Research map / Label-efficient and robust learning / Recognition and long-tail
Dynamic Dual-Attentive Aggregation Learning for Visible-Infrared Person Re-Identification
Mang Ye, Jianbing Shen, David J. Crandall, Ling Shao, Jiebo Luo
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 intra-modality weighted-part attention P (ablation) | two-stream baseline · SYSU-MM01 all-search | Rank-1 (%): 48.18 -> 53.69 | Improved |
| add intra-modality weighted-part attention P (ablation) | two-stream baseline · SYSU-MM01 indoor-search | Rank-1 (%): 49.52 -> 58.08 | Improved |
| add cross-modality graph attention G (ablation) | two-stream baseline · SYSU-MM01 all-search | Rank-1 (%): 48.18 -> 50.75 | Improved |
| add cross-modality graph attention G (ablation) | two-stream baseline · SYSU-MM01 indoor-search | mAP (%): 58.12 -> 62.26 | Improved |
| add G on top of P with dynamic dual aggregation (ablation) | B+P · SYSU-MM01 all-search | mAP (%): 51.37 -> 53.02 | Improved |
| add G on top of P with dynamic dual aggregation (ablation) | B+P · SYSU-MM01 indoor-search | Rank-1 (%): 58.08 -> 61.02 | Improved |
| use learnable weighted part aggregation instead of average (ablation) | part attention (Res) · SYSU-MM01 all-search | Rank-1 (%): 48.34 -> 51.06 | Improved |
| use learnable weighted part aggregation instead of concatenation (ablation) | part attention (Res) · SYSU-MM01 all-search | Rank-1 (%): 50.34 -> 51.06 | Improved |
| use residual BN instead of plain residual connection (ablation) | weighted part attention · SYSU-MM01 all-search | Rank-1 (%): 51.06 -> 53.69 | Improved |
| increase number of selected graph neighbors Ng from 0 to 8 (ablation) | B+G · SYSU-MM01 all-search | Rank-1 (%): 48.18 -> 50.75 | Improved |
| increase number of body parts p beyond 3 (ablation) | DDAG · SYSU-MM01 all-search | accuracy | Mixed |
| increase graph attention heads L beyond 4 (ablation) | DDAG · SYSU-MM01 all-search | accuracy | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HSME — HSME: Hypersphere Manifold Embedding for Visible Thermal Person Re-Identification (AAAI 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Modality Synergy Complement Learning with Cascaded Aggregation for Visible-Infrared Person Re-identification (ECCV 2022 · similarity 0.91)
- Implicit Discriminative Knowledge Learning for Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.91)
- Empowering Visible-Infrared Person Re-Identification with Large Foundation Models (NeurIPS 2024 · similarity 0.90)
- Towards Grand Unified Representation Learning for Unsupervised Visible-Infrared Person Re-Identification (ICCV 2023 · similarity 0.90)
- Syncretic Modality Collaborative Learning for Visible Infrared Person Re-Identification (ICCV 2021 · similarity 0.90)
- Shallow-Deep Collaborative Learning for Unsupervised Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.88)
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.
