Research map / Label-efficient and robust learning / Recognition and long-tail
WHU-MARS: A Multispectral Aerial-Ground Benchmark Towards Any-Scenario Person Re-Identification
Yuxuan Zhao, Zhongao Zhou, Bin Yang, He Li, Jian Liang, Jun Chen, Bo Du, Mang Ye
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Progressive Center Alignment (ProCA) to the baseline (ablation) | UAD baseline · WHU-MARS-1000 AS-ReID | mAP: 8.9 -> 9.3 | Improved |
| add Global Prototype Discrimination (GPD) to the baseline (ablation) | UAD baseline · WHU-MARS-1000 AS-ReID | mAP: 8.9 -> 10.9 | Improved |
| add ProCA on top of GPD (ablation) | UAD · WHU-MARS-1000 AS-ReID | mAP: 10.9 -> 11.0 | Improved |
| add ProCA on top of GPD (ablation) | UAD · WHU-MARS-1000 AS-ReID | Rank-1: 28.7 -> 29.5 | Improved |
| use UAD single unified model versus CLIP-ReID | UAD · WHU-MARS-1000 AS-ReID | Rank-1: 26.6 -> 29.5 | Improved |
| restrict evaluation to ground-daytime subset | UAD · WHU-MARS-1000-GD vs WHU-MARS-1000 | mAP: 11.0 -> 14.5 | Improved |
| scale benchmark from 1000 to 2337 identities | UAD · WHU-MARS-2337 vs WHU-MARS-1000 | mAP: 11.0 -> 9.6 | Hurt |
| use one unified UAD model instead of VI-specific methods on NIR->RGB | UAD · WHU-MARS-1000 VI-ReID (NIR->RGB) | Rank-1: 27.63 -> 25.53 | Hurt |
| use one unified UAD model instead of VI-specific methods on TIR->RGB | UAD · WHU-MARS-1000 VI-ReID (TIR->RGB) | Rank-1: 8.29 -> 9.04 | Improved |
| use UAD instead of AG-specific methods for aerial-to-ground | UAD · WHU-MARS-1000 AG-ReID (A->G) | mAP: 10.1 -> 11.5 | Improved |
Closest papers by meaning
- MCVL: Multi-Space Cross-View Learning for Aerial-Ground Person Re-Identification (ECCV 2026 · similarity 0.90)
- AG-VPReID: A Challenging Large-Scale Benchmark for Aerial-Ground Video-based Person Re-Identification (CVPR 2025 · similarity 0.89)
- SeCap: Self-Calibrating and Adaptive Prompts for Cross-view Person Re-Identification in Aerial-Ground Networks (CVPR 2025 · similarity 0.89)
- GSAlign: Geometric and Semantic Alignment Network for Aerial-Ground Person Re-Identification (NeurIPS 2025 · similarity 0.89)
- View-decoupled Transformer for Person Re-identification under Aerial-ground Camera Network (CVPR 2024 · similarity 0.88)
- Learning Commonality, Divergence and Variety for Unsupervised Visible-Infrared Person Re-identification (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
