Research map / 3D vision / Human pose estimation
Towards Highly Accurate and Stable Face Alignment for High-Resolution Videos
Ying Tai, Yicong Liang, Xiaoming Liu, Lei Duan, Jilin Li, Chengjie Wang, Feiyue Huang, Yu Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| Preserve fractional landmark coordinates in heatmap generation and recover them by sampling three heatmap points (ablation) | Stacked-hourglass face alignment · 300W test set | NRMSE: 4.07 -> 3.80 | Improved |
| Preserve fractional coordinates in heatmap regression instead of rounding them (ablation) | Stacked-hourglass face alignment · 300W, face-scale groups | RMSE: 2.66/$\bm{2.42}$ ($0.24$~$\downarrow$); $3.71$/$\bm{3.35}$ ($0.36$~$\downarrow$); $5.06$/$\bm{4.55}$ ($0.51$~$\downarrow | Improved |
| Add the time-delay term to the stabilization loss (ablation) | FHR stabilization algorithm · Video face alignment | NRMSE | Improved |
| Add the time-delay term to the stabilization loss (ablation) | FHR stabilization algorithm · Video face alignment | temporal stability | Improved |
| Replace the proposed smooth loss with the midpoint-based smooth loss (ablation) | FHR stabilization algorithm · Video face alignment | stability loss | Hurt |
| Add the stabilization algorithm | FHR · 300-VW Scenario1, 7 landmarks | NRMSE: 4.44 -> 3.93 | Improved |
| Add the stabilization algorithm | FHR · 300-VW Scenario2, 7 landmarks | NRMSE: 3.96 -> 3.82 | Improved |
| Add the stabilization algorithm | FHR · 300-VW Scenario3, 7 landmarks | NRMSE: 5.45 -> 4.91 | Improved |
| Add the stabilization algorithm | FHR · Talking Face, 7 landmarks | NRMSE: 2.06 -> 2.14 | Hurt |
| Add the stabilization algorithm | FHR · 300-VW Scenario1, 68 landmarks | NRMSE: 5.07 -> 4.42 | Improved |
| Add the stabilization algorithm | FHR · 300-VW Scenario2, 68 landmarks | NRMSE: 4.34 -> 4.18 | Improved |
| Add the stabilization algorithm | FHR · 300-VW Scenario3, 68 landmarks | NRMSE: 7.36 -> 5.98 | Improved |
Closest papers by meaning
- Adaptive Wing Loss for Robust Face Alignment via Heatmap Regression (ICCV 2019 · similarity 0.87)
- Heatmap Regression without Soft-Argmax for Facial Landmark Detection (ICCV 2025 · similarity 0.87)
- Towards Fast, Accurate and Stable 3D Dense Face Alignment (ECCV 2020 · similarity 0.86)
- Heatmap Regression via Randomized Rounding (TPAMI 2021 · similarity 0.86)
- RED-Net: A Recurrent Encoder–Decoder Network for Video-Based Face Alignment (IJCV 2018 · similarity 0.86)
- FaithfulFaces: Pose-Faithful Facial Identity Preservation for Text-to-Video Generation (NeurIPS 2026 · similarity 0.83)
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.
