Smart Surveillence Robot for Weapon Detection
Savitha Acharya N · International Journal for Research in Applied Science and Engineering Technology · 2019
The present day surveillance and control systems still require human supervision and intervention.Also most of the surveillance systems are static and are used to monitor only limited area (CCTV).Hence we need a dynamic, robust system which can detect and alert about hand-held weapons (such as pistols and knives) and also the system needs to be mobile.This work presents a novel smart surveillance system which consists of a robot that is used for streaming live videos and can be remotely controlled .This is accomplished through the use of Raspberry-pi which controls the robot streaming the live video and artificial neural network trained to determine the presence of a threat.We reformulate this detection problem into the problem of minimizing false positives and solve it by building the key training data-set guided by the results of a neural network classifier.Histogram of Gradients (HOG) features have been chosen for the feature extraction phase and the extracted features are fed to neural network with a few hidden layers to achieve detection.Once the weapon has been detected , the system generates an alert message and automatically sends an alert message to the designated phone number.This system also aims in reducing the realtime processing time to deliver almost lag-free experience.