What is Video Analytics?

Video analytics refers to two quite different kinds of analysis of videos, image data that can be displayed on a variety of devices, such as mobile devices, computer monitors, and television screens. For one thing, when videos are posted online, video analytics can refer to an analysis of information about those videos and their viewers. The pioneer in this use seems to be YouTube® Insight™. The second meaning of video analytics is the use of computers to make identifications of various kinds without human aid.

The first meaning of video analytics, analyzing video performance, provides data that is meant to be of benefit to video posters, advertising partners, and content partners of YouTube®. Information supplied by Insight™ starts with simple and straightforward information, such as number of views shown over time, and geographical popularity shown on a world map. More complex analyses are also available. These include an analysis of the referrers, that is, insight into how one’s videos were discovered, and the demographics of viewers by age and gender. Beyond that, one can discover more about people who favorited the video and day-by-day subscribers increases.

The second meaning of video analytics encapsulates several very different technology solutions. Video analytics is used in security to detect perimeter violations using Video Motion Detection (VMD), for license plate recognition using Automatic Number Plate Recognition (ANPR), and to count people. VMD uses Closed Circuit TV (CCTV) cameras to detect activity at places and times that there should be none, signaling an intruder. They are rigged to trigger an alarm. False alarms can be caused by birds or animals, adverse weather conditions, and blowing debris.

ANPR, also referred to as License Plate Recognition (LPR), is a particular type of Optical Character Recognition (OCR) in which the pixels on the digital image of a license plate are transformed into ASCII text. ANPR cameras are equipped with Infrared (IR) illumination because this provides the best image quality, particularly with fast-moving vehicles. There are many other names for this technology, including car plate recognition, automatic plate identification, and automatic license recognition.

People counting can serve business ends as well as security purposes. In at least some versions, it can track bidirectional flow past an identified boundary that the user can define. It is useful for educational institutions, prisons, airports, banks, recreation facilities, and retail stores.