Research map / Label-efficient and robust learning / Recognition and long-tail
Counterfactual Intervention Feature Transfer for Visible-Infrared Person Re-identification
Xulin Li, Yan Lu, Bin Liu, Yating Liu, Guojun Yin, Qi Chu, Jinyang Huang, Feng Zhu, Rui Zhao, Nenghai Yu
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 Graph Feature Transfer (GFT) (ablation) | baseline backbone · SYSU-MM01 all-search single-shot | mAP: 66.58 -> 72.12 | Improved |
| add Graph Feature Transfer (GFT) (ablation) | baseline backbone · SYSU-MM01 all-search single-shot | rank-1: 70.49 -> 72.01 | Improved |
| add unbalanced scenario simulation (UBS) to GFT (ablation) | GFT · SYSU-MM01 all-search single-shot | mAP: 72.12 -> 71.97 | Hurt |
| add unbalanced scenario simulation (UBS) to GFT (ablation) | GFT · SYSU-MM01 all-search single-shot | rank-1: 72.01 -> 72.90 | Improved |
| add homogeneous/heterogeneous graph module (H2G) to GFT+UBS (ablation) | GFT+UBS · SYSU-MM01 all-search single-shot | mAP: 71.97 -> 73.79 | Improved |
| add homogeneous/heterogeneous graph module (H2G) to GFT+UBS (ablation) | GFT+UBS · SYSU-MM01 all-search single-shot | rank-1: 72.90 -> 72.29 | Hurt |
| add Counterfactual Relation Intervention (CRI) (ablation) | H2FT · SYSU-MM01 all-search single-shot | rank-1: 72.29 -> 74.08 | Improved |
| add Counterfactual Relation Intervention (CRI) (ablation) | H2FT · SYSU-MM01 all-search single-shot | mAP: 73.79 -> 74.79 | Improved |
| add CRI (ablation) | CIFT · SYSU-MM01 all-search single-shot | affinity error ratio (%): 5.16 -> 3.90 | Improved |
| add CRI (ablation) | CIFT · SYSU-MM01 indoor-search multi-shot | affinity error ratio (%): 6.15 -> 4.66 | Improved |
| use transferred graph features instead of backbone features at inference | CIFT · SYSU-MM01 all-search single-shot | rank-1: 71.77 -> 74.08 | Improved |
| use CIFT instead of GNN rerank on the HCT backbone | HCT backbone · SYSU-MM01 all-search single-shot | rank-1: 60.52 -> 66.18 | Improved |
Closest papers by meaning
- Prototype Topology Consistency for Visible-Infrared Lifelong Person Re-Identification (NeurIPS 2026 · similarity 0.88)
- Joint Color-irrelevant Consistency Learning and Identity-aware Modality Adaptation for Visible-infrared Cross Modality Person Re-identification (AAAI 2021 · similarity 0.88)
- Unsupervised Visible-Infrared Person Re-identification under Unpaired Settings (ICCV 2025 · similarity 0.87)
- Dual-Level Matching With Outlier Filtering for Unsupervised Visible-Infrared Person Re-Identification (TPAMI 2025 · similarity 0.87)
- High-Order Structure Based Middle-Feature Learning for Visible-Infrared Person Re-identification (AAAI 2024 · similarity 0.87)
- Unsupervised Visible-Infrared Person Re-Identification via Progressive Graph Matching and Alternate Learning (CVPR 2023 · 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.
