Investigation, Evaluation, and Dynamic Monitoring of Traditional Chinese Village Buildings Based on Unmanned Aerial Vehicle Images and Deep Learning Methods

The investigation, evaluation, and dynamic monitoring of traditional village buildings are
crucial for the protection and inheritance of their architectural styles. This study takes traditional
villages in Shandong Province, China, as an example, employing UAV images and deep learning
technology. Utilizing the YOLOv8 instance segmentation model, it introduces three key features re-
flecting the condition of traditional village buildings: roof status, roof form, and courtyard vegetation
coverage. By extracting feature data on the condition of traditional village buildings and constructing
a transition matrix for building condition changes, combined with corresponding manual judgment
assistance, the study classifies, counts, and visually outputs the conditions and changes of buildings.
This approach enables the investigation, evaluation, and dynamic monitoring of traditional village
buildings. The results show that deep learning technology significantly enhances the efficiency and
accuracy of traditional village architectural investigation and evaluations, and it performs well in
dynamic monitoring of building condition changes. ... mehrThe “UAV image + deep learning” technical
system, with its simplicity, accuracy, efficiency, and low cost, can provide further data and technical
support for the planning, protection supervision, and development strategy formulation of traditional
Chinese villages.