基于深度学习的人脸识别算法研究  

Face recognition algorithm based on deep learning

作　　者：胡亚洲 周亚丽[1] 张奇志[1] Hu Yazhou;Zhou Yali;Zhang Qizhi(School of Automation,Beijing Information Science&Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学自动化学院,北京100192

出　　处：《计算机应用研究》2020年第5期1432-1436,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(11672044,11172047);北京信息科技大学教改项目(2016JGYB09)。

摘　　要：考虑到现实环境中的人脸图片在角度、光线、分辨率上的复杂程度,对Inception-ResNet-V1网络结构进行了改进,同时完成了数据集制作、超参数调节等相关工作,并在家庭服务机器人平台上进行了实验研究。实验结果表明,改进的网络结构在LFW测试集上准确率达到99.22%,高于原始网络结构的99.05%;在亚洲人脸数据集上准确率达到99.20%,高于原始网络结构的97.10%;在自建非匹配人脸数据集上误识别率为3.43%,低于原始网络结构的12.28%。可以看出,与原始网络结构相比,改进网络结构提升了人脸识别的准确率且降低了误识别率。Considering the complexity of the angle,light and resolution of face images in real environment,this paper improved the network structure of Inception-ResNet-V1,completed the related work of dataset production and hyper-parameter adjustment,and carried out experimental research on the home service robot platform.The experimental results show that the improved network achieves 99.22%accuracy in LFW test set,which is higher than 99.05%of the original network structure.It reaches 99.20%accuracy on the Asian face dataset,which is higher than 97.10%of the original network structure.The false recognition rate on self-built mismatched face dataset is 3.43%,which is lower than 12.28%of the original network structure.It can be seen that compared with the original network structure,the improved network structure improves the accuracy of face recognition and reduces the false recognition rate.

关 键 词：家庭服务机器人 人脸识别 深度学习 Inception-ResNet-V1 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]