Precision agriculture with YOLO-Leaf: advanced methods for detecting apple leaf diseases

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Accurate detection of apple leaf diseases is vital for crop health and yield.
• Current detection methods struggle with variable lighting, shadows, and scale complexities.

Purpose of the Study:

• To develop an advanced model for detecting apple leaf diseases with improved accuracy and generalization.
• To overcome the limitations of existing detection methods in challenging environmental conditions.

Main Methods:

• Proposed a novel model named YOLO-Leaf.
• Integrated Dynamic Snake Convolution (DSConv) for robust feature extraction.
• Employed BiFormer to enhance the attention mechanism.
• Introduced IF-CIoU for improved bounding box regression.

Main Results:

• YOLO-Leaf achieved high detection accuracy on FGVC7 and FGVC8 datasets.
• Demonstrated superior performance compared to existing models.
• Attained mAP50 scores of 93.88% (FGVC7) and 95.69% (FGVC8).

Conclusions:

• The YOLO-Leaf model significantly enhances apple leaf disease detection accuracy.
• The proposed methods (DSConv, BiFormer, IF-CIoU) contribute to improved detection performance.
• YOLO-Leaf shows strong potential for practical application in agricultural disease surveillance.