Research map / Detection and segmentation / CNN features
A Deep Regression Architecture With Two-Stage Re-Initialization for High Performance Facial Landmark Detection
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 global and local two-stage re-initialization to the VGG-S baseline (ablation) | VGG-S facial landmark detector · 300-W Full Set | NME: 5.69 -> 4.99 | Helped |
| Add global and local two-stage re-initialization to the TCDCN baseline (ablation) | TCDCN facial landmark detector · 300-W Full Set | NME: 6.12 -> 4.99 | Helped |
| Add the local re-initialization and regression stage (ablation) | Proposed− global-stage model · 300-W Full Set | NME: 5.27 -> 4.99 | Helped |
| Add the local re-initialization and regression stage (ablation) | Proposed− global-stage model · AFLW | NME: 2.33 -> 2.17 | Helped |
| Use two-stage re-initialization instead of the VGG-S baseline (ablation) | Facial landmark detector · AFLW | NME: 2.68 -> 2.17 | Helped |
| Use two-stage re-initialization instead of the TCDCN baseline (ablation) | Facial landmark detector · AFLW | NME: 2.99 -> 2.17 | Helped |
| Use two-stage re-initialization under increasingly extended face-box scales (ablation) | Facial landmark detector · 300-W Full Set, scale 0.1 to 0.5 | NME: 4.99 -> 5.93; Baseline1 6.12 -> 24.43 | Mixed |
| Use two-stage re-initialization under increasingly shifted face-box centers (ablation) | Facial landmark detector · 300-W Full Set, translation 0.05 to 0.25 | NME: 5.01 -> 5.77; Baseline1 6.29 -> 18.67 | Mixed |
| Use two-stage re-initialization under increasingly rotated face images (ablation) | Facial landmark detector · 300-W Full Set, rotation 5° to 25° | NME: 5.13 -> 6.20; Baseline1 6.35 -> 11.71 | Mixed |
Closest papers by meaning
- Facial Landmarks Detection by Self-Iterative Regression Based Landmarks-Attention Network (AAAI 2018 · similarity 0.88)
- Robust Facial Landmark Detection via a Fully-Convolutional Local-Global Context Network (CVPR 2018 · similarity 0.88)
- Towards Accurate Facial Landmark Detection via Cascaded Transformers (CVPR 2022 · similarity 0.87)
- Deep Structured Prediction for Facial Landmark Detection (NeurIPS 2019 · similarity 0.87)
- Direct Shape Regression Networks for End-to-End Face Alignment (CVPR 2018 · similarity 0.87)
- Dynamic Attention-Controlled Cascaded Shape Regression Exploiting Training Data Augmentation and Fuzzy-Set Sample Weighting (CVPR 2017 · 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.