Development of Cabbage Classification System by Machine Learning

Abstract

No cabbage-sorting machine exists that distinguishes fresh cabbages with 2–3 outer leaves and processes cabbages without outer leaves. We developed a cabbage-sorting system based on machine learning using a convolutional neural network (CNN) approach to support people with disabilities. We also fabricated a cabbage-sorting device using a web camera, load cell, and microcomputer. We created a “fine-tuned” CNN by modifying the VGG16 architecture. For the fine-tuned CNN, the discrimination accuracy and all other evaluation metrics were high. The time required to complete the operation by the cabbage-sorting device was 5.8 seconds for the fine-tuned CNN. These evaluations indicate that the cabbage discrimination program should incorporate the fine-tuned CNN. This cabbage discrimination system can be used to assist beginners and people with disabilities, which could help reduce the labor shortage.