Research map / Video / Video segmentation and tracking
Hyperbolic Hierarchical Alignment for Video-Based Visible-Infrared Person Re-Identification
Shuang Li, Changjiang Kuang, Jiaxu Leng, Mingpi Tan, Zhanjie Wu, Shuanglin Yan, Xinbo Gao
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 HGI hyperbolic temporal interaction (ablation) | CLIP baseline · HITSZ-VCM I2V | Rank-1 / mAP: 64.4/49.8 -> 69.4/53.2 | Improved |
| add DGF dual-geometry fusion on top of HGI (ablation) | HHSA · HITSZ-VCM I2V | Rank-1 / mAP: 69.4/53.2 -> 72.0/58.3 | Improved |
| add HMA modality centroid alignment (ablation) | HHSA · HITSZ-VCM I2V | Rank-1 / mAP: 72.0/58.3 -> 74.1/59.5 | Improved |
| add HPA prototype alignment (ablation) | HHSA + HMA · HITSZ-VCM I2V | Rank-1 / mAP: 74.1/59.5 -> 76.0/63.2 | Improved |
| make HGI hyperbolic instead of Euclidean (Euclidean-only DGF) (ablation) | HHSA · HITSZ-VCM I2V | Rank-1 / mAP: 72.1/57.1 -> 75.6/62.5 | Improved |
| use dual-geometry DGF instead of Euclidean-only DGF with Euclidean HGI (ablation) | HHSA · HITSZ-VCM I2V | Rank-1: 72.1 vs 71.7 | No clear effect |
| insert HGI every layer instead of every second layer (ablation) | HHSA · HITSZ-VCM I2V | Rank-1 / mAP: 76.0/63.2 -> 73.1/58.6 | Hurt |
| restrict HGI to shallow or deep layers instead of interleaved (ablation) | HHSA · HITSZ-VCM I2V | Rank-1: 76.0 -> 72.1 | Hurt |
| curvature c=0.5 or 2.0 instead of 1.0 (ablation) | HHA · HITSZ-VCM I2V | Rank-1: 76.0 -> 72.2 (c=0.5), 74.7 (c=2.0) | Hurt |
| use HHA instead of Euclidean counterpart (ablation) | HHA · HITSZ-VCM tree-likeness | mean delta max: 0.0627 -> 0.0419 | Improved |
| increase sequence length from 6 to 10 (ablation) | HHA · HITSZ-VCM I2V | Rank-1: 76.0 -> 77.5 | Improved |
Closest papers by meaning
- Hi-CMD: Hierarchical Cross-Modality Disentanglement for Visible-Infrared Person Re-Identification (CVPR 2020 · similarity 0.88)
- Learning by Aligning: Visible-Infrared Person Re-Identification Using Cross-Modal Correspondences (ICCV 2021 · similarity 0.88)
- Dynamic-Static Collaboration for Unsupervised Domain Adaptive Video-Based Visible-Infrared Person Re-Identification (AAAI 2026 · similarity 0.88)
- High-Order Structure Based Middle-Feature Learning for Visible-Infrared Person Re-identification (AAAI 2024 · similarity 0.88)
- Syncretic Modality Collaborative Learning for Visible Infrared Person Re-Identification (ICCV 2021 · similarity 0.88)
- TokenMatcher: Diverse Tokens Matching for Unsupervised Visible-Infrared Person Re-Identification (AAAI 2025 · 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.
