Hybrid Defect Detection Framework for Ceramic Tiles
The traditional method for quality control is manual visual inspection, which is inefficient and subjective. Therefore, we designed an automated quality control system to identify surface defects on ceramic tiles by leveraging deep learning and computer vision. Built upon the YOLO architecture, we optimized the model by integrating an attention mechanism and an SPD convolution module, significantly enhancing detection precision while keeping the model lightweight for efficient deployment.