Research map / Image generation / Editing and style transfer
FVeinSyn: Synthetic Finger Vein Image Generator
Yifan Wang, Jie Gui, Adams Wai Kin KONG, Baosheng Yu, Changsheng Chen, Qi Li, Zhenan Sun, James Tin-Yau Kwok, Alex C. Kot
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 on FVeinSyn instead of real data with same number of identities | recognition model · Open-set recognition (average) | TAR@FAR=1e-3: 0.6428 -> 0.7976 | Improved |
| scale FVeinSyn to 10k identities x 50 samples | recognition model · Open-set recognition (average) | TAR@FAR=1e-3: 0.7976 -> 0.9171 | Improved |
| increase samples per identity from 10 to 50 at 1.9k identities | FVeinSyn · Open-set recognition (average) | TAR@FAR=1e-3: 0.7976 -> 0.8625 | Improved |
| increase identities from 1.9k to 10k at 10 samples each | FVeinSyn · Open-set recognition (average) | TAR@FAR=1e-3: 0.7976 -> 0.8576 | Improved |
| pretrain with FVeinSyn before training on real data | recognition model · Cross-dataset evaluation (average over test sets) | TAR@FAR=1e-3: 0.2106 -> 0.7622 | Improved |
| pretrain with FVeinSyn on HKPU-FV test set | recognition model · Cross-dataset HKPU-FV | TAR@FAR=1e-3: 0.1823 -> 0.7937 | Improved |
| use FVeinSyn pretraining with only 100 identities instead of real data with 1,969 | recognition model · Limited identities | TAR@FAR=1e-3: 0.6222 -> 0.6924 | Improved |
| use FVeinSyn instead of real data at 50 identities | recognition model · Limited identities (50 IDs) | TAR@FAR=1e-3: 0.0008 -> 0.3148 | Improved |
| use FVeinSyn instead of BTUSyn for one-shot enrollment | recognition model · Few-shot N=1, PLUS-FV3 | TAR@FAR=1e-3: 0.3834 -> 0.8874 | Improved |
| use FVeinSyn instead of BTUSyn for one-shot enrollment | recognition model · Few-shot N=1, THU-FV3 | TAR@FAR=1e-3: 0.6213 -> 0.5826 | Hurt |
| use all mixed intra-class variations instead of shift only (ablation) | ICDG · ICDG ablation (average) | TAR@FAR=1e-3: 0.8789 -> 0.9171 | Improved |
| use all mixed intra-class variations instead of skin scattering only (ablation) | ICDG · ICDG ablation (average) | TAR@FAR=1e-3: 0.8801 -> 0.9171 | Improved |
Closest papers by meaning
- PVTree: Realistic and Controllable Palm Vein Generation for Recognition Tasks (AAAI 2025 · similarity 0.86)
- vesselFM: A Foundation Model for Universal 3D Blood Vessel Segmentation (CVPR 2025 · similarity 0.80)
- A Simple Baseline for Efficient Hand Mesh Reconstruction (CVPR 2024 · similarity 0.80)
- Fine-Grained Multi-View Hand Reconstruction Using Inverse Rendering (AAAI 2024 · similarity 0.80)
- VasTSD: Learning 3D Vascular Tree-state Space Diffusion Model for Angiography Synthesis (CVPR 2025 · similarity 0.80)
- Hand1000: Generating Realistic Hands from Text with Only 1,000 Images (AAAI 2025 · similarity 0.79)
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.
