Research on the identification of the ancient building roof based on the object detection algorithm

Abstract

In recent years, the digital protection of ancient buildings based on modern technology has received more and more attention. The digital protection of ancient buildings and the establishment of the database can enhance the ability of ancient buildings to resist risks. However, the degree of automation in the process of database establishment is not high, which requires expert identification and statistics, which is time and force-consuming. Therefore, the construction of the automatic identification model of the roof of ancient buildings is of great significance for the maintenance and management, dating and three-dimensional reconstruction of ancient buildings. This paper develops a method for establishing the database of ancient building roof components, and develops a deep learning model framework based on the target detection algorithm to identify the specific types of ancient building roof. The automatic roof identification method proposed in this paper performs well in the hip roof, and the results can meet the needs of the roof digital protection and maintenance management.