Research on Drainage Pipe Defect Detection Method Based on Improved YOLOv8

With the continuous advancement of urbanization, the underground drainage pipeline network is becoming increasingly large and dense. Regular inspection and maintenance of these pipelines are essential to effectively reduce the socioeconomic losses caused by pipeline defects. At present, closed-circuit television (CCTV) detection has become the mainstream underground pipeline detection method. This method uses underground pipeline robots to shoot in real time, and then professional inspectors make judgments. However, manual inspection of a large number of pipeline videos has problems such as fatigue, strong subjectivity and low efficiency. At the same time, due to the complexity of the pipeline environment, a large number of false detections may occur. In order to solve the above problems, this study uses deep learning methods to study underground drainage pipeline defects. An improved underground drainage pipeline defect detection algorithm based on YOLOv8 is proposed. Experimental results show that the improved algorithm can effectively identify eight defects: joint dislocation, cracks, attached sediments, settled sediments, tree roots, obstacles, branch pipes and intrusion sealing materials. The mAP reaches 95.1%, which is 3.4% higher than the original YOLOv8 algorithm. The quality improvement effect is obvious, and it has significant theoretical significance and application prospects.