One-pass person re-identification by sketch online discriminant analysis. (September 2019)

• Record Type: Journal Article
• Authors: Li, Wei-Hong; Zhong, Zhuowei; Zheng, Wei-Shi
• Is Part Of: Pattern recognition. Volume 93(2019:Sep.)
• Page Start: 237
• Page End: 250
• Publication Date: 2019-09
• Subjects: Online learning -- Person re-identification -- Discriminant feature extraction; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2019.03.015
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: SoDA is proposed as an online person re-identification (re-id) approach to effectively alleviate the high cost on space and computational complexity when person re-id methods have to be trained on large-scale or high dimensional data or streaming data. SoDA can efficiently keep the main data variations of all passed samples in a low rank sketch matrix when processing sequential data samples, and estimate the approximate within class variance from the sketch data information. The feature dimension reduction is naturally embedded in the proposed SoDA, and no extra online dimension reduction algorithm is required for high-dimensional data. Solid theoretical analysis on how the optimal feature transformation learned by SoDA sequentially approximates its offline version that is learned on all observed data samples. Extensive experimental results have shown the effectiveness of our SoDA and empirically support our theoretical analysis. Abstract: Person re-identification (re-id) is to match people across disjoint camera views in a multi-camera system, and re-id has been an important technology applied in smart city in recent years. However, the majority of existing person re-id methods assumes all data samples are available in advance for training. However, in a real-world scenario person images detected from multi-camera system are coming sequentially, and thus these methods are not designed for processing sequential data in an online way. While there is a few work on discussing online re-id, most of them require considerable storage of all passed labelled data samples that have been ever observed. In this work, we present an one-pass person re-id model that adapts the re-id model based on each newly observed data and no passed data are required for each update. More specifically, we develop a Sketch online Discriminant Analysis (SoDA) by embedding sketch processing into Fisher discriminant analysis (FDA). SoDA can efficiently keep the main data variations of all passed samples in a low rank matrix when processing sequential data samples, and estimate the approximate within-class variance (i.e. within-class covariance matrix) from the sketch data information. We provide theoretical analysis on the effect of the estimated approximate within-class covariance matrix. In particular, we derive upper and lower bounds on the Fisher discriminant score (i.e. the quotient between between-class variation and within-class variation after feature transformation) in order to investigate how the optimal feature transformation learned by SoDA sequentially approximates the offline FDA that is learned on all observed data. Extensive experimental results have shown the effectiveness of our SoDA and empirically support our theoretical analysis.