Highway bridges are an important part of modern traffic construction, and their maintenance and preservation are increasingly concerned by the industry. As an important component of bridge, bearing is directly related to the stress state of the overall structure and the overall traffic safety. However, at present, the detection of bridge bearing is usually carried out manually, which not only consumes enormous manpower and material resources, but also affects the normal traffic operation, and the safety of the inspectors is also unavailable. Through the rational use of advanced science and technology, and under the guidance of in-depth learning and image processing technology, this research carries out software development to detect and identify bridge bearing diseases in an efficient and reasonable way. The application and experiment in a specific engineering case show that the detection and recognition method based on convolution neural network has strong intuition, and can realize the prediction of multiple images under the same folder, and describe the disease situation in detail.

CICTP