Introduction to the Milestone Systems MIPS 2018, Hanoi, Vietnam
by Chris Cubbage, Executive Editor
APAC LEADING THE SURVEILLANCE WORLD
As of 2016, the global video surveillance market was valued at $15.4 billion and mostly driven by the China market with 42% market share, exceeding $6.4B. Across the world, eight countries have higher growth rates than the global average, with five in the APAC region, being China, Indonesia, Vietnam, India and Thailand. The remaining three are Mexico, Brazil and Argentina.
The APAC region will be the gravitational pull for continued growth of the video surveillance segment and its dominance in the physical security sector. Globally, physical security’s convergence with ICT infrastructure will drive growth in video and system analytics, hybrid deployment from Artificial Intelligence (AI) edge to AI cloud infrastructure and most importantly, will need to be increasingly supported by cybersecurity to protect privacy, accuracy and capability.
By 2021 the APAC market forecast for video management systems (VMS) is to double to $663M, with China demonstrating a much larger video channel concentration, with 250 and over channels and 1000 and over channels representing over a third of the total VMS license revenue.
In opening the MIPS2018 Conference for Milestone Systems today (24 January 2018) in Hanoi, Vietnam, Monica Wang, senior analyst with IHS Markit provided an overview of the physical security market, with a focus on the major trends in video surveillance. The physical security sector is separated into security equipment and security services, with consumer video surveillance and video analytics showing clear growth trends. Indeed, the fastest growing sector in security equipment is the consumer video surveillance segment and in security services, video analytics is showing the greatest potential for continued growth. For the APAC region, enterprise storage and video surveillance are two of the fastest growing segments.
ANALYTICS & AI – MARKET DRIVERS IN VIDEO SURVEILLANCE
The key market trends driving growth is the continued transition from analogue to digital cameras and this trend will continue with the advent of network cameras, representing 71% of the market by 2021. The next transition for digital cameras is moving to higher resolution cameras. As camera resolution improves there is corresponding growth in video analytics, which has forecast triple digit growth rates with capability to automate the video monitoring process. Growth is being supported with the advent of a new generation of video analytics that is building market confidence with its accuracy. By 2022, the number of cameras with inbuilt video analytics will grow by a factor of 4 and for video recorders with inbuilt video analytics, it is expected to grow by a factor of 5.
Why is there a growing market demand for deep learning video analytics? Digital video streams can create big data pools and with cloud computing and better deep learning algorithms, the benefits of rapid image processing across cloud platforms will allow the algorithms to focus on accuracy and have a capability to improve the efficiency of the video analytics solution, processing video images in a fraction of the time and in much higher volumes.
For the market segment verticals, in particular for transportation, government and retail, this capability creates new opportunities and will see the strongest growth. City Surveillance capabilities are being demonstrated in China. With a need to cover expansive geography, China is seeing network cameras deployed on a massive scale with video analytics increasingly applied for extracting facial recognition. It is now a requirement in China for tender documents to have facial recognition as standard for all City Surveillance applications.
The additional major trend is the deployment of Artificial Intelligence (AI) technology in video surveillance, using a mix of technologies at the AI edge and integrated to the AI cloud. AI at the edge saves on bandwidth and relieves on computing capacity at the headend with object detection, crowd monitoring and feature extraction done by the camera before image transmission. As new chip vendors enter the market, AI camera prices will also continue to decline and video surveillance will continue to converge with ICT equipment, bringing significantly improved capability for feature extractions of humans and vehicles, object searching and data mining.
The final trend is in cybersecurity and the key challenge to overcome before video surveillance becomes a major aspect of the IoT revolution. Video surveillance and cybersecurity will become critical in terms of privacy and the path to cybersecurity for video surveillance includes connectivity, data collection, data computation and the creation of biometrics, such as face, gate, movement, behaviour, as well as object and people correlations. Cybersecurity has become a major cornerstone of the video surveillance sector and includes the need to have pre-defined processes in dealing with and responding to identified vulnerabilities, effective vulnerability notifications and software patch delivery, best practices in standards for vendors and camera product features relating to camera encryption of images and certification by third parties.
ERA OF AI
The era of AI is changing every environment, including in web services, intelligent machines, healthcare, security and finance. By 2025 there will be a 1000x GPU-computing improvement in performance over the current CPU. The integration of big data, neural networks and AI platforms will see massive increase in capability and the video surveillance market is a key area where these capabilities will be demonstrated. By 2025 it is forecast there will be a billion AI cameras deployed in world cities. With this level of intelligence and video analytics operating across these cities, this will bring a far greater understanding and insight into city activity, with the capability to drill down to the individual and specific objects. With a billion security cameras operating at 30 frames a second, this creates over 30 billion frames a second every day. The human ability to process and understand these images will not match the capability of AI. Currently, a human can understand about 5 frames a second, where the Tesla V100 system can handle 900 frames a second. The 8x Tesla can do 7,000 frames per second with less cost and greater accuracy. Over a short time, with machine learning, the systems will become faster and more accurate in applying biometric and movement algorithms for people and object matching on a scale never seen before, let alone imagined.
MILESTONE SYSTEMS TRANSFORMATION
“It is inevitable that all devices will be connected”, stated Milestone CTO Bjørn Skou Eilertsen. As an industry leader, Milestone Systems is focusing on the three key trends of aggregation, automation and augmentation, with aggregation of devices, automation of systems and augmentation with humans.
Milestone is transforming the way it thinks about its solutions, products and platforms. With 22 solution partners, the company is maintaining its attention on the customer requirements with the Milestone video technology platform consisting of the presentation interface, device hardware, video services interface and cybersecurity. Each segment sets out to meet the corresponding key trend, with how devices are aggregated, with automation occurring in the video service interface and the augmentation provided via the presentation interface.
One of the key investments for Milestone has been on expanding the driver framework into the IoT framework and to aggregate all of the sensor information with the video services interface, through building greater video processing services at the GPU level and improving compute capacity exponentially better and faster, with more innovations coming to market in 2019. One such innovation will be in the Mobile market, with a Mobile SDK to allow customers to adapt mobile applications to their own requirements. And in cybersecurity Milestone will be certified in data privacy and security to create more confidence in its approach to security by design, security by default and security by deployment.
