Enhanced YOLOv8-ECCI Algorithm for High-Precision Detection of Purple Spot Disease in Soybeans

Area of Science:

• Agricultural Science
• Plant Pathology
• Computer Vision

Background:

• Seed-level disease detection in soybeans faces challenges like small samples and occlusions.
• Existing methods struggle with the complexities of identifying diseases directly on seeds.

Purpose of the Study:

• To introduce YOLOv8-ECCI, an enhanced algorithm for high-precision soybean purple spot disease identification on seeds.
• To address limitations in current seed-level disease detection techniques.

Main Methods:

• Utilized YOLOv8 architecture, enhancing it with the ECCI module for improved performance.
• Conducted experiments comparing YOLOv8-ECCI against the baseline YOLOv8n model.
• Performed cross-dataset validation using the African Wildlife dataset.

Main Results:

• YOLOv8-ECCI achieved significant improvements: +5.7% precision, +6.5% recall, +8.0% mAP@0.5, and +7.1% mAP@0.5:0.95 over YOLOv8n.
• Demonstrated superior generalization, outperforming conventional methods by +6.0% precision and +2.9% mAP@0.5 on a separate dataset.
• Effectively overcame challenges in seed-level pathology detection.

Conclusions:

• YOLOv8-ECCI provides a robust and accurate solution for soybean purple spot disease detection at the seed level.
• The enhanced algorithm shows strong potential for practical in-field agricultural disease detection and quality control.
• Validated generalization capability confirms the model's reliability across different datasets.