Deep Learning Architect: Classification for Architectural Design Using Artificial Intelligence

In this paper, we use cutting-edge deep learning and computer vision to compare architectural designs from various businesses. We trained a squeeze net model using different classification algorithms to identify 1404 architects’ photos with 88.6% accuracy by integrating a dataset of web-scraped photographs with our collection of architectural works. By analyzing the weights in the trained Neural Network model, we may determine the extent to which the computer understands common aesthetic traits across architects. We use this metric to categorize buildings by various architects into groups sharing commonalities in their design. We next evaluate our results against the standard categorization used by architects and historians. It is surprising that our classification of architectural styles corroborates commonly held beliefs in the field, and that the acquired knowledge of architectural characteristics is in line with a common appreciation for these styles.