Research map / Video / Video-language understanding
Spatio-Temporal Person Retrieval via Natural Language Queries
Masataka Yamaguchi, Kuniaki Saito, Yoshitaka Ushiku, Tatsuya Harada
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 whole-frame context to tube features (ablation) | CCA with VGG16 RGB tube features · Person tube retrieval | R@5: 0.572 -> 0.623 | Improved |
| add optical flow features (ablation) | CCA with VGG16 RGB tube+context · Person tube retrieval | R@1: 0.279 -> 0.285 | Improved |
| add C3D features (ablation) | CCA with VGG16 RGB tube+context · Person tube retrieval | R@1: 0.279 -> 0.300 | Improved |
| add C3D features (ablation) | CCA with VGG16 RGB tube+context · Person tube retrieval | R@10: 0.751 -> 0.788 | Improved |
| add optical flow on top of RGB plus C3D (ablation) | CCA · Person tube retrieval | R@1: 0.300 -> 0.300 | No clear effect |
| replace CCA with DSPE | multimodal retrieval · Person tube retrieval | R@1: 0.300 -> 0.347 | Improved |
| replace DSPE with proposed DSPE++ loss | multimodal retrieval · Person tube retrieval | R@1: 0.347 -> 0.357 | Improved |
| replace DSPE with proposed DSPE++ loss | multimodal retrieval · Person tube retrieval | R@10: 0.783 -> 0.795 | Improved |
| replace clip-level baseline with max over tube scores from the trained model | clip retrieval · Clip-level person retrieval | R@1: 0.373 -> 0.415 | Improved |
| vary number of candidate tubes Nc (100 to 700) (ablation) | CCA retrieval · Person tube retrieval (val) | R@K | Mixed |
Closest papers by meaning
- An Empirical Study of CLIP for Text-Based Person Search (AAAI 2024 · similarity 0.85)
- Exploring the Potential of Large Vision-Language Models for Unsupervised Text-Based Person Retrieval (AAAI 2025 · similarity 0.85)
- Automatic Synthetic Data and Fine-grained Adaptive Feature Alignment for Composed Person Retrieval (NeurIPS 2025 · similarity 0.84)
- Causality-Inspired Invariant Representation Learning for Text-Based Person Retrieval (AAAI 2024 · similarity 0.84)
- Chat-based Person Retrieval via Dialogue-Refined Cross-Modal Alignment (CVPR 2025 · similarity 0.84)
- Graph-Based Cross-Domain Knowledge Distillation for Cross-Dataset Text-to-Image 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.
