18 сентября 2017
Students and graduates of Innopolis University developed the best solution for unmanned vehicles for identification of road humps, thus leaving behind the teams of Massachusetts Institute of Technology, the University of Cambridge, and the University of Science and Technology of Beijing in the team rating.
VisionHack 2017, the first International Hackathon on Artificial Intelligence and Computer Vision with available prize fund of more than $30,000 has finished in Moscow. The purpose of the competition is to create an Advanced Driver Assistance System (ADAS) for automated identification of emergency signals emitted by vehicles, pedestrians, road humps and events affecting road-traffic safety.
To create the system, developers the IT University used a data array from 10-second videos coming from driving recorders. Analysis of these records helped to define the character of movement i.e. driving along the pedestrian crossing or under the overhead road, moving in/out of the city, facing speed bumps, movement of screen wipers, etc.
“We tackled each task individually. For instance, we looked for the relevant road signs to detect pedestrian crossings. To deal with overhead roads, we focused on the moment when the upper part of the road turned dark. Meanwhile, a moment the camera started moving up and down helped us to detect speed bumps” — explained Sergey Verentsov, the team leader.
According to members of the IT University, experience of their precursors — graduates of Innopolis University working on a driver support system at RoadAR — helped them to develop the solution: “We used methods for detecting road signs to identify the location of pedestrian crossings and road humps”, — added a Master’s degree student of Innopolis University.
The team of Innopolis University developed the best solution for detecting road humps and won a $2,000 prize. Representatives of the Russian IT University took the 7th position in the team rating leaving behind the teams of Massachusetts Institute of Technology, the University of Cambridge, and the University of Science and Technology of Beijing, Polytechnic University of Catalan, and Arizona State University.
VisionHack 2017 was organized by National University of Science and Technology MISIS and Cognitive Technologies under supervision of AMD, Vneshekonombank, Jaguar, Nvidia, Uber, Land Rover, Skolkovo, United Electronic Market Place. The team of Innopolis University included 2nd-year Master’s degree students with major in Data Science: Anton Kiselev, Sergey Verentsov, Savely Vorontsov, and graduates of the University — Sergey Solonets and Konstantin Sozykin.