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.
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.
 | 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
The volume of the global market for cameras with computer vision for the year grew to $7.41 billion
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
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |