Image Processing in identifying the Industrial Automation Impact using RestNet with Convolutional Neural Network

Abstract :

The integration of Image Processing (IP) and Deep Learning (DL) techniques within smart Internet of Things (IoT)-based industrial automation systems has significantly advanced manufacturing efficiency. In industrial manufacturing, the precise mechanical inspection of components such as gears and bearings is critical; however, human factors often compromise the stability, efficiency, and accuracy of conventional testing methods. To address these challenges, this study proposes a novel edge detection approach leveraging a Convolutional Neural Network (CNN) with ResNet-152 for multidirectional edge detection of mechanical parts, enhancing feature detection precision. The method improves productivity, predictive maintenance, quality control, and overall operational excellence. The proposed model was evaluated against various DL methods and achieved an edge detection accuracy of 92.53%, surpassing traditional approaches. These results demonstrate the potential of the ResNet-152-based CNN in delivering high-quality, reliable defect detection in industrial environments.