Research map / Image generation / Autoregressive and latent generation
CR-FIQA: Face Image Quality Assessment by Learning Sample Relative Classifiability
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 |
|---|---|---|---|
| Use relative classifiability (CR) as the learned quality target rather than CCS alone (ablation) | CR-FIQA(S) · Adience, AgeDB-30, CALFW, CFP-FP | ERC verification error (FNMR at FMR=1e-3) | Helped |
| Learn CR simultaneously with optimizing the FR class centers (ablation) | CR-FIQA(S) · Adience, AgeDB-30, CALFW, CFP-FP | ERC verification error (FNMR at FMR=1e-3) | Helped |
| Use CR-FIQA(L) instead of SER-FIQ for quality-based sample rejection | ArcFace face verification · XQLFW | ERC AUC (FMR=1e-3): 0.1812 vs 0.2084 | Hurt |
| Use CR-FIQA quality scores as embedding weights | ArcFace face verification · IJB-C 1:1 mixed verification | TAR (%) at FAR=1e-6: 89.85 -> 90.16 | Helped |
| Use CR-FIQA quality scores as embedding weights | ElasticFace face verification · IJB-C 1:1 mixed verification | TAR (%) at FAR=1e-6: 89.15 -> 90.30 | Helped |
| Use CR-FIQA quality scores as embedding weights | MagFace face verification · IJB-C 1:1 mixed verification | TAR (%) at FAR=1e-6: 85.67 -> 87.12 | Helped |
| Use CR-FIQA quality scores as embedding weights | CurricularFace face verification · IJB-C 1:1 mixed verification | TAR (%) at FAR=1e-6: 90.46 -> 90.82 | Helped |
| Use CR-FIQA(S) for quality ranking instead of MagFace | ArcFace face verification · LFW | ERC AUC (FMR=1e-3): 0.0016 vs 0.0020 | Hurt |
Closest papers by meaning
- CLIB-FIQA: Face Image Quality Assessment with Confidence Calibration (CVPR 2024 · similarity 0.92)
- DSL-FIQA: Assessing Facial Image Quality via Dual-Set Degradation Learning and Landmark-Guided Transformer (CVPR 2024 · similarity 0.87)
- RepFace: Refining Closed-Set Noise with Progressive Label Correction for Face Recognition (AAAI 2025 · similarity 0.85)
- Vec2Face: Scaling Face Dataset Generation with Loosely Constrained Vectors (ICLR 2025 · similarity 0.82)
- Integrating Global Context Contrast and Local Sensitivity for Blind Image Quality Assessment (ICML 2024 · similarity 0.82)
- UIFace: Unleashing Inherent Model Capabilities to Enhance Intra-Class Diversity in Synthetic Face Recognition (ICLR 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.