Video management for the future

The VMS has been around for many years and in various forms as the basis for surveillance projects. The ‘various forms’ in the past have included full enterprise-level systems that can

manage hundreds or even thousands of cameras, down to free versions that are developed and supplied by camera manufacturers, as well as free versions from the larger vendors that limit the number of cameras one can use.

The VMS as it is today does its job well and we have seen the focus moving from managing cameras to including more analytics and AI functionality to add value to the user. These value-adding functions can be included in the systems as a whole, but more often they are sold as additional add-ins and the question here has become whether you rely on one company for this additional functionality, or whether you adopt a platform approach that allows you to integrate third-party applications into your base VMS.

The cloud is also starting to become a feature in the VMS space, allowing users to avoid the IT infrastructure costs and complexity of hosting your servers in-house. This changes the game for hardware vendors, but allows the VMS vendors to adapt their systems to a hosted model – hosted by the vendor or third parties.

The question also needs to be raised as to how important these complex applications will be in the future. With an increase in edge storage and analysis capabilities on modern cameras, much (or some, depending on who you ask) of the processing done by the VMS in the past can be deployed to the edge. Using the S&ST; model, most of the VMS processing is removed, which could make cloud-based services more attractive.

Naturally, few organisations will be prepared to throw all their eggs into one basket, which we believe will lead to a more ‘hybrid’ approach, making use of the best features cloud, edge and on-site systems can provide.

Additionally, we have seen the appliance model holding ground and even growing. The model sees vendors supplying an appliance that includes a server, storage and VMS and makes it simple to plug in a screen and the cameras and use it without any hassles (or very few). While these solutions are more common in small and medium surveillance projects, they still remove much of the infrastructure complexity to make it a viable and attractive approach for many.

And let us not forget incorporating additional security and IoT monitoring into the same platform for broader security and operations management. In this scenario, the existing platforms have an advantage as they can integrate features themselves, or from third parties if their systems are open.

So, what are we looking at when it comes to the future of the VMS as we know it today? Hi-Tech Security Solutions asked four companies either developing or representing some of the leading global VMS systems for their insights. We received input from:

Hi-Tech Security Solutions: What do you see as the future of the VMS? Will these systems still be the large installations requiring lots of server power, or will the functionality be broken up to take advantage of the available processing, storage and other benefits of servers, edge and cloud?

Gus Brecher: There is one thing that is certain and that is that there will always be some form of VMS to manage video and information, regardless of where it resides. If you think about it, historically, existing sites have had a dedicated network connecting to devices on the network with centralised management and a storage repository, which is in effect a private, site-based ‘cloud’. What has happened of late is that, because of available bandwidth and services like AWS, there is the possibility of moving this to a remote site in the cloud. The proliferation of IoT devices has also created opportunities to connect many different devices in remote locations.

Our recent research has shown that most customers still do not want to rely on a remote IT infrastructure for something as important as safety and security, so for a lot of mission-critical sites, the existing infrastructure is going to be around for quite some time. Having said this, there is no doubt that customers are becoming more accepting of this architecture and will increasingly do so over the next few years. Cloud services/processing and availability has opened up doors for the rollout of surveillance solutions that would, in the past, have been extremely difficult to implement.

Some examples would be street surveillance solutions with a pay-per-use model, or centrally managed residential solutions with thousands of users connecting via the cloud. It has also meant that central backups and redundancy can be achieved and hybrid solutions can be maximised, with on-site and cloud management co-existing.

The other movement towards processing on devices (the edge) has also meant that the storage and some intelligence (for example, video analytics) can be distributed to some degree. This has the ability to reduce centralised storage and CPU requirements at the central location.

Perhaps the most important effect of the cloud and IOT has been to shift the mindset of surveillance companies and developers.

At the end of the day though, you still need the VMS to manage, store and make sense of the video and various information/triggers that are being generated in order to provide an effective operating and/or command centre environment.

Felipe Detoni: Fuelled by trends like sustainability and managing data most efficiently, video management in the broadest sense of the word (VMS software, recording devices, cameras etc.) will be a combination of server, edge and cloud-based solutions.

