Concrete structure apparent defect identification method and device based on inspection robot

Abstract

The application discloses a concrete structure apparent defect identification method and device based on a patrol robot, and comprises the following steps: collecting an apparent image of a target concrete structure in real time by using a patrol robot in response to an apparent defect identification instruction of the target concrete structure; obtaining a defect identification model based on YOLOv5; inputting the processed apparent image into the defect identification model; extracting global image features and context image features of the apparent image by each Transform module in a backbone network; merging and processing the global image features and the context image features by an SPPF module; enhancing key features of the merged image features by a channel attention layer and a spatial attention layer of each CBAM module in a neck network; performing feature fusion on the enhanced merged image features; obtaining fusion image features corresponding to each CBAM module; performing defect identification on each fusion image feature by a head network; and obtaining a defect identification result output by each prediction head module.

Images