In the past, machine vision was limited to highlight-controlled environments, costly sensor technology, and restrictive feature detection. Today, artificial intelligence (AI) is set to change the market, creating new classes of applications and significant new opportunities.
Machine vision is in the process of transition and dramatic expansion. Deep learning (DL) techniques are taking machine vision systems to next level, driving the mass adoption in several industries -- including the automotive, retail, consumer, industrial, and surveillance sectors.
Machine Vision Tech Market Development
DL-based machine vision marks a departure from other approaches used in the sector, which were more limited regarding their application. ABI Research now forecasts that machine vision technology will see a CAGR of 53 percent between 2018 and 2023 -- with $193.8 billion of annual revenue generated from services and hardware by the end of the forecast period.
Machine Vision vendors previously relied on hardcoded feature detection techniques, which meant they could only be applied in highly controlled environments -- such as inspecting a single type of object on a production line.
DL-based machine vision systems are far more flexible. One system can recognize many object types and be deployed in a range of circumstances. Also, cashier-less stores -- like Amazon Go -- demonstrate where cameras can track the movement of both customers and items around the retail store.
Another example of innovation would be the machine vision systems being employed to support autonomous driving. These systems can make distinctions between multiple types of road users.
"It is these new DL-based applications, among others, that are set to drive growth in the machine vision space, which would have been impossible using traditional machine vision techniques," said Jack Vernon, analyst at ABI Research.
If we look at some of the applications increasing adoption of machine vision systems, we will see that it is the innovations in deep learning that are driving their growth. Take, for instance, advanced driver assistance systems (ADAS), which are a core technology in autonomous driving.
By 2023, 37 million vehicles shipped will contain between level 2 to 5 ADAS. Over half of the 34.446 million level 2 ADAS systems shipped in that year will use DL-based machine vision, while the remaining level 3-5 vehicles will all use the approach -- this represents a massive growth in adoption of machine technology and will contribute enormously to the growth.
The same DL-based image recognition techniques used in machine vision are also being applied to sensors outside of traditional RGB (primary color) cameras, these will also have a transformative effect in those markets, and likely significantly increase adoption on those technologies.
For instance, the use of LiDAR systems will be incorporated into autonomous driving systems, on the back of the fact that deep learning enables machines to interpret LiDAR data in a more sophisticated way, allowing software to identify features of the landscape and other road users.
DL-based image recognition techniques are also going to change how many different sensor systems are going to be used. In the healthcare space, a number of startups and large research entities are building DL-based image recognition software that can identify health issues directly from MRI, radar, x-ray data.
These examples demonstrate how DL-based machine vision techniques are transforming not only the growth of RGB camera systems, but also how many other different sensors will be used in future.
Outlook for Machine Vision Applications
Few companies have fully settled on their favored hardware and software technology for machine vision applications across different verticals, creating opportunities and competition for many vendors in both spaces.
Consequently, savvy vendors are competing aggressively across the technology stack as potential customers for their solutions chase the high-value applications -- such as autonomous driving.
The scale of the opportunities have attracted significant investments in machine vision over the past four years. That's a trend that looks set to continue for another two years. As an example, in 2017, venture capitalists invested $2.7 billion in machine vision startups.
Machine vision is in the process of transition and dramatic expansion. Deep learning (DL) techniques are taking machine vision systems to next level, driving the mass adoption in several industries -- including the automotive, retail, consumer, industrial, and surveillance sectors.
Machine Vision Tech Market Development
DL-based machine vision marks a departure from other approaches used in the sector, which were more limited regarding their application. ABI Research now forecasts that machine vision technology will see a CAGR of 53 percent between 2018 and 2023 -- with $193.8 billion of annual revenue generated from services and hardware by the end of the forecast period.
Machine Vision vendors previously relied on hardcoded feature detection techniques, which meant they could only be applied in highly controlled environments -- such as inspecting a single type of object on a production line.
DL-based machine vision systems are far more flexible. One system can recognize many object types and be deployed in a range of circumstances. Also, cashier-less stores -- like Amazon Go -- demonstrate where cameras can track the movement of both customers and items around the retail store.
Another example of innovation would be the machine vision systems being employed to support autonomous driving. These systems can make distinctions between multiple types of road users.
"It is these new DL-based applications, among others, that are set to drive growth in the machine vision space, which would have been impossible using traditional machine vision techniques," said Jack Vernon, analyst at ABI Research.
If we look at some of the applications increasing adoption of machine vision systems, we will see that it is the innovations in deep learning that are driving their growth. Take, for instance, advanced driver assistance systems (ADAS), which are a core technology in autonomous driving.
By 2023, 37 million vehicles shipped will contain between level 2 to 5 ADAS. Over half of the 34.446 million level 2 ADAS systems shipped in that year will use DL-based machine vision, while the remaining level 3-5 vehicles will all use the approach -- this represents a massive growth in adoption of machine technology and will contribute enormously to the growth.
The same DL-based image recognition techniques used in machine vision are also being applied to sensors outside of traditional RGB (primary color) cameras, these will also have a transformative effect in those markets, and likely significantly increase adoption on those technologies.
For instance, the use of LiDAR systems will be incorporated into autonomous driving systems, on the back of the fact that deep learning enables machines to interpret LiDAR data in a more sophisticated way, allowing software to identify features of the landscape and other road users.
DL-based image recognition techniques are also going to change how many different sensor systems are going to be used. In the healthcare space, a number of startups and large research entities are building DL-based image recognition software that can identify health issues directly from MRI, radar, x-ray data.
These examples demonstrate how DL-based machine vision techniques are transforming not only the growth of RGB camera systems, but also how many other different sensors will be used in future.
Outlook for Machine Vision Applications
Few companies have fully settled on their favored hardware and software technology for machine vision applications across different verticals, creating opportunities and competition for many vendors in both spaces.
Consequently, savvy vendors are competing aggressively across the technology stack as potential customers for their solutions chase the high-value applications -- such as autonomous driving.
The scale of the opportunities have attracted significant investments in machine vision over the past four years. That's a trend that looks set to continue for another two years. As an example, in 2017, venture capitalists invested $2.7 billion in machine vision startups.