Research map / 3D vision / Face and shape
Nasal Patches and Curves for Expression-Robust 3D Face Recognition
Mehryar Emambakhsh, Adrian N. Evans
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 spherical patches instead of nasal curves | nasal descriptor · FRGC v2.0 | rank-one rate: 94.1% -> 97.9% | Improved |
| use spherical patches instead of nasal curves | nasal descriptor · Bosphorus | R1RR: 86.1% -> 95.35% | Improved |
| use spherical patches instead of nasal curves | nasal descriptor · BU-3DFE expressions | R1RR: 69.8% -> 81.8% (disgust) | Improved |
| add GA feature selection (ablation) | patches and curves · Bosphorus (neutral gallery vs non-neutral probe) | CMC rank 1-2: average improvement of approximately 1% | Improved |
| increase gallery samples per subject from 1 to 2 | nasal curves · FRGC merged | R1RR: 91.6% -> 96.8% | Improved |
| increase gallery samples per subject from 1 to 7 | spherical patches · FRGC merged | R1RR: 96.2% -> 99.75% | Improved |
| switch probes from neutral to non-neutral | spherical patches · FRGC | R1RR: 98.45% -> 98.5% | No clear effect |
| use nose region only instead of full face | spherical patches · FRGC ROC III | 0.1% FAR verification: 93.5% vs 98.04% (Wang et al. 2010 full face) | Hurt |
| nose-only descriptor vs earlier nose-based methods | spherical patches · FRGC v2.0 | rank-one rate: at least 3% better, up to 6% | Improved |
| proposed landmarking vs Creusot et al. | landmark detector · Bosphorus | nose tip accuracy: 95.47% -> 97.52% | Improved |
| proposed landmarking vs Creusot et al. | landmark detector · Bosphorus | eye corner accuracy: 98.82% -> 96.69% | Hurt |
Closest papers by meaning
- Facial Landmark Detection: A Literature Survey (IJCV 2018 · similarity 0.84)
- RetinaFace: Single-Shot Multi-Level Face Localisation in the Wild (CVPR 2020 · similarity 0.84)
- Joint Face Alignment and 3D Face Reconstruction with Application to Face Recognition (TPAMI 2018 · similarity 0.83)
- Inequality-Constrained and Robust 3D Face Model Fitting (ECCV 2020 · similarity 0.83)
- Learning to Regress 3D Face Shape and Expression From an Image Without 3D Supervision (CVPR 2019 · similarity 0.82)
- REALY: Rethinking the Evaluation of 3D Face Reconstruction (ECCV 2022 · similarity 0.82)
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.
