printer friendly version

Surveillance on the edge

Edge processing, a practical solution that has been available for some time, has proven its utility in various scenarios, tailored to the unique requirements of each user. Today, we see the evolution of edge processing with the integration of AI capabilities, empowering camera operating systems to load and modify AI modules for specific tasks, similar to managing apps on a smartphone.

Edge processing has many benefits in remote locations, such as reduced bandwidth, lower latency, privacy, and scalability, but are these needed in locations where infrastructure is plentiful? SMART Security Solutions asked two companies involved in edge (and traditional surveillance operations and products) about the market’s general view and uptake of edge processing.

Our questions went to:

• Marcel Bruyns, Axis Communications.

• Rodney Taylor, Guardian Eye.

SMART Security Solutions: How have you experienced the market adoption of edge processing in surveillance? Is it commonly used or only in specific scenarios?

Bruyns: Over the past few years, we have seen the adoption of edge processing increase dramatically. We have seen a larger uptake in specific scenarios, for example, our perimeter defender solution allows for scalability and expansion with a reduction of risk from server failure to ensure a maximum up time of the perimeter security.

Marcel Bruyns

Taylor: The market and especially video monitoring companies, welcomed the progress in edge analytics. Traditionally massive amounts of video had to be processed by control room operators, or guards, to filter relevant events from general footage. Today it is not a question of analytics as an option, but rather a necessity to effectively manage video monitoring.

SMART Security Solutions: What exactly is defined as ‘edge processing’? Is it a camera operating on its own, or can it be an ‘edge appliance’? Would an edge appliance not be the same as having a server/NVR onsite? What would the customer benefit be?

Bruyns: At Axis, we offer an open platform that allows our products to be used in multiple applications. Our Axis Camera Application Platform (ACAP for Short) allows Axis and integration partners to develop applications to run on the edge. These could be AI applications for human, vehicle, or object classification or complicated applications like facial recognition or ANPR systems. We have used this platform for many years to perform our basic functional applications, such as motion detection.

Our products can operate independently. We have our own edge-processing VMS system called Axis Camera Companion. This VMS uses camera processing to run the application with a secure link to a cloud-based platform, negating the need for an on-premises server. Axis Camera Companion has been part of our portfolio for many years. It allows for the management of multiple sites with a simple-to-use platform.

ACAP allows for complete flexibility of our products; it allows for complete stand-alone systems or complicated hybrid systems. This allows for scalability and future-proofing of your system, allowing applications to be added at a later stage, or the flexibility to change the system as the needs change. With the processing happening on the edge, it allows for less risk from a single point of failure; for example, if a server running analytics fails, the system using that server is no longer operational, but on the edge, if the camera fails, you would only lose the functionality of that device. Along with the edge process, we have a reliable edge storage function making use of SD card storage.

Rodney Taylor

Taylor: Edge processing can be on the camera or the NVR. When using more advanced analytics, a separate processing device can be placed onsite to do the initial filtering. Edge analytics on the camera itself is advancing rapidly, and added features are constantly being upgraded. The processing of the video stream in the camera allows for filtering of events to only send the required events through to a monitoring station. This saves a massive amount of video storage and greatly reduces costs. From an operational point of view, the monitoring staff can focus on real events as the false events are greatly reduced. Added value is that the data in the form of events can be processed much faster to facilitate accurate reporting and review of historical events.

SMART Security Solutions: Edge solutions need processing power for AI and analytics to do their job. Does this mean edge devices (cameras specifically) are the more expensive equipment because of the hardware and software required?

Bruyns: It is true that AI and analytics require more processing power to operate on the edge. To do this we continually invest in developing our System on Chip platform to cope with the demands of edge processing. We do, however, see a natural increase in processing ability from chip manufacturers with a reduced overall cost. The investment in processing does add benefits to the camera as a stand-alone device in terms of more efficient power consumption and improved video quality, for example. Based on this, we have not seen a large increase in the cost of the hardware, and it has allowed us to improve the on-device operating system, Axis OS, which adds further benefits to a quality product.

Taylor: With the development of AI and the massive competition between camera manufacturers, camera prices are affordable as there is no need for expensive NVRs and massive amounts of storage capabilities. Additional AI processing devices are also not required if the correct cameras are selected. All of this drives the installation costs down, resulting in a saving passed onto the client.

SMART Security Solutions: Would surveillance operations not be enhanced by edge cameras talking to cloud servers, eliminating the control room/server room/technology requirements (except for networking equipment)?

Bruyns: Yes, better edge processing will enhance operations. One example is pre-classification on the edge of objects, which allows cloud solutions to receive the information without a larger file format video attached. Processing on the edge also reduces the size requirement of the servers, either onsite or in the cloud.

Taylor: Surveillance operations are set up for failure without analytics. Whether the analytics are camera-based or additional edge devices, they have become a necessity rather than a ‘nice to have’.

The massive expansion of network capabilities has enhanced the capability to transfer more video to cloud servers, reducing the requirement for onsite equipment. This also has the benefit of centrally managed servers and equipment. Cloud servers are more secure as no equipment onsite can be damaged or manipulated.

SMART Security Solutions: What services/solutions does your company offer in the edge surveillance market?

Bruyns: We offer a wide range of solutions:

• Axis Camera Companion (on edge VMS).

• Axis Object Analytics (our AI-based object classification application).

• Axis Retail Suite (Axis retail analytics for people counting, queue monitoring, etc.).

• Axis Perimeter Defender (edge-based perimeter solution).

• Axis Licence Plate Recognition.

• Audio analytics, barcode reader (for cameras), Axis Face Detector, Axis Live Privacy Shield, Fence Guard, Motion Guard and Loitering Detection, Radar Visualizer, and speed monitoring.

Taylor: Guardian Eye offers edge devices in the form of cameras, NVRs and onsite edge devices as well as cloud-based servers with cloud storage of events. Each solution design is built from scratch to facilitate the best option depending on the client risk analysis and network capabilities.

The Guardian Eye Nerve Centre capability is built to accommodate any form of event and alarm into a centralised platform to facilitate optimal operational performance.

Share this article:

Further reading: