Attributes in Multiple Facial Images

TL;DR

This paper investigates the inconsistency of facial attribute predictions across multiple images of the same subject and proposes methods to improve attribute estimation accuracy in such scenarios.

Contribution

It introduces two novel approaches to address attribute inconsistency across multiple images and demonstrates their effectiveness on large public datasets.

Findings

Methods improve attribute estimation consistency across multiple images.

Proposed approaches correct incorrect attribute labels.

Effective on both still images and video frames.

Abstract

Facial attribute recognition is conventionally computed from a single image. In practice, each subject may have multiple face images. Taking the eye size as an example, it should not change, but it may have different estimation in multiple images, which would make a negative impact on face recognition. Thus, how to compute these attributes corresponding to each subject rather than each single image is a profound work. To address this question, we deploy deep training for facial attributes prediction, and we explore the inconsistency issue among the attributes computed from each single image. Then, we develop two approaches to address the inconsistency issue. Experimental results show that the proposed methods can handle facial attribute estimation on either multiple still images or video frames, and can correct the incorrectly annotated labels. The experiments are conducted on two large public databases with annotations of facial attributes.