基于MaskR一CNN的轻量级草莓实例分割算法

打开文本图片集

DOI：10.13733/j.jcam.issn.2095-5553.2025.07.018

Abstract：Inresponse tothecomplexityoftheorchardpicking environment，thedificultyofacuratesegmentation betweenstrawberies andthesurrounding environment，and the inabilityof the existing model processing speedtorealize fastsegmentation，asegmentationalgorithmoflightweight strawberry instance based on Mask R—CNNisproposed.Onthe basis of theoriginal MaskR—CNNalgorithm，theMobileNetV3network isusedtoreplace theoriginal ResNet101 backbonenetwork，thealgorithmislightweight，andthechannelatentionmechanismintheoriginalMobileNetV3 residualstructure isreplacedbythecolaborativeatentionmechanism module，which iscombinedwith the feature pyramidnetworkarchitecturetoperformthefeatureextraction，andthestrawberyindividualsarerealized.The preciseand fastlocalizationsegmentationofindividual strawberiesisachieved.Finallycomparison experimentsareperformedontheselflabeleddataset.TheexperimentalresultsshowthattheproposedimprovedMaskR—CNNalgorithmimprovesborder （2号 mAP and mask mAP by 75% and 4.05% respectively，and the detection speed by 2o.O9 frames/s compared with the original Mask R—CNN model，which reduces the dependence of the model on hardware storage space and arithmetic power.

Keywords：strawberry image；instance segmentation;improved Mask R—CNN;CA atentionmechanism；lightweight network

0 引言

随着城镇化不断推进，劳动力人数减少，人工成本增长等问题日益严重，果蔬采摘时间占整体生产时间的 25% ，采摘成本占总成本约 50%[1] 。（剩余9570字）