Research map / 3D vision / Face and shape
3D Face Reconstruction with Dense Landmarks
Erroll Wood, Tadas Baltrušaitis, Charlie Hewitt, Matthew Johnson, Jingjing Shen, Nikola Milosavljević, Daniel Wilde, Stephan Garbin, Toby Sharp, Ivan Stojiljković, Tom Cashman, Julien Valentin
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 |
|---|---|---|---|
| train landmarks with Gaussian NLL loss (predict uncertainty) instead of L2 on coordinates (ablation) | dense landmark CNN (ResNet-101) · 300W | NME | Improved |
| fit with 703 landmarks instead of 320 or 68 (ablation) | 3DMM fitting · NoW validation | 3D reconstruction error | Improved |
| use predicted landmark uncertainty sigma in model fitting instead of fixed sigma (ablation) | 3DMM fitting with 703 landmarks · NoW validation | 3D reconstruction error | Improved |
| fit one face model to all frames/views jointly instead of single view | dense-landmark fitting · MICC (cooperative) | Mean error (mm): 1.64 -> 1.43 | Improved |
| fit one face model to all frames/views jointly instead of single view | dense-landmark fitting · MICC (indoor) | Mean error (mm): 1.62 -> 1.42 | Improved |
| fit one face model to all frames/views jointly instead of single view | dense-landmark fitting · MICC (outdoor) | Mean error (mm): 1.61 -> 1.42 | Improved |
| use real-time MobileNet V2 with 320 landmarks instead of offline ResNet-101 with 703 | dense-landmark pipeline · monocular reconstruction | fidelity and speed: 150FPS | Mixed |
Closest papers by meaning
- Large Pose 3D Face Reconstruction From a Single Image via Direct Volumetric CNN Regression (ICCV 2017 · similarity 0.89)
- FaceScape: 3D Facial Dataset and Benchmark for Single-View 3D Face Reconstruction (TPAMI 2023 · similarity 0.89)
- A Simple, Fast and Highly-Accurate Algorithm to Recover 3D Shape from 2D Landmarks on a Single Image (TPAMI 2017 · similarity 0.88)
- FreeEnricher: Enriching Face Landmarks without Additional Cost (AAAI 2023 · similarity 0.88)
- Continuous Landmark Detection With 3D Queries (CVPR 2023 · similarity 0.88)
- Generalizable Face Landmarking Guided by Conditional Face Warping (CVPR 2024 · similarity 0.86)
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.
