https://github.com/thoth000/shape-aware-refinement
[JSAI2025] Research code for shape-aware refinement in segmentation using PDEs.
https://github.com/thoth000/shape-aware-refinement
anisotropic-diffusion pytorch segmentation
Last synced: 29 days ago
JSON representation
[JSAI2025] Research code for shape-aware refinement in segmentation using PDEs.
- Host: GitHub
- URL: https://github.com/thoth000/shape-aware-refinement
- Owner: thoth000
- Created: 2025-03-30T03:05:48.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-05-17T17:46:48.000Z (about 1 year ago)
- Last Synced: 2025-07-23T07:31:44.498Z (11 months ago)
- Topics: anisotropic-diffusion, pytorch, segmentation
- Language: Python
- Homepage:
- Size: 3.54 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Shape-Aware Refinement
This research proposes a **PDE-based trainable refinement module** for **tubular-structure segmentation** to enhance shape continuity.
## Paper / Slides
- [Presentation Slides](https://github.com/thoth000/Shape-Aware-Refinement/blob/main/docs/presentation_ja.pdf)
- [JSAI 2025 Abstract (Japanese)](https://github.com/thoth000/Shape-Aware-Refinement/blob/main/docs/jsai_abstract.md)
## Overview
- PDE-based refinement improves mask continuity **while maintaining thin shape**
- Our approach is applied to **various architectures** for tubular-structure segmentation because of **post-process**
## Dataset
- [DRIVE dataset](https://github.com/zhengyuan-liu/Retinal-Vessel-Segmentation/tree/master/DRIVE)
## Environment
- Ubuntu 24.04.2 LTS
- NVIDIA TITAN RTX(Memory: 24GB) * 4
## How to run
1. setup the environment.
```bash
$ pip install -r requirements.txt
```
2. Download DRIVE dataset to `/dataset`.
3. Run `/dataset/data_process.py` to fit dataset for training models.
```bash
$ cd dataset
$ python data_process.py -dp DATASET_PATH -dn DATASET_NAME
```
4. train and test model.
```bash
$ cd pde-shape-refiner
$ main_ddp.sh
```