/CycleGAN-for-medical-image-enhancement

learning code based on CycleGAN-medical-image-segmentation

Primary LanguagePython

CycleGAN

The framework of network is based on the paper below: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks.

Some changes are taken during learning, and matlab module folder is increased

Prerequisites

  • tensorflow r1.1 or tensorflow 2.6
  • numpy 1.11.0
  • scipy 0.17.0
  • pillow 3.3.0

Train procession

python main.py --dataset_dir=med_image

Test procession

python main.py --dataset_dir=med-image --phase=test --which_direction=AtoB

References