Is the machine vision market set to undergo explosive growth?
The global machine vision market is still in a phase of steady, rapid growth. According to forecasts released by BBC Research, the global machine vision market was about US$16 billion in 2018, and is slated to grow at a compound annual growth rate of 9.2% to US$24.8 billion in 2023.
This growth rate may not seem to be particularly eye-catching, but the market's strong momentum is apparent once you take into consideration the fact that it has averaged double digit growth since 2002.
What’s even more noteworthy is that the overall machine vision market is currently going through two profound changes expected to bring the market to turning points that lead to even more accelerated growth.
One of the changes is that the mainstream technology for machine vision is shifting from PCs to embedded frameworks. After years of development, conventional PC-based vision systems are already quite mature and have several distinctive advantages, including high performance, the ability to implement relatively complex system functions, and access to a greater abundance of software and hardware resources. In contrast, embedded vision systems have limited performance and are often constrained by cost, power consumption, size and appearance, not to speak of the many technological barriers that have also yet to be overcome. However, the ultimate goal for developing machine vision is to incorporate vision functions into various embedded systems, so that machine vision—once considered “abstract and inaccessible”—can be applied to different aspects of life and become within reach and readily available. The prospect of a "ubiquitous" market is quite appealing indeed, which is why a considerable number of players have been focusing on embedded vision in recent years, jointly driving changes in the mainstream technologies for machine vision.
The second change in the machine vision market is driven by artificial intelligence (AI). Vision processing has always been a classic application scenario for AI technologies such as machine learning. The application of machine learning and eventually deep learning will engender profound changes in the role of embedded vision products. For instance, the purpose of embedded vision products will evolve from simple environment sensing to smarter, vision-guided automated functions. A classic example is the technological advancement of autonomous driving: in lower-level driver assistance systems, embedded vision usually serve passive safety functions such as detection and reminders. In more advanced automated driving systems, embedded vision is to a greater degree incorporated into active safety functions, making rapid, reliable responses on the driver’s behalf.
These two "changes" will consistently contribute toward the amassing of energy in the machine vision market within the foreseeable future and the subsequent formation of a "bursting point" in the broader market.
Security and surveillance: Security and surveillance has always been a major application market for machine vision, and has been the main force driving the industry's development over the last decade. The above two new sources of growth momentum give this market great potential. From the perspective of an incremental market, the development of embedded vision technology will increase the possibility of consumer-level applications, such as smart door locks and video doorbells, ensuring the market penetration of these applications. From the perspective of a stock market, security and surveillance equipment that are upgraded into smarter versions will possess the ability to not just "see" but to "understand", a step-up that also has potential for massive demand.
Smart manufacturing: Industrial applications can perhaps be regarded as the starting point of the machine vision market, and the increasing use of AI technology-based embedded vision systems will inevitably make the entire manufacturing process smarter, accelerating the deployment and realization of Industry 4.0. For example, in the past, robots on the production line were confined in a "cage" due to their limited visual perception and judgment abilities, and were expected to complete individually only a limited number of specific tasks. Smarter embedded vision will lead to the future development of collaborative robots, which will have the ability to work together with people in more complex scenarios—an ideal state of smart manufacturing.
Figure 1: Avnet released an object identification embedded vision development kit that can be extensively applied in robot positioning and video surveillance.
Self-driving cars: Automobile electronics is without a doubt a rapidly growing embedded vision market. In most schematics for self-driving cars, embedded vision systems not only take on the task of external environment sensing along with LiDAR and mmWave radar, but are also omnipresent within the vehicle to provide additional functions such as fatigue testing and hand gesture recognition applications, making intelligent cockpits a reality.
Figure 2: Avnet Telematics Box supports advanced anti-theft and onboard vision functions, and can be seamlessly synchronized with the vehicle's operating system to realize intelligent driving.
The application furthermore brings vehicle safety to a whole new level, and has attracted great interest from private car owners. With the support of the advanced anti-theft system and the surveillance function of built-in cameras, Telematics Box lays the foundation for smart gateway systems—systems that collect and synchronize data from smart devices installed on the vehicle. They can be seamlessly synchronized with the operating system of your car, and their ability to manage and track vehicles will attract the interest of many enterprises.
Furthermore, the application brings vehicle safety to a whole new level, and has attracted great interest from private car owners. With the support of the advanced anti-theft system and the surveillance function of built-in cameras, Telematics Box lays the foundation for smart gateway systems—systems that collect and synchronize data from smart devices installed on the vehicle.
Drones: There is a natural connection between drones and vision procession, and new drone applications are constantly emerging. Despite the significant difference in market capacity compared with cars, drones will most definitely become an important vertical market for certain embedded vision companies.
Figure 3: e-Commerce giants are testing the waters for drone logistics, accelerating the growth of this embedded vision market segment.
(Image source: Internet)
New retail: Amazon Go is currently a trailblazer in this field, employing hundreds of embedded cameras and back-end vision processing systems to capture and analyze user behavior in its unmanned stores. Despite the experimental nature of Amazon Go and the difficulty in determining whether or not it will eventually succeed, its endeavors will surely pave the way into the fields of retail and commerce for embedded vision.
Figure 4: Densely installed surveillance cameras on the ceiling of an Amazon Go unmanned store capture and analyze user behavior in real-time.
(Image source: Internet)
Entertainment and gaming: When it comes to embedded vision applications in entertainment and gaming, the Microsoft Kinect and Nintendo Wii definitely come to mind. These devices are however just preliminary examples of embedded vision testing the waters in consumer-level applications; the ultimate goal is to find a killer application one day by perfectly combining embedded vision functions with smartphones.
VR/AR: After the hype we have all experienced over the past two years, VR/AR has clearly cooled down. These two technologies have however established a presence in people’s imaginations of future technologies. The influence of VR/AR on the way people interact with machines and how information is transmitted in the future is immeasurable. Embedded vision, as a foundational technology for VR/AR, will certainly not be missing the blowout that’s soon to come.
The above is our conclusion after roughly surveying the future target markets of embedded vision. As technology solutions continue to improve and mature, embedded vision is sure to continue wowing us with new possibilities and unforeseen applications. Without a doubt, we are currently at the tipping point for exponentially accelerated market growth.
Xilinx SoCs and the reVISION Stack Accelerate Embedded Vision Integration
SoCs with programmable logic are an essential element of real-time embedded vision systems. Designer...
Autonomous vehicles are the future
Self-driving cars, unmanned aircraft or driverless tractors—autonomous vehicles stopped being merely...
How all programmable technology revolutionizes embedded vision
Autonomous driving is just the start of EV solutions...