## 🎉 UPDATE: Congratulations to Our Winners! 🎉

We're pleased to announce the winners of the First International AG-VPReID 2025 Competition:

### 🥇 **First Place: X-TFCLIP** 
**Universidad Autónoma de Madrid & Pázmány Péter Catholic University**
- 🎯 **Overall Performance:** Rank-1: 71.56%, mAP: 73.60%
- 👥 **Team:** Tamás Endrei, Ivan DeAndres-Tame, Ruben Tolosana, Ruben Vera-Rodriguez, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia

### 🥈 **Second Place: TFCLIP-AG-VPReID**
**Dalian University of Technology & Wuhan University of Technology**
- 🎯 **Overall Performance:** Rank-1: 71.89%, mAP: 72.70%
- 👥 **Team:** Zijing Gong, Yuhao Wang, Xuehu Liu, Pingping Zhang

### 🥉 **Third Place: ACP-VPReID**
**University of Beira Interior**
- 🎯 **Overall Performance:** Rank-1: 67.72%, mAP: 69.34%
- 👥 **Team:** Md Rashidunnabi, Hugo Proença, Kailash A. Hambarde

---

🙏 **Congratulations to our winners**, and thank you to all of those that competed for making the competition such a success! 

📚 Stay tuned for the paper summarizing the results and methods, which will be presented at the **IEEE International Joint Conference on Biometrics (IJCB 2025)** in September.

**Regards,**  
*The AG-VPReID 2025 Challenge Organizing Committee*

---
# AG-VPReID 2025 Competition
## The Second Aerial-Ground Person Re-identification Challenge

Welcome to AG-VPReID 2025, the second installment of the Aerial-Ground Person Re-identification Challenge series! This competition addresses the critical challenge of matching individuals between aerial drone footage and ground-level cameras—a crucial capability for modern surveillance systems, security applications, and public safety operations.

### The Challenge

Person re-identification (ReID) involves matching individuals across non-overlapping camera views. While ground-based ReID has been extensively researched, aerial-ground ReID—especially at high altitudes—remains largely unexplored due to unique challenges:

- Extreme viewpoint differences between aerial (80-120m altitude) and ground cameras
- Significant scale variations and resolution disparities
- Motion blur and occlusions in dynamic aerial footage
- Diverse environmental conditions and illumination changes

### The Dataset

AG-VPReID is a novel large-scale dataset featuring:
- 3,027 identities (including 1,693 distractors)
- 13,507 tracklets with approximately 3.7 million frames
- Video data captured from drones at high altitudes (80-120m)
- Multiple ground camera perspectives (CCTV and wearable cameras)
- 15 soft-biometric attributes annotated for each identity

### Your Goal

Your challenge is to develop robust algorithms for video-based person re-identification between high-altitude aerial footage and ground-level cameras. Participants will train models to accurately match individuals across these drastically different viewpoints, leveraging both spatial and temporal information in the video sequences.

### Evaluation Metrics

Performance will be evaluated using:
- Mean Average Precision (mAP)¹
- Cumulative Matching Characteristics (CMC)² at Rank-1, Rank-5, and Rank-10
- Altitude-specific metrics to assess robustness across different flying heights

### Important Dates
- Competition Start: March 2025
- Competition End: June 3, 2025
- Winners Announcement: By June 23, 2025

This competition is organized in conjunction with the International Joint Conference on Biometrics (IJCB) 2025.

Join us in pushing the boundaries of cross-view person re-identification and developing solutions for this challenging real-world surveillance scenario!

---

¹ Automatically updated on the leaderboard  
² Manually updated through announcements

### Results
Results are updated daily at 10am (AEST) and can be viewed [here](https://www.kaggle.com/competitions/agvpreid25/overview/result).

### Citation

If you use our dataset or code in your research, please cite our paper:

```bibtex
@inproceedings{nguyen2025agvpreid,
  title={AG-VPReID: A Challenging Large-Scale Benchmark for Aerial-Ground Video-based Person Re-Identification},
  author={Nguyen, Huy and Nguyen, Kien and Pemasiri, Akila and Liu, Feng and Sridharan, Sridha and Fookes, Clinton},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2025}
}
```