We see an ever-increasing demand for artificial intelligence (AI) to keep data manageable and useable. AI like intelligent video analytics, Camera Trainer (machine learning) and latest analytics based on deep neural networks are either offered at the edge, near the edge (gateway/server), or in the cloud, whatever makes most sense considering the application and keeping efficiency in mind. At Bosch, we strongly believe that every video security or video management solution will always offer AI at the edge to support pre-processing of data to substantially reduce network strain and to enhance the robustness and flexibility of the overall system. This is why all IP cameras from Bosch have been equipped with built-in AI as a standard since 2016.

Nevertheless, we also need to consider AI at other locations throughout the whole installation, since future solutions are increasingly data-driven and require consolidation of data. This requires additional software solutions that help consolidate data (audio and video) from multiple sensors (cameras). As a consequence, video management software (primarily focused on security) will increasingly offer plug-ins for data management, like Intelligent Insights from Bosch. This software offers widgets to make data from multiple sensors visible and useable, increasing overall situational awareness. Intelligent Insights can be used stand-alone to enhance third-party video management solutions or combined with BVMS from Bosch.

Another trend to consider is that we are often confronted with brownfield installations that require flexibility. Therefore it makes sense, depending on the existing installation, that manufacturers offer a flexible concept that allows customers to pick and choose multiple video management components like alarm monitoring, video analytics and the like that are added at the edge, near the edge or in the cloud.

Nick Grange: We feel that the traditional VMS is here to stay for larger integrated projects requiring a VMS or physical security information management (PSIM) system that allows for the control of a site or multiple sites and systems from a single operator platform. Many of the systems offered in the market today prompt the operator with an SOP that needs to be followed in handling alarms and contacting standby personnel.

There are also many larger systems which require technologies such as video walls and video event management system (VEMS) as well as third-party system integrations that will be accommodated by means of I/O, SDKs and APIs etc. We believe that the functionality between on-site and cloud services will be split going forward and alarm functions and processing will be handled at the edge, on the camera or device, then pushed to the cloud for analysis, confirmation, logging and remote monitoring of the alarms and events. In addition, it will then be sent to the on-site control room for similar processing and response.

Cloud -based solutions will allow these events to be pushed off-site thus providing easy access to events from anywhere by any or multiple authorised persons, cloud servers will also ensure data is stored safely offsite in the event of a disaster or theft. It will also allow for more detailed reporting of events and actions as these events would have been filtered and prequalified.

George Psoulis: The move towards distributed computing is gaining ever more momentum. Edge devices, such as cameras, have been capable of providing basic analytic functions at a low price point for some time. In the last year especially, many manufacturers have released models that can perform more advanced analytics such as facial recognition on the edge and some can even run true artificial intelligence. This removes the need for expensive centralised hardware, but more importantly, decentralises the decision-making process and speeds up workflows.

The adoption of cloud is one of the trends that depends heavily on this type of model for wide adoption. The economies of cloud start to kick in when dealing with mid to high camera counts, but the limiting factor in South Africa is still, unfortunately, bandwidth availability. By utilising the ‘fog computing’ model where more intelligence resides on the edge, the bandwidth can be managed much more intelligently. This opens up cloud as well as distributed storage architectures not possible before because the decision of where to send the footage can be made at the source.

Hi-Tech Security Solutions: Are we going to see more VMS players open their platforms to third parties to allow, as an example, for specialised software developers to make their analytics or BI (business intelligence) available on multiple existing platforms? Or will these developers need to pick a platform and stick to it, or perhaps standardise on the OSSA/S&ST; model?

Gus Brecher: This is something that all the manufacturers have been grappling with for some time. The building management/automation industry has achieved some form of success by adopting standard protocols, but even that sector is very messy and cluttered with protocols like OPC, BACnet, Modbus, LonTalk, KNX and many others, which has meant that the ultimate objective has not really been realised.

As a company, we have always been proponents of integrated solutions and this makes up a huge part of our offering, so we would welcome standardisation. It will also enable us to offer more intelligent solutions without having to continuously develop new protocol drivers (although there is always some development to be done).

