ATTN-FIQA: Interpretable Attention-based Face Image Quality Assessment with Vision Transformers

2026-04-21

AI Summary

This work proposes a training-free face image quality assessment (FIQA) method that leverages the intrinsic properties of pre-trained Vision Transformer models for face recognition. It reveals, for the first time, that the magnitudes of multi-head attention in the pre-softmax layer of the final transformer block naturally encode image quality information. By aggregating attention heads and averaging their responses through a single forward pass, the method produces an image-level quality score without requiring fine-tuning or additional training. Extensive experiments across eight benchmark datasets and four mainstream architectures demonstrate strong correlation with perceptual image quality, while also enabling visualization of facial regions most influential to the quality prediction. The approach achieves high efficiency, interpretability, and broad generalizability.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Qualitative Reasoning

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Face Image Quality Assessment (FIQA) aims to assess the recognition utility of face samples and is essential for reliable face recognition (FR) systems. Existing approaches require computationally expensive procedures such as multiple forward passes, backpropagation, or additional training, and only recent work has focused on the use of Vision Transformers. Recent studies highlighted that these architectures inherently function as saliency learners with attention patterns naturally encoding spatial importance. This work proposes ATTN-FIQA, a novel training-free approach that investigates whether pre-softmax attention scores from pre-trained Vision Transformer-based face recognition models can serve as quality indicators. We hypothesize that attention magnitudes intrinsically encode quality: high-quality images with discriminative facial features enable strong query-key alignments producing focused, high-magnitude attention patterns, while degraded images generate diffuse, low-magnitude patterns. ATTN-FIQA extracts pre-softmax attention matrices from the final transformer block, aggregate multi-head attention information across all patches, and compute image-level quality scores through simple averaging, requiring only a single forward pass through pre-trained models without architectural modifications, backpropagation, or additional training. Through comprehensive evaluation across eight benchmark datasets and four FR models, this work demonstrates that attention-based quality scores effectively correlate with face image quality and provide spatial interpretability, revealing which facial regions contribute most to quality determination.

Problem

• Face Image Quality Assessment
• Vision Transformers
• Attention Mechanism
• Interpretability
• Training-free

Innovation

• attention-based FIQA
• Vision Transformers
• training-free
• spatial interpretability
• face image quality assessment