Detecting anomalies in surveillance videos is an important research problem in computer vision. In this thesis, we propose two deep network architectures for anomaly detection, Anomaly Detection Network (ADNet) and Anomaly ...