Pipeline Leak Detection Using Infrared Cameras: A Convolutional Neural Network Approach

Resumo

Pipeline leak detection is crucial for maintaining pipeline safety, particularly in complex environments. This study proposes a novel approach that integrates infrared cameras with a convolutional neural network model, specifically VGG16, utilizing infrared cameras that do not inherently measure temperature. Our results indicate that this approach is highly effective, with the model achieving 100% accuracy on both the training and validation datasets, and a near-zero validation loss in a laboratory environment. The confusion matrix confirmed that there were no misclassifications, and the Receiver Operating Characteristic (ROC) curve demonstrated an Area Under the Curve (AUC) of 1.0. These findings underscore the model's potential for real-world pipeline monitoring applications.