---
title: Snowleopard reID
emoji: 🐆
colorFrom: red
colorTo: indigo
sdk: gradio
sdk_version: 5.49.1
python_version: 3.11
app_file: app.py
pinned: false
short_description: Snow Leopard reID using AI
---

# Snow Leopard Re-Identification

AI-powered snow leopard re-identification system using computer vision and deep learning.

## Features

- **Single Image Matching**: Upload an image to identify individual snow leopards against a catalog
- **Batch Processing**: Process multiple images at once with filtering options
- **Multiple Detection Methods**: YOLO or Grounding DINO + SAM HQ for segmentation
- **Multiple Matching Algorithms**: SIFT, SuperPoint, DISK, or ALIKED feature extractors

## Local Development

### Prerequisites

- Python 3.11+
- [uv](https://github.com/astral-sh/uv) package manager

### Setup

```bash
# Install dependencies
make install

# Run the app locally
make run
```

The app will be available at `http://localhost:7860`.

## Data

- **Catalog**: Pre-computed features for known snow leopards (stored with Git LFS)
- **SAM HQ Model**: Downloaded automatically at runtime from HuggingFace Hub

## Tech Stack

- Gradio for the web interface
- PyTorch for deep learning
- Grounding DINO for zero-shot object detection
- SAM HQ (Segment Anything Model High Quality) for segmentation
- LightGlue for feature matching
- Wasserstein distance for match scoring