Low-Power Convolutional Neural Network Processor for a Face-Recognition System

被引:24
作者
Bong, Kyeongryeol [1 ]
Choi, Sungpill [1 ]
Kim, Changhyeon [1 ]
Yoo, Hoi-Jun [2 ]
机构
[1] Korea Adv Inst Sci & Technol, Daejeon, South Korea
[2] Korea Adv Inst Sci & Technol, Dept Elect Engn, Daejeon, South Korea
关键词
application-based system; authentication; face recognition; image processing and computer vision; low-power design; neural nets; special-purpose system;
D O I
10.1109/MM.2017.4241350
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
The authors propose a low-power convolutional neural network (CNN)-based face recognition system for user authentication in smart devices. The system comprises an always-on functional CMOS image sensor (CIS) for imaging and face detection, and a low-power CNN processor (CNNP) for face verification. Implemented in 65-nm CMOS technology, the system consumes 0.62 mW to evaluate one face at 1 fps and achieves 97 percent accuracy.
引用
收藏
页码:30 / 38
页数:9
相关论文
共 10 条
[1]  
Andrews G., 2015, KEEPING ALWAYS ON SY
[2]  
[Anonymous], 0749 U MASS COLL INF
[3]   Review of CMOS image sensors [J].
Bigas, M ;
Cabruja, E ;
Forest, J ;
Salvi, J .
MICROELECTRONICS JOURNAL, 2006, 37 (05) :433-451
[4]  
Bong K, 2017, ISSCC DIG TECH PAP I, P248, DOI 10.1109/ISSCC.2017.7870354
[5]   Always-On CMOS Image Sensor for Mobile and Wearable Devices [J].
Choi, Jaehyuk ;
Shin, Jungsoon ;
Kang, Dongwu ;
Park, Du-Sik .
IEEE JOURNAL OF SOLID-STATE CIRCUITS, 2016, 51 (01) :130-140
[6]  
Dongsuk Jeon, 2015, 2015 Symposium on VLSI Circuits (VLSI Circuits), pC48, DOI 10.1109/VLSIC.2015.7231322
[7]  
Evans D., 2011, CISCO
[8]  
Jaderberg M., 2014, ARXIV 1405 3866
[9]  
Moons B, 2017, ISSCC DIG TECH PAP I, P246, DOI 10.1109/ISSCC.2017.7870353
[10]   Rapid object detection using a boosted cascade of simple features [J].
Viola, P ;
Jones, M .
2001 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOL 1, PROCEEDINGS, 2001, :511-518