Improving Domain-Adaptive Person Re-Identification by Dual-Alignment Learning With Camera-Aware Image Generation

Domain adaptation in person re-identification (re-ID) has always been challenging, especially for the lack of supervision information on the target domain. Existing methods generally introduced extra supervision by adversarial learning techniques, then added all the augmented data in the training process to optimize the re-ID model. However, the direct utilization of all the generated data not only increases additional computational cost but also ignores the potential correlation between the origin and generated data. In this article, we propose a novel dual-alignment learning framework (DAL) with camera-aware image generation to efficiently and effectively tackle this issue. Specifically, we propose a camera transfer matching module to generate additional training images with different camera styles, and construct the matching pairs with each containing a origin image and one corresponding camera transferred image. To strengthen the correlation of images for each matching pair, we align the pseudo-labels via clustering algorithm to reduce the pseudo-labels distribution discrepancy between the origin and generated images. Besides, to avoid model degeneration affected by some inaccurate pseudo-labels on unlabelled data, we maximize the mutual information to align the image feature representations of matching pair. The DAL allows us to decrease the camera variance and enhance the discrimination ability of re-ID model. Extensive experiments on three large-scale benchmarks demonstrate the superiority of DAL over state-of-the-art methods.

I. Introduction

Person re-identification (re-ID) [1]–[4] has attracted considerable academic attention, attributable to its tremendous capability in security applications, such as person tracking [5], behavior analysis [6] and so on. The scope of person re-ID is to match the images of the same target person captured by non-overlapping cameras [7]. Although deep convolutional neural networks [8]–[10] have significantly prompted the development of person re-ID, the vast majority of existing methods typically require large-scale labelled training data that entails high annotation cost, restricting the practicability and expansibility of person re-ID in real-world applications. Nowadays, how to perform person re-ID in an unsupervised manner is becoming a new research hotspot.