Research map / Vision-language and multimodal / Human and embodied modeling
Person Search With Natural Language Description
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 VGG-16 person re-identification pre-training (ablation) | GNA-RNN · CUHK-PEDES test | top-1 accuracy: 8.93 -> 19.05 | Helped |
| Add VGG-16 person re-identification pre-training (ablation) | GNA-RNN · CUHK-PEDES test | top-10 accuracy: 32.32 -> 53.64 | Helped |
| Add word-level gates (ablation) | GNA-RNN · CUHK-PEDES test | top-1 accuracy: 13.86 -> 19.05 | Helped |
| Add word-level gates (ablation) | GNA-RNN · CUHK-PEDES test | top-10 accuracy: 44.27 -> 53.64 | Helped |
| Use unit-level attention instead of average pooling (ablation) | GNA-RNN · CUHK-PEDES test | top-1 accuracy: 4.85 -> 19.05 | Helped |
| Use unit-level attention instead of average pooling (ablation) | GNA-RNN · CUHK-PEDES test | top-10 accuracy: 27.16 -> 53.64 | Helped |
| Vary the number of visual units from 512 to 128, 256, 1024, or 2048 (ablation) | GNA-RNN · CUHK-PEDES test | top-1 accuracy: 16.15, 16.75, 19.05, 18.62, 18.25 | Hurt |
| Vary the number of visual units from 512 to 128, 256, 1024, or 2048 (ablation) | GNA-RNN · CUHK-PEDES test | top-10 accuracy: 48.58, 49.25, 53.64, 52.39, 51.59 | Hurt |
| Mask nouns in the description (ablation) | Manual person search · 20-image retrieval pools | top-1 accuracy: 0.59 -> 0.38 | Hurt |
| Mask adjectives in the description (ablation) | Manual person search · 20-image retrieval pools | top-1 accuracy: 0.59 -> 0.44 | Hurt |
| Mask verbs in the description (ablation) | Manual person search · 20-image retrieval pools | top-1 accuracy: 0.59 -> 0.57 | No clear effect |
Closest papers by meaning
- Diverse Person: Customize Your Own Dataset for Text-Based Person Search (AAAI 2024 · similarity 0.86)
- Visual Perturbation for Text-Based Person Search (AAAI 2025 · similarity 0.85)
- An Empirical Study of CLIP for Text-Based Person Search (AAAI 2024 · similarity 0.84)
- Exploring the Potential of Large Vision-Language Models for Unsupervised Text-Based Person Retrieval (AAAI 2025 · similarity 0.84)
- Divide and Conquer: Hybrid Pre-training for Person Search (AAAI 2024 · similarity 0.84)
- Modeling Thousands of Human Annotators for Generalizable Text-to-Image Person Re-identification (CVPR 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.