Research on rice disease and pest detection algorithm based on EBP-YOLOv11

Abstract

Field-based rice disease and pest images usually contain small lesions, cluttered backgrounds, and large scale differences, which make accurate detection difficult. To improve the detection ability of YOLOv11, this study constructs EBPYOLOv11 by modifying the attention module, feature fusion structure, and prediction branch. EMA is used to emphasize disease-related responses and reduce irrelevant background information. BiFPN is introduced to strengthen feature interaction among different levels, and an additional P2 head is used for fine-scale target prediction. Experimental results show that EBP-YOLOv11 improves Precision, Recall, mAP@0.5, and mAP@0.5:0.95 by 6.6%, 3.4%, 7.7%, and 8.0%, respectively, showing its value for rice field monitoring.