Mural damage detection and the repair have always been an important research topic in the field of ancient cultural research.Traditional manual restoration methods rely on the experience of researchers and professional painters and superior paint-ing techniques and knowledge of historical materials to repair the missing areas,and this method is not only time-consuming,but al-so has high requirements for the workers.Earlier fresco restoration works directly on the fresco itself,increasing the possibly unre-coverable and damage.Deep learning can effectively extract the implicit features of images,and meanwhile,the application in image recognition of cultural relics is rapidly developing.Taking Jiaoshan temple grottoes in Datong,Shanxi province as an example,for the solution to the problems of breaking and cracking of cave murals,a method of mural damage detection based on YOLOv5 algo-rithm is proposed in view of the typical damage diseases such as shedding cracking and peeling.The experimental results show that the proposed mural damage detection method can accurately locate the damage detection part in the mural and accurately and effi-ciently complete the damage classification.
万方数据