Machine vision

At VTT, we develop next generation machine vision solutions for demanding industrial applications by combining novel imaging methods, optical innovation, and AI-based image analysis. Our agile process makes it possible to rapidly develop ideas into prototypes and tailored machine vision systems, solving problems beyond traditional machine vision.
Key facts
We customise our imaging and measurement solutions and hardware to fit the needs of each customer and case, and we are not fixed to any specific technique or component. We then use AI in image analysis to understand images at varying levels of complexity.
We develop machine vision solutions for manufacturing and process industries, and work with customers with a wide range of different challenges and use cases from quality control to inspection of materials.
The multidisciplinary expertise in our team allows us to carry out projects from start to finish. Our experts in electronics, optics, mechanics, and AI algorithms work together to develop and build tailored on-line prototypes.

On this page
Next generation machine vision for demanding industrial applications
Measurement technologies that provide accurate information about industrial processes are crucial for successful automation and digitalisation. By bringing an alternative to the human eye to the production line, companies can make processes such as in-line measurement of critical parameters, quality control, condition monitoring, and identifying maintenance needs more systematic, precise, and cost-efficient. With accurate and standardized data straight from the production line, it is possible to optimise processes and achieve cost savings.
We develop machine vision solutions for vastly different industries and use cases. Our unique expertise is in identifying and developing imaging systems that are customized to fit the specific needs and challenges. We are not limited to any specific technologies or products, and we always opt for the best possible solution considering customer specific needs. We are able to develop novel solutions that go beyond the capabilities of traditional machine vision.

We have experience working with the following industries:
- Metal, steel, paper and pulp
- Mining, mineral processing, scrap metal sorting
- Pharmaceutical manufacturing, process monitoring
- Consumer applications
- Defense, intelligent traffic, police

Expertise through three different approaches
Our approach to machine vision systems combines expertise in novel imaging methods, optical innovation, and AI-based image analysis. By bringing together these different aspects, we are able to develop viable solutions for even the most demanding applications.
Novel imaging methods and optical innovations
Often the measurement environments and accuracy requirements of industrial processes are too demanding for traditional camera technology, so to carry out the measurements, we need to overcome limits of existing technologies in terms of both measurement approaches and mechanics. At VTT, we focus specifically on new solutions for challenging applications where novel research and innovations are needed in both imaging methods and image analysis.
Machine vision can utilise, for example, areas of the light spectrum invisible to the human eye to highlight the desired elements of the object being measured. With innovative lighting and optical solutions, it is possible to emphasise elements such as the topography of the object or the properties of the surface.
Our optical innovations and novel imaging methods include:
- Instrument development and optical design
- 3D imaging
- Hyperspectral imaging
- Special imaging methods, including LIDAR/TOF and novel X-ray imaging
- Particle size measurement for different processes.
AI in image analysis for industrial use
By using artificial intelligence and machine learning in the image analysis, we can build tools that are more objective and transferrable than traditional image analysis methods and allow faster and more agile development of new applications.

Our team works together in developing the AI algorithms that are most suitable for each case and imaging method. By applying neural networks to detect anomalies already on the production line, we can support companies in optimising their processes and minimizing waste.
For example, as part of a joint EU-funded X-MINE project, we have used X-ray analysis combined with machine learning in the setting of a mine to separate rocks in the mine based on their mineral content.
Image: Example of AI based particle size analysis
Agile and rapid development from idea to prototype
While we focus on developing effective and innovative machine vision solutions for challenging use cases and environments, we always want to make sure that our solutions can be easily integrated into the customer’s existing production processes.
Our customer projects always begin with a real-world industry challenge, and we work in close cooperation with the customer’s experts to fully understand their individual needs. We then select the most suitable imaging and analysis technologies and take an agile approach in confirming the feasibility of our proposed solution. We first conduct prototyping and testing in a laboratory setting before taking our solution to the real environment. Actual testing takes place at the production facility to ensure it fulfills the customer’s on-line measurement requirements.
With a tailored machine vision solution, companies can obtain valuable information on their manufacturing processes, allowing them to optimise different aspects of the work more accurately and effectively from process and quality control to determining maintenance needs. With these solutions, businesses benefit from removing bottle necks from the production process, optimising quality of the product and minimising waste.
Cases


VTT’s machine vision innovation sped up production of crucial medical devices during the covid-19 pandemic
As part of Business Finland -funded Reboot IOT Factory -project, VTT developed a machine vision -based quality control tool for breath gas analysis components used in GE Healthcare’s patient monitoring devices. The time required for manual quality control had created a bottle neck in the production, which became a problem when the global need for the devices suddenly multiplied as the result of the covid-19 pandemic.
VTT’s innovative machine vision solution measures the performance of the critical components without a need for manual inspections. The solution, which is based on analysing different spectral ranges, was tested at GE Healthcare’s production facility in Vallila, Finland, and it soon became part of the new production process allowing faster, around-the-clock production.
Demanding conditions in mines put machine vision to a test
VTT has worked extensively with mining and minerals companies to develop tools to support more efficient processes and help decrease the footprint of the industry. As part of jointly funded Apassi consortium, VTT has developed machine vision solutions for minerals analysis for Metso Outotec.
For example, as part of the project VTT and Metso Outotec decided to tackle a uniquely demanding application, where machine vision was used to measure particle size for a valuable mineral that was almost translucent, creating additional challenge to both imaging and analysis techniques.
VTT developed a solution based on neural networks which allowed the segmentation of the particles and estimating the particle size from the images.
Meet our experts
How to work with us
-
Problem definition
We discuss with the customer to develop a clear understanding of their problem or measurement need ar a meeting or during on-site visit. -
Initial analysis
In the initial analysis, we discuss different aspects of the project, including specifications, feasibility, possible technologies, concepts, risks, samples, and commercial options. -
Feasibility study
We carry out evaluation of measurement technologies in laboratory environment. A setup is built to measure samples to assess and compare measurement technologies, evaluate risks etc. -
Prototype
On-line prototype is built and tested to verify the measurement concept in the real environment.This stage consists of specification phase, building the prototype, measurement phase, and results analysis and reporting
-
Technology transfer
Together with the customer, we discuss and make a plan for technology transfer. This can be done by either the customer, VTT, or a third party who takes commercial responsibility of the technology.
Contact us here!
Scientific publications
Andrei Baraian, Vili Kellokumpu, Janne Paaso, Lauri Koresaar and Jani Kaartinen: “Computing Particle Size Distribution using Deep Learning”, EUVIP2022
Andrei Baraian, Vili Kellokumpu, Tomi Räty and Leena Kallio. ”Automatic fracture detection and characterization in borehole images using Deep Learning-based semantic segmentation”, accepted to VISAPP 2023