Smart surveillance systems may soon detect violent behavior

If implanting employees with RFID chips violates practical HR policies, and rolling out eagle-eyed drones to monitor defensive behaviors seems a bit too intrusive, researchers at the University of Texas in Austin are hoping its smart surveillance system can lend a hand in detecting that pent-up rage. The "computer vision system" can reportedly analyze human movements as they occur, and distinguish between "friendly behaviors such as shaking hands, and aggressive actions like punching, pushing," or launching pocket rockets. The hope is that this creation will make the oh-so-platitudinal jobs of security guards even less demanding by alerting those on duty of violent fits automatically, curing the problem of sleeping through a battle royale in the east parking deck. Developers created the baseline by breaking down CCTV films and "examining the interplay of different clusters of pixels in order to classify interactions between individuals." The aptly-dubbed "semantic analysis" allows the software to assign a probability that a particular activity (like exchanging trade secrets) is being observed. While the current system has been "80 percent" accurate in testing, a computer vision guru at the University of Leeds, UK claims that it needs a bit of refinement before loosed in security bunkers, but it could probably thrive when keeping watch over those sure-to-be-tumultuousWiidemo kiosks.