Detecting abandoned packages in a multi-camera video surveillance system

Robust Unattended and Stolen Object Detection by Fusing Simple Algorithms

“…This module takes into account pixel differences between consecutive frames and the frame-byframe decisions accumulation. In [4], a scheme based on chromaticity distortion is presented. This system is able to detect shadows, highlights and foreground regions.…”

“…For example, in [3] a system that accumulates temporal foreground results and a neural network classifier are used to discriminate between people, lighting effects and unattended objects. In [4] a multiple-state model of an unattended package provides the ability to detect realistic unattended package events. In [7] a long-term logic is used to differentiate between unattended objects and stationary people, and is robust to temporary occlusion of potential unattended objects.…”

Occlusion and Abandoned Object Detection for Surveillance Applications

“…Stauffer and Grimson [27] present an event detection module that classifies objects, including abandoned objects, using a neural network, but is limited to detecting only one abandoned object at a time. In [28] a multicamera video surveillance system is proposed to detect the owner of each abandoned object is determined and tracked using distance and time constraints and multiple cameras with overlapping fields of view are exploited to cope with occlusion of various types.…”

Automatic Classification of Abandoned Objects for Surveillance of Public Premises

“…Generally, there are three main approaches to classification: shape-based classification (e.g., [1,[8][9][10]12]), motion-based classification (e.g., [3]) or combined shape-motion classification (e.g., [2,6,13]). …”

“…To date, in the visual surveillance research area, the literature reports attempts to analyze four main categories of objects, namely, person, vehicle, group of people, and package (e.g., [1,2,6,[8][9][10][12][13]). …”