Defect Detection of Urban Drainage Pipeline Based on Improved YOLO-V8

This paper innovatively proposes the YOLO-V8++ algorithm, which significantly improves the accuracy and stability of defect detection by fusing RFB and SE attention mechanisms. The innovation of the algorithm is that the SE attention module and the new residual attention module are added to the backbone network to reduce the background interference. The RFB receptive field module was used to improve the accuracy of feature extraction and the efficiency of multi-scale processing. The upgraded wavelet de-noising algorithm is used to eliminate image noise. These improvements not only enhance the understanding of the application of attention and multi-scale features, but also open up the way for the development of high-performance defect detection technology and promote the intelligent process of drainage pipe maintenance.