Activation Template Matching Loss for Explainable Face Recognition

TL;DR

This paper introduces an Explainable Channel Loss (ECLoss) that enables face recognition networks to learn facial part-based features without manual annotations, improving explainability and verification performance.

Contribution

The paper proposes a novel ECLoss that allows networks to learn facial parts in an explainable manner without additional datasets or manual labels.

Findings

ECLoss achieves superior explainability metrics.

ECLoss improves face verification performance.

Visualization confirms effectiveness of ECLoss.

Abstract

Can we construct an explainable face recognition network able to learn a facial part-based feature like eyes, nose, mouth and so forth, without any manual annotation or additionalsion datasets? In this paper, we propose a generic Explainable Channel Loss (ECLoss) to construct an explainable face recognition network. The explainable network trained with ECLoss can easily learn the facial part-based representation on the target convolutional layer, where an individual channel can detect a certain face part. Our experiments on dozens of datasets show that ECLoss achieves superior explainability metrics, and at the same time improves the performance of face verification without face alignment. In addition, our visualization results also illustrate the effectiveness of the proposed ECLoss.