Research map / Architectures and efficiency / CNN design
Hierarchical Pyramid Diverse Attention Networks for Face Recognition
Qiangchang Wang, Tianyi Wu, He Zheng, Guodong Guo
Check 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 HPDA to LS-CNN | LS-CNN · IJB-A, FAR=0.01 | accuracy: 87.2 -> 87.6 | Improved |
| Add HPDA to LS-CNN | LS-CNN · IJB-A, FAR=0.001 | accuracy: 77.5 -> 80.3 | Improved |
| Add HPDA to LS-CNN | LS-CNN · CALFW | accuracy: 95.63 -> 95.90 | Improved |
| Add HPDA to LS-CNN | LS-CNN · CPLFW | accuracy: 89.69 -> 92.35 | Improved |
| Add HPDA to LS-CNN | LS-CNN · LFW | accuracy: 99.83 -> 99.80 | Hurt |
| Add diverse learning (ablation) | HPDA · CPLFW | accuracy: 85.82 -> 86.07 | Improved |
| Add hierarchical features (ablation) | HPDA · CPLFW | accuracy: 85.63 -> 86.07 | Improved |
| Combine global and local CNNs (ablation) | HPDA · CPLFW | accuracy: 79.30 -> 86.07; 85.73 -> 86.07 | Improved |
| Use 3 pyramid scales instead of 1, 2, or 4 (ablation) | HPDA · LFW | accuracy: 99.30; 99.17; 99.33; 99.15 | Mixed |
| Use 3 local branches instead of 1 or 5 (ablation) | HPDA · LFW | accuracy: 99.08; 99.33; 99.07 | Mixed |
| Set diversity margin t to 1.00 rather than 0.25, 0.50, 2.00, or 4.00 (ablation) | HPDA · LFW | accuracy: 99.20; 99.15; 99.33; 98.97; 99.20 | Mixed |
| Use HBP instead of concatenation or addition (ablation) | HPDA · LFW | accuracy: 99.17 -> 99.33; 99.18 -> 99.33 | Improved |
Closest papers by meaning
- CQA-Face: Contrastive Quality-Aware Attentions for Face Recognition (AAAI 2022 · similarity 0.85)
- HyperFace: A Deep Multi-Task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognition (TPAMI 2017 · similarity 0.85)
- PyramidBox: A Context-assisted Single Shot Face Detector (ECCV 2018 · similarity 0.85)
- Learning Oracle Attention for High-Fidelity Face Completion (CVPR 2020 · similarity 0.85)
- Attentional Feature-Pair Relation Networks for Accurate Face Recognition (ICCV 2019 · similarity 0.85)
- Teaching Where to Look: Attention Similarity Knowledge Distillation for Low Resolution Face Recognition (ECCV 2022 · 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.
