Joint Discriminative and Metric Embedding Learning for Person Re-Identification

TL;DR

This paper proposes a joint learning approach combining discriminative and metric embedding losses, along with attribute prediction, to improve person re-identification accuracy across challenging datasets.

Contribution

It introduces a novel combined loss function that unifies discriminative and metric learning, removing the need for margin tuning and enhancing invariance to nuisance factors.

Findings

Outperforms state-of-the-art on three challenging datasets

Joint loss removes the need for margin in softmax loss

Holistic representations are more practical for deployment

Abstract

Person re-identification is a challenging task because of the high intra-class variance induced by the unrestricted nuisance factors of variations such as pose, illumination, viewpoint, background, and sensor noise. Recent approaches postulate that powerful architectures have the capacity to learn feature representations invariant to nuisance factors, by training them with losses that minimize intra-class variance and maximize inter-class separation, without modeling nuisance factors explicitly. The dominant approaches use either a discriminative loss with margin, like the softmax loss with the additive angular margin, or a metric learning loss, like the triplet loss with batch hard mining of triplets. Since the softmax imposes feature normalization, it limits the gradient flow supervising the feature embedding. We address this by joining the losses and leveraging the triplet loss as a proxy for the missing gradients. We further improve invariance to nuisance factors by adding the discriminative task of predicting attributes. Our extensive evaluation highlights that when only a holistic representation is learned, we consistently outperform the state-of-the-art on the three most challenging datasets. Such representations are easier to deploy in practical systems. Finally, we found that joining the losses removes the requirement for having a margin in the softmax loss while increasing performance.