The participants were challenged to develop software that uses artificial intelligence to produce maps, based on both aerial images and laser data.
- Aerial photos and laser data offer a huge potential for creating maps. The data is relatively inexpensive to obtain, and in the long run we believe that with the help of machine learning, we will be able to create maps from aerial photographs to match anything we can produce today, says Eivind.
A total of eleven teams comprising doctoral research fellows and other consultancy experts took part in the competition, with Lars Martin and Eivind emerging as winners. They both work with machine learning and artificial intelligence on a daily basis at Norconsult Digital.
- It was great to be part of such a strong resource pool, and we picked up many useful tips during the competition, says Eivind, before adding:
- In the future, it will also be interesting to see how our clients can benefit from the insights we have gained from the competition, as machine learning is a constantly evolving field. It’s great to have an employer who lets us explore and play with new technology during working hours.
Presented the winning entry at an international machine learning conference
Lars Martin and Eivind both studied at the Norwegian University of Science and Technology, majoring in artificial intelligence. It was therefore particularly interesting l to present the winning solution to an audience of 250 researchers and machine-learning experts from 18 countries during the “Northern Lights Deep Learning Conference” in Tromsø earlier this year.
- As well as the award ceremony and the presentation, we also attended lectures from world-leading experts during the conference. They presented key research on several types of machine learning. It couldn’t have been better, says Eivind.
International interest
Many countries do not have adequate enough systems to produce detailed maps of buildings and terrain. The team from Norconsult Digital believe that the solutions they have developed may be particularly relevant in countries in which it is difficult to map large areas.
- AI map solutions are of interest to only a few specialists here in Norway, but we have encountered major interest from abroad. During the competition we met a professor from Tanzania who was particularly interested in our solution, says Lars Martin.
He emphasises that Norway may also benefit from the solution:
- Although Norway is a clear leader in the development of maps, it is still possible to identify various errors and omissions using the new technology. In Norway, aerial photography, combined with machine learning, has been used to detect new outbuildings or garages that have not previously been mapped.
Oxana Eide
Gruppeleder Maskinlæring Trondheim
Contact us
