Deep Surveillance – Towards Data Science

A report commissioned by the British Security Industry Association in 2013 concluded that 4.1–5.9 million closed-circuit television cameras are in the Britain.[2] The highest estimate amounts to one surveillance camera for every 11 people. In addition to the policy questions surrounding state-owned and private cameras, video surveillance is creating challenges for big data and machine learning. One paper has estimated that 600,000 cameras in the Chinese city of Tianjin create 50 petabytes of data every day.[3] Many surveillance systems still require human supervision. Recent advances in computer vision are, thus, seen as a important trend in video surveillance that could lead to dramatic efficiency gains.