets/real_lrct2hrct_9to5 --name real_lrct2hrct_9to5_cyclegan --model test --no_dropout --input_nc 1 --output_nc 1 --load_size 256 --crop_size 256 --num_test 1 --gpu_ids 0