README.md

Dual-path multi-scale context dense aggregation network for retinal vessel segmentation

This is an implementation of the MCDAU-Net.Due to memory limitations on GitHub, each person needs to perform the chunking operation themselves. For detailed instructions on the chunking process, please refer to the paper.

Environment Configuration：

* Python3.9
* Pytorch1.10
* Best trained with GPUs

File structure：

  ├── core.models: Build MCDAU-Net Model Code
  ├── core.utils: Read the dataset and calculate the mean and standard deviation
  ├── core.blocks: Build the MCDAU-Net module code
  ├── main.py: Model training
  ├── image2patch: Split the image into patches and perform preprocessing operations
  ├── patch2iamge.py: Merge patch into image and perform test operation
  ├── test: Storage location of merged images
  └── test.py: Model testing and save results