基于改進(jìn)MobileNetV2的牛只面部識(shí)別方法研究

打開(kāi)文本圖片集

中圖分類(lèi)號(hào)：TP391.4 文獻(xiàn)標(biāo)識(shí)碼：A 文章編號(hào)：2096-4706（2025）11-0020-05

Research on Cattle Facial Recognition Method Based on Improved MobileNetV2

TIANHuijuan, CAO Mengqi,RU Chunrui,REN Zhaohui (YanglingVocational and TechnicalCollege,Xianyang712199,China)

Abstract:In viewofthe problems thatcattle are easily frightenedand consume manpower and material resources in the processoftraditioalcatleindividualidentifcatio,tispaperproposesanon-contactcatleindividualidentifcationmethodby meansofcomputervisiontechnology.Firstly,thecalefacephotosaretakenintherealcatlefrmevironment,andtheattle facerecognitiondatasetisconstructed.econdlycombinedwiththeactualnetworkconditionsandcomputingpowerlevelofthe catlefarm,thelightweightNuralNetworkmodelMobileNetV2isselectedastebasicnetworkmodel.Finaly,theCoordinate AttentionMechanismisintroducedtoimprovetheMobileNetV2model,soastoimprovetheaccuracyofthemodelandehance its featureextractionabilityforkeypositions.The experimentalresults show thatthe improved MobileNetV2 modelperforms well in the cattle face recognition task. The model size is only 2.86MB ,and the recognition accuracy is as high as 95.81% , which can fully meet the needs ofnon-contact cattle individual recognition in the actual cattle farm environment.

Keywords:MobileNetV2;Attention Mechanism; cattle identificatio1

0 引言

近年來(lái)，我國(guó)牛養(yǎng)殖產(chǎn)業(yè)發(fā)展勢(shì)頭迅猛，市場(chǎng)需求的增長(zhǎng)促使養(yǎng)殖規(guī)模不斷擴(kuò)大。（剩余9433字）