Segmentation algorithm and quantitative calculation of various defects in bridge images based on perceptual analysis

Abstract

The present invention discloses a segmentation algorithm and quantitative calculation of multiple defects of bridge images based on perceptual analysis. The present invention relates to the technical field of bridge defect detection, including the source and construction of a bridge data set, data set division, construction of a multi-task semantic segmentation algorithm based on perceptual analysis and a quantitative calculation method of bridge defect parameters, and a step of calculating bridge defect attribute values. The present invention has the advantages that: by designing a dynamic sparse convolution operator and introducing a multi-group mechanism at the encoder stage, the memory occupancy of the model is reduced, and the ability of the model to capture and distinguish multiple defects is enhanced. At the decoder stage, the patch feature expansion layer and the jump connection channel-spatial attention module are used to reduce the complexity of the model, while improving the precise positioning and segmentation performance of different types of defect areas. In addition, by fusing multi-scale features and unifying the output, the comprehensive perception and analysis of bridge defects are realized, and the accuracy and robustness of the overall detection are improved.

Images