Computer vision Machine vision

What is computer vision?

Computer vision is a technology by which machines can find, track, classify and identify objects by extracting data from images and analyzing the information they receive^[1]

Computer vision is used for object recognition, video analytics, description of the content of images and videos, gesture recognition and handwriting, as well as for intelligent image processing.

How does machine vision differ from computer vision?

Machine vision uses image analysis to solve industrial problems. Machine and computer vision - areas connected

It may seem to beginners that these are different names for the same technology, but this is not the case, since computer vision is the common name for a set of technologies, and machine vision is the field of application.

Machine Vision Problems

Machine vision allows you to abandon manual labor, because a robot can control the assembly of products, read and measure objects, read text, numbers and identify objects.

Machine vision is used in various fields. In medicine - in order to more accurately diagnose, in industry - to reduce the cost of goods through automation. In the automotive industry - for navigating drones, and in retail - for reading barcodes or counting visitors.

Machine Vision Systems

Since machine vision is used to solve various industrial problems, depending on which problem needs to be solved, special machine vision systems are created.

Typical machine vision systems consist of cameras, software, processors, light sources, software applications, and various sensors.

For example, the sensor determined that the part on the conveyor needs to be checked, started the camera and took a picture of this part. After that, the image is sent to the computer, where the software for machine vision processes the resulting picture.

After the image is processed, depending on the state of the part, the program skips or does not skip the part along the conveyor further. That is, if the part is damaged, the software will signal the device to deflect it, stop production, or warn the person that there is a part with a defect.

2024: How Russian enterprises are introducing machine vision

Russian industrial enterprises are actively mastering machine vision technologies, integrating them with artificial intelligence to increase production efficiency. The growing interest in this area became known in October 2024 from the results of a study conducted by the Institute for Statistical Research and Knowledge Economics of the Higher School of Economics.

According to the report, 78% of the 2.3 thousand Russian organizations surveyed show interest in products based on computer vision. The introduction of these technologies takes place in several stages and requires significant investment.

𝓲

Фото: Freepik

The process of integrating machine vision begins with the installation of specialized equipment. Enterprises use IPv cameras of outdoor video surveillances and high-speed cameras capable of shooting up to 40 thousand frames per second. Modern devices can detect fractions up to 1 micrometer, which allows you to control the composition of materials with high accuracy.

The next step is to set up a server infrastructure that handles large amounts of data as high-resolution photos. To reduce the load on servers, neural networks are used that analyze only those images where deviations from the specified parameters are detected.

The final stage is the development and implementation of special software that includes machine learning models and image processing algorithms. The head of the department of machine vision systems and neural networks Nord Clan Pyotr Khvesyuk notes that in order to train neural networks, it is necessary to collect at least 50 thousand photographs that can be obtained both in real conditions and synthesized in the laboratory.

The cost of implementing machine vision technologies varies depending on the complexity of the project. According to the expert of the Yuztech Group of Companies Ilya Smirnov, the basic solution can cost ₽8 -15 million, and personalized software - up to ₽20 million. The payback of such projects is about two years^[2]

2022: Artificial intelligence recognises images worse than humans

Artificial intelligence recognizes images worse than humans.

Computer vision does not have the physiological features that a person has, so it recognizes images worse. This conclusion was reached by scientists from the Higher School of Economics and Moscow Polytechnic University. The HSE announced this on September 7, 2022. Read more here. .

2021: Computer vision experts close projects due to lack of training data

According to a new study by Datagen, 99% of teams of computer vision specialists faced the need to close projects using machine learning due to a lack of data to train their models. Moreover, for the same reason, all (100%) study participants were forced to postpone projects. This became known on December 27, 2021.

As the researchers found out, problems with training data are of a very different nature and affect teams of specialists equally. The most important issues are insufficient annotation (48%), inappropriate domain coverage (47%), and data deficits (44%).

The lack of reliable data for training in a particular subject area is compounded by the fact that computer vision lacks well-defined standards and best practices.

When asked how training data are collected in their organizations, the respondents presented a whole "solyanka team" from various sources and methodologies. Synthetic or real, collected within an organization or obtained from public sets of data - as it turned out, absolutely all data is used to train computer vision models of an organization, regardless of their origin.

However, teams of specialists in the field of computer vision seem to have found a solution to the problem in the form of synthetic data. 96% of respondents already use synthetic data to train their artificial intelligence models. However, the quality, source and proportion of synthetic data used still vary widely by area, with only 6% of teams currently using exclusively synthetic data.

The widespread shift to synthetic data is in line with the number of new projections that 2022 will be a breakthrough for synthetic data.

The Datagen online survey was attended by 300 specialists in the field of computer vision, representing 300 individual enterprises^[3].

