Development of an Intelligent Defect Detection System for Gummy Candy under Edge Computing

Abstract

Gummy candies are one of the products of the food industry. It has invested more resources in all aspects of the food production chain to improve production processes. The defective candies cause the unevenness of the product that will cause the appearance, taste and flavor poor. That will lead to economic losses for the company. Most traditional candy companies set up product inspection personnel to eliminate defective product. In this paper, an intelligent defect detection system for gummy candy industry under edge computing environment is proposed. It can replace manual visual inspection, even shorten the processing time to reduce production costs, thereby improving product quality, the efficiency of the production line, and the number of inspections. The system includes: (1) The intelligent defect detection system by deep learning algorithms. (2) The edge computing architecture with AIoT. The proposed system adopted the YOLO deep learning algorithm. The results show that the Precision is 93%, Recall is 87% and the F1 Score is 90. It has certain empirical reference significance for the intelligent defect detection system of candies products. By adopting deep learning algorithm in the detection system, it can reduce the inspection man-power needs and long-term data collection.