A Method for Automatically Locating Defects in CCTV Inspection Data of Sewer Pipes

Abstract

Identifying and locating sewer defects is crucial for minimizing the risk of sewer-related accidents. Currently, spatial localization of sewer defects in closed circuit television (CCTV) inspection data is primarily performed through manual visual inspections by professional technicians, which is time-consuming and costly. Consequently, there is an urgent need to develop automated solutions. This study proposes an innovative method for automatic defect location, which involves defect tracking based on rules, calculating the distance of the defect relative to the pipe endpoint, and converting this distance into physical world coordinates. Following experimental validation, the multiple object tracking accuracy (MOTA) metric of defect tracking within the proposed method ranged from -1.46 to 0.89 across ten test videos. The proposed method can be easily integrated into practical engineering applications, thereby alleviating the workload of professional technicians in obtaining the geographic locations of sewer defects in CCTV inspection data and reducing both time and labor costs associated with this process.

Keywords

• Sewer defects
• defect detection
• anomaly detection
• object tracking
• computer vision