Research on Foreign Matter Detection in Press Molds Using AI Vision System

Abstract

This study presents an artificial intelligence (AI) vision system for detecting foreign matter in press molds during bevel gear forming. To enhance the productivity of bevel gear manufacturing, the lower mold is fixed to the base of the press, while the upper mold is moved vertically to print; any foreign material in the lower mold can cause damage. A small camera mounted at the center of the upper mold captures consistent images of the lower mold, which are labeled as normal or abnormal (containing foreign matter) and used to train a YOLOv3-based model. The trained system automatically identifies foreign objects inside the lower mold in real time. The developed AI program demonstrates accurate detection of foreign substances and potential for practical deployment in press operations.