Machine vision: illuminating "intelligence" and creating a new "vision" world

As the first of the five senses , In the information that human beings receive from the outside world 80% It is obtained through vision .

With artificial intelligence 、 The rapid development of edge computing and other emerging technologies , People have also given machines “ know ” and “ reform ” The power of the world , So as to replace the human eye to measure the external environment 、 Identify and judge , Complete the assigned task without contact .

This technology is called machine vision （MV）, Known as the “ Eyes of industrial automation ”, It is the most important link to realize intelligent manufacturing .

 

At the key node of intelligent innovation , The traditional manufacturing industry is facing many transformation challenges .

One side , The cost of manpower is rising , The factory faces “ Recruitment difficulties 、 Difficult to stay at work ” The pressure of the ; On the other hand , Under the wave of Intelligent Manufacturing , The demand for quality inspection and production is increasing , The production line has higher requirements for industrial equipment .

In addition, the country pays more and more attention to the development of intelligent manufacturing , Issued a number of policies and plans to vigorously promote product R & D and market expansion in the field of intelligent manufacturing . stay 《“ the 14th Five-Year Plan ” Digital economy development plan 》 It is clearly put forward in ： To 2025 year , The added value of core industries in digital economy accounts for GDP Share to 10%. The manufacturing industry, which is the most important part of the country , The application penetration rate of industrial Internet platform should start from 2020 Year of 14.7%, achieve 45%.

Machine vision quality inspection is an important application scenario of industrial Internet .

At the key node of intelligent innovation , The traditional manufacturing industry is facing many transformation challenges .

One side , The cost of manpower is rising , The factory faces “ Recruitment difficulties 、 Difficult to stay at work ” The pressure of the ; On the other hand , Under the wave of Intelligent Manufacturing , The demand for quality inspection and production is increasing , The production line has higher requirements for industrial equipment .

In addition, the country pays more and more attention to the development of intelligent manufacturing , Issued a number of policies and plans to vigorously promote product R & D and market expansion in the field of intelligent manufacturing . stay 《“ the 14th Five-Year Plan ” Digital economy development plan 》 It is clearly put forward in ： To 2025 year , The added value of core industries in digital economy accounts for GDP Share to 10%. The manufacturing industry, which is the most important part of the country , The application penetration rate of industrial Internet platform should start from 2020 Year of 14.7%, achieve 45%.

Machine vision quality inspection is an important application scenario of industrial Internet .

 

The broad application space breeds sufficient market potential . China is the processing center of the global manufacturing industry , It is becoming one of the most active areas of machine vision in the world .Grand View Research forecast , To 2023 year , The compound growth rate of China's machine vision industry will reach 28%.

In a large number of application scenarios , How does machine vision work ？ In what ways can we help the manufacturing industry achieve intelligent upgrading ？

Intel recently released 《 Special issue of machine vision 2022》 The application of machine vision scheme in the actual scene is explained in detail , It also shows how Intel can play an important role “ The empowerer ” role , Join hands with Chinese machine vision industry partners , Help industrial enterprises accelerate their digital transformation 、 Intelligent upgrade .

                                                  Intel releases 《 Special issue of machine vision 2022》 

                                                            What is the “ The eye of wisdom ”？

 

01 Intelligent quality inspection , Make the field bus more accurate

Machine vision quality inspection is a part of industrial field inspection . In the electronic 、 automobile 、 Metalworking 、 Packaging and other industries , Simple 、 reliable 、 The economical and practical fieldbus technology is becoming more and more popular , Also for process automation 、 Manufacturing automation puts forward higher requirements . therefore , Need to introduce machine vision , Replace the traditional manual detection method .

Machine vision can recognize 、 Extract the key features of the target to be detected , Include dimensions 、 shape 、 Color , At the same time, check whether the product conforms to the design parameters and quality parameters , Once the problem is discovered , Machine vision can also identify the type and location of defects , Assist in production line adjustment .

Although the traditional computer vision technology has been applied in this field , But it is difficult to achieve high accuracy . Machine vision system for industrial environment , It has higher visual perception efficiency and accuracy than human beings , It reduces the possibility of human error , So as to improve the stability of surface defect detection and other applications 、 Accuracy and efficiency .

