Target detection method for dam defect image

Abstract

The invention discloses a target detection method for a dam defect image, and the method comprises the steps: enabling deformable convolution to be used for a VGG16 network, enlarging the convolutionfeeling range, capturing the deformation features of a dam defect through learning convolution offset, and obtaining a feature map of the defect image; when multi-scale feature map detection is carried out, modifying the dimension proportion of a priori frame in an anchor point mechanism, and improving the detection precision of strip-shaped defect features and the generalization ability of the model; adopting an improved non-maximum suppression algorithm to screen and subtract redundant negative samples, and ensuring the diversity of training samples as much as possible on the premise of balancing the ratio of positive samples to negative samples. According to the invention, the dam defect image is effectively detected, the detection of multi-deformation defect features can be realized, and the generalization ability of strip-shaped defect detection is further improved. The method has high detection precision and good convergence performance in target detection of dam defect images.

Description

technical field [0001] The invention belongs to the field of dam defect image target detection, in particular to a target detection method for dam defect images. Background technique [0002] In the field of construction engineering, the inspection items or inspection points that do not meet the specified requirements in the construction quality of the project are defined as defects. With the long-term operation of the hydropower dam, the aging of materials, environmental impact and other reasons lead to the formation of defects to varying degrees. When the defect is relatively light, corresponding measures can be taken to deal with the defect in a timely manner to meet the load-bearing requirements of the structure. Once the defect is not processed and remedied in time, it will pose a major threat to the safe operation of the dam. Therefore, using automatic inspection equipment to detect and troubleshoot defects in time can effectively maintain the structural safety of the ...

AI Technical Summary

Problems solved by technology

[0004] Purpose of the invention: In order to overcome the problems existing in the prior art, the present invention provides a target detection method for dam defect images. The target detection algorithm using deformable convolution to extract features can not only achieve efficient detection, but also accurately identify and Detects dam imperfections with variable geometries

Images