YOLOv8-DBW: An Improved YOLOv8-Based Algorithm for Maize Leaf Diseases and Pests Detection
Xiang Gan1, Shukun Cao1, Jin Wang1
1College of Mechanical Engineering, University of Jinan, Jinan 250022, China.
Sensors (Basel, Switzerland)
|August 14, 2025
Summary
An improved YOLOv8 algorithm enhances maize pest and disease detection accuracy while reducing model complexity. This solution offers efficient, accurate, and deployable capabilities for mobile devices.
Area of Science:
- Agricultural Science
- Computer Vision
- Machine Learning
Background:
- Maize pests and diseases pose significant threats to crop yield.
- Existing detection methods suffer from low accuracy and computational complexity.
- Deployment on mobile or embedded devices is challenging for current models.
Purpose of the Study:
- To develop an improved YOLOv8 algorithm for accurate and efficient maize pest and disease detection.
- To enhance model performance and reduce computational load for practical deployment.
- To provide a superior solution for real-time agricultural monitoring.
Main Methods:
- Modified the YOLOv8n backbone with DSConv modules to reduce parameters and computational load.
- Integrated BiFPN for enhanced feature fusion across different scales.
- Utilized Wise-IoU loss function to improve training convergence and regression accuracy.
Main Results:
- Achieved 1.4% increase in precision, 1.1% in recall, and 1.5% in mAP0.5 compared to YOLOv8n.
- Reduced model parameters by 6.6% and computational costs by 7.3%.
- Demonstrated improved detection accuracy and efficiency.
Conclusions:
- The improved YOLOv8 algorithm offers a significant advancement in maize pest and disease detection.
- The model is accurate, efficient, and suitable for deployment on resource-constrained devices.
- This provides a practical solution for agricultural applications.
