<![CDATA[Manual classification of citrus fruit varieties is time-consuming and error-prone, making an automated digital image-based system essential to support farmers and traders in distribution, pricing, and quality standardization. Previous research has shown that CNN and YOLO are effective in fruit classification, but the models still have difficulty distinguishing classes with visual similarities and under different lighting conditions. This research focuses on analyzing the performance of a YOLOv5-CNN-based citrus fruit classification model, with YOLOv5 used to detect objects and generate bounding boxes, and then the detected areas are classified by EfficientNet-B0. Testing on an independent dataset of 960 images showed an accuracy of 96%. These results indicate that the developed model is effective for automatic citrus fruit classification, providing a basis for the development of a digital image-based fruit sorting and identification system.]]>
Copyrights © 2026
