Self-supervised learning for fully unsupervised re-identification in real-world applications

Re-Identification (ReID) enables real-world applications such as AI-powered surveillance, criminal identification, event understanding, and smart city development. However, it remains challenging due to occlusions, viewpoint changes, and background similarities. Supervised methods perform well but rely on costly, biased annotations, limiting scalability. To address this, we propose self-supervised algorithms for Unsupervised ReID (U-ReID), extendable to modalities such as Text Authorship Verification, tackling high intra-class variation and low inter-class distinction. Our work introduces three fully unsupervised ReID methods: one using camera labels, one without side information, and one scalable to large datasets. We also present a fourth hybrid method for long-range recognition under distortions. These solutions enhance generalization and enable real-world applications in forensics and biometrics.