Within this task we needed to get actual information from photos about geo-located objects like buildings, rivers, lakes, roads, forests, fields and place it on a generated map. 

The task also included the roofs layout classification and recognition. 

We utilized state-of-the-art and our own technologies based on computer vision, object detection, image segmentation and image processing to bring AI into this field. Our solution is a part of big pipeline that can do the following:

•  detect objects;
• determine its types and geometry on images from drones or satellites;
•  recognize buildings and their geometry, including roofs layout, building type,its height and size; 
• recognize roads, lakes, rivers, and put them on the map; 
• analyze forests configuration, 
• detect changes over time by comparing them with previous observations.