农业装备与车辆工程2025，Vol.63Issue(10)：19-26,8.DOI:10.3969/j.issn.1673-3142.2025.10.003
基于改进YOLOv11n的马铃薯外部缺陷识别研究
Research on potato external defect detection based on YOLOv11n
摘要
Abstract
With the continuous expansion of potato cultivation in China,typical external defects such as greening,cracking,and mildew spots frequently occur during harvesting,storage,and transportation.These defects not only affect the appearance quality of potatoes but may also reduce their nutritional value or even produce toxins,leading to economic losses.To address the demand for external defect detection,this study proposes an improved method based on YOLOv11n.Specifically,a Context Gated Linear Unit(CGLU)was embedded into the C3k2 module of the backbone network to enhance contextual feature extraction,and a lightweight SimAM attention mechanism is introduced in the Neck stage to highlight defect-related feature representations.This approach enables more accurate localization and recognition of defects such as greening,cracking,and mildew spots,significantly improving the model＇s capability to detect subtle defects.Experimental results demonstrate that,compared with the baseline YOLOv11n,the improved model achieves gains of 2.5%,3.0%,and 3.2%in mAP@0.5,precision,and recall,respectively,validating the effectiveness and superiority of the proposed method in potato external defect detection.关键词
马铃薯缺陷/品质检测/通道引导的轻量级卷积单元/SimAM注意力机制Key words
potato defects/quality inspection/channel-guided lightweight unit/SimAM attention mechanism分类
农业科技引用本文复制引用
陈科,李琦颖,汪迁,刘晶,邓伟刚..基于改进YOLOv11n的马铃薯外部缺陷识别研究[J].农业装备与车辆工程,2025,63(10):19-26,8.基金项目
内蒙古自治区一流学科科研专项项目(YLXKZX-NND-047) （YLXKZX-NND-047）
内蒙古自然科学基金项目(2022MS05027) （2022MS05027）
内蒙古农业大学青年教师科研能力提升专项(BR220127) （BR220127）
