TMI_SwitchableCycleGAN

Paper

Continuous Conversion of CT Kernel using Switchable CycleGAN with AdaIN (arXiv.org, Serin Yang et al.)

Code

We implemented this code based on the original CycleGAN codes.

Sample Data

The sample data is from Low Dose CT Grand Challenge. The data should be located on './dataset' and './dataset_M' for 2 domain and 3 domain, respectively.

Model

The trained model weights can be downloaded here. The model should be located on './result/checkpoints/'

Train

You can use train.py for 2 domain Switchable CycleGAN, and train_M.py for 3 domain Switchable CycleGAN.

Test

You can use inference.py for 2 domain Switchable CycleGAN, and inference_M.py for 3 domain Switchable CycleGAN.