In the context of sustainability, monitoring and maintenance of traditional buildings is crucial. The concept of artificial intelligence can be an important instrument for the preservation and sustainability of tangible cultural heritage assets and can be considered to be of great importance for buildings such as historic houses. This study investigates the implications of applying machine learning methods in the preservation-oriented maintenance of selected local masonry houses in Menteşe, Muğla (Türkiye). In this study, the physical condition of traditional houses is analyzed through facade photographs to test whether they can provide preliminary architectural information to the users before preparing maintenance or repair planning. The tags used in this study were categorized into two groups: those in need of maintenance and those in need of simple repair. The machine learning model is trained at different parameter values, and then confusion matrix tables, validation and loss plots are obtained. By categorizing vernacular houses according to their physical condition, this study attempts to contribute to the growing body of knowledge and prioritize a new approach for sustainable design interventions. Thus, it is thought that AI-supported analyses will have the potential to provide information about the current state of traditional structures.