Research map / Architectures and efficiency / Attention and transformers
Bilinear Attention Networks for Person Retrieval
Pengfei Fang, Jieming Zhou, Soumava Kumar Roy, Lars Petersson, Mehrtash Harandi
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 Bi-attention w/ AiA to appearance extractor (ablation) | Fa · Market-1501 (single query) | mAP: 80.7 -> 83.4 | Improved |
| add Bi-attention w/ AiA to appearance extractor (ablation) | Fa · CUHK03 detected | mAP: 64.5 -> 67.4 | Improved |
| add part feature extractor (ablation) | Fa · Market-1501 (single query) | mAP: 80.7 -> 85.1 | Improved |
| add Bi-attention w/ AiA to Fa + Fp (ablation) | BAT-net · Market-1501 (single query) | mAP: 85.1 -> 87.4 | Improved |
| add Bi-attention w/ AiA to Fa + Fp (ablation) | BAT-net · CUHK03 detected | mAP: 67.8 -> 73.2 | Improved |
| add AiA mechanism to Bi-attention | BAT-net · CUHK03 labeled | mAP: 72.0 -> 76.1 | Improved |
| add AiA mechanism to Bi-attention | BAT-net · MSMT17 (single query) | mAP: 50.4 -> 56.8 | Improved |
| add AiA mechanism to Bi-attention | BAT-net · Market-1501 (single query) | mAP: 85.5 -> 87.4 | Improved |
| place Bi-attention at p2 instead of p4 (ablation) | BAT-net · CUHK03 detected | mAP: 68.9 -> 73.2 | Improved |
| insert Bi-attention at p4 (ablation) | BAT-net · CUHK03 detected | mAP: 67.8 -> 68.9 | Hurt |
| place Bi-attention at p2 instead of p1 (ablation) | BAT-net · Market-1501 (single query) | mAP: 86.8 -> 87.4 | Improved |
| decrease dimensionality reduction factor r from 8 to 2 (ablation) | Bi-attention · CUHK03 detected | mAP: 73.2 -> 72.3 | Hurt |
Closest papers by meaning
- Attention in Attention Networks for Person Retrieval (TPAMI 2021 · similarity 0.87)
- Bilinear Attention Networks (NeurIPS 2018 · similarity 0.86)
- Mixed High-Order Attention Network for Person Re-Identification (ICCV 2019 · similarity 0.85)
- Hadamard Product for Low-rank Bilinear Pooling (ICLR 2017 · similarity 0.85)
- Revisiting Bilinear Pooling: A Coding Perspective (AAAI 2020 · similarity 0.84)
- Low-Rank Bilinear Pooling for Fine-Grained Classification (CVPR 2017 · 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.