2020: Companies start replacing guards with machine vision cameras

On April 27, 2020, it became known that retail stores, construction and production companies began to equip their premises and sites with video surveillance systems with artificial intelligence. They monitor whether visitors and employees wear medical masks and keep distance from each other, as required by the COVID-19 pandemic. Installing such systems, the cost of which reaches $1 thousand per year, is cheaper than having additional guards, according to market participants interviewed by Reuters.

Representatives of several companies said that artificial intelligence video surveillance will remove any claims related to non-compliance with the instructions of health and human well-being authorities. With such a solution, not only visitors and employees of stores and other enterprises will be able to see that the safety rules are fully observed, but insurers and regulators will also be able to do so.

Retail stores, construction and manufacturing companies began equipping their premises and sites with artificial intelligence video surveillance systems

Least of all, we want the governor to shut us all down simply because no one is following the prescribed guidelines, "said Jen Suerth, vice president of the Chicago-based construction company Pepper Construction. In April 2020, she launched "smart" video surveillance based on the SmartVid.io software product to track worker behavior.

Diamond maker Samarth Diamond and U.S. outdoor mall owner RPT Realty plan to introduce similar technologies. The first plans to use a solution from Glimpse Analytics, the second - RE Insight.

Customers are confident in the capabilities of the technology because they have already used similar tools to study customers entering stores, as well as to detect employees on construction sites who neglect basic safety rules.^[4]

How computer vision systems help control product quality

• How computer vision systems help control product quality
• How Machine Vision-Based Software Works for Product Quality Control and Process Engineering

Computer Vision Market Research

2024: Global Machine Vision Market Growth by 9% to $15.44 Billion

In 2024, costs in the global market for machine vision and robotic systems with visual control reached $15.44 billion. The indicator of the previous year, when expenses in this area were estimated at $14.21 billion, was exceeded by 9%. This is stated in a study by Market Research Future, the results of which TAdviser got acquainted with in mid-January 2025.

One of the key drivers of the industry in question, analysts call the rapid development of automation technologies. Enterprises around the world are striving to improve operational efficiency and reduce production costs. Machine vision tools provide accurate quality control and defect recognition. At the same time, the intervention of employees of enterprises is minimized. As a result, production line productivity increases and the human factor is eliminated. This is particularly important in sectors such as automotive, electronics, consumer goods and pharmaceuticals, where the accuracy and reliability of manufacturing processes is paramount.

The industry is also being stimulated by advances in artificial intelligence and deep learning. These tools enhance the capabilities of machine vision systems, allowing them to cope with complex tasks that were previously difficult or impossible to automate. AI helps to carry out more accurate and faster analysis of images in various conditions. As technology advances, machines are able to perform operations that require high visual acuity, for example, to detect the smallest defects in products on the production line with greater efficiency.

The growing use of robotics in various sectors is another important driver for the machine vision market. Businesses from industries such as agriculture, logistics and healthcare are increasingly using robots for a variety of applications. In addition, the introduction of collaborative robots, or cobots, that can work together with humans is growing.

The authors of the report identify four key segments in the market under consideration: quality control; navigation; identification and inspection; measurement and sizing. In 2023, the costs in these areas amounted to $3.56 billion, $3.1 billion, $4.23 billion and $3.32 billion, respectively. Among the significant players in the industry are named:

• Slamcore;
• National Instruments;
• Keyence;
• Datalogic;
• Sick AG;
• Advantech;
• Flir Systems;
• Amnon Automotive;
• Cognex;
• Microchip Technology;
• Basler;
• Teledyne Technologies;
• Linefollower;
• Omron;
• Euresys.

Geographically, North America is leading with a result of $5.25 billion at the end of 2023: many large enterprises with advanced assembly and production lines on which machine vision systems are used are concentrated in this region. Europe is in second place with costs of $4 billion, and the Asia-Pacific region closes the top three with $3.5 billion. South America secured a contribution of $0.75 billion, the Middle East and Africa - approximately $0.71 billion.

In general, as noted, the current situation on the market indicates a stable growth trajectory due to technological advances, automation needs and the need to control the quality of products in various sectors. Market Research Future analysts believe that in the future, the CAGR will be 8.65%. As a result, by 2032, costs on a global scale could increase to $30 billion.^[5]

2023: Global Computer Vision Camera Market Size Grows to $7.41 Billion for the Year

In 2023, sales of cameras with computer vision on a global scale reached $7.41 billion. For comparison, a year earlier, the market volume was estimated at $6.14 billion. Thus, growth was recorded at 20%. This is stated in a study by Market Research Future, the results of which were published in early November 2024. Read more here.

Computer Vision: Technology, Market, Outlook

Computer Vision: Technology, Market, Outlook

• Video Analytics Systems Video Analytics Systems and Projects Catalog
• Basics of computer vision
• Technologies used
• Applications
• Global Computer Vision Market
• Russian Computer Vision Market

AI eyes: what computer vision systems see today and what they will see tomorrow

AI eyes: What do computer vision systems see today and what will they see tomorrow?

Notes