基于Keras卷积神经网络的多人脸识别抗干扰研究  

Research on Anti-jamming of Multi-face Recognition Based on Keras Convolutional Neural Network

作　　者：储汇 宋陈 汪晨灿 CHU Hui;SONG Chen;WANG Chencan(School of Mechanical Engineering,Anhui University of Science&Technology,Huainan 232001,China)

机构地区：[1]安徽理工大学机械工程学院,安徽淮南232001

出　　处：《洛阳理工学院学报（自然科学版）》2022年第1期62-67,共6页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

摘　　要：人脸识别的训练预测模型是多样而复杂的,影响人脸识别准确率和稳定性的因素也很多。人脸识别的抗干扰设计是构建人脸识别模型不可忽视的重要内容。通过获取更高质量的人脸图像数据源,选择效果更好的人脸识别优化器以及部分超参数的调整,来提高对大量人脸数据进行处理的能力。利用卷积神经网络减少人工干预,提高特征提取的算法精度,从而提高多人脸识别的精度。There are many complicated face recognition training prediction models,with many factors influencing the accuracy and stability of face recognition.Anti-jamming technology research is essential to construct the face recognition models,which aim to get the quality data,selects face recognition optimizer and adjusts other parameters in order to improve the face data processing.Also,convolutional neural network is applied to reduce human intervention and improve the accuracy of feature extraction,realizing the multi-face recognition.

关 键 词：人脸识别 卷积神经网络 优化器 抗干扰 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]