Surveillance Data Analytics

Abstract

In surveillance, object analysis refers to treat an object as a whole, while object analytics means taking each object into multiple parts and analyzing the components from various aspects. In this chapter, we mainly discuss those computable features of a visual object and how to use them for visual object analysis which includes object segmentation, detection and recognition, classification, tracking Murray and Basu (IEEE Trans Pattern Anal Mach Intell 16(5):449–459, 1994), annotation, labeling and ontology, search or retrieval, etc. Bimbo (1999); Hampapur et al. (IEEE Signal Process 22(2):38–51, 2005). We also will introduce deep learning-based visual object analysis in this chapter Goodfellow et al. (2016); LeCun and Bengio (255–258, 1995); LeCun et al. (Nature 521:436–444, 2015).