elcaiseri / Vision-Transformer-Segmentation

Implement ViT Segmentation PyTorch for HuBMAP kaggle competition.

Vision Transformer Segmentation

This repository applies the Vision Transformer (ViT) in PyTorch to identify glomeruli in human kidney tissue images for the HuBMAP Kaggle competition.

Table of Contents

• About
• Datasets and Weights
• Evaluation
• Score
• Usage
• Training and Inference Notebooks
• Contribution and Upvote

About

The Vision Transformer Segmentation project implements ViT in PyTorch for the HuBMAP Kaggle competition. The goal is to identify glomeruli in human kidney tissue images using the power of transformers in computer vision tasks.

Datasets and Weights

• Training Dataset: hubmap-pytorch-vit-segmentation-train-datasets
• Inference Dataset: hubmap-pytorch-vit-segmentation-inference-datasets

Evaluation

This competition is evaluated on the mean Dice coefficient, which compares the pixel-wise agreement between a predicted segmentation and its corresponding ground truth.

LeadBoard-Score

• Private Score: 0.9354
• Public Score: 0.9064

Usage

To use this Vision Transformer Segmentation project, follow these steps:

1. Clone the repository: git clone https://github.com/elcaiseri/Vision-Transformer-Segmentation.git
2. Explore the implementation and code for ViT in PyTorch.

Training and Inference Notebooks

• Training Notebook
• Inference Notebook

Contribution and Upvote

If you find this project helpful or interesting, consider giving it a star! ⭐

© 2023 elcaiseri