It will be interesting to see how much traction OSSA and S&ST; achieve and how quickly. It has certainly taken ONVIF many years to get to where they are now, but by-and-large it has been successful.

Felipe Detoni: OSSA has released a first technology stack that harmonises some key components of video security cameras. The advantage is that third-party software developers can more easily develop software solutions that can be deployed across camera devices of different manufacturers. It offers a higher market potential to software developers typically making business cases more attractive.

S&ST; supports software developers by offering a tool environment to the developer and a marketplace to sell their software solutions. In the first instance, the apps developed and deployed on ‘Driven by OSSA’ cameras add video analytic functionality to the cameras.Consequently, depending on the application of third-party software, it typically will require some level of integration to make it work with the video management software chosen.

Nick Grange: We feel there has already been a shift to VMS players opening their platforms to accommodate third-party edge analytics from devices such as cameras etc. Arteco now make use of Open Connector to make the process of integration to third party devices easier and quicker. Additionally, most devices are ONVIF-compliant, which allows for a standard integration with supported functionality based on ONVIF conformance.

The VMS and PSIM platforms that wish to stay relevant will have to change to accommodate the value add being offered by third-party devices such as cameras, at the normal camera cost, to remain cost-effective in the market.

George Psoulis: It is difficult to predict what the market will do in this regard. Milestone took the decision several years ago to open the platform up completely and R&D; works hard on ensuring integrations are backward and forward compatible. It may be tough for some competitors in the market to open up their platforms to products that may compete directly with their offerings. With enough demand from the market, however, they may well be forced to open up, whether willingly or not.

The OSSA/S&ST; is an exciting development with a great deal of potential. The barrier to entry is a lot lower than ever before and skills become transferable across manufacturers for the first time. We also anticipate that developers from industries outside of security will start to make their presence felt.

Hi-Tech Security Solutions: Apart from the ability to receive multiple camera feeds, store them and allow for easy retrieval, what do you see as the best benefits your platform and your services have to offer in a world where AI-based systems and solutions are appearing everywhere?

Gus Brecher: Processing availability and the availability of open-source neural network libraries has made AI available to developers globally. Unfortunately, if these are poorly implemented, the results are very poor. Add to this the fact that AI is often over-sold and you will find a lot of confusion out there around how and when to use AI. But AI is real and a valuable asset that can add massive value when applied correctly.

We have many successful implementations using our AI suite, applied prudently and in the right scenarios. We are continuously enhancing our own in-house AI solution and also integrating third-party solutions as and where required.

But remember, AI is just a small part of the puzzle in your surveillance solution.

Felipe Detoni: Our video management software offering is capable of combining multiple security domains like video security, intrusion and access control and data management solutions like Intelligent Insights. It is a highly resilient system that focuses on making data available under the most severe circumstances. We, however, have a more holistic approach to video management and consider multiple components required to build a video security solution. We offer a flexible concept that can start with our resilient BVMS video management system, or a third party VMS and subsequently we can start to add various edge, near-edge or cloud components to offer the level of AI needed to support data-driven solutions.

Nick Grange: Arteco, the product we distribute and represent in Africa and South America, is more than your traditional VMS, it is a VEMS, meaning it is geared for video event management and is developed for ease of integrating multiple third-party products to create a seamless single events platform. Although Arteco has its own server-based software-licenced AI services, it also allows third-party analytics integration from multiple OEMs related to cameras and devices offering AI detection, LPR functions, people counting, Covid compliance etc.

Arteco licences each device being added, and while we may lose the analytics licence, we are able to offer the installer a cheaper product, which is easier to sell and justify to the end user, who expects value for money combined with performance.

George Psoulis: One of the benefits of the Milestone XProtect platform derives from its openness and its ability to easily integrate various AI solutions as they come to market. AI is still a new frontier and we haven’t begun to see all of the benefits it will provide in the future. One of the benefits that Milestone offers is the chance for customers to try out several options and then choose the best-of-breed solutions for their specific needs. As customers’ needs evolve, it is also easy to adapt their solutions accordingly without starting over again from zero.

