Deep Learning For Image Classification Of Manufacturing Product

Abstract

Product inspection is essential for ensuring the quality and dependability of manufactured goods. Traditional manual examination procedures are timeconsuming, subjective, and error prone. As manufacturing complexity and production volumes increase, there is a rising need for automated inspection systems with accurate defect detection and classification. This research presents a deep learning-based quality inspection approach for submersible pump impellers. Three convolutional neural network (CNN) architectures, VGG16, ResNet50, and a custom model, are employed. A graphical user interface (GUI) is developed for real-time inspection. The approach achieves up to 99.8% accuracy in identifying defects, including surface scratches, corrosion, and geometric irregularities. It improves the quality assurance process by reducing manual inspection efforts. The GUI improves usability and decision-making. This study contributes to industrial quality control by introducing a novel deep learning application. Future research could explore advanced techniques like anomaly detection to further enhance system performance and versatility. In future research, improving the accuracy and applicability of quality inspection models can be achieved by incorporating additional datasets and pre-trained models, along with fine-tuning techniques. Moreover, developing hardware solutions for real-time deployment in industrial processes will enable seamless integration.