A Real-time web based suspicious behaviour detection system using deep learning and fuzzy logic

Abstract

Many public places, such as shopping malls, avenues, and banks, now have security cameras to ensure the safety of individuals, but manually monitoring them to detect suspicious activity is difficult. When artificial intelligence, machine learning, and deep learning were included into many systems, the technology had progressed too far. So , there is a need for intelligent surveillance systems that can help detect different suspicious activities from live footage tracking. In this project we propose a real time web based suspicious behaviour detection system using deep learning and fuzzy logic, where we used deep learning to extract the features and fuzzy system to classify suspicious actions. The system was developed in a distributed architecture where the classification takes place in the fog layer whereas the detection model to extract features is deployed in the edge layer which boosts the performance of the system significantly.