Awesome-Aerial-Ground-Object-Re-Identification

Welcome to the Awesome-AGReID Repository! This repository curates cutting-edge methods and benchmarks dedicated to Aerial–Ground Object Re-Identification (AG-ReID).

Our Papers

• [SD-ReID] SD-ReID: View-aware Stable Diffusion for Aerial-Ground Person Re-Identification Paper (arXiv)

• [LATeX] LATeX: Leveraging Attribute-based Text Knowledge for Aerial–Ground Person Re-Identification Paper

Methods

• [Drones2025]
  Unsupervised aerial-ground person to group re-identification for UAV surveillance.
  Paper

• [IJCB2025-AG-VPReID-VIR]
  Aerial-ground visible-infrared video person ReID dataset.
  Paper

• [RS2025-AG-VehicleReID]
  Aerial-ground cross-view vehicle ReID dataset and baseline.
  Paper

• [Sensors2025-CVNet]
  Lightweight cross-view vehicle ReID with multi-scale localization.
  Paper

• [SPL2025-3D-HumanReID]
  Omni-directional view person ReID through 3D human reconstruction.
  Paper

• [CVPR2025-AG-VPReID-Net] AG-VPReID-Net: Video-based Aerial–Ground Re-ID framework Paper (arXiv, CVPR 2025)

• [ArXiv2025-MP-ReID] Multi-modal Multi-platform Person Re-Identification Paper

• [CVPR2025-SeCap]
  SeCap: Self-Calibrating and Adaptive Person Re-Identification in Aerial-Ground Networks
  Paper

• [ICME2025-DTS]
  Dynamic Token Selective Transformer for Aerial-Ground Person Re-identification
  Paper

• [ECCV2024-CrossPlatform-ReID]
  Cross-platform video person ReID dataset and adaptation method.
  Paper

• [CVPR2024-VDT] View-decoupled Transformer (VDT) for AGPReID Paper (arXiv)

• [ICME2023-AG-ReID] AG-ReID (v1): Aerial–Ground Person Re-Identification dataset and baseline Paper & Dataset (arXiv, GitHub)

Datasets

• DetReIDX (2025) – Large-scale stress-test AG-ReID dataset with 13M+ bounding boxes, 509 IDs across 7 campuses. Dataset & Paper (arXiv)

• AG-VPReID (2025) – Video-based AG-ReID dataset: 6,632 IDs, 32,321 tracklets, 9.6M frames. Dataset & Paper (arXiv, CVF开放访问)

• AG-ReID.v2 (2024) – Large-scale image-based AG-ReID: 1,615 IDs, 100,502 images, rich attribute annotations. Dataset & GitHub (arXiv, GitHub)

• AG-ReID.v1 (2023) – First AG-ReID dataset: 388 IDs, 21,983 images, with attribute-level annotations. Dataset & GitHub (arXiv, GitHub)

• CARGO (2024) – Synthetic large-scale AGPReID: 5,000 IDs, 108,563 images from 13 cameras (5 aerial, 8 ground). Dataset & Paper (GitHub)

Challenges & Benchmarks

• IJCB2025-AG-VPReID Challenge – Video-based Re-ID challenge (AG-VPReID). [Landing & data via Kaggle] (AG-VPReID 2025)

• IJCB2023-AG-ReID Challenge – AG-ReID image-based challenge results. Details (西维吉尼亚大学视觉中心)

Contact

Feel free to reach out if you have any questions, suggestions, or collaboration proposals:

• Email: 924973292@mail.dlut.edu.cn
• Web: 924973292.github.io

Star History

Acknowledgments

I want to express my gratitude to the academic community and everyone contributing to the advancement of AGReID research.

Citation

If you find our work useful in your research, please consider citing our papers:

```bibtex @article{hu2025sdreid, title={SD-ReID: View-aware Stable Diffusion for Aerial-Ground Person Re-Identification}, author={Hu, Xiang and Zhang, Pingping and Wang, Yuhao and Yan, Bin and Lu, Huchuan}, journal={arXiv preprint arXiv:2504.09549}, year={2025} }

@article{hu2025latex, title={LATeX: Leveraging Attribute-based Text Knowledge for Aerial-Ground Person Re-Identification}, author={Hu, Xiang and Wang, Yuhao and Zhang, Pingping and Lu, Huchuan}, journal={arXiv preprint arXiv:2503.23722}, year={2025} }

@article{nguyen2025ag, title={AG-VPReID 2025: Aerial-Ground Video-based Person Re-identification Challenge Results}, author={Nguyen, Kien and Fookes, Clinton and Sridharan, Sridha and Nguyen, Huy and Liu, Feng and Liu, Xiaoming and Ross, Arun and Michalski, Dana and Endrei, Tam{\'a}s and DeAndres-Tame, Ivan and others}, journal={arXiv preprint arXiv:2506.22843}, year={2025} } ```