Construction period is a core item of information for the heritage conservation of traditional dwellings, supporting building archive compilation, historical character zoning, conservation grading, and restoration strategy formulation in traditional villages. However, accurately determining the construction period of traditional dwellings remains difficult. Existing methods rely heavily on expert judgment and field surveys and are limited by low efficiency, subjectivity, and weak scalability. This study focuses on surface-built traditional dwellings in the Guanzhong region. UAV image acquisition and field photography were used to collect dwelling images, while resident interviews were used to verify construction-period information. Architectural components with chronological relevance were then annotated. On this basis, a construction-period classification model was developed using a support vector machine (SVM). Model performance was evaluated through classifier comparison, RBF parameter optimization, and repeated dwelling-level test-set evaluation. The optimized SVM model achieved an average classification accuracy of 81.9%. Qing-dynasty and post-1980 dwellings showed the highest and nearly identical classification performance, with F1-scores of 87.2% and 87.0%, respectively. Republican-period dwellings also achieved relatively good performance, with an F1-score of 82.9%, whereas dwellings built between 1949 and 1979 showed the lowest performance, with an F1-score of 75.4%. These findings indicate that construction-period classification based on architectural component features and SVM can support preliminary screening of dwelling construction periods, providing auxiliary technical support for digital archiving, period-based classification, character control, conservation planning, and targeted restoration in traditional villages.