Application of Computer Vision and Deep Learning for Automated Defect Detection in High-Speed Manufacturing Lines

Abstract

In high-speed manufacturing environments, maintaining consistent product quality while minimizing inspection time is a persistent challenge. Conventional manual inspection methods are labor- intensive, error-prone, and unsuitable for modern automated production lines operating at high throughput. This research presents a comprehensive study on the application of computer vision and deep learning techniques for automated defect detection in high-speed manufacturing systems. A deep convolutional neural network-based framework was developed to identify surface and structural defects in real time using high-resolution image data. Multiple defect categories, including cracks, scratches, misalignments, and material inconsistencies, were considered. The proposed system was evaluated using industrial image datasets under varying lighting and speed conditions. Performance was assessed in terms of detection accuracy, precision, recall, inference speed, and robustness to noise. Experimental results demonstrate that deep learning-based inspection significantly outperforms traditional machine vision approaches, achieving high detection accuracy while meeting real-time operational constraints. The study highlights the potential of intelligent vision systems to enhance quality assurance, reduce production downtime, and support the transition toward smart manufacturing.