Review of Automatic Detection & Tracking of Abandoned Object

Abstract

This paper presents a framework for detection of abandoned objects from various surveillance videos by
matching the reference frame with the next successive frames of the video sequence. The first frame of video is
considered as reference frame & then by using background subtraction algorithm the foreground is separated from the
background. Gaussian Mixture Model is used to detect static foreground regions. Features are extracted to classify the
object as human or non-human objects. SVM classifier is used to classify the object as human, bag, mobile phone or any
other object from extracted features. Proposed method can locate very small consistent items within the surveillance
videos