Research map / Label-efficient and robust learning / Recognition and long-tail
Cross-Modality Transformer for Visible-Infrared Person Re-identification
Kongzhu Jiang, Tianzhu Zhang, Xiang Liu, Bingqiao Qian, Yongdong Zhang, Feng Wu
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 modality embeddings (ME) (ablation) | ViT baseline · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 54.28 -> 57.40; mAP +2.19 | Improved |
| add modality-aware center loss (MAC) on top of ME (ablation) | BASE+ME · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 57.40 -> 59.53; mAP +3.24 | Improved |
| add modality-aware ID loss (MAID) on top of ME (ablation) | BASE+ME · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 57.40 -> 60.53; mAP +2.57 | Improved |
| combine MAC and MAID as MAE loss (ablation) | BASE+ME · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 57.40 -> 62.58; +5.18% Rank-1, +5.17% mAP | Improved |
| decrease patch stride from 16 to 8 (overlapping patches) (ablation) | CMTR · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 57.38 -> 62.58 | Improved |
| use MAE loss instead of center loss (ablation) | BASE+ME · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 58.09 -> 62.58 | Improved |
| use MAE loss instead of hetero-center loss (ablation) | BASE+ME · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 58.21 -> 62.58 | Improved |
| use fully connected mapping instead of identity mapping for modality removal (ablation) | MAE loss · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 55.59 -> 62.58 | Improved |
| use cosine distance instead of L1 (ablation) | MAE loss · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 57.65 -> 62.58 | Improved |
| use cosine distance instead of Smooth-L1 (ablation) | MAE loss · SYSU-MM01 all-search single-shot (ablation) | Rank-1: 59.89 -> 62.58 | Improved |
| vary MAE weight lambda from 0 to 1-6 (ablation) | CMTR · SYSU-MM01 all-search single-shot (ablation) | Rank-1 | No clear effect |
Closest papers by meaning
- Modality Unifying Network for Visible-Infrared Person Re-Identification (ICCV 2023 · similarity 0.94)
- CM-NAS: Cross-Modality Neural Architecture Search for Visible-Infrared Person Re-Identification (ICCV 2021 · similarity 0.94)
- Cross-Modality Person Re-identification with Memory-Based Contrastive Embedding (AAAI 2023 · similarity 0.94)
- Shallow-Deep Collaborative Learning for Unsupervised Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.92)
- Implicit Discriminative Knowledge Learning for Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.91)
- TokenMatcher: Diverse Tokens Matching for Unsupervised Visible-Infrared Person Re-Identification (AAAI 2025 · similarity 0.91)
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.
