GSEYOLOX-s: An Improved Lightweight Network for Identifying the Severity of Wheat Fusarium Head Blight

Fusarium head blight (FHB) is one of the most detrimental wheat diseases. The accurate identification FHB severity significant to sustainable management and guarantee food production security. A total 2752 images with five infection levels were collected establish an grading dataset (FHBSGD), a novel lightweight GSEYOLOX-s was proposed automatically recognize FHB. simple, parameter-free attention module (SimAM) fused into CSPDarknet feature extraction network obtain more representative disease features while avoiding additional parameters. Meanwhile, ghost convolution model (G-head) designed achieve speed improvements. Furthermore, efficient intersection over union (EIoU) loss employed accelerate convergence improve positioning precision. results indicate that only 8.06 MB parameters achieved mean average precision (mAP) 99.23% detection 47 frames per second (FPS), which best performance compared other models, such as EfficientDet, Mobilenet-YOLOV4, YOLOV7, YOLOX series. deployed on mobile terminals assist farmers in real-time facilitate precise crop