PolypSegmentation

Introduction

Built based on ACSNet. I have extended the code to support more models and make it easier to use. On top of the original code, I added some function: progress bar display, log logging, multi-scale training for all networks, etc.

Currently supported models:

• Unet
• ACSNet
• PraNet

Requirements

• torch
• torchvision
• tqdm
• opencv
• scipy
• skimage
• PIL
• numpy

Usage

1. Training

# To train PraNet, for example.
python train.py  --config pranet --exp_name pranet --gpu_ids 0
# You can resume traning by:
python train.py  --config pranet --continue_train

More network and training configurations need to be modified in configs.

2. Inference

python test.py  --config pranet