农业机械学报2025，Vol.56Issue(9)：657-666,728,11.DOI:10.6041/j.issn.1000-1298.2025.09.056
基于改进YOLO v10n的苹果叶片病害检测方法
Apple Leaf Disease Detection Method Based on Improved YOLO v10n
摘要
Abstract
Early detection of apple leaf diseases is very important to ensure apple yield and quality.However,due to the characteristics of different disease forms and small lesion areas on apple leaves,the disease detection accuracy is not high,which is difficult to meet the needs of accurate prevention and control in modern agriculture.To address this issue,an improved YOLO v10n-based method for apple leaf disease detection was proposed.The standard convolution(Conv)in the backbone network was replaced with receptive field attention convolution(RFAConv)to enhance the network's ability to capture important features of diseased leaves.The reparameterized generalized feature pyramid network(RepGFPN)was introduced in the Neck network to fuse low-level features with high-level semantic information,thereby improving the model's ability to extract features of diseases at different scales on apple leaves.A spatial pyramid pooling-fast_attention(SPPF_attention)module was constructed to enable the model to focus more effectively on critical attention information.A spatial and channel reconstruction convolution_v10Detect(SC_C_v10Detect)head was designed to enhance the model's detection capability for lesions of varying sizes.Results showed that the proposed model achieved a precision of 86.3％,a recall of 86.8％,an mAP@0.5 of 90.8％,and an mAP@0.5:0.95 of 62.9％.Compared with the original YOLO v10n model,the performance metrics were improved by 1.2,4.2,2.1,and 2.8 percentage points,respectively.Furthermore,the proposed model outperformed Faster R-CNN,SSD,YOLO v8n,YOLO v9s,YOLO v10n,and YOLO 11n in terms of mAP@0.5,with improvements of 25.9,12.2,2.8,1.6,2.1,and 1.4 percentage points,respectively.The improved method proposed significantly improved the detection accuracy of apple leaf diseases,and provided reliable technical support for early warning and precise prevention of orchard diseases,which was of great significance for promoting the intelligent development of agriculture.关键词
苹果/叶片病害/目标检测/YOLO v10/多尺度融合/注意力机制Key words
apple/leaf disease/target detection/YOLO v10/multiscale fusion/attention mechanism分类
信息技术与安全科学引用本文复制引用
赵晓,杨梦婷,张懿丹..基于改进YOLO v10n的苹果叶片病害检测方法[J].农业机械学报,2025,56(9):657-666,728,11.基金项目
国家自然科学基金项目(61971272、61601271) （61971272、61601271）
