Implementation of Human Detection on Raspberry Pi for Smart Surveillance

Abstract

With the rapidly increasing rate of theft and insecurity, surveillance cameras have become a piece of conventional household equipment for many. Although these cameras just record the footage and make it conceivable to review these recordings after a sense of malicious activity being spotted by somebody. There have been approaches to robotize this procedure by introducing a movement discovery framework and utilizing other IoT based sensors. Such solutions were not reliable as, in adverse conditions, these systems could not adequately detect the human presence and during the mishap the individuals deemed necessary were not alerted. In this paper, we introduce a compact and complete solution to this problem, all packed in single Raspberry Pi by implementing an effective object detection model using Machine Learning and integrating it with advanced Image Enhancement technology to detect only the human activity even under adverse conditions. Also, the entire data that is being collected by this system has been end-to-end encrypted by using Cryptography in order to prevent Man in the middle attacks.