AI-Based CCTV Monitoring of Road Drainage Inlet Blockage: Effects of Observation Height and Angle on Detection Performance

Blockage of road drainage inlets by debris is a major cause of urban flooding. Manual inspection of urban drainage facilities is time-consuming and costly, highlighting the need for automated inspection technologies. In this study, a closed-circuit television (CCTV)- and AI-based technology was developed to automatically monitor the drainage inlet blockage status by varying camera height and observation angle to examine their influence on detection performance. The YOLOv11l and RT-DETR-L models classified the drainage inlet conditions in the testbed into three categories: fully open (clean), partially blocked (trash), and fully blocked (clogged). Both models achieved F1 scores of 0.994. Under clogged conditions, RT-DETR-L achieved a higher confidence score, whereas YOLOv11l achieved a slightly higher intersection over union (IoU). A three-way analysis of variance (ANOVA) revealed that the inlet condition was the most influential factor affecting both the confidence score and IoU, with a significant height–angle interaction under the clogged condition: performance improved with height at a front-facing angle but declined at oblique angles. These findings indicate that observation height and angle should be considered jointly when evaluating CCTV configurations for drainage inlet monitoring. Further validation using independent real-world CCTV data is required before extending these findings to broader field applications.