How to make use of video's metadata treasure trove

Nov. 20, 2017
Potential uses for video data within security and beyond will be unlimited in the future

Businesses today demand far more detail from their video surveillance footage than ever before. Higher resolutions, higher frame rates, better light sensitivity and excellent dynamic ranges all matter when it comes to capturing images that can distinguish individuals or objects for irrefutable evidence. But the growing demand for more detail and the increasing number of networked cameras has other consequences. Not only does the amount of data skyrocket, all that data has to be searched through when looking for evidence or monitored live. Yet statistics say that only 10 percent of data is ever used.

Video analytics is a powerful tool that can help to manage and improve usability of the vast quantities of video data. It acts as a never-tiring operator assistant within the security system, capturing data on everything that happens in each monitored scene around the clock. Content analysis information, in the form of metadata, is stored alongside the video images. This adds sense and structure to video footage, enabling network video cameras to describe what they’re seeing and alert operators to any potential threats the moment they happen. Operators can take the appropriate action faster, easier and more efficiently.

With the generation of metadata, each camera in effect becomes smart, analyzing the scene and providing all kinds of statistics. The camera classifies all objects entering or leaving the monitored scene as a person, a bike, a car, or a truck. It also adds object attributes like speed, geolocation, direction, color, size and more. For even more specific alarm detection, trigger rules can be configured to capture specialized behavior patterns like loitering, following a path, left luggage, entering an area, and more. The alarm engine in each camera can filter on these attributes and can logically combine trigger rules. This allows for very robust and specific detection of alarm conditions while preventing false and missed alarms.

The recorded metadata also enables forensic search on previously recorded video. This enables the operator to quickly find video of that red car whose driver did not pay at the gas station or how many people passed through a door during a certain period, even when these searches were not known during the recording of the video and metadata. The metadata is very compact, hardly adds to the size of the recordings and can be searched through very quickly.

Common Uses of Video Analytics

Video analytics adds value to video footage, as it alerts operators as soon as certain alarm rules are met, such as someone approaching or climbing over a fence. It can even be configured to recognize changes in speed (running), shape (crouching) or aspect ratio (falling).

Common uses of video analytics include detecting loitering, left objects, and a person or object entering a pre-defined field. This can help alert operators to perimeter breaches, people loitering near a sensitive area, theft or unauthorized removal of objects, and much more. Analytics are also being used to enforce health and safety regulations – such as monitoring no-parking zones or detecting blocked emergency exits – as well as for ship tracking and traffic monitoring, such as wrong-way detection, traffic counts, and monitoring roadsides for parked cars.

Video analytics is also being used to optimize business processes based on, for example, people counting or crowd density information to better control check-out lines in large stores. In addition, retailers use video analytics to count people entering and exiting stores to determine numbers of shoppers and days with peak traffic.

Future Uses for Metadata

In the future, the metadata automatically and continuously generated by video analytics will bring added value for both security and non-security applications. In-camera analytics for traffic applications will produce metadata that can help intersection controller systems to improve traffic throughput and pedestrian safety, and reduce pollution and CO2 emissions, or to help traffic planners identify potentially dangerous traffic situations that need to be addressed.  It can also help analyze usage patterns of parking lots. In more and more applications, the camera will be used as an intelligent sensor of which only the metadata will be used, not the video itself.

For example, for retailers, metadata can be used to identify shoppers’ movements throughout a store. Through sophisticated cloud-based processing of the metadata, this information can be gathered without the actual video streams ever leaving the retailer’s premise, protecting the privacy of individual shoppers. The metadata can be mined into performance results to enable retailers to see how shoppers are moving through the store – where they walk, stop and dwell – and to determine engagement rates at a display. This can help merchandisers evaluate the performance of each display in a store, providing them the information they need to increase customer engagement with products, which can lead to increased sales and revenue.

In the realm of smart cities, metadata gathered from video cameras could be used to control lighting in a city based on the presence of people or to improve transportation planning based on people counting. Metadata can also be used for logistics optimization to improve container flow for shipping and delivery services.

These are just a few examples of more advanced uses of metadata. However, the possibilities within and beyond security are unlimited. The door is open for the development of new IoT applications that can assist in gathering and uncovering trends and actionable data from metadata generated by intelligent video devices. It’s time to rethink what video security can do!

About the Author:

Sean Murphy is the director of regional marketing—video systems for Bosch Security Systems. He can be reached at [email protected]