TSUAB and TSU Faculty of Mechanics and Mathematics scientists are developing algorithms to train an AI to recognize road surface defects using video and photo images. Jointly developed by the University of Tomsk scientists, the technology is viewed as promising in light of the import substitution trend.
“We task ourselves to use photo and video information, as well as AI algorithms, to recognize roads and bridges, and defects on their surfaces. It is an important feature that road and bridge construction workers would find very useful when making decisions on maintenance works and full structural overhauls,” says the head of the TSUAB Department of Engineering Geology, Bridges, and Road Constructions, Pavel Elugachev.
According to Elugachev, currently the defects by video and photography are found manually, which takes a lot of time and effort. TSUAB scientists and scientists at the TSU Department of Mathematical Analysis of the Faculty of Mechanics and Mathematics are training an AI and conducting an experiment on using computer vision in road construction. To do that kind of job, a car or a drone with a smartphone is more than enough.
As explained by the research engineer Aleksei Bannikov, the passage of a car or a flight of a drone are recorded on a camera, then the computer vision algorithm detects potential road defects on the recording, may it be cracks, pits, or others.
According to the representatives of the universities, the research is part of the import substitution trend to develop domestic AI algorithms.
“After a year of work, we got sufficient results, which showed that this technology is indeed promising. It also allows us to speculate that in the future the monitoring of roads and various structures with the help of artificial intelligence will be universally accepted”, Pavel Elugachev sums up.
Original article by: press service of TSUAB, RIA Tomsk