Evaluating Deep Learning Models for Architectural Image Classification: A Case Study on the UC Davis Campus

Deep learning has become a cornerstone of machine learning, excelling in tasks that demand intricate data analysis, such as image classification. This study delves into the use of deep learning for architectural images recognition. In this work, we introduce a unique dataset of 25 labeled classes including architectural buildings and departments at the Univeristy of California, Davis. By employing and adapting a range of Convolutional Neural Networks (CNNs)—including MobileNet, ResNet, Xception, Inception, and AlexNet—we comprehensively evaluate their performance in identifying and categorizing distinctive architectural features of the campus. Through a detailed comparative analysis, we highlight the strengths and limitations of each model, shedding light on their practical applications for automated campus management and navigation. Our findings underscore the robustness and adaptability of deep learning in architectural recognition while contributing to the broader discourse on its application in real-world academic settings. This work not only establishes a foundation for future studies but also proposes innovative pathways for integrating deep learning in architectural and campus-focused technologies.