Learning Deep Feature Representations with Domain Guided Dropout for Person Re-identification

TL;DR

This paper introduces a pipeline and a novel Domain Guided Dropout algorithm for learning robust deep features across multiple datasets in person re-identification, significantly outperforming existing methods.

Contribution

The paper proposes a domain-guided dropout technique and a multi-domain training pipeline to enhance deep feature learning for person re-identification.

Findings

Outperforms state-of-the-art on multiple datasets

Improves feature robustness across domains

Demonstrates effectiveness of domain-guided dropout

Abstract

Learning generic and robust feature representations with data from multiple domains for the same problem is of great value, especially for the problems that have multiple datasets but none of them are large enough to provide abundant data variations. In this work, we present a pipeline for learning deep feature representations from multiple domains with Convolutional Neural Networks (CNNs). When training a CNN with data from all the domains, some neurons learn representations shared across several domains, while some others are effective only for a specific one. Based on this important observation, we propose a Domain Guided Dropout algorithm to improve the feature learning procedure. Experiments show the effectiveness of our pipeline and the proposed algorithm. Our methods on the person re-identification problem outperform state-of-the-art methods on multiple datasets by large margins.