Research map / 3D vision / Face and shape
Physically-Based Face Rendering for NIR-VIS 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 |
|---|---|---|---|
| Add generated paired NIR-VIS images to training (ablation) | LightCNN-29 face recognition model · LAMP-HQ | VR@FAR=0.01%: 67.28±1.9 -> 91.97±1.5 | Helped |
| Increase generated-data share from 10% to 50% (ablation) | LightCNN-29 face recognition model · LAMP-HQ | VR@FAR=0.01%: 82.13±1.7 -> 90.64±1.5 | Helped |
| Increase generated-data share from 50% to 100% (ablation) | LightCNN-29 face recognition model · LAMP-HQ | VR@FAR=0.01%: 90.64±1.5 -> 91.97±1.5 | Helped |
| Use ID-MMD instead of domain-level MMD (ablation) | LightCNN-29 face recognition model · LAMP-HQ | VR@FAR=0.01%: 90.93±1.6 -> 91.97±1.5 | Helped |
| Use ID-MMD instead of pairwise feature loss (ablation) | LightCNN-29 face recognition model · LAMP-HQ | VR@FAR=0.01%: 90.41±1.5 -> 91.97±1.5 | Helped |
| Add transformed NIR normals to NIR rendering (ablation) | NIR renderer · NIR rendering (LAMP-HQ reconstructions) | PSNR: 17.32 -> 17.34 | Helped |
| Use the full VIS-to-NIR reflectance transformation and rendering method rather than monochrome VIS rendering (ablation) | NIR face rendering · NIR rendering (LAMP-HQ reconstructions) | PSNR: 16.61 -> 21.03 | Helped |
| Generate paired faces with physically based rendering instead of DVG-Face | NIR-VIS face generation · LAMP-HQ generated faces | MS (1 v N): 0.244 -> 0.411 | Helped |
| Fine-tune on target data with ID loss and ID-MMD rather than ID loss alone (ablation) | LightCNN-29 · LAMP-HQ | VR@FAR=0.01%: 94.5±0.8 -> 98.6±0.4 | Helped |
| Fine-tune on target data with ID loss and ID-MMD rather than use generated data without target fine-tuning | LightCNN-29 · LAMP-HQ | VR@FAR=0.01%: 92.0±1.5 -> 98.6±0.4 | Helped |
Closest papers by meaning
- Multi-Energy Guided Image Translation with Stochastic Differential Equations for Near-Infrared Facial Expression Recognition (AAAI 2024 · similarity 0.87)
- Hypergraph-Guided Disentangled Spectrum Transformer Networks for Near-Infrared Facial Expression Recognition (AAAI 2024 · similarity 0.86)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.86)
- Monocular Identity-Conditioned Facial Reflectance Reconstruction (CVPR 2024 · similarity 0.85)
- SFDM: Robust Decomposition of Geometry and Reflectance for Realistic Face Rendering from Sparse-view Images (CVPR 2025 · similarity 0.84)
- TokenMatcher: Diverse Tokens Matching for Unsupervised Visible-Infrared Person Re-Identification (AAAI 2025 · similarity 0.84)
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.