2026 Volume 7 Issue 2 Pages 160-170
Sewerage facilities are simultaneously facing infrastructure aging and a reduction in maintenance personnel, making the efficiency of routine monitoring an urgent challenge under limited financial resources. Although deep-learning-based image analysis is considered a promising solution, its practical deployment is hindered by variations in imaging conditions and by the burdens of training-data preparation and model updating. This paper proposes a practical image-analysis method for sewerage facility management that does not depend entirely on deep learning. By combining low-level image features, including color, brightness, and contours, with temporal changes, the proposed method reduces the burden of model development and updating while enabling practical sewerage facility management. In addition, focusing on sewer pipelines, this study demonstrates the potential effectiveness of integrated monitoring, in which multiple monitoring indicators— specifically water-quality anomalies and water level—are acquired from a single camera and monitored simultaneously.
