Research on an Algorithm for Detecting Defects on the Inner Wall of PE Pipelines

Low recognition accuracy for existing pipeline defect detection; Problems such as slow detection, To improve the efficiency and accuracy of testing, Proposing a new pipeline defect detection algorithm YOLO-gas. first, Designing lightweight modules in the feature extraction layer C2f-Faster, Significantly reduce the number of parameters and computation of the model, and increase the speed of model detection,Suppressing Background Information Interference and Enhancing Pipeline Defect Features by Embedding Convolutional Attention Module CBAM in Neck Networks, Modify the loss function to Focal -EIoU Finally, the YOLO-gas improved model was compared to the YOLOv8n baseline model,Verify the superiority of the improved algorithm.The number of parameters in the improved model is drastically reduced,mAP50 improved by 2 percentage points compared to the YOLOv8 model, The improved model has high classification efficiency and detection accuracy of 94.8%; Deep learning-based target recognition technical support is provided for the realization of real-time detection of pipeline defects.