Edge computing based surveillance framework for real time activity recognition

Closed Circuit Television (CCTV) based Surveillance has become the fundamental part of the security Systems. In most cases, surveillance feeds are only used as evidence. The emergence of Edge Computing gives hope for enabling real time surveillance systems that focuses on prevention of crimes. The proposed architecture consists of a Convolutional Neural Network (CNN) enabled in an edge device, with reduced computational complexity, which classifies various actions like Pulling, pushing and other hand movements and locates the identified activities in the image frame using bounding boxes. The proposed architecture gives an alert whenever a suspicious activity is detected. The system was found efficient when validated against the Dataset taken from the SRM IST Campus.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.