Research map / Label-efficient and robust learning / Recognition and long-tail
Person Tube Retrieval via Language Description
Hehe Fan, Yi Yang
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 scene representation to person feature (ablation) | MSSP (HGLMM) · ActivityNet-PTRLD | Top-1 accuracy: 27.8 -> 29.1 | Improved |
| add action representation (3D ResNet) to person feature (ablation) | MSSP (HGLMM) · ActivityNet-PTRLD | Top-1 accuracy: 27.8 -> 38.6 | Improved |
| add scene to person+action (ablation) | MSSP (HGLMM) · ActivityNet-PTRLD | Top-1 accuracy: 38.6 -> 41.3 | Improved |
| increase visual structure preservation weight lambda_xx beyond 0.001 (ablation) | MSSP (HGLMM) · ActivityNet-PTRLD | Top-1 accuracy | Hurt |
| increase textual structure preservation weight lambda_yy beyond 0.1 (ablation) | MSSP (HGLMM) · ActivityNet-PTRLD | Top-1 accuracy | Hurt |
| MSSP (bidirectional ranking + structure preservation) vs triplet cosine embedding, HGLMM | HGLMM text embedding · ActivityNet-PTRLD | Top-1 accuracy: 33.8 -> 41.3 | Improved |
| MSSP vs DSPE++ (adds multi-scale sub-tube and visual internal structure preservation) | HGLMM text embedding · ActivityNet-PTRLD | Top-1 accuracy: 38.8 -> 41.3 | Improved |
| MSSP vs DSPE with HGLMM on video retrieval | HGLMM text embedding · MSR-VTT | Top-1 accuracy: 33.1 -> 33.1 | No clear effect |
| HGLMM instead of CNN or GRU text embedding | MSSP · ActivityNet-PTRLD | Top-1 accuracy: 35.0 -> 41.3 | Improved |
Closest papers by meaning
- Spatio-Temporal Person Retrieval via Natural Language Queries (ICCV 2017 · similarity 0.88)
- Diverse Knowledge Distillation for End-to-End Person Search (AAAI 2021 · similarity 0.84)
- A Survey on Text-Based Person Search (TPAMI 2026 · similarity 0.83)
- Human-centered Interactive Learning via MLLMs for Text-to-Image Person Re-identification (CVPR 2025 · similarity 0.83)
- Human Semantic Parsing for Person Re-Identification (CVPR 2018 · similarity 0.83)
- Automatic Synthetic Data and Fine-grained Adaptive Feature Alignment for Composed Person Retrieval (NeurIPS 2025 · similarity 0.82)
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.
