An improved YOLOv5-based apple leaf disease detection method.

Area of Science:

• Agricultural Science
• Computer Vision
• Plant Pathology

Background:

• Accurate fruit tree leaf disease identification is vital for reducing pesticide use, boosting yields, and promoting ecological agriculture.
• Computer vision offers potential for plant disease detection, but existing models lack accuracy and diversity handling.
• Current limitations hinder the application of plant pest detection technologies in real-world agricultural settings.

Purpose of the Study:

• To develop an efficient and accurate detection model for apple leaf disease spots.
• To enhance the performance of the Yolov5 detection network for agricultural applications.
• To address the limitations of existing models in terms of disease diversity and detection accuracy.

Main Methods:

• An improved Yolov5 network, termed A-Net, was developed for apple leaf disease spot detection.
• The Wise-IoU loss function, incorporating attention and dynamic focusing mechanisms, was integrated into the Yolov5 model.
• The RepVGG module replaced the original convolutional module within the Yolov5 architecture.

Main Results:

• The A-Net model demonstrated effective suppression of error weights compared to the baseline.
• The improved model achieved a Mean Average Precision (mAP) of 92.7% at an IoU threshold of 0.5.
• A-Net outperformed several other object detection models in detecting apple leaf diseases.

Conclusions:

• The enhanced A-Net model significantly improves the accuracy and efficiency of apple leaf disease detection.
• The integration of Wise-IoU and RepVGG modules contributes to superior performance in identifying plant diseases.
• This advancement holds promise for practical applications in precision agriculture and sustainable farming practices.