Mobile-CBSD: A Lightweight Apple Leaf Disease Detection Model Based on Improved YOLOv11

Apple leaf diseases can significantly affect the yield and quality of apple crops. However, conventional manual detection methods are inefficient and highly susceptible to subjective judgment, rendering them inadequate for large-scale agricultural production. To address these limitations, this study proposes Mobile-CBSD, a lightweight deep learning model for apple leaf disease detection based on an enhanced version of YOLOv11. First, the original backbone network of YOLOv11 is replaced with MV4_CBAM, a lightweight architecture that improves feature representation capability while reducing model size. Second, the SE attention mechanism is redesigned and integrated into the network to strengthen multi-scale feature fusion. Furthermore, the traditional CIoU loss function is replaced with SIoU to accelerate convergence and enhance localization precision. Experimental results demonstrate that, while maintaining model compactness, Mobile-CBSD achieves an mAP@0.5 of 90.89%, representing a 2.16% improvement over the baseline, along with a 1.02% increase in overall precision. The model size is reduced from 5.4 MB to 4.8 MB. These findings indicate that Mobile-CBSD achieves an effective balance among accuracy, inference speed, and deployability, offering a practical and scalable solution for the efficient monitoring of apple leaf diseases.