Abstract: This study introduces a strawberry disease detection and segmentation system based on an improved YOLOv8 model. The system aims to address the limitations of traditional manual monitoring methods in strawberry disease detection by leveraging deep learning technology to enhance detection accuracy and efficiency. We employ YOLOv8 n-FasterNet, a lightweight and high-speed variant of YOLOv8 optimized for real-time applications. By replacing the backbone of YOLOv8 with FasterNet, the model achieves significantly faster speed while maintaining accuracy. Additionally, we apply the Lion optimizer to enhance training efficiency and model performance. Experimental results show that the system achieves remarkable performance in strawberry disease detection and segmentation on the publicly available Kaggle strawberry disease dataset, with an accuracy of 0.969, a mAP of ~0.932, and an FPS of 133, fulfilling real-time detection demands. Given its strong performance, the system has great potential for smart agriculture,agricultural research,agricultural insurance,agricultural product quality control,and agricultural education, offering promising prospects for practical implementation and broad adoption.