Detecting and Alerting of Threatening Objects

Author(s):

Ashitha Naik , Nitte Meenakshi Inst, of Technology, Bengaluru, India; Darshan P, Nitte Meenakshi Inst, of Technology, Bengaluru, India; Ajay Vastrad, Nitte Meenakshi Inst, of Technology, Bengaluru, India; Kuruba Sumanth, Nitte Meenakshi Inst, of Technology, Bengaluru, India; Sai Charan, Nitte Meenakshi Inst, of Technology, Bengaluru, India

Keywords:

Deep Learning, Crime Detection, GSM Technology, Anomaly Detection Technique

Abstract:

During last few decades in surveillance cameras have been installed in different locations. The examination of the information was captured using the cameras that can play actual part in incidence prediction, online monitoring and goal driven analysis of applications which includes anomalies and intrusion detection. In recent days many Artificial intelligence techniques are been used to detect anomalies among them CNN using deep learning techniques was enhanced in the detection accuracy significantly. The main aim of this project is to recommend a new technique using on deep learning techniques for the anomaly detection in surveillance cameras. This method has been estimated in the UCSD dataset and showed in increase rate of correctness of the anomaly detection. This paper introduces Crime detection method based on deep learning technique. The architecture of this method has two main sectors which are called train network and detection classifier. The first phase aims for feature extraction and is consisted of five components with a deep structure. The aim of the second phase is detection. This phase is consisted of five deep neural network classifiers and reconstruction network. Each component in detection phase produces a detected class and a score. At last, by these detection classes and scores, the ensemble classifier performs the final detection and announces it..

Other Details:

Manuscript Id	:	IJSTEV6I12013
Published in	:	Volume : 6, Issue : 12
Publication Date	:	01/07/2020
Page(s)	:	22-27

Download Article