Pavement Distress Identification Based on Computer Vision and Controller Area Network (CAN) Sensor Models

Recent technological developments have attracted the use of machine learning technologies and sensors in various pavement maintenance rehabilitation studies. To avoid excessive road damages, which cause high costs, reduced mobility, vehicle safety concerns, periodic roads is necessary. As part works, conditions should be monitored continuously. This monitoring possible using modern distress detection methods that are simple to use, comparatively cheap, less labor-intensive, faster, safer, able provide data on a real-time basis. paper proposed developed two models: computer vision sensor-based. The model was You Only Look Once (YOLOv5) algorithm for detecting classifying distresses into nine classes. sensor-based combined eight Controller Area Network (CAN) bus available most new vehicles predict distress. research employed an extreme gradient boosting (XGBoost) train model. results showed achieved 98.42% 97.99% area under curve (AUC) metrics training validation datasets, respectively. attained accuracy 81.28% F1-score 76.40%, agree with past indicated both models proved highly efficient predicting can used complement each other. Overall, tools cheap practical condition compared traditional manual instruments.