For decades, most cameras placed in public spaces were used for crime deterrence and evidence collection. They primarily provided surveillance of property, parking lots, and vehicles. The use of video cameras at traffic intersections and other areas of high pedestrian traffic rapidly expanded the number of deployed devices used for investigations of past events. The effectiveness of these applications was almost entirely dependent on the ability of humans to locate, view, and gather facts from archived video (Robin, L., Peterson, B. E., & Lawrence, D. S. (2020). Public Surveillance Cameras and Crime. ). The required effort was so great that most recorded data was never accessed or used.
Beginning around 2012, the development of relatively low cost and powerful hardware and software systems for automated video analysis expanded the possible use cases for cameras beyond surveillance. The availability of accurate object detection using deep learning techniques with data from cameras combined with the evolution of AI hardware accelerators has spawned several video analysis applications in the fields of retail business, medicine, and sports.