Research map / Vision-language and multimodal / Human and embodied modeling
Pose-Guided Multi-Granularity Attention Network for Text-Based Person Search
Ya Jing, Chenyang Si, Junbo Wang, Wei Wang, Liang Wang, Tieniu Tan
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 |
|---|---|---|---|
| Use PMA instead of Dual Path with VGG-16 features | Text-based person search · CUHK-PEDES test | top-1 accuracy: 32.15 -> 47.02 | Improved |
| Use PMA instead of Dual Path with ResNet-50 features | Text-based person search · CUHK-PEDES test | top-1 accuracy: 44.40 -> 53.81 | Improved |
| Use the full 9408-word vocabulary instead of the reduced vocabulary (ablation) | Text representation · CUHK-PEDES test | top-1 accuracy: 48.67 -> 46.51 | Hurt |
| Use bidirectional rather than unidirectional LSTM encoding (ablation) | Text representation · CUHK-PEDES test | top-1 accuracy: 47.02 -> 48.67 | Improved |
| Add pose confidence maps to the visual CNN input (ablation) | Global-feature baseline · CUHK-PEDES test | top-1 accuracy: 48.67 -> 50.06 | Improved |
| Add coarse alignment to pose-augmented visual features (ablation) | Con-pose · CUHK-PEDES test | top-1 accuracy: 50.06 -> 52.20 | Improved |
| Add fine-grained alignment to Con-pose + CA (ablation) | Con-pose + CA · CUHK-PEDES test | top-1 accuracy: 52.20 -> 53.81 | Improved |
| Use pose guidance in fine-grained alignment (ablation) | PMA fine-grained alignment · CUHK-PEDES test | top-1 accuracy: 52.54 -> 53.81 | Improved |
| Use noun phrases rather than every sentence word in fine-grained alignment (ablation) | PMA fine-grained alignment · CUHK-PEDES test | top-1 accuracy: 52.71 -> 53.81 | Improved |
| Use adaptive hard attention rather than soft attention over regions (ablation) | PMA coarse alignment · CUHK-PEDES test | top-1 accuracy: 52.35 -> 53.81 | Improved |
| Select a fixed 5, 10, or 20 regions rather than using adaptive hard attention (ablation) | PMA coarse alignment · CUHK-PEDES test | top-1 accuracy: 51.24; 53.56; 53.21; adaptive: 53.81 | Mixed |
Closest papers by meaning
- Visual Perturbation for Text-Based Person Search (AAAI 2025 · similarity 0.87)
- Adaptive Uncertainty-Based Learning for Text-Based Person Retrieval (AAAI 2024 · similarity 0.86)
- Exploring the Potential of Large Vision-Language Models for Unsupervised Text-Based Person Retrieval (AAAI 2025 · similarity 0.86)
- Multi-Modal Disordered Representation Learning Network for Description-Based Person Search (AAAI 2024 · similarity 0.85)
- Automatic Synthetic Data and Fine-grained Adaptive Feature Alignment for Composed Person Retrieval (NeurIPS 2025 · similarity 0.85)
- Coarse-to-Fine Latent Diffusion for Pose-Guided Person Image Synthesis (CVPR 2024 · similarity 0.85)
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.
