Smart Video Surveillance Using Deep Learning

Abstract

There is a lot of assessment happening in the business about video observation among them; the occupation of CCTV accounts has been blocked. CCTV cameras are put all around the spots for perception and security. Somewhat recently, there have been progressions in profound learning calculations for reconnaissance. These progressions have shown a fundamental pattern in profound reconnaissance and guaranteeing radical proficiency. The normal employments of deep learning are theft, violence revelation, and area of the chances of impact. This project aims to distinguish strange occasions or peculiarities in recordings utilizing Spatio-temporal autoencoder. We will present a Spatio-temporal autoencoder for this video observation project, which depends on a 3D convolution organization. We train an auto-encoder in this deep learning project for unusual occasion detection on normal videos. We recognize the strange occasion’s dependent on the Euclidean distance of the custom video feed and the casings anticipated by the auto-encoder.