Revolutionizing hyperspectral imaging
This is the first video of the FCAI success stories series. In the video series, we explain why fundamental research in AI is needed, and how research results create solutions to the needs of people, society and companies.
Researchers at FCAI have developed an artificial intelligence based spectral imaging technology that will make color determination accurate, universally true and accessible to everyone with a smartphone.
Professor Arto Klami, doctoral students of computer science Mikko Toivonen and Chang Rajani have designed computer vision algorithms under the spexel.ai project. These algorithms can be used to determine colors to a very high degree of precision.
The new technology is based on the combination of a smartphone, a peripheral device attached to the phone, and an application that determines the color in a cloud-based service with the help of computer vision algorithms. This technology can soon turn your smartphone into an extremely accurate camera.
Accurate color detection can help, for example, interior designers or those who work with paint. With the spexel.ai technique, they can take a picture of their subject and obtain accurate information on its color shade from the application. The technique could also be useful for online shoppers looking for a product of a specific hue.
Spexel´s vision is to bring hyperspectral imaging to consumers. In this, color detection is a natural starting point, as the technical solutions required are easier to implement. Spexel technology allows for the development of other solutions as well.
“It could be used to perform regular self-checks to accurately analyze and track skin changes via your smartphone, e.g. detecting and identifying rashes, moles, or just monitoring general skin health. It could also be used to develop a smartphone-based device to provide information on a plant’s physiology, or for the quality assessment of fruits and vegetables”, says Ville Kurri, commercial lead for spexel.ai.
The technology involved in hyperspectral imaging is already in use, but most of the devices with hyperspectral imaging capacity currently available are expensive specialist equipment.
“We have been able to replace a large part of the hyperspectral technology previously done optically or mechanically with artificial intelligence methods. This enables low-cost spectral imaging solutions, and for the technology to be included in existing cameras including smartphones”, says Kurri.
Text and video production by Mia Paju.
Spexel.ai is based on technology developed at the Department of Computer Science of University of Helsinki, building on long-term artificial intelligence research of the Multi-source Probabilistic Inference group being part of FCAI.
Read more about research behind Spexel.ai solutions:
Snapshot hyperspectral imaging using wide dilation networks, Machine Vision and Applications (2021)
Practical Camera Sensor Spectral Response and Uncertainty Estimation, Journal of Imaging (2020)
Multiscale Cloud Detection in Remote Sensing Images Using a Dual Convolutional Neural Network, IEEE Transactions on Geoscience and Remote Sensing (2020)