Research map / Vision-language and multimodal / Cross-modal learning
Tackling Alignment Ambiguity in Person Retrieval through Conversational Attribute Mining
Hao Zou, Runqing Zhang, Jin Ding, Xue Zhou, Jianxiao Zou, Mingzhu Cai
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 BCM bidirectional cross-attention mixer (ablation) | CECA · CUHK-PEDES | Rank-1: 75.26 -> 78.30 | Improved |
| add BCM bidirectional cross-attention mixer (ablation) | CECA · RSTPReid | Rank-1: 69.35 -> 71.40 | Improved |
| add CAWL confidence-aware weighting loss (ablation) | CECA · CUHK-PEDES | Rank-1: 77.66 -> 78.30 | Improved |
| add CAWL confidence-aware weighting loss (ablation) | CECA · RSTPReid | mAP: 49.59 -> 52.07 | Improved |
| add CAM conversational attribute mining to baseline (ablation) | CLIP baseline · ICFG-PEDES | Rank-1: 65.70 -> 69.12 | Improved |
| add CAM conversational attribute mining to baseline (ablation) | CLIP baseline · CUHK-PEDES | mAP: 65.26 -> 64.80 | No clear effect |
| increase dialogue rounds (ablation) | CECA · CUHK-PEDES / ICFG-PEDES / RSTPReid | Rank-1 | Improved |
| swap MLLM from Qwen2-VL-7B to LLaVA-1.5-7B or BLIP-2 (ablation) | CECA · CUHK-PEDES / ICFG-PEDES / RSTPReid | Rank-1: 78.30 -> 78.12 (LLaVA), 77.60 (BLIP-2) on CUHK-PEDES | No clear effect |
| add ReID-domain pre-trained CLIP weights | CECA · CUHK-PEDES | Rank-1: 78.30 -> 79.65 | Improved |
| use CECA instead of RDE in cross-domain transfer | CECA · CUHK-PEDES to ICFG-PEDES | Rank-1: 48.18 -> 47.76 (CUHK to ICFG); 38.11 -> 44.54 (ICFG to CUHK) | Mixed |
Closest papers by meaning
- Chat-based Person Retrieval via Dialogue-Refined Cross-Modal Alignment (CVPR 2025 · similarity 0.91)
- Adaptive Uncertainty-Based Learning for Text-Based Person Retrieval (AAAI 2024 · similarity 0.89)
- Exploring the Potential of Large Vision-Language Models for Unsupervised Text-Based Person Retrieval (AAAI 2025 · similarity 0.88)
- Human-centered Interactive Learning via MLLMs for Text-to-Image Person Re-identification (CVPR 2025 · similarity 0.87)
- Automatic Synthetic Data and Fine-grained Adaptive Feature Alignment for Composed Person Retrieval (NeurIPS 2025 · similarity 0.87)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · 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.
