Deep learning for image classification of submersible pump impeller

Abstract

This study presented a deep learning-based model in the submersible pump impellers quality inspection process. The proposed method aimed to relieve worker workload, automate the system, as well as increase the accuracy in defect detection and classification. The proposed approach aims to be implemented on systems with low investment cost and limited resources, i.e., small single-board computers, enabling flexible deployment in industrial environments. The model consisted of three convolutional neural network (CNN) models, i.e., visual geometry group 16 (VGG16), ResNet50, and a custom model. The outputs of three networks were either synthesized later through an ensemble stage or used separately. A graphical user interface (GUI) was also developed for real-time inspection and user-friendly interaction. The approach achieved up to 99.8% accuracy in identifying defects, including surface scratches, corrosion, and geometric irregularities. The proposed method improved the quality assurance process by reducing manual inspection efforts. Future research could explore advanced techniques like anomaly detection to further enhance system performance and versatility.