Ukrainian NGO uses AI to identify reusable materials to rebuild war-battered Bucha

Artificial intelligence recognized the materials of damaged and destroyed buildings in Bucha with an accuracy of up to 86%.

The survey was developed by the NGO ReThink team, which contributes to Ukraine's transition to a closed-loop economy. The obtained results were published on an interactive web platform, Rubryka reports.

Data from the Damaged In UA organization became the basis for AI training. The trained model classified 20 materials, most of which are slate, metal tiles, and concrete slabs.

The goal of the project is to create an open database with records of the materials of the destroyed buildings in the city of Bucha, which were determined using image processing by trained artificial intelligence (AI). 

Its results make it possible to assess the potential of secondary use of construction materials of the studied buildings and update the applied methods for other damaged regions of Ukraine that suffered from Russian aggression.

The AI identifies metal sheets in the best way and brick the worst, which is the least present in the images, with 37% definition clarity. The results of the analysis have the potential for a better assessment of the level of destruction in other territories of Ukraine, the team says.

UAV images of all Bucha buildings and files with an assessment of the level of destruction in the city provided by the Damaged in UA organization, as well as data from open sources, were used to analyze the buildings.

The project was implemented within the UNDP project "Strengthening Recovery in Ukraine Based on Actual Data," a component of which is executed by the Kyiv School of Economics.