Intelligent Identification for Imbalanced Oil–Gas Pipeline Defects Based on Triple Attention Mechanism and Dual-Modal Deep Convolution Network

Abstract

AbstractThe class imbalance of crack defects in oil and gas pipelines presents a significant challenge to the diagnosis of pipeline integrity, and the scarcity of high-risk cracks can easily lead to the recognition models tending to predict them as the ...