Research map / 3D vision / Face and shape
Towards a Complete 3D Morphable Model of the Human Head
Stylianos Ploumpis, Evangelos Ververas, Eimear O’ Sullivan, Stylianos Moschoglou, Haoyang Wang, Nick E. Pears, William A. P. Smith, Barış Geçer, Stefanos P. Zafeiriou
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 |
|---|---|---|---|
| add ear model fusion (UHM instead of CFHM-ref) (ablation) | CFHM-ref · Head shape estimation (fitted facial meshes) | AUC: 0.751 -> 0.875 | Improved |
| add ear model fusion (UHM instead of CFHM-ref) (ablation) | CFHM-ref · Head shape estimation (ground-truth facial meshes) | AUC: 0.880 -> 0.912 | Improved |
| add ear model fusion (UHM instead of CFHM-ref) (ablation) | CFHM-ref · Head shape estimation (fitted facial meshes) | failure rate (%): 3.64 -> 1.51 | Improved |
| add ear model fusion (ablation) | CFHM-ref ear · Ear shape estimation (ground-truth ear meshes) | AUC: 0.697 -> 0.802 | Improved |
| add Gaussian-process refinement with MeIn3D scans (CFHM-ref vs CFHM-reg) (ablation) | CFHM-reg · Head shape estimation (fitted facial meshes) | AUC: 0.693 -> 0.751 | Improved |
| add Gaussian-process refinement (CFHM-ref vs CFHM-GP) (ablation) | CFHM-GP · Head shape estimation (ground-truth facial meshes) | AUC: 0.844 -> 0.880 | Improved |
| regression combination vs Gaussian-process combination (ablation) | CFHM-reg vs CFHM-GP · Head shape estimation (fitted facial meshes) | AUC: 0.693 vs 0.681 | No clear effect |
| combine LSFM face with LYHM head (CFHM-reg instead of LYHM) | LYHM · Head shape estimation (fitted facial meshes) | AUC: 0.605 -> 0.693 | Improved |
| combine LSFM face with LYHM head (CFHM-ref instead of LYHM) | LYHM · Head shape estimation (ground-truth facial meshes) | failure rate (%): 14.10 -> 0.62 | Improved |
| use bespoke demographic models instead of global CFHM-ref | CFHM-ref · Generalization error (300 test heads) | generalization error | Hurt |
| fuse ear model into CFHM-ref (UHM) (ablation) | CFHM-ref · Specificity | specificity error | No clear effect |
| end-to-end eye 3DMM regression network instead of Eye3DMM fitting | Eye3DMM · Eyediap gaze estimation | mean gaze error (degrees): 9.44 -> 8.85 | Improved |
Closest papers by meaning
- Combining 3D Morphable Models: A Large Scale Face-And-Head Model (CVPR 2019 · similarity 0.94)
- A 3D Morphable Model of Craniofacial Shape and Texture Variation (ICCV 2017 · similarity 0.93)
- i3DMM: Deep Implicit 3D Morphable Model of Human Heads (CVPR 2021 · similarity 0.90)
- A 3D Morphable Model Learnt From 10,000 Faces (CVPR 2016 · similarity 0.90)
- Large Scale 3D Morphable Models (IJCV 2017 · similarity 0.90)
- 3D Gaussian Head Avatars with Expressive Dynamic Appearances by Compact Tensorial Representations (CVPR 2025 · similarity 0.88)
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.
