AG-VPReID 2025: Aerial-Ground Video-based Person Re-identification Challenge Results
This addresses the problem of cross-viewpoint person tracking for surveillance and public safety applications, building incrementally on prior challenges.
The paper introduced the AG-VPReID 2025 Challenge, the first large-scale video-based competition for person re-identification across aerial and ground viewpoints, using a new dataset with 3,027 identities and 3.7 million frames. The winning method achieved 72.28% Rank-1 accuracy in aerial-to-ground ReID and 70.77% in ground-to-aerial ReID, surpassing existing baselines.
Person re-identification (ReID) across aerial and ground vantage points has become crucial for large-scale surveillance and public safety applications. Although significant progress has been made in ground-only scenarios, bridging the aerial-ground domain gap remains a formidable challenge due to extreme viewpoint differences, scale variations, and occlusions. Building upon the achievements of the AG-ReID 2023 Challenge, this paper introduces the AG-VPReID 2025 Challenge - the first large-scale video-based competition focused on high-altitude (80-120m) aerial-ground ReID. Constructed on the new AG-VPReID dataset with 3,027 identities, over 13,500 tracklets, and approximately 3.7 million frames captured from UAVs, CCTV, and wearable cameras, the challenge featured four international teams. These teams developed solutions ranging from multi-stream architectures to transformer-based temporal reasoning and physics-informed modeling. The leading approach, X-TFCLIP from UAM, attained 72.28% Rank-1 accuracy in the aerial-to-ground ReID setting and 70.77% in the ground-to-aerial ReID setting, surpassing existing baselines while highlighting the dataset's complexity. For additional details, please refer to the official website at https://agvpreid25.github.io.