Researchers, through the Army-funded research dubbed Mind's Eye, have created intelligent software that recognizes human activities in video and can predict what might just happen next; sounding an alarm if it detects anomalous behavior.
The high-level artificial visual intelligence system that has been presented in the paper [PDF], titled “Using Ontologies in a Cognitive-Grounded System: Automatic Action Recognition in Video Surveillance," focuses on artificial intelligence cognitive engine that automatically detects and interprets a person's actions using a surveillance feed.
The systems uses years of advancement in the field of surveillance and specifically in object recognition. Researchers behind the system have also incorporated activity forecasting [PDF] that tries to understand what humans will do based on their physical trajectories.
CCTV systems of today are more or less reactive in a way. Cameras tend to only relay the captured feed to some place where analyses of feed happen; that too by humans with associated risks of distraction and drowsiness. Researchers behind this intelligent software have designed the software to be proactive such that it helps in preventing crimes and dangerous events from happening.
Surveillance systems have become economical to the point that almost all public places have got cameras installed. Decision makers are of the opinion that beyond the CCTV and other related systems, the main cost associated with surveillance systems are personnel employed to analyze the imagery. This particular research and the system that has resulted from it, if mature enough, will probably tempt decision makers into replacing humans involved in imagery analysis.