Assessing corn quality classification with deep-learning module

Abstract

This study presents a new classification framework for corn quality assessment using the you only look once version 8 (YOLOv8) deep learning architecture. Traditional corn-sorting methods are labor-intensive, time-consuming, and prone to human error. To address these challenges, this study proposes the corn quality classification (CQC) algorithm, an artificial intelligence (AI)-driven approach for automatically recognizing and categorizing pest- and disease-related defects. The proposed model uses YOLOv8, which integrates a convolutional neural network (CNN) backbone for feature extraction and object detection. The system was trained on labelled corn images to classify them into three categories: healthy, water rot, and bug. The model achieved an overall classification accuracy of 92.7%, with class-specific accuracies of 88.9% for healthy, 70% for water rot, and 100% for bug. Key performance metrics included a recall of 93%, an F1-score of 62%, a mean average precision (mAP) of 92.7%, and an intersection over union (IoU) of 52.3%. Additionally, the F1-confidence and recall-confidence curves provided further insights into the model’s performance. The proposed classification framework and CQC algorithm demonstrate the feasibility of AI-powered agricultural monitoring systems, offering a reliable and efficient approach to corn quality assessment. By enabling consistent and accurate classification, the proposed system has the potential to reduce labor requirements and support precision agriculture practices for improved crop quality management.