Research map / 3D vision / Face and shape
Learning From Millions of 3D Scans for Large-Scale 3D Face Recognition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Fine-tune FR3DNet on the single-scan LS3DFace gallery | FR3DNet · LS3DFace | Rank-1 accuracy (%): 95.51 -> 98.75 | Helped |
| Use a 160×160 input instead of 96×96 (ablation) | FR3DNet · Validation set | Accuracy (%): 82.27 -> 86.33 | Helped |
| Increase input size from 160×160 to 224×224 (ablation) | FR3DNet · Validation set | Accuracy (%): 86.33 -> 86.85 | No clear effect |
| Use kernel size 7 rather than 5 in the first two convolutional layers (ablation) | FR3DNet · Validation set | Accuracy (%): 86.60 -> 88.73 | Helped |
| Use kernel size 7 rather than 3 in the first two convolutional layers (ablation) | FR3DNet · Validation set | Accuracy (%): 86.33 -> 88.73 | Helped |
| Increase kernel size from 7 to 9 in the first two convolutional layers (ablation) | FR3DNet · Validation set | Accuracy (%): 88.73 -> 88.92 | No clear effect |
| Fine-tune FR3DNet on the gallery | FR3DNet · LS3DFace open world, 4% openness | Average Rank-1 accuracy (%): 92.41 -> 97.22 | Helped |
| Fine-tune FR3DNet on the gallery | FR3DNet · LS3DFace open world, 50% openness | Average Rank-1 accuracy (%): 57.02 -> 61.20 | Helped |
Closest papers by meaning
- Led3D: A Lightweight and Efficient Deep Approach to Recognizing Low-Quality 3D Faces (CVPR 2019 · similarity 0.91)
- Regressing Robust and Discriminative 3D Morphable Models With a Very Deep Neural Network (CVPR 2017 · similarity 0.90)
- FaceScape: A Large-Scale High Quality 3D Face Dataset and Detailed Riggable 3D Face Prediction (CVPR 2020 · similarity 0.89)
- Enhancing Face Recognition With Self-Supervised 3D Reconstruction (CVPR 2022 · similarity 0.88)
- WebFace260M: A Benchmark Unveiling the Power of Million-Scale Deep Face Recognition (CVPR 2021 · similarity 0.88)
- Virtual Fully-Connected Layer: Training a Large-Scale Face Recognition Dataset With Limited Computational Resources (CVPR 2021 · 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.