Towards Dry Fruit and Nut Sorting Automation: A Comparative Analysis of YOLO Models for Object Detection

Dry fruit and nut detection is an important part of the food processing industry for product sorting, quality control, and allergen detection.This study presents the results of various YOLO models to detect and classify common dry fruits and nuts.By improving the accuracy and reliability of automated systems in the food processing industry, we address the challenges posed by visually similar objects.Accurate detection of various nuts and dry fruits ensures product quality and enables efficient sorting and packaging processes.At the same time, In food manufacturing facilities it can help to prevent allergic reactions and safeguard consumer health.In our research, we compared the performance of different YOLO architectures-v5, v8, v9, and v10 on a dataset of 3,776 annotated images, utilizing mixed dry fruits and nuts.Therefore, This identifies the best performance model among these four models under an industrial setting that requires high accuracy for detecting an object.The capability of each model in detecting objects was analyzed based on accuracy, precision, recall, and mAP.The results of the experiment identified that YOLOv9 has better precision and recall of 99.5% with a mean average precision of 99.2% per class.This shows that the system remains a preferable option for application in cases where high demand is placed on accuracy in detection.Other Yolo models also performed well, but a bit less than YOLOv9.Highlighting the comparative analysis has shown not only the progressive trend of the YOLO model architecture but also assisted in seeking out optimizations of models against complex operational requirements of the food processing industry.The study can provide a future benchmark for developing improved technologies of automatic object detection with enhanced accuracy and operational efficiencies.Creating a dataset specially annotated for dry fruits and nuts detection highlights the study's importance for real-world applicability and efficiency.