Research map / Label-efficient and robust learning / Recognition and long-tail
PropagationNet: Propagate Points to Curve to Learn Structure Information
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 |
|---|---|---|---|
| Add the Propagation Module (ablation) | Stacked hourglass baseline · WFLW | NME: 4.81 -> 4.48 | Helped |
| Replace bottleneck blocks with Multi-view Hourglass (ablation) | Stacked hourglass baseline · WFLW | NME: 4.81 -> 4.67 | Helped |
| Use anti-aliased CNN with Gaussian kernel size 2 (ablation) | BL · WFLW | NME: 4.81 -> 4.79 | Helped |
| Use anti-aliased CNN with Gaussian kernel size 3 (ablation) | BL · WFLW | NME: 4.81 -> 4.67 | Helped |
| Use anti-aliased CNN with Gaussian kernel size 5 (ablation) | BL · WFLW | NME: 4.81 -> 4.75 | Helped |
| Replace Adaptive Wing Loss with Focal Wing Loss (ablation) | PropagationNet training · WFLW | NME: 4.81 -> 4.64 | Helped |
| Use PropNet instead of Wing | Face alignment · WFLW testset | NME: 5.11 -> 4.05 | Helped |
| Use PropNet instead of AWing | Face alignment · 300W fullset, inter-ocular normalization | NME: 3.07 -> 2.93 | Helped |
| Use PropNet instead of Wing | Face alignment · 300W fullset, inter-pupil normalization | NME: 4.04 -> 4.10 | Hurt |
| Use PropNet instead of Wing | Face alignment · 300W challenging subset, inter-pupil normalization | NME: 7.18 -> 5.75 | Helped |
Closest papers by meaning
- Disentangling 3D Pose in a Dendritic CNN for Unconstrained 2D Face Alignment (CVPR 2018 · similarity 0.87)
- Direct Shape Regression Networks for End-to-End Face Alignment (CVPR 2018 · similarity 0.86)
- Sparse Local Patch Transformer for Robust Face Alignment and Landmarks Inherent Relation Learning (CVPR 2022 · similarity 0.86)
- InverseFaceNet: Deep Monocular Inverse Face Rendering (CVPR 2018 · similarity 0.85)
- Unsupervised Depth Estimation, 3D Face Rotation and Replacement (NeurIPS 2018 · similarity 0.85)
- 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.