Research map / Graph learning / GNN architectures
Multiscale Dynamic Graph Representation for Biometric Recognition With Occlusions
Min Ren, Yunlong Wang, Yuhao Zhu, Kunbo Zhang, Zhenan Sun
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 graph module (feature graphs with dynamic matching) (ablation) | CNN backbone · ND-LG4000 iris | FRR@FAR=0.1%: 2.27% -> 1.09% | Improved |
| add graph module on occluded iris (ablation) | CNN backbone · ND-LG4000 occluded | FRR@FAR=0.1%: 31.85% -> 5.09% | Improved |
| add graph module on occluded iris (ablation) | CNN backbone · ND-LG4000 occluded | EER: 3.66% -> 1.72% | Improved |
| use multiscale instead of single scale 2 (ablation) | MS-DGR · ND-LG4000 iris | FRR@FAR=0.1%: 1.43% -> 1.09% | Improved |
| use multiscale instead of single scale 1 (ablation) | MS-DGR · ND-LG4000 iris | FRR@FAR=0.1%: 1.92% -> 1.09% | Improved |
| use 32 nodes instead of 8 (ablation) | medium-scale FG · ND-LG4000 iris | FRR@FAR=0.1%: 1.77% -> 1.43% | Improved |
| use 32 nodes instead of 128 (ablation) | medium-scale FG · ND-LG4000 iris | FRR@FAR=0.1%: 1.57% -> 1.43% | Improved |
| add SE layer to residual SE-GAT (ablation) | SE-GAT · ND-LG4000 iris | FRR@FAR=0.1%: 1.35% -> 1.09% | Improved |
| add residual structure to SE-GAT (ablation) | SE-GAT · ND-LG4000 iris | FRR@FAR=0.1%: 2.89% -> 1.09% | Improved |
| add residual and SE layer to GAT (ablation) | SE-GAT · ND-LG4000 iris | EER: 1.01% -> 0.58% | Improved |
| mask occluded areas before feature extraction | CNN frameworks · ND-LG4000 50% occlusion | FRR/EER | Hurt |
| feed occluded areas instead of cropping them | Ours · ND-LG4000 30% occlusion | FRR@FAR=0.1%: 2.28% -> 1.48% | Improved |
Closest papers by meaning
- Dynamic Graph Representation for Occlusion Handling in Biometrics (AAAI 2020 · similarity 0.96)
- Robust Facial Landmark Detection via Occlusion-Adaptive Deep Networks (CVPR 2019 · similarity 0.85)
- FC-DGCN: Deep Graph Convolutional Network for Face Clustering and Recognition (NeurIPS 2026 · similarity 0.84)
- Sequential Motif Profiles and Topological Plots for Offline Signature Verification (CVPR 2020 · similarity 0.84)
- Dynamic Graph Message Passing Networks for Visual Recognition (TPAMI 2022 · similarity 0.84)
- Attributes Shape the Embedding Space of Face Recognition Models (ICML 2025 · 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.
