Tunnel crack detection and measurement method based on dual-depth learning model

• Main Authors: CHENG BAOXI, ZHANG JIANQING, LYU BO, WEI DONGXIAO, ZHAO XIANGNAN, LIU WEIJUN, WEI HANMING, HU LINA, XIN ZHITAO, PANG HONGYUAN, ZHANG GUOHUA, LIU TAO
• Format: Patent
• Language: Chinese; English
• Published: 16.03.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention belongs to the technical field of tunnel detection, and particularly relates to a tunnel crack detection and measurement method based on a dual-depth learning model, which comprises thefollowing steps: acquiring a tunnel image; creating a training set of target detection; creating a training set of semantic segmentation; training a target detection model; training a semantic segmentation model; detecting an input image to be detected by using the trained target detection model to judge whether a crack exists or not; inputting the image with the crack into a trained semantic segmentation model for image segmentation; predicting the type of the image, the coordinates of the crack in the image and the crack length information; and outputting and storing a model prediction result. Compared with a pure image processing technology, the invention is high in detection accuracy and high in detection speed; compared with a pure deep learning algorithm, the invention has the advantages of a deep learning a