Over the past decade, the reliability and functionality of video analytics have advanced to a point where, for many security operations, it is no longer a question of “should we use video analytics?” But rather, “How should we implement video analytics within our technical architecture?”
There are two main options for implementing video analytics software and, depending on the environment, either can be effective. The software containing the video analytics algorithms can be located in one of two broad areas: inside an IP camera (referred to as the edge, as it is at the edge of the network) or on a centralized server (referred to as the center or server-based). A derivative of the edge approach can be used for analog cameras that lack the computing power to run embedded analytics. This is done by placing the video analytics software on video encoders, which are also at the edge of the network.
There are pros and cons to both edge- and server-based approaches. Understanding these differences can help system integrators advise their customers on the optimal approach. For some projects, a combination of both approaches can yield superior results. This method is especially applicable at electric utilities or other organizations that have a number of large facilities as well as many smaller remote locations and limited bandwidth.
The "Tale of the Tape" table describes the criteria for making a decision on whether to run video analytics on the edge or in the center. For an integrator, understanding your customer’s specific environment and requirements will help determine the relative importance of each criterion.
In addition to the parameters outlined, there are other factors that system integrators should consider, such as how the analytics will be used and where the captured video will be stored. Storage is especially important when using advanced forms of video analytics, since they rely on captured video.
Additional considerations with advanced video analytics
In simple video analytics applications, the software analyzes the video, checks for anomalies, and generates an alarm. Advanced video analytics, on the other hand, also record summaries of what the analytics analyze. A good example of this is using video analytics to review the past 24 hours of recorded video to show all of the instances where a person wearing a reddish jacket walked through the scene. The same concept can be applied to size, object, speed, trajectory (direction), aspect ratio or license plate numbers. These more advanced types of video analytics applications generate this hidden information, called metadata, for future use.
In order to query this metadata as part of a forensic search, it must be stored with the video on the NVR, and the VMS must have the interface to query it. This is why it is necessary for the video analytics vendor to partner closely with the VMS vendor; in fact, for highly sophisticated video analytics applications, it is not uncommon for all of the components to be offered as a single-source solution.
What to choose?
Edge analytics are particularly advantageous for remote locations with limited bandwidth. They can be a very effective solution at an attractive price. With the advantage of more CPU power, server-based analytics can run more analytics per camera. Since they are camera agnostic, you have the flexibility to select different cameras based on desired functionality and location requirements.
Criterion |
Edge-based |
Server-based |
---|
Reliability |
Typically, IP cameras have less processing power unless they have a dedicated on-board processor. However, since they have access to raw, uncompressed video, the reliability of the analytics can be high. Reliability would only be problematic if the embedded processer was not powerful enough for the video analytics application.
|
Servers have a lot of processing power, and you can define a maximum number of cameras per server to reduce overload. Since only the frame rate and resolution are transmitted from the camera, server-based solutions are generally very reliable. The only exception to this would be if the quality of the incoming video stream was poor.
|
Video analytics options
|
Most video analytics are able to run on the edge, but due to limited processing power, each camera is usually restricted to running one analytics application at a time. So you might set up a train platform camera to detect either loitering or an abandoned bag, but not both. This is because analytics use a lot of horsepower, which is usually limited within the camera.
|
The video analytics are unrestricted by processing power, which means that multiple analytics applications can run on any camera simultaneously.
|
Selfsufficiency
|
In isolated locations, each with just one or two cameras, you wouldn’t put excessive hardware at such remote locations, especially servers. Such locations would likely be hot, cold, dirty and hard to access for maintenance. With intelligence at the edge, the camera can do everything, even record to embedded storage. Such autonomy means a remote camera can detect threats and trigger actions at an isolated site. With analytics on the edge, the network could be down and the analytics would still run fine.
|
With analytics at the center, cameras at remote sites aren’t self-sufficient.
|
Bandwidth
|
Analytics on the edge are ideal for low-bandwidth connections since analytics occur within the camera and only a simple alarm is transmitted to the center.
|
Server-based analytics can only be used when the incoming video stream meets a minimum quality level, which is defined by frame rate and resolution.
|
Camera choice
|
Camera manufacturers typically only make analytics available on a subset of camera models. This is because: (1) the increased cost of manufacturing such cameras can be prohibitive; and (2) cameras running analytics are a small percentage of all cameras sold. From an end-user standpoint, this translates into fewer camera choices.
|
Being server-based, the analytics are camera agnostic. You can use any camera model from any manufacturer.
|
Substitution capability
|
Most manufacturers use the same analytics software in their entire camera portfolio, so if a customer finds that the analytics don’t work well, they will likely have to replace all the existing cameras with a different brand.
|
If the analytics don’t perform to expectation, you can simply replace the software on a server. Just make sure that the server-based analytics software is compatible with the chosen video management system (VMS) or is a licensable feature of the VMS.
|
Ease of installation
|
A manufacturer’s camera portfolio will typically use the same analytics software, so once you learn to configure one, you can configure all. However, different brands use different software, so if a customer’s surveillance network uses various types of cameras, your installers must be certified and trained to configure them all.
|
Since the software is camera agnostic, it can be used with all brands and models of cameras, and even encoders. This makes it much easier for installers to gain experience and become experts.
