In this blog article, we'll take a closer look at the topic of computer vision, looking at the applications of machine learning and deep learning.
Computer vision (CV), also known as computer vision, influences our working world and our everyday lives without most of us being consciously aware of it. This technology in the field of artificial intelligence (AI) enables machines to 'see' in a similar way to humans, recognizing and interpreting complex visual data.
With Computer vision can automatically recognize and extract large amounts of unstructured data from image and video formats. The combination with machine learning, in particular with deep neural networks (deep learning), is essential here. This enables computer vision to recognize objects as well as faces and now even emotions in a matter of seconds.
Here is a short video from the cloud on the subject in English:
By loading the video, you accept YouTube's privacy policy.
Read more
The areas of application are diverse and range from the recognition and processing of documents to the recognition of traffic signs and the analysis of X-ray images. The only basic requirement for computer vision to work: The data sources, i.e. images, recordings or other documents, must have been digitized in advance.
With advancing digitalization, the The technology is therefore being used in more and more industries, from agriculture and the automotive industry to the insurance sector. Microsoft also uses this technology as part of its Azure cloud computing platform. It is used to support a wide range of services.
CV also contributes to the improvement of intelligent video analysis software. It does so by supporting complex tasks such as scene reconstruction and object recognition. Smart companies have recognized the enormous potential of this technology and are using it to improve both business processes and data analysis. This saves them time and money.
Finally, you can also read about an example where a pro-gamer and world champion was beaten by the use of computer vision in his game.
Definition - What is computer vision?
Computer vision is a specialized field of artificial intelligence (AI). It finds its application in simulating and in some cases even surpassing human vision. At its core, it deals with the automated acquisition, processing, analysis and interpretation of visual data (image and video).
The developed algorithms and techniques in computer vision enable computer systems to understand and interpret visual information in a similar way as humans do. This ranges from simple applications for image analysis and text reading to complex scene understanding and scene reconstruction.
Computer vision has made considerable progress in recent years and remains an active area of research and development. This can also be seen in more comprehensive vision AI approaches, in which pure image processing and deep learning are supplemented by other AI aspects such as semantic analyses. With the advent of more powerful AI techniques, the potential for the application of visual data analysis has expanded considerably.
Likewise, the cloud has ensured more intensive use. These advances enable computer vision to be used in many application areas. From text recognition to face recognition to autonomous vehicle navigation, a wide range of tasks are covered, constantly expanding learning.
Areas of application for computer vision
Computer vision is used by various industries and organizations. The automation and the associated time and cost savings are the greatest advantage in the application. For a better understanding, here are some practical examples of applications in selected industries:
Automotive industry
Automakers use computer vision for driver assistance systems, autonomous driving, traffic sign and pedestrian detection, and monitoring the vehicle's interior.
Public health
In medicine, computer vision is used to analyze medical images, improve diagnostic procedures and detect diseases. Examples are the analysis of X-ray images, CT scans or MRI images.
Retail
Retail companies are using computer vision to analyze customer shopping behavior, automate inventory tracking, provide customer-focused recommendations, and improve theft prevention systems.
Agriculture
Computer vision is used in agriculture to detect plant diseases, monitor crop growth, automate harvesting processes and optimize the use of fertilizers or pesticides.
Authorities and banks
Here, computer vision is used for document processing to automatically read documents such as passports, ID cards or driving licenses and extract relevant information. This speeds up administrative processes such as identity verification or document creation.
The application of computer vision in government and banking helps improve safety, efficiency and customer experience.
These examples illustrate that computer vision is being used in a variety of areas to improve processes, increase efficiency, enhance safety and develop new innovative solutions.
Computer vision as an IT subfield
Artificial Intelligence (AI)
Artificial intelligence (AI) refers to the ability of computers or machine systems to perform tasks that would normally require human thought. It involves the development of algorithms and techniques that enable computers to analyze data, recognize patterns, draw conclusions and solve problems.
Computer Vision (CV)
Computer vision uses machine learning and deep learning to analyze and interpret visual data. This involves tasks such as object recognition, image classification, face recognition, image segmentation, motion tracking, and more. By using Deep Learning models, especially Convolutional Neural Networks (CNNs), computer vision systems can handle complex visual tasks with high accuracy.
The use of deep learning in computer vision has led to significant advances in image recognition, analysis, and processing. By training large neural networks with large amounts of data, computer vision systems can recognize and understand complex patterns and features in images.
Computer vision is therefore a field of application within artificial intelligence that is based on machine learning and deep learning in particular.
Machine Learning (ML)
Machine learning (ML) is an umbrella term for various algorithms and techniques that allow a computer system to learn from experience and recognize patterns in data. It enables the computer to perform tasks or make predictions without being explicitly programmed for the application.
Deep Learning (DL)
Deep Learning is a special approach to machine learning based on artificial neural networks. These networks consist of several layers that are interconnected. This is where the term "deep" comes from, from the English word "deep". Deep learning models are able to automatically learn abstract representations of data by extracting hierarchical features in the data. Very often, these applications are used in the so-called cloud.
Differences - computer vision vs. machine vision
Computer vision and machine vision are terms that are often used interchangeably. Mainly because they refer to similar concepts and technologies. However, there are some subtle differences between the two terms.
Computer Vision
Computer vision refers to the scientific and technical field concerned with the automated processing, analysis and interpretation of visual information. The goal of computer vision is to enable computers to see like humans.
It encompasses a wide range of techniques to capture, understand and interpret visual data (images or videos). These include image processing, pattern recognition, object recognition, image segmentation, 3D reconstruction, and more. Computer vision finds application in various fields such as document recognition, face recognition, medical imaging and analysis, surveillance systems in security with image and video, etc.
Machine Vision
Machine vision technology helps industrial plants in operation to make decisions. The application of machine vision is used in visual inspection and defect detection, positioning and recognition, object sorting and so on.
Machine vision is one of the founding techniques of industrial automation and has helped improve product quality, speed up production and optimize manufacturing.
In summary, computer vision is a broader concept. It deals with the entire range of processing and interpretation of visual information, both in the form of image and video. Machine vision is a specific subset of computer vision. Machine vision often uses techniques from computer vision to achieve its goals.
Emergence of computer vision technology
The following are some milestones and developments in the history of computer vision. The general public became aware of the impressive capabilities in 2017. Because at that time, the neural network AlphaGo against the world champion in Board game Go in winning use, thereby demonstrating the capabilities of AI systems in real-life visual perception.
- 1960s
- 1966 - The Summer Vision Conference developed by Marvin Minsky is considered one of the earliest milestones in the history of computer vision. Fundamentals and challenges were discussed there.
- 1970s
- 1970 - The first image processing system capable of recognizing geometric shapes was developed by Lawrence Roberts.
- 1973 - Michael Fischler and Robert Elschlager developed the Pictorial Structures Model for detecting and tracking objects in images.
- 1980s
- 1980 - David Marr developed a theory of visual perception and presented a mathematical model for image analysis.
- 1983 - The Scale-Invariant Feature Transform (SIFT) method by David Lowe enabled robust feature detection and description in images.
- 1986 - The Optical flow processes by Berthold K. P. Horn and Brian G. Schunck enabled the motion tracking of objects in image sequences.
- 1990s
- 1994 - The Active Appearance Model (AAM) by Timothy Cootes enabled the modeling and tracking of facial features.
- 1999 - The Speeded Up Robust Features Procedure (SURF) by Herbert Bay was developed and enabled fast and robust feature extraction in images.
- 2000s
- 2001 - The Viola Jones Procedure for real-time face recognition was developed by Paul Viola and Michael Jones.
- 2012 - AlexNet, a deep neural network, won the ImageNet competition and significantly improved the performance of image classification.
- 2010s until today
- 2014 - The Generative Adversarial Network (GAN) by Ian Goodfellow enabled the generation of realistic images.
- -2015 - The Convolutional Neural Network (CNN) ResNet achieved very high accuracy in image classification.
- 2017 - The neural network AlphaGo defeated the world champion in the board game Go and demonstrated the capabilities of AI systems in visual perception.
- 2020 - The Transformer-Model, originally developed for speech processing, was applied to computer vision and led to major advances in image processing and text-image interaction.
These milestones have shaped the development and progress of computer vision and have led to a variety of applications in areas such as autonomous driving, medicine, security, entertainment (film, video), and document processing.
Outlook and conclusion
Computer vision remains an active research area with high development and innovation potential for the future. Learning is continuously developing due to the high amount of data read out daily.
Excursus: Computer vision and the computer vision syndrome
Computer vision has nothing to do with the Computer vision syndrome (CVS). Here, the term is used in a completely different context: namely, when it comes to eyestrain caused by looking at the screen for too long.
FAQ
What is Computer Vision?
Computer vision is a field of computer science that deals with the automated processing, analysis, and interpretation of visual data to enable computers to see like humans.
Where is computer vision used?
Industrial automation, document processing, robotics, automatic document verification, augmented reality (AR), facial recognition, medical imaging, surveillance and security, document management, etc. Microsoft, by the way, uses CV in its Azure cloud platform.
EN 
DE 
ES 
FR