Boxiang Zhang, Baijian Yang, Xiaoming Wang, Corey Vian
PatchFlow improves defect detection in die casting using local patch features and a flow-based model, reducing error rates significantly on multiple datasets.
Die casting is a manufacturing process used to create precise metal parts, but it often suffers from surface defects that are hard to detect. This study introduces a new computer vision technique, PatchFlow, which combines detailed image patches with advanced machine learning to better identify these defects. By using a specialized model that adapts to industrial images, this method significantly improves the accuracy and efficiency of defect detection compared to existing methods. The results show a substantial reduction in error rates on several benchmark datasets, demonstrating the potential impact of this technology on quality control in manufacturing.
