Object Movement Detection by Real-Time Deep Learning for Security Surveillance Camera
Abdolreza Abhari, Sepideh Banihashemi, Jason Li · 2017
Developing a smart Web Video Player application connected to a security surveillance camera to keep track of the object of interest is an ongoing research. This paper presents a methodology to real time data mining of the sequence of frames from a live stream collected by security camera by processing trajectories of an object of interest. Two classifiers and a clustering method are implemented all working in real-time. Real-time Deep Learning and Support Vector Machines (SVM) machine learning algorithms are implemented on a local server without the use of cloud computing. This is a popular architecture for many buildings and industries who want to have an in-house smart security camera application.