Review on applications of computer vision techniques for pipeline inspection

Pipeline inspection is critical for ensuring the safety, integrity, and efficiency of energy and water transportation systems. Anomalies such as leaks, corrosion, and structural deformations can result in severe issues for the environment, economy, and public safety. Traditional inspection methods, such as manual visual inspection, magnetic flux leakage, and ultrasonic testing, face challenges such as high cost, labor intensity, and limited automation. Computer vision has emerged as a promising solution to overcome these limitations by enabling automated, efficient, and cost-effective anomaly detection in pipelines. This paper systematically reviews the application of computer vision techniques for pipeline inspection, including data collection, image processing, deep learning-based analysis, digital twin-based management. Key influencing factors, challenges, and opportunities are discussed. The findings reveal that, with appropriate preprocessing, deep learning models achieved high accuracy in detecting cracks, corrosion, deformation, and dents, but their generalization performance was dependent on dataset quality, quantity, and preprocessing techniques. This review highlights the potential and barriers of computer vision to revolutionize pipeline inspection, providing a structured foundation for future research and applications.