Tyumen Researchers Train AI to Design Buildings Inspired by Local Architectural Heritage
Researchers at Tyumen Industrial University are developing a generative AI model that will help architects design buildings while taking the architectural character of a specific region into account.
The three-year project is funded by a grant from the Russian Science Foundation.
The research team is led by Maria Gaiduk, Ph.D. in Architecture and Deputy Director for Research and Innovation at the Institute of Architecture and Design. According to Ivan Velizhanin, an expert in parametric design and neural networks, conventional AI design tools are often "stylistically blind" to local architectural traditions and do not always take a region's cultural context into account.
To address this limitation, the researchers will create a digital database of Tyumen's architectural heritage by digitizing more than 1,000 images of historic buildings and annotating their stylistic features. Rather than copying historical structures, the AI model will generate contemporary architectural concepts that reflect the city's architectural identity.
The research team also includes early-career researchers Karolina Turanskaya and Kira Kolesnikova. In the future, the methodology could be applied in education and tourism and expanded to other Russian cities.
