SlimFocal-YOLO: A Lightweight Grape Disease Detection Model Based on YOLOv8

Given the intricate nature and diverse manifestations of grape diseases, as well as the growing demand for mobile device deployment in practical detection scenarios, the challenge of lightweighting detection models without compromising accuracy has become increasingly prominent and challenging. To address this issue, this study delves into grape disease detection using an enhanced YOLOv8 model. We introduce a novel model architecture named SlimFocal-YOLO by incorporating the slim-neck structure and the FocalModulation module into the YOLOv8 framework, refining the YOLO neck while integrating hierarchical contextualization and gated aggregation steps. Experimental results demonstrate that the improved SlimFocal-YOLO model improves the mAP of disease image prediction from 0.937 to 0.940, while simultaneously reducing FLOPs by 8.5%. These findings provide important references and guidance for further research and application in the field of grape disease detection.