Safety has always been a concern in transportation on the road but also in on the public transport network.
Transportation authorities have invested massively in video based systems to monitor and enforce safety regulations. We aim to improve two of the limitations these systems present. Computer vision image signature technology can monitor the flow of people and vehicles whilst preserving citizen privacy when no violation is detected. Our vision research on generic scene understanding will be able to convert existing cameras in the field into video monitoring systems trained to detect multiple scenarios where safety is at risk.