Retina Vessel Segmentation Project
This repository contains the code and resources for a deep learning-based approach to segment retinal vessels using the RAVIR dataset.
This study focuses on the early detection of retinal vascular changes, employing a U-Net variant named "RetinaSegmentor." The model achieved a Mean Intersection over Union (mIoU) of 0.72 and a Dice coefficient of 0.823 on the RAVIR dataset.
- Install dependencies:
pip install -r code/requirements.txt - Download the RAVIR dataset and place it in the
data/RAVIR/directory.
- Use the jubeter notebook in 'code/RetinaSegmentation.ipynb' to train and test the model.
Our model demonstrated exceptional performance, with a high mIoU and Dice coefficient, showcasing its potential for early detection of retinal vascular changes.
This project is licensed under the MIT License.
For questions or feedback, contact us at youssefma27@gmail.com.