Research map / 3D vision / Face and shape
Learning Dense Facial Correspondences in Unconstrained Images
Ronald Yu, Shunsuke Saito, Haoxiang Li, Duygu Ceylan, Hao Li
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 |
|---|---|---|---|
| use dense-correspondence network instead of Zhu et al. cascaded CNN regression | 3DMM face fitting · AFLW2000 visible inner landmarks (all images) | NMS error: 4.60 -> 3.58 | Improved |
| use dense-correspondence network instead of Zhu et al. | 3DMM face fitting · AFLW2000 68 landmarks (yaw 60-90) | NMS error: 7.93 -> 9.56 | Hurt |
| use dense-correspondence network instead of SDM | 3DMM face fitting · AFLW2000 68 landmarks (yaw 30-60) | NMS error: 4.94 -> 6.06 | Hurt |
| use dense-correspondence network instead of SDM | 3DMM face fitting · AFLW2000 68 landmarks (yaw 0-30) | NMS error: 3.67 -> 3.62 | No clear effect |
| use dense-correspondence network instead of Zhu et al. | 3DMM face fitting · AFLW visible landmarks (yaw 0-30) | NMS error: 5.00 -> 5.94 | Hurt |
| use dense-correspondence network instead of Zhu et al. | 3DMM face fitting · AFLW visible landmarks (yaw >60) | NMS error: 6.74 -> 7.96 | Hurt |
| use dense-correspondence network instead of Jourabloo et al. | 3DMM face fitting · AFLW-PIFA 21 landmarks | NMS error: 4.72 -> 5.42 | Hurt |
| replace iterative cascaded regression (Zhu et al.) with single-pass flow network | face alignment · Runtime large-pose alignment | ms per image: 75.7 -> 9.35 | Improved |
| replace Jourabloo et al. with single-pass flow network | face alignment · Runtime large-pose alignment | ms per image: 1666 -> 9.35 | Improved |
| fine-tune the synthetic-trained network on real 300W-LP photographs (ablation) | dense correspondence network · real images | qualitative | Improved |
Closest papers by meaning
- “Look Ma, no landmarks!” – Unsupervised, Model-based Dense Face Alignment (ECCV 2020 · similarity 0.91)
- DensePose: Dense Human Pose Estimation in the Wild (CVPR 2018 · similarity 0.89)
- Dense 3D Face Correspondence (TPAMI 2017 · similarity 0.89)
- DenseReg: Fully Convolutional Dense Shape Regression In-The-Wild (CVPR 2017 · similarity 0.89)
- A Sparse and Locally Coherent Morphable Face Model for Dense Semantic Correspondence Across Heterogeneous 3D Faces (TPAMI 2021 · similarity 0.89)
- 3D Face Tracking from 2D Video through Iterative Dense UV to Image Flow (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.
