Neural network against garbage: technologies help to clean hard-to-reach shores

      Coastal pollution is a serious environmental problem, especially in remote regions. To solve such situations, specialists from the Yandex School of Data Analysis and the Far Eastern Federal University have created the Clean Coast neural network, which is able to find and classify garbage on aerial photographs of the coast.

      How does it work?

      The neural network analyzes aerial images and recognizes 6 types of debris: fishing nets, iron, rubber, coarse plastic, concrete and wood. Then it determines the coordinates of the waste, its composition and weight, which helps to plan the cleaning: how many people and special equipment are needed, the route.

      The system has already been tested by an ecological expedition in the South Kamchatka Federal Reserve. There, the cleaning of 5 tons of waste was four times faster. The neural network is also planned to be used in other national parks and reserves of the Far East and the Arctic. In addition, the system can be trained to find other types of waste.

      This is a great example of how modern technology helps solve environmental problems. This makes it possible to preserve nature in the most remote corners of our planet, which corresponds to the goals and objectives of the national project "Ecological Well-being".