The best way to conference proceedings by Francis Academic Press

Web of Proceedings - Francis Academic Press
Web of Proceedings - Francis Academic Press

Face Recognition Research Based on Fully Convolution Neural Network

Download as PDF

DOI: 10.25236/icmit.2017.25


YangWang, Jiachun Zheng

Corresponding Author

Jiachun Zheng


Face recognition technology has always been a hot topic, and has a widely application in our daily life. In recent years, with the rapid development of deep learning and the birth of various frameworks, face recognition is provided a new platform and ushered in a new opportunity. In this paper, a fully Convolution Neural Network(CNN) based on the Caffe framework and GPU is put forward for face recognition. The accuracy rate of it reaches 99% in the test set. Compared with the traditional extraction feature combined with the classifier method, CNN has a unique advantage, and doesn’t need to manually extract features.


Face recognition,Caffe,CNN,feature extraction.