Research of CCTV Drainage Pipeline Defect Recognition Method Based on YOLOv5s
Urban drainage pipelines are often blocked,aged,corroded and damaged,which seriously affects the healthy development of the city and the normal life of citizens.By analyzing limitations of the current drainage pipeline defect recognition method,a drainage pipeline defect recognition model was conducted based on YOLOv5s to learn,train and verifie 10 common drainage pipeline defects.The intelligent identification was finished by four steps of data collection,data processing,model training and target detection.The experimental results show that the mean average precision of the drainage pipeline defect recognition model reaches 85.42％(the average precision of the four defect recognitions of dislocation,foreign body penetration,leakage and rupture reaches 93.6％,91.7％,91.7％and 91.3％,respectively),which verifies the effectiveness of the model.In addition,compared with the Faster R-CNN model,the YOLOv5s model has higher mean average precision and the obvious advantages of smaller model memory and faster detection speed,which significantly improve the engineering applicability of the model.
国家科技期刊平台
NSTL
万方数据
