Learning Steadily: Accumulating Relative Point Margin Scores for Face Image Quality Assessment
CARPM-FIQA stabilizes face-image quality scores by accumulating relative point margins across training.
The paper introduces CARPM-FIQA, a stabilization method for face-recognition-integrated face image quality assessment. Instead of single-epoch estimates, it accumulates relative point margin scores—the ratio of intra-class compactness to inter-class separation—over the entire training trajectory. On SynFIQA and eight benchmarks with four face-recognition models, CARPM-FIQA(L) and CARPM-FIQA(S) rank 4th and 6th of 17 methods by averaged pAUC-EDC and AUC-EDC, staying within a few percent of the best method after per-benchmark normalization.
- CARPM-FIQA accumulates relative point margins across the full training trajectory.
- Single-epoch FR-integrated quality estimates fluctuate as the feature space evolves.
- On eight benchmarks it ranks 4th and 6th of 17 methods by error-versus-discard curves.
- Cumulative averaging reduces variance and improves ranking stability with convergence guarantees.
From the original article
1 min read at huggingface.co
Face Image Quality Assessment determines the suitability of captured face images for automated face recognition (FR), a critical capability for reliable biometric systems. Existing state-of-the-art FR-integrated FIQA methods suffer from temporal instability: as the feature space evolves during training, single-epoch quality estimates fluctuate, creating a moving target that undermines reliable quality prediction.
By Guray Ozgur, Tahar Chettaoui, Eduarda Caldeira · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.