02 Machine vision , Make the robot softer

With the continuous development of China's economy , Our needs are becoming more and more diverse . Small batch 、 Customization 、 flexible manufacturing , It has become an irreversible trend in China's manufacturing industry .

2020 year , China newly installed 16.8 Ten thousand industrial robots , Of the world's new installations 43.85%. Machine vision can bring “ flexible ”.IDC forecast ： Integrating machine vision into robot systems , The operation flexibility can be increased 50%, The quality of the products has been improved 35%.

From logistics 、 Metalworking 、 Car to 3C market , Robots equipped with machine vision , You can grab irregular packages 、 Welding non-standard metal parts 、 Grab randomly placed and stacked components 、 Assemble customized product parts , At the same time, it can also realize navigation and obstacle avoidance , Walk through the manufacturing plants with complex environment with a soft body .

 

03 Predictive maintenance , Lower production line cost and longer life

To improve production efficiency 、 Reduce operating costs , Factory production lines in many industries are often 24 Hour run , Unplanned sudden downtime often results in heavy losses . here , Rely on sensors to collect key component performance and operator behavior , And analyze it , To achieve predictive maintenance , The manufacturer can more accurately judge when the machine needs to be maintained 、 Optimize production steps .

according to IoTAnalytics Published data ： To 2026 year , The predictive maintenance market will continue to maintain high growth , Compound annual growth rate (CAGR) by 31%.

As technology matures , In the past, it can only be used for machine vision and other equipment in high-end fields , It can now be deployed at the edge , Analyze the pictures 、 video 、 Time series and other massive data , And carry out real-time analysis , It can not only improve the yield of the production line , At the same time, it can prolong the service life of the machine , Increase cost effectiveness .

                                                          Intel   framework

                                                        Accelerate machine vision applications

 

COVID-19 highlights the importance of robots 、 Automation represented by machine vision 、 The important value of intelligent production system —— By applying a machine vision system , Enterprises can reduce the impact of employees' inability to work on the production rhythm and production order , At the same time, reduce the pressure on epidemic prevention and control . And as the epidemic gradually subsided , Manufacturing is expected to accelerate recovery , The market demand for machine vision will also be growing .

A large number of demands have undoubtedly raised the standards and demands of the manufacturing industry for software and hardware technology . In order to speed up the implementation of machine vision in the industrial Internet , Intel offers a range of Hardware 、 Software and toolkits , Help the implementation of more schemes , Accelerate the upgrading of industrial automation industry , Unleash the potential of data at the edge .

At the hardware level , Third generation intel   To the strong   Scalable processor built-in AI To accelerate the technology , Intergenerational average performance gain 1.46 times , Image classification AI Improved reasoning performance 1.56 times , Provide strong support for the server-side load required by machine vision . Plus Intel Lexus  、 core  、 To the strong   Wait for the processor and FPGA And network chips , The processor hierarchy of different computing power levels , At the same time, it has rich expandable high-speed IO And peripheral interface circuit , Can satisfy from the foundation 、 Main stream 、 Advanced and other different machine vision systems .

At the software level , Intel   Industrial edge insight platform （EII）、OpenVINO  Tool set 、oneAPI tool kit , And the Intel machine vision development suite , The Intel software infrastructure that makes up machine vision ：

 ■ EII It includes a set of mature pre integrated software stacks . Extract at the edge 、 Analyze and store video data and timing data , built-in AI function , Provide... For the edge end AI bonus , Improve operational and productivity .

■ OpenVINO  The tool suite aims to accelerate deep learning reasoning , Based on the latest generation of artificial neural network , Scalable across Intel   Computer vision and non visual workloads for hardware , And greatly improve performance .

■ oneAPI Toolkit as a cross industry 、 to open up 、 A unified programming model based on standards , Provide a common developer experience across accelerator architectures , It can help the solution development of manufacturing industry achieve faster performance 、 Higher productivity , And broader innovation .

■  Intel   Machine vision development kit is a software and hardware tool kit , Based on Intel Lexus  、 core  、 Or Xeon   Processor development , Some of them adopt the new generation modular core board hardware architecture — Industrial edge node reference architecture , It can adapt to complex machine vision scenes .

