From an article in The Age on CCTV monitoring to reduce crime: two interesting issues emerge wrt out work. The first relates to information overload monitoring what is going on on all the cameras, and the second is interpreting violent intent from distal visual information alone.
The immediate challenge when monitoring CCTV is information overload. With so many cameras and crowds of people passing by, identifying where trouble may occur is prohibitively complex. The absence of sound greatly compounds this problem (thankfully plans to attach microphones to cameras have so far met with implacable opposition). Is that man talking to a girlfriend or intimidating a vulnerable stranger? Are those two men about to start throwing punches or are they simply mucking around? In the absence of more detailed information, camera operators are unable to make an informed judgment.