Sewer pipeline condition assessment and defect detection using computer vision

Abstract

The structural integrity and operability of sewer pipeline systems are crucial for society's health, urban environment, and economic stability. Advancements in computer vision (CV) have revolutionized sewer defect inspection, offering unprecedented accuracy and efficiency in identifying and assessing pipeline failures. While prior reviews exist, they often lack systematic comparisons of models, detailed dataset analyses, or comprehensive severity assessment frameworks. This paper presents a comprehensive review of CV implementations for sewer defect detection, location, and characterization. It thoroughly evaluates main inspection techniques, diverse datasets, and key performance metrics. State-of-the-art CV models and their applications are critically reviewed, alongside defect severity assessments and their link to maintenance strategies. Key challenges and limitations are identified, leading to recommendations for enhancing inspection efficiency and accuracy. The paper consolidates findings on methodological trends, data analysis advancements, algorithm performance variations, and improved severity assessment approaches.