Machine vision, deep learning, sensor technology, intelligent manufacturing, autonomous driving, and many other new vocabulary are not unfamiliar to many people in this era of rapid technological progress. In fact, with the deep integration of machine vision technology with more application scenarios, it is often difficult to clearly define the strict boundaries between the different concepts mentioned above. Simply put, machine vision is the use of machines instead of human eyes to recognize, measure, and make judgments on objects. As the proportion of advanced manufacturing in China increases, the accuracy and efficiency of human eyes in industrial production lines can no longer meet the requirements of industrial upgrading. Machine vision has already been used in various industries to achieve higher production efficiency. Machine vision requires the support of AI technologies such as deep learning and has wide applications in fields such as intelligent manufacturing. When it comes to a comprehensive technology, it should not be distinguished by boundaries. Machine vision is widely used in various industries, and with the continuous increase in demand for automation and intelligence in various industries, the development of machine vision contains infinite potential. Recently, Yiou Think Tank, in collaboration with Alibaba Cloud Accelerator, released the "2022 China Machine Vision Industry Application Research Report", analyzing the current development status of the machine vision field and making predictions on future application trends based on technological pain points.
Widely applicable fields
The downstream application fields of machine vision are diverse, such as consumer electronics, semiconductors, agricultural harvesting, photovoltaics, warehousing and logistics, medical, heavy industry and metal processing, and so on. According to GGII statistics, consumer electronics was the largest application market for machine vision in 2021, accounting for approximately 31.6% of the total market share.
Machine vision is mainly used in the field of consumer electronics for PCB/FPCAOI inspection, component and whole machine appearance inspection, assembly guidance, and other applications. In addition, the penetration rate of machine vision in scenarios such as connector inspection, SMT, hard disk inspection, online component classification and screening, and QR code reading is gradually increasing.
Due to the fragmented application scenarios of industrial machine vision, there are multiple and repetitive characteristics in the application of scenes in production and manufacturing. Yiou Think Tank believes that due to the blurred boundaries of various application scenarios of industrial machine vision, it can be divided into four types of solutions based on specific functions.
The four major solutions are just one classification, and machine vision has more functions. The initial application of machine vision in various industries often involves detection in the production process. With the popularization of technology and the decrease in costs, machine vision is expected to strengthen its application depth in the industry and enter other stages such as recognition, measurement, and positioning.
Machine vision is actually very different from robots. Many robots in the industrial field do not even have visual functions and mainly handle relatively simple and repetitive processes. With the advancement of machine vision technology, the field of robotics has also benefited from it, such as introducing 2D and 3D vision technologies into robots to achieve more functions.
Garrett Place, the head of development for IFM's robot perception business, pointed out that the advancement of 3D vision and machine vision technology has also formed widespread applications in the field of robotics, allowing them to work in more unstructured environments. This flexibility can reduce the overall cost of robot integration and provide more application areas.
However, due to the widespread application of machine vision technology, many complex problems in various scenarios cannot be solved by a single technology supplier. In machine vision solutions, it is common to see a combination of multiple suppliers and technologies, because each supplier or technology solution has its own advantages, and industry collaboration can open up more fields that were not previously involved in machine vision.
With the application of machine vision technology in more industries, its market size is gradually increasing. According to a survey, the sales of machine vision devices and systems in China increased from 1.98 billion yuan in 2012 to 16.1 billion yuan in 2021, with a compound growth rate of approximately 31.7%, and a particularly significant growth rate of 56% in 2021. The sales revenue of China's machine vision industry is expected to exceed 20 billion yuan in 2022.
The Three Major Trends of Machine Vision in the Future
With the arrival of Industry 4.0, the demand for machine vision technology in industrial scenarios continues to drive the development of industrial machine vision technology. Yiou Think Tank also pointed out three major trends in the future development of visual technology, including the improvement of 3D technology requirements, the gradual dominance of domestic substitution, and the integration of fragmented scenes.
Trend 1: In the vast application scenarios, 3D technology will develop more maturely and widely
Along with the development of 2D vision, 3D technology is also on the rise. From the current trend, the development momentum of 3D machine vision is far greater than that of 2D. Although the current development of industrial machine vision is relatively mature, 3D technology is still in the early stages of development in China. Foreign companies and products are leading the domestic market in high-precision micro inspection and occupy a large share of the market. Domestic enterprises are gradually replacing foreign enterprises in hardware and software.
Trend 2: Localization substitution becomes the main theme, competition between domestic and foreign manufacturers further intensifies, and market structure is reconstructed
The current replacement rate of machine vision is 50%, but it is mostly limited to the field of 2D machine vision. In the future, with the cooperation and win-win of domestic brands, the product functions will be professional and the types will be refined, and the replacement rate will gradually increase. The industry will tend towards specialized division of labor, and autonomous visual platforms, visual systems, and equipment will work together to gradually surpass foreign brands and grow into the main force of visual technology in China's intelligent manufacturing industry.
Trend 3: Integrating fragmented scenes and standardizing integrated equipment will be the next generation development direction of machine vision
With the gradual expansion of industrial automation, products under the intelligent manufacturing mode are produced in multiple varieties, small batches, and personalized ways. Enterprises are beginning to shift towards mass customization production, but machine vision technology can only recognize and classify a few products in automated production lines, making it difficult to meet fragmented scenarios.
To address this situation, integrating fragmented scenarios and creating integrated devices will greatly promote the rapid integration of information technology and operational technology. At the same time, building a digital closed-loop throughout the entire process will also be one of the important measures.
Of course, in addition to the three major trends, the development of machine vision has also entered a higher level with breakthroughs in many industries and played an important role in the progress of certain industries.
For example, the semiconductor industry is facing great difficulties in upgrading Chinese manufacturing processes as the US becomes increasingly tight. For the Chinese semiconductor industry, due to the increasing size of wafers, the thinning of internal circuits, and the decreasing volume of connectors, there are more and more devices that require precise recognition and positioning through machine vision.
The requirements for production efficiency and defect rate are also becoming increasingly strict. The upgrading of the semiconductor industry heavily relies on the guarantee of machine vision products.
At this point, the development of domestically produced AOI testing equipment is a key factor in the success of semiconductor investment. For domestic new investment semiconductor manufacturers who are striving to catch up with foreign levels, it is crucial to improve their processes and quality. Through the images collected by AOI, the yield and output of new products, as well as whether the manufacturer's capabilities and bottlenecks have the opportunity to be calculated, all affect the development trend of the entire industry.
The development level of the machine vision industry has become a necessary step in determining the further development of many other industries. Whether it's semiconductors or new energy vehicles, whether it's intelligent mining or smart agriculture, machine vision is becoming a cornerstone technology type in the industry.