With the blessing of the above software and hardware , Intel offers 3D Visual acceleration scheme , It can provide visual guidance and detection for robots ：

Visual guidance ： With the help of 3D Visual positioning and recognition of workpiece , Instruct the robot to grasp the workpiece , Used for sorting and stacking 、 Loading and unloading, welding, etc . be based on OpenVINO  Intelligent visual guidance and grasping of the tool suite can shorten the reasoning time 6-15 times ,CPU Load reduction 25%-73%.

Visual inspection ： Use the workstation camera 、 Laser scanner, etc , Perform contour detection on the workpiece 、 Surface defect detection and 3D reconstruction

  Besides , Intel SigOpt AI The training platform can be used to automatically optimize the parameters of the deep learning training pipeline , So as to improve the modeling efficiency , Speed up production .

Of course , These Intel solutions , Not on paper , It has been implemented in many enterprises in all walks of life .

                                                          Machine vision makes flowers on the ground

                                                          China's intelligent manufacturing has been upgraded by leaps and bounds

In the traditional wheel hub production process , Wheel hubs of different models are often completed manually , Low efficiency , And the training cost is high , It's easy to make mistakes . Made by Xinjie electric AI Hub classification / Segmentation scheme , Use the 11th generation Intel   core   processor 、OpenVINO  Tool kits and oneAPI tool kit , The reasoning accuracy can reach 99%, The reasoning speed ranges from 531 Milliseconds down to 33 millisecond , At the same time, the labor cost will be reduced 75%.

As another manufacturing giant , Foxconn has set up an industrial interconnection factory （Fii）, Automation system is widely used , But data congestion and network delay caused by massive data , Prevents device synchronization , Reduce production line efficiency . Foxconn uses the industrial Internet TSN Solution , One of its core objectives is to introduce more AI function , Build an intelligent system to monitor products . Now? , Use 11th generation Intel   core   processor 、 Intel   To the strong   And Lingdong   processor 、 Intel  FPGA Intel   Ethernet controller (i210) As a key equipment component ,TSN It can predict and analyze the service life and service condition of the cutting machine , And realize three advantages on the whole ：

1, The production line yield in the manufacturing process is improved 8%;

2, The transmission success rate of the industrial Internet of things has increased to 99%

3, Anomaly detection 、 Predictive analysis, etc AI Application improves cost effectiveness and productivity

Besides , In China Telecom 5G MEC Edge cloud platform 、 In the scenarios of junction box positioning welding in the photovoltaic industry , Intel's hardware and software play a decisive role , Help enterprises improve efficiency , Eliminate safety hazards , Reduce operating costs .

Of course , The use of machine vision is not limited to these , It has become an integral part of the intelligent upgrading of infrastructure . With the development of Intelligent Manufacturing , Machine vision as its key element , We will continue to promote the automation of industrial systems 、 Intelligent , Improve the system at cost 、 efficiency 、 Security 、 Stability, etc .

                                                                  Industrial insight

                                                                    Wisdom changes the future

Next few years , Manufacturing machine vision systems are expected to integrate more accurate vision sensors , More powerful 、 Diversified computing power , And integrate more powerful machine vision algorithms , Derive more scenario based applications . At the same time, the machine vision system will also be applied in cooperation with industrial robots and other systems , Help build flexibility 、 Intelligent production line , Realizing the vision of intelligent manufacturing . 

 

It is a continuous process to lead the development of machine vision through technological innovation and strategic transformation , Improve the efficiency of the enterprise 、 The stability of the supply chain depends on the quality of the products . Driven by digital innovation technology , More and more manufacturing enterprises will conduct independent research and development 、 Ecological cooperation, etc , Apply machine vision to industrial inspection 、 Industrial robot 、 Predictive maintenance and other scenarios , Transform traditional industrial processes into intelligent processes , It also lays a solid foundation for the mining and utilization of enterprise data resources .

Intel will continue to strengthen and ISV、SI、OEM、ODM And other ecological partners , Provide optimized machine vision solutions , take The end-to-end technological advantages empower partners , Assist in the scene implementation of machine vision solutions , And help more manufacturing enterprises to benefit from these programs , Continue to expand the capability range and service boundary of intelligent manufacturing , Turn massive amounts of industrial data into actionable insights .