templeblock / tip_mumfordshahloss

Paper

• Mumford–Shah Loss Functional for Image Segmentation With Deep Learning
  • Authors: Boah Kim and Jong Chul Ye
  • published in IEEE Transactions on Image Processing (TIP)

Implementation

A PyTorch implementation of deep-learning-based segmentation based on original cycleGAN code. [https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix] (*Thanks for Jun-Yan Zhu and Taesung Park, and Tongzhou Wang.)

• Requirements
  • Python 2.7
  • PyTorch 1.1.0

Main

• Training: LiTS_train_unet.py which is handled by scripts/LiTS_train_unet.sh
• A code for Mumford-Shah loss functional is in models/loss.py.
  • 'levelsetLoss' and 'gradientLoss2d' classes compose our Mumford-Shah loss function.