Identify and Track Important Assets

ObjectVideo proposes a video system to identify faces on non-cooperative subjects at a far distance. The system will be deployed hidden in hostile terrain to monitor people who travel through the camera's view, and it will report back to a central station only if it sees a known fugitive. The output will be a video clip of the person who just crossed the camera's field of view, and a database face shot of the likely match. The hardware will be light enough, small enough, and energy efficient enough to perform these tasks independently. There are three key technical challenges. (1) Acquisition - Development of suitable pan-tilt-zoom (PTZ) algorithms so that the system can find a target's face over a long distance, and enhance it for better identification. (2) Identification - Improvement of current face recognition software to handle different angles and lighting so that it can match against a rudimentary watch list. (3) Design of the hardware that will enable all algorithms to run in an insubstantial form factor. ObjectVideo already has significant experience with computer vision-based automated video surveillance technologies and their application to real-world scenarios, including porting them to hardware.