Research map / Vision-language and multimodal / Human and embodied modeling
Quota-Calibrated Fine-Grained Alignment with Context-Aware Marginals for Text-based Person Retrieval
Dongsheng Li, Xinyuan Guo, Huijie Zhang, Pingting Hao, Qiushi Xia
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 QC-Align training regularizer | Baseline-CLIP · CUHK-PEDES | Rank-1: 68.71 -> 70.89 | Improved |
| add QC-Align training regularizer | Baseline (ResNet-50) · RSTPReid | Rank-1: 38.75 -> 43.35 | Improved |
| add QC-Align on top of IRRA | IRRA · CUHK-PEDES | Rank-1: 73.38 -> 74.62 | Improved |
| add QC-Align on top of CADA | CADA · CUHK-PEDES | Rank-1: 78.37 -> 79.31 | Improved |
| add SCAN local alignment (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 68.71 -> 67.79 | Hurt |
| add SCAN local alignment (ablation) | Baseline (ResNet-50) · CUHK-PEDES | Rank-1: 57.19 -> 58.69 | Improved |
| add standard OT (uniform marginals) (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 68.71 -> 67.02 | Hurt |
| add unbalanced OT (UOT) (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 68.71 -> 67.67 | Hurt |
| add MLM auxiliary loss (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 68.71 -> 69.46 | Improved |
| add Sinkhorn divergence to standard OT (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 67.02 -> 69.31 | Improved |
| add CAME non-uniform marginals (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 67.02 -> 69.49 | Improved |
| combine SD and CAME (full QC-Align) versus either alone (ablation) | Baseline-CLIP · CUHK-PEDES | Rank-1: 69.49 -> 70.89 | Improved |
Closest papers by meaning
- Adaptive Uncertainty-Based Learning for Text-Based Person Retrieval (AAAI 2024 · similarity 0.90)
- Automatic Synthetic Data and Fine-grained Adaptive Feature Alignment for Composed Person Retrieval (NeurIPS 2025 · similarity 0.88)
- Chat-based Person Retrieval via Dialogue-Refined Cross-Modal Alignment (CVPR 2025 · similarity 0.88)
- UFineBench: Towards Text-based Person Retrieval with Ultra-fine Granularity (CVPR 2024 · similarity 0.87)
- DM-Adapter: Domain-Aware Mixture-of-Adapters for Text-Based Person Retrieval (AAAI 2025 · similarity 0.87)
- Exploring the Potential of Large Vision-Language Models for Unsupervised Text-Based Person Retrieval (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.
