An Artificial IoT-Enabled Smart Production Line for 360° Visual Defect Detection and Classification of Cherry Tomatoes
Yu‐Huei Cheng, Che-Nan Kuo, Yu‐Da Lin
Abstract
The aging of the world’s population is causing older farmers familiar with agricultural practices to retire, resulting in a growing shortage of related labor. The ever-increasing labor shortage in traditional agriculture has reduced the manual sorting of agricultural products during processing and packaging. Delicate products such as cherry tomatoes pose a severe problem. This study proposes an artificial intelligence of things (AIoT) intelligent production line that leverages deep learning technology, the internet of things, and data analytics for fully automated and efficient 360-degree visual defect detection and classification of cherry tomatoes. By integrating image recognition and AIoT technology, the system can prevent the occurrence of defects during packaging, ensure consistent product quality during grading, and improve the reputation of agricultural producers. We achieved an outstanding mean average precision of 97.4% in the AIoT intelligent production line actual application in defect detection. The results showed that AIoT and intelligent screening plate mechanisms were advantageous for the detection of cherry tomatoes and help promote the development of delicate agricultural products. This study contributed to the development of intelligent agriculture by reducing labor and time costs, quickly detecting defects, effectively controlling sorting quality, and improving crop management through data analysis.