中国农机化学报2026，Vol.47Issue(1)：67-72,6.DOI:10.13733/j.jcam.issn.2095-5553.2026.01.010
基于改进YOLOv8的巨峰葡萄检测方法研究
Research on improved YOLOv8-based detection method for Kyoho grapes
摘要
Abstract
Kyoho grapes exhibit diverse shapes,complex backgrounds,and varying illumination conditions in natural environments,making traditional object detection methods insufficient to meet practical demands.To address this issue,this paper proposes a Kyoho grape object detection method based on an improved YOLOv8 model.The model structure is optimized by introducing SPD—Conv,SPPF—LSAK,and CloFormer modules to enhance the model's ability to represent object features in complex scenes.Experimental results demonstrate that the improved model significantly outperforms the original YOLOv8 model in terms of detection precision(P),recall rate(R),and mean Average Precision(mAP),with mAP increasing from 86.89%to 92.24%,precision from 87.34%to 90.77%,and recall rate from 82.47%to 84.68%.Compared with existing YOLOv5,YOLOv6,YOLOv8,and YOLOv10 models,the proposed model achieves an improvement of 2.14 to 6.60 percentage points in detection accuracy under the same experimental conditions,while also exhibiting higher real-time performance in detection speed(FPS),reaching 87 frames per second.Finally,in a comparative analysis of images from the test set,the improved YOLOv8 model outperforms the original model in detection accuracy and detail capture capabilities,significantly enhancing detection effectiveness in complex backgrounds.This study provides an efficient and accurate solution for Kyoho grape detection in precision agriculture,laying a solid foundation for the application of fruit object detection tasks in intelligent agricultural systems.In the future,the method proposed in this paper can be further applied to other fruit detection tasks and further improve detection performance in complex environments by incorporating multimodal data.关键词
巨峰葡萄/YOLOv8/深度学习/目标检测/模型改进Key words
Kyoho grapes/YOLOv8/deep learning/object detection/improved model分类
农业科技引用本文复制引用
Zhao Xin,Xiong Haifeng,Liu Huawei..基于改进YOLOv8的巨峰葡萄检测方法研究[J].中国农机化学报,2026,47(1):67-72,6.基金项目
国家自然科学基金(61901292) （61901292）
山西省教育科学"十四五"规划项目(GH—220011) （GH—220011）
