One-shot Face Recognition by Promoting Underrepresented Classes

TL;DR

This paper addresses imbalanced large-scale face recognition by introducing a regularizer and a novel promotion loss to improve recognition of underrepresented classes, achieving state-of-the-art results on a benchmark.

Contribution

It proposes a new regularizer and a promotion loss to enhance one-shot face recognition in imbalanced datasets, improving generalization for underrepresented classes.

Findings

Achieved 94.89% recognition accuracy on MS-Celeb-1M low-shot benchmark.

Outperformed all previous methods on the same benchmark.

Enhanced recognition of underrepresented classes with the proposed loss.

Abstract

In this paper, we study the problem of training large-scale face identification model with imbalanced training data. This problem naturally exists in many real scenarios including large-scale celebrity recognition, movie actor annotation, etc. Our solution contains two components. First, we build a face feature extraction model, and improve its performance, especially for the persons with very limited training samples, by introducing a regularizer to the cross entropy loss for the multi-nomial logistic regression (MLR) learning. This regularizer encourages the directions of the face features from the same class to be close to the direction of their corresponding classification weight vector in the logistic regression. Second, we build a multi-class classifier using MLR on top of the learned face feature extraction model. Since the standard MLR has poor generalization capability for the one-shot classes even if these classes have been oversampled, we propose a novel supervision signal called underrepresented-classes promotion loss, which aligns the norms of the weight vectors of the one-shot classes (a.k.a. underrepresented-classes) to those of the normal classes. In addition to the original cross entropy loss, this new loss term effectively promotes the underrepresented classes in the learned model and leads to a remarkable improvement in face recognition performance. We test our solution on the MS-Celeb-1M low-shot learning benchmark task. Our solution recognizes 94.89% of the test images at the precision of 99\% for the one-shot classes. To the best of our knowledge, this is the best performance among all the published methods using this benchmark task with the same setup, including all the participants in the recent MS-Celeb-1M challenge at ICCV 2017.