Adaptive L2 Regularization in Person Re-Identification

TL;DR

This paper proposes an adaptive L2 regularization method for person re-identification that updates regularization factors during training, leading to improved performance on multiple datasets, especially MSMT17.

Contribution

It introduces trainable, adaptively updated regularization factors in person re-identification models, replacing fixed regularization parameters.

Findings

Achieves state-of-the-art results on MSMT17 dataset.

Demonstrates improved performance on Market-1501 and DukeMTMC-reID datasets.

Validates the effectiveness of adaptive regularization through extensive experiments.

Abstract

We introduce an adaptive L2 regularization mechanism in the setting of person re-identification. In the literature, it is common practice to utilize hand-picked regularization factors which remain constant throughout the training procedure. Unlike existing approaches, the regularization factors in our proposed method are updated adaptively through backpropagation. This is achieved by incorporating trainable scalar variables as the regularization factors, which are further fed into a scaled hard sigmoid function. Extensive experiments on the Market-1501, DukeMTMC-reID and MSMT17 datasets validate the effectiveness of our framework. Most notably, we obtain state-of-the-art performance on MSMT17, which is the largest dataset for person re-identification. Source code is publicly available at https://github.com/nixingyang/AdaptiveL2Regularization.