README.md

---
title: CoWTracker
emoji: 🐮
colorFrom: green
colorTo: yellow
sdk: gradio
sdk_version: 6.2.0
app_file: app.py
pinned: false
license: cc-by-nc-4.0
suggested_hardware: a10g-small
short_description: Dense Point Tracking with Cost-Volume Free Warping
---

# 🐮 CoWTracker

**Cost-Volume Free Warping-Based Dense Point Tracking**

Zihang Lai<sup>1,2</sup>, Eldar Insafutdinov<sup>1</sup>, Edgar Sucar<sup>1</sup>, Andrea Vedaldi<sup>1,2</sup>

<sup>1</sup>Visual Geometry Group, University of Oxford &nbsp;&nbsp; <sup>2</sup>Meta AI

---

Upload a video and CoWTracker will track every pixel through time, visualizing the motion with colorful trajectories.

## Features

- **Dense Tracking**: Track all pixels simultaneously
- **Bidirectional**: Track forwards and backwards from any query frame
- **Interactive**: Choose query frame and visualization settings
- **Fast**: Efficient warping-based architecture

## Links

- [Project Page](https://cowtracker.github.io/)
- [GitHub](https://github.com/facebookresearch/cowtracker/)
- [Paper](#)

## Usage

1. Upload a video (or select an example)
2. Click "Process Video"
3. Select query frame using the slider
4. Click "Start Tracking"
5. Adjust visualization settings as needed

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference