TYUMEN, July 20 – Trend Wire. Scientists from Tyumen Industrial University will develop a generative neural network model for designing buildings taking into account the architectural features of a particular region, the university’s press service reported to Trend Wire.
According to Ivan Velizhanin, an expert in the field of parametric design and neural networks, whose opinion is cited by the press service, existing universal neural network solutions are still often “stylistically blind” to local specifics and are not always able to work with the cultural context of a particular territory.
“The university team, led by the candidate of architecture, deputy director for science and innovation of the Institute of Architecture and Design, Maria Gaiduk, will develop a generative neural network model to preserve the architectural identity of the urban environment (using the example of Tyumen),” the press service said.
The developers plan to teach algorithms to “understand” cultural and historical context. They will form a database of the architectural heritage of Tyumen, for which they will digitize more than a thousand images of ancient objects and carry out a parametric marking of their stylistic features, reflecting the recognizable architectural code of the city.
According to the idea of the project’s authors, the university clarified, this approach will allow creating not copies of historical buildings, but modern concepts that are stylistically related to the urban environment and take into account regional characteristics. In the future, the technology can be used in the educational process and for commercial and tourism purposes.
The project, according to the press service, is designed for three years and will be implemented with a grant from the Russian Science Foundation. In addition to Gaiduk and Velizhanin, the interdisciplinary project team includes young researchers Karolina Turanskaya, who combines competencies in the field of AI, architecture and cultural studies, and Kira Kolesnikova, specializing in architectural visualization.
The university noted that the project materials will be able to be used not only by scientists and teachers, but also by design organizations, city services, and in the future, commercial structures to create AI assistants for architects.
Also, the press service noted, it is possible to scale the methodology to other historical cities of Russia. This will make it possible to create a unified catalog of “Architectural Codes of Russia” – a system of digital profiles that will help preserve the unique appearance of cities in conditions of active urban development.
Source: RIA