Comparative Analysis of YOLOv5–YOLOv11 Models for Automated Wheat Quality Assessment with Detection of Sunn Pest Damage and Foreign Materials

In the agricultural sector, product quality and food safety are of the utmost importance, particularly for staple food products such as wheat. Wheat quality control is approached as a process that involves identifying weevil damage and foreign materials. In traditional methods, these detection processes are mostly carried out using techniques that rely on manual methods, which can lead to time loss and human errors. This study proposes to identify sunn pest damage and foreign materials by replacing manual processing steps with an automated system using an AI-based integrated image processing system. YOLO-based deep learning models were used in the proposed system for object detection. In this context, the YOLOv5, YOLOv7, YOLOv8, and YOLOv11 variants were compared. A unique multi-class dataset consisting of wheat photographs obtained under various conditions was created for this study. In addition to the images obtained from this dataset, data augmentation methods were employed to enhance the model’s generalization capability. Experimental results indicate that the YOLOv11-L model demonstrated the best performance with accuracy of 97.1%, precision of 95.8%, recall of 94.6%, and an F1-score of 95.2%. Furthermore, it demonstrated superior performance compared to other models, achieving a 96.8% [email protected] and a 75.4% [email protected]:0.95. The results obtained demonstrate that the designed system is feasible for real-time implementation and that AI-powered object detection techniques provide an effective solution for agricultural quality control processes.