Moving object detection and tracking using deep learning neural network and correlation filter

Abstract

Object tracking is a key step in computer vision for video surveillance, public safety, and traffic analysis. Object detection and tracking are the two correlated components of Video Surveillance. Object detection in videos is the first step before performing complicated tasks such as tracking. Deep learning neural networks is a powerful programming paradigm which learns multiple levels of representation and abstraction of data such as images, sound, and text. In this paper Gaussian mixture model (GMM) based object detection, deep learning neural network-based recognition and tracking of objects using correlation filter is proposed, which can handle false detections, with improving the efficiency. The algorithm is designed to detect only cars and humans' while the performance is analyzed using True Positive Rate (TPR) and False Alarm Rate (FAR) as probabilistic metrics. The Experimental results of the proposed method are found to be better with an accuracy of 88%.