Most video analytics operate by simply detecting pixel changes or motion on the camera view. This is useful in low activity settings (like a perimeter fence) where not much happens and security personnel want to pay attention to every activity. However, in busy retail, simple motion-based analytics produce too many false positives. Users looking for when an item on a shelf moved will get hits for everyone who walked in front of that shelf.
In Foreground/Background Separation analytics the software analyzes the video and separates foreground and background motion, people vs. assets, and records those movements as different object types. So when a user wants to see when merchandise leaves the shelf, they do an asset search, looking only for background activity. This results in fewer false positives – getting you to the video you need faster.
For instance, a loss prevention Investigator wants to monitor a rack of expensive, high-theft handbags. Simple motion-based alarms alert him every time a person walks between the camera and the shelf. With Foreground/background analysis, the Investigator defines an asset alert (focus on background activity) and only receives alerts when the merchandise moves.
Additional parameters, allow him to tune the alerts to differentiate between normal customer shopping behavior and a shelf wipeout, for example.
What other aspects of Video Analytics would you like to learn about? Post your question in the comments area.