Research map / Label-efficient and robust learning / Recognition and long-tail
Quality Aware Network for Set to Set Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace average pooling with learned quality-weighted aggregation | QAN · PRID2011 common setting | CMC1: 81.3 -> 90.3 | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · iLIDS-VID common setting | CMC1: 60.6 -> 68.0 | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · PRID2011 cross-dataset | CMC1: 29.4 -> 34.0 | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · iLIDS-VID cross-dataset | CMC1: 44.1 -> 47.7 | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · YouTube Face | accuracy: 95.46$\pm$ 0.07\% -> 96.17$\pm$ 0.09\% | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · IJB-A | TPR@FPR 1e-3: 85.30$\pm$3.48\% -> 89.31$\pm$3.92\% | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · IJB-A | TPR@FPR 1e-2: 93.81$\pm$1.44 -> 94.20$\pm$1.53\% | Helped |
| Replace average pooling with learned quality-weighted aggregation | QAN · IJB-A | TPR@FPR 1e-1: 97.85$\pm$0.61\% -> 98.02$\pm$0.55\% | Helped |
| Use volunteer-rated human quality scores instead of network-predicted scores (ablation) | QAN on YouTube Face · YouTube Face | recognition performance | Hurt |
| Connect quality generation to Pool3 rather than earlier or later feature levels (ablation) | QAN on YouTube Face · YouTube Face | accuracy | Helped |
| Jointly train feature generation and quality generation rather than fixing the feature generator (ablation) | QAN on YouTube Face · YouTube Face | accuracy | Helped |
Closest papers by meaning
- FaceCoresetNet: Differentiable Coresets for Face Set Recognition (AAAI 2024 · similarity 0.88)
- Open-Set Recognition: A Good Closed-Set Classifier is All You Need (ICLR 2022 · similarity 0.86)
- Set Prediction in the Latent Space (NeurIPS 2021 · similarity 0.85)
- Generative-Discriminative Feature Representations for Open-Set Recognition (CVPR 2020 · similarity 0.84)
- Blind Image Quality Assessment Based on Geometric Order Learning (CVPR 2024 · similarity 0.81)
- GHOST: Gaussian Hypothesis Open-Set Technique (AAAI 2025 · similarity 0.81)
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.