Research map / Vision-language and multimodal / Human and embodied modeling
Partial Scene Text Retrieval
Hao Wang, Minghui Liao, Zhouyi Xie, Wenyu Liu, 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 DPMA at inference (ablation) | Baseline · Chinese avg PPR | mAP: 42.25 -> 77.34 | Improved |
| add DPMA at inference (ablation) | Baseline · English avg PPR | mAP: 33.63 -> 55.39 | Improved |
| replace bag construction with DPMA (ablation) | Baseline + Bags · English avg PPR | mAP: 47.09 -> 55.39 | Improved |
| replace bag construction with DPMA (ablation) | Baseline + Bags · Chinese avg PPR | mAP: 72.00 -> 77.34 | Improved |
| replace bag construction with DPMA (ablation) | Baseline + Bags · ArT TIR | mAP: 73.99 -> 78.18 | Improved |
| replace bag construction with DPMA (ablation) | Baseline + Bags · Chinese avg NCPP | mAP: 47.53 -> 75.53 | Improved |
| replace bag construction with DPMA (ablation) | Baseline + Bags · ReCTS online speed | FPS: 2.1 -> 9.3 | Improved |
| replace conventional MIL with RankMIL (ablation) | MIL · Chinese avg PPR | mAP: 73.44 -> 77.34 | Improved |
| replace conventional MIL with RankMIL (ablation) | MIL · English avg PPR | mAP: 52.04 -> 55.39 | Improved |
| add noisy pseudo-labelled patches P^r to CMSL training (ablation) | CMSL-B · Chinese avg PPR | mAP: 71.94 -> 69.92 | Hurt |
| add translation labels T^r as extra query words (ablation) | CMSL-A · Chinese avg PPR | mAP: 70.45 -> 71.94 | Improved |
| add boundary points K=7 instead of rectangles (ablation) | Baseline · ArT TIR | mAP: 70.00 -> 77.94 | Improved |
Closest papers by meaning
- Beyond Cropped Regions: New Benchmark and Corresponding Baseline for Chinese Scene Text Retrieval in Diverse Layouts (ICML 2025 · similarity 0.88)
- MSTAR: Box-free Multi-query Scene Text Retrieval with Attention Recycling (NeurIPS 2025 · similarity 0.88)
- Instance-Level Composed Image Retrieval (NeurIPS 2025 · similarity 0.83)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · similarity 0.83)
- LamRA: Large Multimodal Model as Your Advanced Retrieval Assistant (CVPR 2025 · similarity 0.83)
- CoLLM: A Large Language Model for Composed Image Retrieval (CVPR 2025 · similarity 0.83)
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.
