Occluded Cloth-Changing Person Re-Identification via Occlusion-aware Appearance and Shape Reasoning

Existing methods in Person Re-Identification (ReID) often fail when simultaneously confronted with occlusions and clothing changes. In this paper, we introduce a challenging yet practical task called Occluded Cloth-Changing Re-ID (OCCRe-ID/). We propose Occlusion-aware Appearance and Shape Reasoning, the first framework OCCReID. We first propose an occlusion synthesis strategy to expose the model to real-world occlusion variations. We mitigate clothing changes by coupling silhouette-based body shape information with appearance. Unlike previous works that directly leverage unreliable features extracted from occluded images by off-the-shelf backbones, we propose an occlusion-awareness strategy to handle occlusions for ReID. An occlusion detection module is elaborately designed to generate occlusion-aware feature, which is then used to guide the framework to reason robust appearance and shape features. Extensive experiments demonstrate the superiority of our framework over both cloth-changing Re-ID and occluded Re-ID methods.

1. Introduction

Person Re-Identification (Re-ID) involves matching the same person in a non-overlapping camera system. Since the emergence of deep learning, person Re-ID has advanced significantly [30, 34, 20]. These works assume a simplistic Re-ID scenario where the target person reappears with the same clothing and fully observable appearance without occlusions. However, in real-world, occlusions and clothing changes occur often as shown in Figure 1. This leads to unreliable appearance and a dramatic drop in matching accuracy for traditional Re-ID methods [28]. Addressing this shortcoming, we introduce a novel Re-ID task called Occluded Cloth-Changing Person Re-ID (OCCRe-ID).