基于信息多模态融合技术在动物识别模型中的应用  

Application of information multimodal fusion technology in animal recognition model

作　　者：周晓健 段旭 阚继承 ZHOU Xiaojian;DUAN Xu;KAN Jicheng(Hainan Vocational University of Science and Technology,Haikou 571126,China)

机构地区：[1]海南科技职业大学,海南海口571126

出　　处：《中国高新科技》2023年第3期31-33,共3页

摘　　要：国家一级保护动物海南黑冠长臂猿仅分布于我国海南岛,提升对该物种群研究效率且尽可能避免人类行为的干扰成为痛点。文章基于多模态特征融合识别技术,提出一套猿类面部图像特征识别的算法框架,实现海南黑冠长臂猿个体的识别;结合基于时序维度的海南黑冠长臂猿面部特征和猿鸣声纹特征的融合识别模型,实现对长臂猿个体的研究和跟踪。以上框架将神经网络与海南黑冠长臂猿的识别、分类、标记和跟踪等深度结合,提升研究人员的工作效率,同时,也为其他珍稀生物个体多模态特征识别研究提供借鉴。Hainan black-crested gibbon,a national Class I protected animal,is only distributed in Hainan Island.It is a pain point to improve the research efficiency of this species group and avoid human interference as much as possible.In this paper,based on multi-modal feature fusion recognition technology,an algorithm framework for facial image recognition of apes is proposed to realize individual recognition of Hainan black-crowned gibbon.The study and tracking of individual gibbons can be realized by integrating the recognition model of facial features and vocal features of Hainan black-crested gibbon based on time series dimension.The above framework deeply combines the neural network with the recognition,classification,labeling and tracking of Hainan black-crested gibbon,which improves the working efficiency of researchers and also provides reference for the multi-modal feature recognition of otherrareorganisms.

关 键 词：多模态融合 特征识别 海南黑冠长臂猿 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]