Research map / Vision-language and multimodal / Cross-modal learning
VRCLIP: Multimodal Canonical Correlation Alignment for CLIP-Driven Vision-Radio Person Re-Identification
Rui Zhang, Yaqi Wang, Yadong Li, Ruixu Geng, Jianyang Wang, Qijun Ying, Dongheng Zhang, Yang Hu, Yan Chen
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 the RF modality to the RGB branch (ablation) | VRCLIP · VRR | mAP: 78.0 -> 94.1 | Improved |
| add the vision modality to the RF branch (ablation) | VRCLIP · VRR | mAP: 76.4 -> 94.1 | Improved |
| add LoRA adapters plus lighting-aware margin loss to the CLIP encoder (ablation) | VRCLIP · VRR | mAP: 88.8 -> 94.1 | Improved |
| use CCA alignment loss instead of InfoNCE (ablation) | VRCLIP · VRR | mAP: 87.3 -> 94.1 (Table 3); text says 7.8% drop | Improved |
| use CCA alignment loss instead of InfoNCE (ablation) | VRCLIP · VRR localization | mean absolute error (m): 0.236 -> 0.205 | Improved |
| add RF-anchored reliability gating instead of direct concatenation (ablation) | VRCLIP · VRR | mAP: 84.1 -> 94.1 (Table 3); text says 11.9% loss | Improved |
| fuse RF with RGB (VRCLIP) instead of RGB-only CAL | person ReID · VRR indoor occlusion | mAP: 8.4 -> 94.1 | Improved |
| use VRCLIP fusion instead of RF-only RFMamba | person ReID · VRR | mAP: 93.0 -> 93.9 | Improved |
| use VRCLIP instead of generic multimodal fusion baselines DeMo and MambaPro | RGB-RF person ReID · VRR | mAP: 78.8 -> 93.9 (DeMo), 75.9 -> 93.9 (MambaPro) | Improved |
| use VRCLIP instead of RF-only RFMamba | person ReID · THP dataset | mAP: 97.6 -> 98.7 | Improved |
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.
- MambaPro — MambaPro: Multi-Modal Object Re-identification with Mamba Aggregation and Synergistic Prompt (AAAI 2025) · better on 3 of 3 shared results
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
- Blurring Modal Boundaries: A Unified Survey from Single- to Multi-modal Person Re-ldentification (IJCV 2026 · similarity 0.89)
- Learning by Aligning: Visible-Infrared Person Re-Identification Using Cross-Modal Correspondences (ICCV 2021 · similarity 0.88)
- Unsupervised Visible-Infrared Person Re-identification under Unpaired Settings (ICCV 2025 · similarity 0.88)
- Syncretic Modality Collaborative Learning for Visible Infrared Person Re-Identification (ICCV 2021 · similarity 0.88)
- Shallow-Deep Collaborative Learning for Unsupervised Visible-Infrared Person Re-Identification (CVPR 2024 · similarity 0.87)
- TokenMatcher: Diverse Tokens Matching for Unsupervised Visible-Infrared Person Re-Identification (AAAI 2025 · similarity 0.86)
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.
