Video analytics has long been heralded as the cutting-edge of video surveillance, and despite some claiming the technology is too costly, too complex and delivers inconsistent performance, the reality is video analytics has never offered as many benefits. Misconceptions about the capabilities of the technology do exist, many based on out-dated thinking or experience with systems that are either not properly developed or poorly implemented. Today’s video analytics offers a growing number of benefits, and with the advent of faster processing, can now deliver a greater level of flexibility for system integrators and end users.
Let’s start out by considering two facts about edge-based video analytics. The first is that if you purchase edge-based analytics options for the lowest possible cost, they are unlikely to be the best available. Edge-based analytics are typically loaded onto a device such as a camera or a codec. If that device costs little or no more than a typical budget camera or codec, you can’t realistically expect the manufacturer to have invested heavily in the R&D of the analysis set-up.
It should also be remembered that not all video analytics technologies are the same, nor should they be. If a device is targeted at the mainstream market for detecting simple motion-based events, it simply cannot be expected to match the performance of a fully configurable advanced analytics engine. Try to make a budget device with simple on-board analytics cope with rippling water or busy scenes with changing illumination, and you’re not going to get good results.
The second fact is that edge-based analytics devices which require nothing more than the camera powering up and a tick box activating in the menu will again not cope with complex scenes or high level requirements. Equally, a device with advanced analytics will perform badly if it installed with no or very few changes to the default settings.
Edge-based analytics require the video stream to match with the system configurations. Factors such as perspective, range, distances and skew can all impact on the results achieved, even with the most advanced solutions.
This isn’t to say that budget devices with video analytics that are quick and easy to configure don’t have any value; they do. The point here is that integrators and installers need to have realistic expectations of the solutions they are implementing.
The defining factor when selecting video analytics is to specify the right level of performance for the task in hand. For integrators and installers seeking to implement smart solutions, it is vital to select edge-based video analytics that are robust, reliable and capable of performing in challenging environments. A good degree of configuration is essential, and a key feature must be well-developed analytics that empower the user.
Defining the unknown
Video analytics offer a great aide for those seeking to protect a site. Considering that the vast majority of captured video will have nothing of interest in it, video analytics can flag exceptions for the user, making the security on offer more efficient.
Integrators and installers will understand the performance criteria required for successful video analytics, and the purpose here is not to reiterate those. However, in most applications, the typical process is to identify potential threats and use video analytics to protect the relevant areas.
For example, if the entrance to a site is protected by gates or barriers, video analytics might be deployed to ascertain whether there is any traffic, either vehicular or pedestrian, through those gates at certain times of the day. These could be flagged as alerts, highlighting access and egress to the security team. This is a fairly common use of video analytics, and could be achieved with very basic analytics rules.
However, what happens if a more complex situation arises? For the purpose of this example, consider the situation if after entering through the gateway, vehicles and pedestrians had a choice of turning left or right. Following an incident, it becomes necessary for the system to flag all vehicles that turned right in the past 30 days, or all pedestrians who turned left in the same time period, or even both.
Video analytics are sold as a smart technology, so many end users would assume that this would be a simple task. However, the truth is that many systems using edge-based analytics wouldn’t be of any use.
An operator would have to call up all entry events and manually assign them as of interest or not of interest. For an end user who has invested in a smarter solution, such an outcome doesn’t paint video analytics in a very positive light.
Because video analytics rely on a configured definition of the types of behaviour they should be detecting, this can create issues if the security team discovers an unknown risk, or a threat which was not anticipated. Changing the analytics to address the new threat can help going forwards, but it does mean that unexpected behaviours may well have been missed in archived footage.
For a video analytics system to be considered smart, it should be possible for the collected metadata to be examined, thus enabling specified behaviours to be identified after events have occurred. This means users can go back and reanalyse footage, quickly and easily, to identify risks and threats, establish trends and assess footage, even if the video analytics was not configured to do so at the time of the image capture.
This ability adds value to a video analytics system, and creates an enhanced return on investment for the end user. It is possible with edge-devices that capture metadata, such as is the case with the Intelligent Video Analytics and Essential Video Analytics offerings from Bosch Security.
Bosch offers two levels of video analytics: Intelligent Video Analytics and Essential Video Analytics. Both are edge-based and licence-free. The level of the analytics implementation depends upon the specification of the camera being used.
Essential Video Analytics is the ideal analytics solution for small and medium businesses, retail stores, commercial buildings, warehouses and logistics depots. It can be used for advanced intrusion detection (such as loitering alarms and identifying a person or object entering a pre-defined field), enforcing health and safety regulations, parking management, detecting blocked emergency exits or objects that have been left behind, and analysing behaviour in various environments.
Intelligent Video Analytics offers the highest level of built-in video analytics and is designed for demanding environments. This solution is ideal for mission-critical applications such as the perimeter protection of airports, critical infrastructures and government buildings, border security and traffic monitoring (wrong-way detection, traffic counts, and parking violation detection).
Intelligent Video Analytics does everything Essential Video Analytics does, and also has the ability to perform video content analysis over large areas and differentiate between genuine security events and known false triggers such as snow, wind (moving trees), rain, hail and water reflections, all of which can make video data more difficult to interpret with standard analytics engines. Cameras with Intelligent Video Analytics are also resistant to camera shaking caused by wind, city traffic or knocks against the camera pole.
The demand for smarter analytics and the increasing number of video streams that require analysis overstretches resources with many analytics options. Bosch has therefore engineered its video analytics at the edge to keep data manageable. It achieves this by recording the metadata along with the video. If required, metadata can be retained without video, ensuring critical information is preserved in a lightweight format.
Metadata adds sense and structure to video footage at the point of capture. This enables users to stream only what’s relevant, retrieve evidence quickly and simply, and trigger alarms when needed. It eliminates the need to process hours of video data, making it easier to manage than ever before.
Essential Video Analytics and Intelligent Video Analytics deliver a smart option, but metadata archiving adds value for end users.