|
Carbon footprint
|
An IP camera consumes only a few watts of power and little, if any, additional power to run video analytics.
|
If the same server is used for analytics as well as recording video, the additional power required for video analytics is negligible. If, however, a dedicated server is required for analytics, it could easily consume a few hundred watts of power, which is divided among the connected cameras.
|
Total cost of ownership (TCO)
|
TCO of edge-based analytics is effectively zero when factors such as power, cooling and rack space are factored in. That’s because an IP camera with an embedded SD card doesn’t take up space in a rack, consume a lot of power or place a burden on the HVAC.
|
If analytics are running on the Network Video Recorder (NVR), the TCO is negligible because it is not consuming any power beyond what it needs to record the video. However, if the analytics are running on separate dedicated servers, the TCO can be significant.
|
Price
|
In order to differentiate their hardware and gain a competitive edge, many IP camera manufacturers now include embedded analytics at a very low price point.
|
With server-based analytics, the cost per camera is typically higher. The focus of server-based analytics is more on reliability, power, and feature sets than on cost.
|
Licensing
|
Typically, the camera is either sold as a part number with analytics already enabled or the analytics is a software feature that can be unlocked in the camera for an additional fee (license). The license is typically tied to a specific camera.
|
Server-based analytics are typically licensed on a per-camera basis, but usually this virtual license can be reassigned to a different camera if desired at a later point in time.
|
If your customer is unfamiliar with video analytics and wants to test the analytics with an existing camera, a server-based pilot is the simplest way. If the customer uses IP cameras, all it’ll need to try it out is an analytics license on the server. If the customer uses older analog cameras, you can simply send a copy of the analog signal through a single channel encoder and get results instantly.
There are multiple benefits to using video analytics at the edge (i.e., near or inside the camera). For one thing, analytics at the edge provides the ability to process what is happening in a field of view and discern if a relevant alert is triggered. This can be faster and less expensive than the original video analytics model of using a separate dedicated server.
However, there isn’t one right solution, as a video analytics' complexity and a camera’s processing power are not always aligned. Some analytics can begin the analysis at the camera and also utilize a server to balance the workload. Others may be best used in server-only models. Speed of alert is of importance, as results that are not urgent may not dictate a powerful camera.
Another variable is whether the system needs actual video of an event or just information (metadata) from that video. When recorded video is not required at a server, intelligent cameras at the edge help lessen the required bandwidth. Intelligent cameras and the cloud go hand-in-hand. For example, only metadata is needed when counting people; therefore, intelligent cameras can do all the processing in the camera, and only the metadata is sent to the cloud. For security, only a low-bandwidth stream is sent to the cloud, while the high-resolution video is stored at the camera.
When video is required, the edge advantage becomes far less, since the video must reach the server to be recorded. Having analytics such as face and demographics at the server level keeps the cost of the cameras low since the processor on the server does most of the work. Processing power on servers is far cheaper than having a robust processor in each camera. Analytics that require a lot of processing power greatly increase the cost of the cameras, since they must have a robust processor. When the processing takes place at the server level, the customer can keep overall costs down by using far cheaper cameras and using a centralized server-based system.
Sometimes, a combination is optimal. For example, some products has a patented approach that enables analytics processing both at the server and distributed to the edge. These product’s system operates on a server between the camera and the video management system (VMS), analyzing video streams and providing output of that analysis. A software module runs inside video encoders and cameras at the edge. This software completes "preprocessing" at the edge and sends information to the server, which completes the process and provides the output. Unlike strictly edge-based analytics, the approach is not limited by processing power and memory in the camera. Compared to server-only installations, the system is more scalable (by a factor of 10 to 20 compared to server-based systems).
There is a trend in the market of camera vendors which turning their cameras into open platforms to allow software vendors to load analytics (and other applications) onto the cameras. Previously, software vendors had to work closely with camera vendors, even creating special software versions. Today, the cameras are not yet at the level of an iPhone or Android [platform], but they are much more open and there is greater variety in terms of applications you can load.
Experts believe that edge-based analytics are an interesting alternative to traditional server-based (centralized) solutions. Edge deployment lends itself to a distributed solution where infrastructure is not available, hence where transmitting video of high quality to a centralized server is not an option. Transport (road/rail) has been a major beneficiary of edge-based analytics technology. The lack of infrastructure results in a need for a more complex management of rules and possibly more challenging environmental aspects. In order to operate advanced video analytics solutions at the edge, a suitable hardware platform should be provided with enough processing power. However, often at the edge, the system must be rugged and should operate at high temperature extremes; consequently, the availability of such a hardware platform is less likely. Because of these issues, most manufacturers have opted to offer only basic analytics solutions at the edge.
Edge-based analytics run on raw video data as opposed to encoded video on the server, allowing the analytics to gather more sensitive and accurate data. In addition, it allows the analytics to control the sensor and enable optimized video input for the analytic engine. Edge-based analytic cameras offer a host of benefits to facilities that need to monitor large perimeters, complex campus environments or geographically dispersed open spaces. Edge-based analytic devices do not rely on servers or third-party software. This reduces the network bandwidth requirements while maintaining performance at the highest level. In addition, when technology developers offer a complete solution that ties in edge analytics and video management, users benefit from a single, tightly integrated solution, which means there is less opportunity for failure.
Adapted from 3VR and securityinfotech