Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Concordant Attention via Target-aware Alignment for Visible-Infrared Person Re-identification
Jianbing Wu, Hong Liu, Yuxin Su, Wei Shi, Hao Tang
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 target-aware concordant alignment (TCA) to the ResNet-50 baseline (ablation) | CAL · SYSU-MM01 all-search single-shot | Rank-1: 57.82 -> 70.15 | Improved |
| add MatchDistill on top of TCA (ablation) | CAL · SYSU-MM01 all-search single-shot | Rank-1: 70.15 -> 74.66 | Improved |
| add diversity constraint to DRM (ablation) | DRM · SYSU-MM01 all-search single-shot | Rank-1: 72.76 -> 74.66 | Improved |
| use modality-specific DRM instead of shared DRM (ablation) | DRM · SYSU-MM01 all-search single-shot | Rank-1: 72.57 -> 74.66 | Improved |
| use learned DRM regions instead of horizontal splitting or learnable vectors (ablation) | CAL · SYSU-MM01 all-search single-shot | Rank-1: 73.15 -> 74.66 vs HS(n=4); 71.07 -> 74.66 vs learnable vectors | Improved |
| use TSAA instance-norm similarity instead of scaled dot product (ablation) | TSAA · SYSU-MM01 all-search single-shot | Rank-1: 72.57 -> 74.66 | Improved |
| add query-guided channel attention in TSAA (ablation) | TSAA · SYSU-MM01 all-search single-shot | Rank-1: 70.57 -> 72.57 | Improved |
| add part-aligned center loss (PCL) (ablation) | CAL · SYSU-MM01 all-search single-shot | Rank-1: 66.94 -> 74.66 | Improved |
| use PCL instead of triplet, center or ICA&CCA losses (ablation) | CAL · SYSU-MM01 all-search single-shot | Rank-1: 71.48 -> 74.66 vs center loss; 68.09 -> 74.66 vs triplet | Improved |
| add cross-modal query matching (CQM) to feature and logit distillation (ablation) | MatchDistill · SYSU-MM01 all-search single-shot | Rank-1: 73.63 -> 74.66 | Improved |
| CAL versus CMT on multi-shot and RegDB visible-to-infrared Rank-1 | CAL · SYSU-MM01 all-search multi-shot / RegDB V2I | Rank-1: 77.05 vs 80.23 (SYSU multi-shot); 94.51 vs 95.17 (RegDB V2I) | Hurt |
Closest papers by meaning
- Multi-Memory Matching for Unsupervised Visible-Infrared Person Re-Identification (ECCV 2024 · similarity 0.91)
- Empowering Visible-Infrared Person Re-Identification with Large Foundation Models (NeurIPS 2024 · similarity 0.90)
- Visible-Infrared Person Re-Identification via Semantic Alignment and Affinity Inference (ICCV 2023 · similarity 0.90)
- Implicit Discriminative Knowledge Learning for Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.90)
- TokenMatcher: Diverse Tokens Matching for Unsupervised Visible-Infrared Person Re-Identification (AAAI 2025 · similarity 0.89)
- Shallow-Deep Collaborative Learning for Unsupervised Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.89)
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.
