Research map / Label-efficient and robust learning / Recognition and long-tail
HydraPlus-Net: Attentive Deep Features for Pedestrian Analysis
Xihui Liu, Haiyu Zhao, Maoqing Tian, Lu Sheng, Jing Shao, Shuai Yi, Junjie Yan, Xiaogang Wang
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 multi-directional attention (AF-net) to M-net | M-net · CUHK03 re-id | Top-1: 88.2 -> 91.8 | Improved |
| add multi-directional attention (AF-net) to M-net | M-net · VIPeR re-id | Top-1: 51.6 -> 56.6 | Improved |
| add multi-directional attention (AF-net) to M-net | M-net · Market-1501 re-id | Top-1: 73.1 -> 76.9 | Improved |
| add multi-directional attention to M-net | M-net · RAP attributes | mA: 74.44 -> 76.12 | Improved |
| add multi-directional attention to M-net | M-net · PETA attributes | mA: 80.58 -> 81.77 | Improved |
| add multi-directional attention to M-net | M-net · PA-100K attributes | mA: 72.3 -> 74.21 | Improved |
| add multi-directional attention to M-net | M-net · RAP attributes | Precision: 77.83 -> 77.33 | Hurt |
| apply attention maps to adjacent blocks (multi-directional) instead of naive same-block attention (ablation) | naive attention per branch · VIPeR re-id | Top-1 | Improved |
| keep middle-level (block 2) attention maps and features (ablation) | AF-net · VIPeR re-id | Top-1 | Improved |
| add all three MDA modules instead of pruning one or two branches (ablation) | AF-net · VIPeR re-id | Top-1 | Improved |
| use multiple attention channels (L=8) instead of one or two (ablation) | MDA module · pedestrian features | feature diversity | Improved |
| add HP-net over DeepMar on PETA | DeepMar · PETA attributes | mA: 82.6 -> 81.77 | Hurt |
Closest papers by meaning
- What Can Help Pedestrian Detection? (CVPR 2017 · similarity 0.86)
- Mask-Guided Attention Network for Occluded Pedestrian Detection (ICCV 2019 · similarity 0.86)
- Where, What, Whether: Multi-Modal Learning Meets Pedestrian Detection (CVPR 2020 · similarity 0.86)
- NAS-PED: Neural Architecture Search for Pedestrian Detection (TPAMI 2024 · similarity 0.85)
- Selective and Orthogonal Feature Activation for Pedestrian Attribute Recognition (AAAI 2024 · similarity 0.83)
- Pedestrian Attribute Recognition: A New Benchmark Dataset and a Large Language Model Augmented Framework (AAAI 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.