Hi-Tech Security Solutions: Do standards like ONVIF make it easier for VMS providers to ensure they support a wide range of camera brands, or is it just as much work ensuring the latest cameras are fully supported?

Gus Brecher: Without doubt. Almost all camera manufacturers have embraced ONVIF for the most part for offering the basic functions (for example, multiple video streams, audio, I/O, event notifications, edge storage). Where there are more sophisticated algorithms or functions on the cameras, we still need to implement an API integration with the selected cameras, but this is the exception rather than the norm.

Felipe Detoni: ONVIF standards make basic integration easier. Depending on the cameras and their functionality, certain (more advanced) features can require native integration. Nevertheless, ONVIF constantly evolves and introduces new standards. Our expectation is that more advanced features, like understanding metadata generated by video analytic capabilities can be more easily integrated in the future once multiple vendors adhere to standards like ONVIF Profile M.

Nick Grange: Absolutely, it is a great initiative. Even though it is voluntary and not mandatory for industry compliancy, it is an open industry forum that provides and promotes standardised interfaces for effective interoperability of IP-based physical security products. We are finding more and more suppliers participating, this definitely assists the distributors like ourselves and the integrators/installers we sell to, in terms of making their job easier.

ONVIF takes the need away for testing compliancy between products provided the original equipment manufacturer (OEM) holds up their end and remains conformant with the current ONVIF standards.

George Psoulis: ONVIF has been a great boon for the industry and it has come on by leaps and bounds over the years. Provided that the standards have been properly adhered to, ONVIF can now extend the basics we are used to, such as video feeds and I/O, to now also offering metadata support as well as supporting more granular analytics. There are, of course, applications and features that are not covered very well by the ONVIF standard. That is where Milestone puts in a lot of development effort so as to ensure that these extra features and functions are supported through dedicated drivers.

Hi-Tech Security Solutions: What major changes have been made to your system over the past year or two? What do you see as the most important development for your product over the coming year or two? How do you plan to take advantage of the seemingly unstoppable push to cloud-based everything?

Gus Brecher: Our main focus has been on our fundamental architecture and AI. We have re-structured our architecture in order to accommodate a future move to the cloud and a VSaaS offering (Video Surveillance-as-a-Service) as well as offering a more powerful enterprise, multi-site solution. While our research shows that it will be some time before cloud surveillance solutions gain widespread acceptance, we believe that our new architecture will ensure that we are ready when it does and will also open up new markets in the short to medium term.

On the AI side of things, we have implemented new analytics which includes the use of neural networks. This has enabled us to identify specific objects in order to improve the decision-making process, reduce false alerts and offer intelligent analytics that previously would have been impossible. For example, when we track objects across points, we can ensure that we trigger alerts only on people, rather than animals. We can also identify faces and even facial landmarks like tear-ducts, lips and eyes for temperature measurement. This AI also enables us to find specific objects across multiple cameras for forensic searching.

Felipe Detoni: Major changes from Bosch: integration of multiple security domains (video security, access control, intrusion) and data management (Intelligent Insights). Coming developments from Bosch include: assisted object and person tracking with map-based tracking assistant, person identification and VSaaS.

Our plans to take advantage of future developments like cloud-based solutions will be our flexible concept. Depending on customer requirements, we are preparing ourselves to offer video management components where it is needed: edge, near-edge or cloud.

Nick Grange: Arteco is cloud-ready, but is pushing development in three directions in this respect: the traditional on-premises server/client model, hybrid models and also a pure cloud-based offering. We see a market for hosting and offering cloud services in the future and can add value with support to integrators/installers and their end user clients. This will offer better integrity of stored events and associated footage, as well as better reporting on hardware issues and events received from site etc.

George Psoulis: Milestone is constantly adapting with the ever-evolving landscape around deep level integrations into ‘things’ outside of a standard security installation, becoming a data aggregator for edge-based devices and handling facial recognition, AI, IoT, sound intelligence, autonomous drones, body-worn cameras etc.