Efficient Tomato Disease Detection Using MaxMin-Diffusion Mechanism and Lightweight Techniques.

Area of Science:

• Agricultural Science
• Computer Science
• Artificial Intelligence

Background:

• Automated disease detection is crucial for agricultural modernization.
• Traditional models struggle with complex diseases and time-series data.
• Existing methods face accuracy and robustness challenges.

Purpose of the Study:

• To develop an accurate and robust disease detection model for agriculture.
• To introduce the novel maxmin-diffusion mechanism.
• To enable efficient disease detection on mobile devices.

Main Methods:

• Proposed a disease detection model utilizing the maxmin-diffusion mechanism.
• Dynamically adjusted attention weights to focus on disease regions.
• Performed lightweight optimization for mobile deployment.

Main Results:

• Achieved high performance in bacterial spot disease detection (Precision: 0.98, Recall: 0.95, Accuracy: 0.96, mIoU: 0.96).
• Demonstrated superior fine-grained feature extraction and time-series processing compared to traditional mechanisms.
• Showcased enhanced accuracy and robustness in dynamic disease recognition.

Conclusions:

• The maxmin-diffusion mechanism significantly improves disease segmentation accuracy and robustness.
• The lightweight model offers high-precision detection suitable for resource-constrained mobile devices.
• Provides strong technical support for smart agriculture applications.