<![CDATA[The classification of buildings into residential and commercial categories is essential for computer vision applications that support urban analysis and automated decision-making. Accurate identification of building functions from visual data enables downstream applications, such as service differentiation and resource allocation. However, manual image-based classification is inefficient and subjective, particularly when applied to large-scale datasets. This study proposes a deep learning approach to classifying residential and commercial buildings using street-level images. A dataset comprising 882 images was collected from publicly available sources and categorized based on visual characteristics. The dataset was divided into training, validation, and test sets in a 60:20:20 ratio. The preprocessing stages included data cleaning, labeling, cropping, and resizing. Two models were evaluated: a conventional Convolutional Neural Network (CNN) and MobileNetV2 with transfer learning. Model performance was optimized by tuning the batch size and learning rate. The experimental results show that MobileNetV2 achieved the best performance with a batch size of 32 and a learning rate of 0.0001, attaining an accuracy of 92.05% and precision, recall, and F1-score values of 91.46%. Evaluation using a confusion matrix indicated low misclassification rates both building categories. These results demonstrate that deep learning models can effectively classify buildings based on visual data.]]>
Copyrights © 2026
