April 3

SUAI has developed a machine learning model for landscape analysis

Employees of the SUAI School of Engineering have presented a machine learning model that can accurately recognize objects and analyze the environment using data collected from unmanned aerial vehicles.

Intelligent data processing allows for mapping and classifying objects.

The system automatically collects data from drones for analysis. A drone is equipped with a lidar, a device that uses laser beams to measure the distance to objects and creates a 3D point cloud. During the flight, the drone scans the area, and the lidar records millions of points reflected from the ground such as landscape elements such as trees, buildings, rivers, roads, paths, and other objects. Additionally, you can configure the model to recognize specific objects, such as power transmission line supports, cars, and bridges. This data is then transmitted to the server, where it is processed. Using artificial intelligence, the system recognizes key objects and classifies them with an accuracy of up to one unit. Then the landscape features are analyzed.

The advantages of this machine learning model are high accuracy of recognition of several classes of objects, a large analysis area. Analysis of 3D point clouds from drones and automatic recognition of objects has a wide range of applications: forestry and ecology, construction and cadastral work, agriculture, logistics, monitoring of water bodies, and security of objects.