Unequal-Training for Deep Face Recognition With Long-Tailed Noisy Data

TLDR

Large‑scale face datasets are massive, long‑tailed, and noisy, which makes training difficult. This work proposes an unequal‑training strategy that treats head and tail data differently using noise‑robust loss functions to exploit their distinct characteristics. The framework employs two training streams: one uses head data supervised by a Noise Resistance loss to learn discriminative representations, while the other uses tail data to gradually mine stable discriminative information from confusing tail classes. The unequal‑training framework and loss functions improve performance, reduce GPU memory consumption, and achieve the best result on MegaFace Challenge 2 with large‑scale noisy training data.

Abstract

Large-scale face datasets usually exhibit a massive number of classes, a long-tailed distribution, and severe label noise, which undoubtedly aggravate the difficulty of training. In this paper, we propose a training strategy that treats the head data and the tail data in an unequal way, accompanying with noise-robust loss functions, to take full advantage of their respective characteristics. Specifically, the unequal-training framework provides two training data streams: the first stream applies the head data to learn discriminative face representation supervised by Noise Resistance loss; the second stream applies the tail data to learn auxiliary information by gradually mining the stable discriminative information from confusing tail classes. Consequently, both training streams offer complementary information to deep feature learning. Extensive experiments have demonstrated the effectiveness of the new unequal-training framework and loss functions. Better yet, our method could save a significant amount of GPU memory. With our method, we achieve the best result on MegaFace Challenge 2 (MF2) given a large-scale noisy training data set.