Strawberry Disease Detection Algorithm based on YOLO11-Strawberry

Abstract

Strawberry is a high-value fruit that is highly susceptible to various diseases due to its delicate skin and environmental sensitivity. Early and accurate detection of strawberry diseases is crucial for minimizing economic losses and ensuring food safety. However, traditional disease detection methods mainly rely on manual visual inspection, which is time-consuming, labor-intensive, and prone to subjective errors. In recent years, deep-learning-based object detection techniques have shown great potential in the agricultural field. This study proposes an improved YOLO11s model, YOLO11-Strawberry, for real-time strawberry disease detection. First, we introduce the VoV-Faster Block, which replaces high-computation modules in the original model to improve computational efficiency while maintaining accuracy. Then, we propose the Multi-Scale Feature Channel Fusion Attention (MSFCA) module, which enhances the detection of small disease regions while reducing background interference. The experimental results demonstrate that YOLO11-Strawberry achieves outstanding performance in strawberry disease detection, with a precision of 93.2%, recall of 92.5%, and mAP50 of 95.3%. Additionally, the proposed improvements lead to a 10.6% reduction in model parameters and a 17.4% reduction in computational cost, significantly enhancing the model's lightweight characteristics and making it suitable for real-time deployment on edge devices. The proposed YOLO11-Strawberry model provides an efficient and lightweight solution for intelligent strawberry disease detection, contributing to the advancement of agricultural automation and precision farming.