Research map / Label-efficient and robust learning / Recognition and long-tail
Brute-Force Facial Landmark Analysis With a 140,000-Way Classifier
Read the paperCheck 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 |
|---|---|---|---|
| Add pose-class regression post-processing to the classifier (ablation) | Large-scale facial landmark classifier · 300VW Category 1 | AUC: 0.678 -> 0.774 | Helped |
| Add pose-class regression post-processing to the classifier (ablation) | Large-scale facial landmark classifier · 300VW Category 3 | failure rate (%): 3.431 -> 3.189 | Helped |
| Add temporal smoothing instead of pose-class regression post-processing (ablation) | Large-scale facial landmark classifier · 300VW Category 1 | failure rate (%): 2.221 -> 2.462 | Hurt |
| Add temporal smoothing instead of pose-class regression post-processing (ablation) | Large-scale facial landmark classifier · 300VW Category 3 | AUC: 0.709 -> 0.718 | Helped |
| Condition predictions on a user-provided nose landmark (ablation) | Large-scale facial landmark classifier · 300VW Category 3 | failure rate (%): 4.010 -> 1.564 | Helped |
| Use multi-label loss with overlapping pose-class membership (ablation) | Large-scale facial landmark classifier · Facial landmark validation | landmark error | Helped |
| Use multi-label loss without overlapping pose classes (ablation) | Large-scale facial landmark classifier · Facial landmark validation | landmark accuracy | Hurt |
| Use soft targets rather than one-hot softmax targets (ablation) | Large-scale facial landmark classifier · Facial landmark validation | landmark error | Mixed |
| Increase the number of classes beyond 1,000 with naive softmax classification (ablation) | Naive K-way facial landmark classifier · Facial landmark validation | landmark reprojection error | Hurt |
| Add synthetic large-pose training images (ablation) | Large-scale facial landmark classifier · 300VW Category 3 hard frames | robustness to large pose variation | Helped |
Closest papers by meaning
- Self-Reinforced Cascaded Regression for Face Alignment (AAAI 2018 · similarity 0.90)
- LUVLi Face Alignment: Estimating Landmarks' Location, Uncertainty, and Visibility Likelihood (CVPR 2020 · similarity 0.89)
- Look at Boundary: A Boundary-Aware Face Alignment Algorithm (CVPR 2018 · similarity 0.89)
- Semantic Alignment: Finding Semantically Consistent Ground-Truth for Facial Landmark Detection (CVPR 2019 · similarity 0.89)
- Generalizable Face Landmarking Guided by Conditional Face Warping (CVPR 2024 · similarity 0.87)
- LAFS: Landmark-based Facial Self-supervised Learning for Face Recognition (CVPR 2024 · similarity 0.87)
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.