A Fast Face Recognition Algorithm Based on MapReduce

Author

Zhen Zhang ; Wei Li ; HaiTao Jia

Author_Institution

Res. Inst. of Electron. Sci. & Technol., Univ. of Electron. Sci. & Technol. of China, Chengdu, China

Volume

2

fYear

2014

Firstpage

395

Lastpage

399

Abstract

Classifying large-scale face images is a very hot topic in the field of prevention. Our fast face recognition algorithm include two parts: (1) parallel facial feature extraction algorithm based on MapReduce, (2) DSVM that distributed SVM based on MapReuce. Firstly, combine some face images to blocks, and parallel extract HOG feature on every block. DSVM algorithm will be used to train face classification. Inadequate sampling and sufficient sampling will be used to address the problem of uneven samples. We adopt Chain Mapper/Chain Reducer model to improve the efficiency of DSVM. Finally, results show that parallel facial feature extraction algorithm and DSVM both have high processing efficiency, and DSVM can achieve good recognition accuracy.

Keywords

data handling; face recognition; feature extraction; image classification; image sampling; parallel algorithms; support vector machines; ChainMapper model; ChainReducer model; DSVM algorithm; MapReduce; distributed SVM; face classification; fast face recognition algorithm; inadequate sampling; large-scale face image classification; parallel HOG feature extraction; parallel facial feature extraction algorithm; sufficient sampling; Accuracy; Computational modeling; Computers; Face; Feature extraction; Support vector machines; Training; Hadoop; Hog; MapRedcue; SVM; inadequate sampling; sufficient sampling;

fLanguage

English

Publisher

ieee

Conference_Titel

Computational Intelligence and Design (ISCID), 2014 Seventh International Symposium on

Print_ISBN

978-1-4799-7004-9

Type

conf

DOI

10.1109/ISCID.2014.195

Filename

7082015