Research map / Vision-language and multimodal / Cross-modal learning
Achieving Text-Based Person Retrieval With Any Granularity
Jialong Zuo, Hanyu Zhou, Dongyue Wu, Yongtai Deng, Mengdan Tan, Nong Sang, Changxin Gao, Xiang Bai
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 orthogonal-expert visual perception (OE-MVP) (ablation) | CMAM baseline · MG-Eval (Separate) | R@1: 44.98 -> 48.48 | Improved |
| add probabilistic cross-identity alignment (PCIA) on OE-MVP (ablation) | CMAM · MG-Eval (Separate) | R@1: 48.48 -> 50.10 | Improved |
| add granularity-consistent reasoning (GCR) on OE-MVP (ablation) | CMAM · MG-Eval (Separate) | R@1: 48.48 -> 49.14 | Improved |
| add GCR on top of OE-MVP and PCIA (ablation) | CMAM · MG-Eval (Separate) | R@1: 50.10 -> 50.40 | Improved |
| add plain MoE without orthogonality or heterogeneity (ablation) | OE-MVP · MG-Eval (Separate) | R@1: 44.98 -> 45.36 | No clear effect |
| add orthogonality loss to MoE experts (ablation) | OE-MVP · MG-Eval (Separate) | R@1: 45.36 -> 47.60 | Improved |
| make experts structurally heterogeneous (ablation) | OE-MVP · MG-Eval (Separate) | R@1: 45.36 -> 46.68 | Improved |
| raise PCIA attribute overlap threshold from 0.3 to 0.6 (ablation) | CMAM · MG-Eval average rank-1 | R@1 average: 48.32 -> 49.50 | Improved |
| raise PCIA attribute overlap threshold from 0.6 to 0.8 (ablation) | CMAM · MG-Eval average rank-1 | R@1 average: 49.50 -> 49.10 | Hurt |
| train IRRA on multi-grained instead of fixed-granularity data (ablation) | IRRA · CUHK-PEDES | R@1: 73.38 -> 67.12 | Hurt |
| train RDE on multi-grained instead of fixed-granularity data (ablation) | RDE · CUHK-PEDES | R@1: 75.94 -> 69.31 | Hurt |
| train CMAM on multi-grained instead of fixed-granularity data (ablation) | CMAM · CUHK-PEDES | R@1: 76.62 -> 77.12 | 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.
- FGCLIP — FG-CLIP: Fine-Grained Visual and Textual Alignment (ICML 2025) · better on 2 of 2 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
- UFineBench: Towards Text-based Person Retrieval with Ultra-fine Granularity (CVPR 2024 · similarity 0.91)
- Chat-based Person Retrieval via Dialogue-Refined Cross-Modal Alignment (CVPR 2025 · similarity 0.87)
- Adaptive Uncertainty-Based Learning for Text-Based Person Retrieval (AAAI 2024 · similarity 0.87)
- Automatic Synthetic Data and Fine-grained Adaptive Feature Alignment for Composed Person Retrieval (NeurIPS 2025 · similarity 0.87)
- Exploring the Potential of Large Vision-Language Models for Unsupervised Text-Based Person Retrieval (AAAI 2025 · similarity 0.86)
- DM-Adapter: Domain-Aware Mixture-of-Adapters for Text-Based Person Retrieval (AAAI 2025 · similarity 0.84)
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.
