Cognitive Pilot, a subsidiary of Sberbank and Cognitive Technologies, has launched a projecttogether with the agricultural holding
In addition to combines, in testing there will beparticipate tractors, spraying machines and other equipment. Based on the tests, the companies will create an open Cognitive Agro Dataset dataset with images that can be used for training autonomous agricultural machines.
"The goal of our project is to create a unique tool that allows developers of control systems for unmanned agricultural machinery from all over the world to effectively solve the problems of training neural networks on real data and create their own autopilots.Cooperation with the Peschanokop Agrarian Group, the opening of the landfill, will allow us to recreate a wide range of situations, including those that are extremely rare in the conditions of real agricultural work, in different weather conditions with different illumination, which are of the highest value for the preparation of the database, and as a result, ensuring the safe operation of the autonomous control system for agricultural machinery."
Olga Uskova, CEO, Cognitive Pilot
The company notes that the creation of Cognitive Agro Dataset is one of the first attempts in the world to create a dataset for industry developers.
Previouslywas reportedthat the Russian company Cognitive Technologies has developed computer vision technology for drones of the South Korean company Hyundai Mobis, part of the Hyundai Motor Group.