农机使用与维修Issue(6)：1-8,8.DOI:10.14031/j.cnki.njwx.2025.06.001
基于改进YOLOv8n的竹节检测方法
Bamboo Joint Detection Method Based on Improved YOLOv8n
摘要
Abstract
Rapid detection and accurate identification of bamboo nodes are important prerequisites for improving the qual-ity of bamboo primary processing.A YOLOv8n-CSM bamboo joint detection model is proposed to address the difficult problem of avoiding joints in bamboo processing.First,ConvNeXt v2 is used as the backbone feature extraction network to enhance the feature extraction capability of the network;second,the Slim-Neck neck network structure is construc-ted using the Generalized-Sparse Convolution(GSConv)and VoVGSCSP modules,which reduces the number of model parameters while keeping the recognition accuracy of the model;finally,the multi-scale convolutional attention(MSCA)mechanism is embedded into the trunk output of the model to enhance the feature extraction ability of the mod-el for bamboo nodes and weaken the background interference.The results show that the improved YOLOv8n-CSM mod-el has a mean average precision(mAP)of 94.4%.Compared with the target detection models YOLOv3-tiny,YOLOv5,YOLOv6,and YOLOv8n,the mAP of the YOLOv8n-CSM model is 0.5 mAP.CSM model's mAP0.5 is 0.5,1.1,1.3,and 1.5 percentage points higher,respectively,and the improved model has the advantages of small pa-rameter count and fast detection speed,which can provide technical support for the fast and accurate detection of bamboo joints.关键词
目标检测/竹节/YOLOv8n/MSCA注意力机制/Slim-Neck/ConvNeXt v2Key words
target detection/bamboo/YOLOv8n/MSCA attention mechanism/Slim-Neck/ConvNeXt v2分类
农业科技引用本文复制引用
李贵强,陈继飞..基于改进YOLOv8n的竹节检测方法[J].农机使用与维修,2025,(6):1-8,8.基金项目
中国工程院战略研究与咨询项目(2024-XZ-49) （2024-XZ-49）
