The study on surface defect detection of stamped parts based on improved deep learning

Stamped parts are widely used in industries such as automotive and aerospace, where surface defects like scratches and cracks can affect appearance and function. Traditional manual inspection methods are inefficient and prone to errors. This paper proposes an improved deep learning-based approach for detecting surface defects in stamped parts, combining image preprocessing, feature enhancement, and deep neural networks. The YOLOv8 model, enhanced with the Feature Fusion Attention Network (FFA-Net) and Gold-YOLO, was tested on an augmented image dataset. Experimental results show that the improved model achieves higher precision, recall, and detection accuracy compared to baseline methods. The model demonstrates robustness under various environmental challenges, making it suitable for industrial defect detection applications.