基于YOLO11n的轻量级草莓成熟度检测方法

打开文本图片集

中图分类号：S126;S668.4 文献标识码：A 文章编号：1000-4440（2025）10-1997-12

Abstract：In field environments，toaddress thechallenges of limitedcomputing resources in strawbery-picking robots，small target size，aswellasocclusionsandoverlaps causedbyleavesand fruits thathinderaccurate detection，animproved model basedonYOLO11nwas proposed todetect strawberries inthe immature，turming，and mature stages.Firstly， MobileNetV4 was used to replace theoriginal backbone network to reducethenumber of parameters and computational cost. Secondly，anewfeature fusionmethod，Bi-Freq，was proposedtoreplacetheneck network’soriginalfeaturefusionstrategy，enhancing featurerepresentation androbustness.Finally，theSEAMatention mechanismwasadded tothedetection head toimprove the model’scapabilityin procesing spatialand channel information.The improved model（YOLO11nMFBS）achieved 1.713 M parameters and 4.7 G FLOPs，reducing parameters and floating-point computation by 33.9% and 26.6% respectively compared to the original YOLO11n. Compared with other mainstream detection models，YOLO11nMFBS demonstrates superior performance in both lightweight design and detection accuracy.

Key words：object detection；YOLO11n； lightweight design；feature fusion；attention mechanism；strawberr

草莓果实色泽诱人，营养价值高，向来深受大众喜爱，调查结果显示，中国是目前世界上最大的草莓生产国，草莓产量约占全球的 1/3[1] 。（剩余16334字）