Vernacular architecture in the space of a modern city, based on deep learning methods and three-dimensional structural analysis

Abstract

This research paper describes the process and results of a project to automatically classify historical buildings using aerial photography and satellite imagery. New computational scientific methods and the availability of satellite images have created more opportunities to work on automated recognition of pieces of historical architecture. In this regard, the convolutional neural network (CNN) is the main classification approach within the project. As a result, the trained model is tested using a validation data set and has a roughly 98% accuracy. In addition, being affected by urbanization and other factors, local architectural heritage faces the challenge of introducing innovations for sustainable development, with originality and authenticity being preserved in redesign and planning. Thus, this study uses a visualized quantitative analysis to analyze the research trends in Russian vernacular architecture and study new ways of coexistence between vernacular architecture, object perception and cultural ecology. The most important task of this study is to analyze the theory of coordination between the emotion social and cultural structure and the cultural ecosystem in vernacular architecture. The main contribution of this study is the proposed concept of a subjective-cultural eco-design system for vernacular architecture sustainable development to establish a 3D structural analysis design paradigm and evaluation analysis matrix, and to ensure that vernacular architecture demonstrates the ability to self-renew by continuous exchange and revision in the dynamic cycle of the current design system.