<p>This study classifies and modifies the Islamic architectural style for the first time. Transfer learning (TL) is used to identify classes, and neural style transfer (NST) is used to modify the modern style to a modern Islamic style. Google Photos was the source of our data set. They were then processed to get a consensus regarding the quantity, kind, and dimensions of the channels. It was examined to assess its performance and identify the best ways to address its issues, thereby understanding how well this data performed on the confusion network. Overfitting occurred due to the limited amount of data. It also remained after data augmentation. Utilizing transfer learning increased the predicate’s efficiency. The evolution of the various convolutional network structure types and their key concepts were discussed during this study, as well as transfer learning techniques, and the best one was selected for research. Data sets were applied to seven types of architecture, and the training accuracy, prediction accuracy, precision, recall, and F1 score were computed. Also, a statistical analysis is reported as a confidence interval, and the model’s stability and consistency are calculated for each mode. TL is not only a high-precision method for the classification and recognition of shapes, but it’s also an artistic tool for transferring styles. Transfer of architectural style from one design to another. NST can transfer Islamic style to modern style. NST can transfer the features that characterize the Islamic style to modern buildings, as well as colors. The sharp edges are converted to carved edges, as well as the Islamic decoration of windows and walls.</p>