Reinforcement Learning for anomaly detection in real-time camera feed

How to automatically monitor wide critical open areas is a challenge to be addressed. In this project we are looking for using CNN+LSTM technique for identifying anomalies and by using a deep reinforcement learning approach, classify them into one or more groups such as health, crime, accidents etc. This project aims to alleviate this problem by using deep learning reinforcement algorithms to emergency conditions in a video feed. In this way, the intern should work on this real-time data to, at first, finding anomalies from the live video, then, categorize them into relevant classes. Moreover, since the intern currently works on Artificial Consciousness in his PhD studies as well as have a good background in Machine Learning techniques, depends on the provided facilities, wondering he has an idea to apply Consciousness concept to the model in order to increase the ability of anomaly detection like a human.

Faculty Supervisor:

Kin-Choong Yow

Student:

Soheil Ahmadi Vosta Kolaei

Partner:

Intelense Inc

Discipline:

Engineering

Sector:

Information and cultural industries

University:

University of Regina