VanillaFaceNet:一种高精度快速推理的牛脸识别方法  

VanillaFaceNet:A high-precision and rapid inference for bovine face recognition

作　　者：栾浩天 齐咏生[1,2,3] 刘利强 王朝霞[1,2,3] 李永亭 LUAN Haotian;QI Yongsheng;LIU Liqiang;WANG Zhaoxia;LI Yongting(College of Electric Power,Inner Mongolia University of Technology,Hohhot 010051,China;Large-scale Energy Storage Technology Engineering Research Center of Ministry of Education,Hohhot 010080,China;Inner Mongolia Autonomous Region University Smart Energy Technology and Equipment Engineering Research Center,Hohhot 010080,China)

机构地区：[1]内蒙古工业大学电力学院,呼和浩特010051 [2]大规模储能技术教育部工程研究中心,呼和浩特010080 [3]内蒙古自治区高等学校智慧能源技术与装备工程研究中心,呼和浩特010080

出　　处：《农业工程学报》2024年第18期120-131,共12页Transactions of the Chinese Society of Agricultural Engineering

基　　金：国家自然科学基金项目(62363029);内蒙古科技计划项目(2020GG0283,2021GG0256);内蒙古自然科学基金项目(2022MS06018,2024QN06020);呼和浩特市高校院所协同创新项目(XTCX2023-16);自治区直属高校基本科研业务费项目(ZTY2024024)。

摘　　要：快速精准确定牛只身份对于牛只活体贷款,改善牛只骗保等问题具有重要意义。针对不同牛只面部差异小,FaceNet网络层数深,推理速度较慢,模型分类精度不足等问题,该研究提出了基于FaceNet的牛脸识别方法-VanillaFaceNet。该方法首先将FaceNet的主干特征提取网络替换为极简网络VanillaNet-13并提出动态激活和增强型线性变换的激活函数两种方法提高网络的非线性;然后,提出一种新的DBCA(dual-branch coordinate attention)注意力模块,能够更好地反映不同牛只面部特征之间的差异,从而提高网络的识别精度;最后,针对triplet loss仅能减小牛只类间差异的问题,采用center-triplet loss联合监督来减少牛只类内差异,从而提高了相同牛只身份比对的准确性。基于自建的牛脸数据集对该模型进行训练和测试,试验结果表明,VanillaFaceNet对牛只识别的准确率达到88.21%,每秒传输帧数为26.23帧。与FaceNet、MobileFaceNet、CenterFace、CosFace和ArcFace算法相比,本文算法的识别准确率分别提高了2.99、9.58、6.26、3.85和4.49个百分点,推理速度分别提升了2.67、0.77、0.10、1.28和0.94帧/s。该模型对牛只有较为优秀的识别效果,适于在嵌入式设备上部署,实现了牛只面部识别精度和推理速度之间的平衡。Intelligent farming has been an ever-increasing trend in agricultural production,with the development of artificial intelligence(AI)and Internet of Things(IoT).Rapid and accurate identification of cattle identity is of great significance to prevent the insurance fraud for the live cattle loans in the cattle industry.Among them,computer vision can be expected for the cattle face recognition in the modernization transformation of the livestock industry.Smart devices and systems can also be integrated to achieve the intelligent cattle management,feeding,and disease prevention.However,the traditional identification(such as ear tags and collars)has limited the large-scale production in recent years,due to the small differences in facial features among different cattle,the deep layers of the FaceNet network,slow inference speeds,and insufficient classification accuracy.In this study,a cattle face recognition was proposed using FaceNet,called VanillaFaceNet.Firstly,the backbone feature extraction network of FaceNet was replaced with the latest simplified network.VanillaNet-13.Dynamic activation and enhanced linear transformation of activation functions were proposed to improve the non-linearity of the network.Specifically,dynamic activation was fully utilized the expressive power of activation functions during training when dynamically adjusting,in order to flexibly adapt the variations in data distribution at different stages of training.Dynamic activation was used to merge the convolutional layers during inference phase.The computational load was reduced to improve the inference speed of networks.The performance and efficiency of model were then enhanced during training and inference.Activation functions with linear transformations were significantly enhanced the non-linearity through parallel stacking.Multiple activation functions were stacked in parallel,thus enabling each layer to capture more complex features.Additionally,spatial context information was embedded within the activation functions.The spatial relation

关 键 词：识别 特征 提取 牛脸 FaceNet 注意力机制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]