IG-FIQA: Improving Face Image Quality Assessment through Intra-class Variance Guidance robust to Inaccurate Pseudo-Labels

TL;DR

IG-FIQA introduces intra-class variance guidance and data augmentation to improve face image quality assessment, achieving state-of-the-art results while being robust to inaccurate pseudo-labels.

Contribution

The paper proposes a novel intra-class variance guidance method and on-the-fly data augmentation for FIQA, enhancing robustness and generalization.

Findings

Achieved state-of-the-art performance on multiple benchmarks.

Robust to inaccurate pseudo-labels due to intra-class variance guidance.

Efficient implementation with minimal computational overhead.

Abstract

In the realm of face image quality assesment (FIQA), method based on sample relative classification have shown impressive performance. However, the quality scores used as pseudo-labels assigned from images of classes with low intra-class variance could be unrelated to the actual quality in this method. To address this issue, we present IG-FIQA, a novel approach to guide FIQA training, introducing a weight parameter to alleviate the adverse impact of these classes. This method involves estimating sample intra-class variance at each iteration during training, ensuring minimal computational overhead and straightforward implementation. Furthermore, this paper proposes an on-the-fly data augmentation methodology for improved generalization performance in FIQA. On various benchmark datasets, our proposed method, IG-FIQA, achieved novel state-of-the-art (SOTA) performance.