Real-Time Intelligent Surveillance using Ethical Anomoly Detection

S. Reid, C. Neff
United States

Keywords: surveillance, AI, Autonomy, Privacy, ethics


Real-time situational awareness is crucial for public and private security and safety. Our technology uses state-of-the-art research in computer vision to perform complex cognitive tasks such as action recognition and anomalous behavior detection. This technology is vital for enabling effective real-time video surveillance at scale, where current methods fail to deliver. First, our technology uses artificial intelligence to detect objects, persons, their movement, and to track those people through the camera environment. Then, those poses and movements are used to determine what actions people are taking, and if any anomalous behaviors are present in the environment. This allows for vast amounts of video feed to be processed in real-time without spending exorbitant amounts of human resources monitoring each individual camera feed. Additionally, by only using movement information, these two tasks do not have access to any information relating to appearance or any other factors that can be used to discriminate based on race, gender, sexuality, socio-economic status, or any other protected class. Beyond the privacy benefits, this also removes any inherent biases that might be learned from historical data or biased interpretations of behaviors, leading to a truly unbiased system.