Research on the Application of Computer Vision in the Field of Intelligent Manufacturing

• A novel lightweight CNN with attention mechanism achieves 98.5% average precision for defect detection, surpassing state-of-the-art methods. • The proposed method operates at 30 FPS, enabling real-time inspection in production lines without compromising accuracy. • Deployment in a pilot line reduced inspection time by 40% and improved product quality consistency, demonstrating practical industrial applicability. • The study provides a framework for integrating computer vision into intelligent manufacturing, offering a scalable solution for quality control.