Metadata is data about data.
It sounds confusing, but it does not need to be. Metadata plays a critical role in the video surveillance industry, and without it we would not have video analytics or AI, the innovations that are shaping the industry today.
The easiest way to think about metadata is to compare it to a real world experience. Pretend you found a box with an old painting in it. You might wonder who painted it, where it is from, and who owned it. Without this information, you do not know what the painting is worth or what you should do with it. The information you are missing is not actually part of the painting, but without it the painting is a lot less meaningful.
This is exactly how metadata works for digital files. It is information about the file, or data (information) about data (the file). Without metadata, digital files would be completely disorganized and useless to most. To solve the problem of organization, digital files use metadata, which is divided into three categories: descriptive, structural and administrative.
Pretend you examined the painting and found the artist’s name signed in the corner, that would be descriptive data. It acts as a description of the painting, telling you it is a work of art from a certain painter.
Now imagine you flip over the painting and on the back it says 2/5, meaning the artwork is part of a series of five paintings with this one being number two in the series. The series number represents structural data — it indicates how this object is organized and where it belongs in the complete set of paintings.
Finally, as you inspect the box the painting came in you see a series of shipping labels with different addresses on them. You now know the history of where that painting has been, which is administrative data, the type of information that tells you the when and how of an object.
The painting analogy is a great method for thinking of metadata in the offline world. Of course, the digital version of that same analogy could be done using digital photos. In fact, if you have ever tagged someone in a digital photo, you have created a small piece of descriptive metadata.
In video surveillance, metadata is vital to sorting and searching the enormous volume of video data that is collected every day. In its simplest form, administrative metadata is attached to every video file, so the system knows exactly what time it was created. In addition, camera IDs are a form of structural metadata that allows the systems to associate the video clip to the correct camera. While both administrative and structural metadata are the foundations of video data management, descriptive data is where things have become the most interesting.
Thanks to AI, enormous volumes of descriptive metadata are being created that can be attached to video files to make them infinitely more valuable. AI can identify object types, along with their individual properties like colour, speed, direction of travel, and more. This information is attached to each video file in the form of descriptive metadata and enables the advanced search functionality available in today’s surveillance systems. Want to find the man in the red shirt traveling south on a bicycle? Simply search the metadata.
Without metadata our industry would not be what it is today, and video surveillance footage would simply be a linear recording of events, no different to how we once collected footage on VCR tapes.
Thanks to metadata, we can quickly analyze large volumes of video data in a fraction of the time it would take without it, and while data about data sounds like a terrible definition, it’s a lot easier to understand than most people would think.
Colin Bodbyl is the chief technology officer of Stealth Monitoring (www.stealthmonitoring.com).
Print this page
