Video analytics thought about it

rabiakhatun

Local digital technology with the function of analyzing data “on board”, before transmitting information to the server, is today commonly referred to as Internet of Things (IoT) devices or edge computing solutions.

These definitions are quite applicable to IP cameras with built-in video analytics. They collect, process and transmit data without human intervention – like other IoT devices. And at the same time, they bring computing devices and data storage as close as possible to information sources and consumers – like any Edge Computing tools.

In essence, these are different content writing service names for the same thing. We refer to analytics and information processing directly "on board" the camera as edge computing, while the interaction of such cameras with the central event processing server is called the Internet of Things.

When it comes to an IP camera that streams video for processing on a remote server, it acts as a source of raw video data. Cloud video analytics therefore does not involve using the camera for any edge computing.



Transmitting video streams to a central server puts a significant load on the data exchange network. And an intelligent IP camera transmits only useful information to the monitoring center - for example, a read vehicle number, a motion signal, a short clip or a photo with changes. Even the most modern codecs for compressing video data are not able to unload communication channels as effectively as the computational analytics "on board" the camera does.

The demand for such solutions first appeared at the dawn of the development of cellular communications, when transmission channels were relatively weak.

Today, when high-quality video traffic amounts to gigabytes, in-camera analytics can reduce it to megabytes or even kilobytes of data.

A modern IP camera with built-in video analytics can also be delegated a number of additional security system functions to respond to local events. For example, a camera that detects motion can be programmed to trigger an alarm or an intrusion message. If built-in analytics can determine the type of moving object, it can, depending on the situation, play a message about prohibiting entry of cars, pedestrians, and at the same time ignore interference such as birds and small animals.

It is necessary to take into account that centralized video analytics directly depends on the performance of the server and communication lines. Failure of any component of such a system is a serious accident. At the same time, a camera with built-in video analytics, in the absence of communication with the center, will perform its functions autonomously and will continue to transmit data immediately after the problems are resolved.

Most modern cameras with built-in video analytics based on artificial intelligence (AI) have flexible data storage settings, which ensures the safety of important information even during long-term failures.

DATA SECURITY

Video analytics in a camera can be divided into two main types. Basic analytics is motion detection using a software image processing unit. Its mathematical algorithms analyze the change of pixels in the frame. The downside is that it reacts to almost any movement, such as a gust of wind.