Research map / 3D vision / Face and shape
Unconstrained Face Alignment via Cascaded Compositional Learning
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 |
|---|---|---|---|
| use cascaded compositional learning rather than the strongest listed baseline | face alignment · AFLW-Full | mean error (% normalised by face size): 3.73 -> 2.72 | Helped |
| use cascaded compositional learning rather than the strongest listed baseline | face alignment · AFW | mean error (% normalised by face size): 3.25 -> 2.45 | Helped |
| replace head-pose-selected CFSS with compositional prediction | AFLW-Full face alignment · AFLW-Full | mean error (% normalised by face size): 3.12 -> 2.72 | Helped |
| optimize domain composition for landmark-shape error rather than classification scores (ablation) | CCL · AFLW-Full | mean error (% normalised by face size): 3.86 -> 2.72 | Helped |
| partition the optimisation space into 16 domains instead of 1 (ablation) | CCL · AFLW-Full | mean error (% normalised by face size): 3.73 -> 2.72 | Helped |
| increase the number of domains from 16 to 64 (ablation) | CCL · AFLW-Full | mean error (% normalised by face size): 2.72 -> 2.69 | Helped |
| add local landmark-visibility encoding to the feature mapping (ablation) | CCL · AFLW-Full | mean error (% normalised by face size): 3.78 -> 2.72 | Helped |
| exclude zero-composition domains on the fly (ablation) | CCL inference · AFLW-Full | FPS: 90 -> 350 | Helped |
| exclude zero-composition domains on the fly (ablation) | CCL inference · AFLW-Full | accuracy | No clear effect |
Closest papers by meaning
- Mnemonic Descent Method: A Recurrent Process Applied for End-To-End Face Alignment (CVPR 2016 · similarity 0.89)
- Self-Reinforced Cascaded Regression for Face Alignment (AAAI 2018 · similarity 0.89)
- Large-Pose Face Alignment via CNN-Based Dense 3D Model Fitting (CVPR 2016 · similarity 0.87)
- Direct Shape Regression Networks for End-to-End Face Alignment (CVPR 2018 · similarity 0.86)
- Generalizable Face Landmarking Guided by Conditional Face Warping (CVPR 2024 · similarity 0.84)
- Neural Implicit Morphing of Face Images (CVPR 2024 · 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.