Unsupervised Person Re-identification by Deep Learning Tracklet Association

TL;DR

This paper introduces TAUDL, an unsupervised deep learning framework that automatically learns person re-identification from video tracklets without manual labeling, improving scalability and performance across multiple datasets.

Contribution

The paper proposes a novel unsupervised deep learning approach that jointly learns within-camera tracklet association and cross-camera correlation, eliminating the need for manual pairwise labeling.

Findings

Outperforms state-of-the-art unsupervised re-id methods

Effective across six benchmark datasets

Demonstrates strong scalability and accuracy

Abstract

Mostexistingpersonre-identification(re-id)methods relyon supervised model learning on per-camera-pair manually labelled pairwise training data. This leads to poor scalability in practical re-id deployment due to the lack of exhaustive identity labelling of image positive and negative pairs for every camera pair. In this work, we address this problem by proposing an unsupervised re-id deep learning approach capable of incrementally discovering and exploiting the underlying re-id discriminative information from automatically generated person tracklet data from videos in an end-to-end model optimisation. We formulate a Tracklet Association Unsupervised Deep Learning (TAUDL) framework characterised by jointly learning per-camera (within-camera) tracklet association (labelling) and cross-camera tracklet correlation by maximising the discovery of most likely tracklet relationships across camera views. Extensive experiments demonstrate the superiority of the proposed TAUDL model over the state-of-the-art unsupervised and domain adaptation re- id methods using six person re-id benchmarking datasets.