Intelligent detection method for apparent diseases of sound barrier of viaduct based on computer vision

The invention relates to the technical field of bridge sound barrier detection, in particular to an elevated bridge sound barrier apparent disease intelligent detection method based on computer vision, and the method specifically comprises the following steps: carrying out the preprocessing of an original data set; the method comprises the following steps: carrying out structure identification and labeling on a sound barrier shot from a long distance, classifying the appearance types of the sound barrier, carrying out classification and labeling on parts of the sound barrier, building a classification deep neural network model II of the types of the parts of the sound barrier, and training and identifying the classification deep neural network model II; disease pictures in daily detection of the viaduct sound barrier are classified and labeled, and a sound barrier apparent disease classification deep neural network model III is built, trained and recognized; and intelligent evaluation. Compared with the prior