DeepFace: Closing the Gap to Human-Level Performance in Face Verification

In modern face recognition, the conventional pipeline consists of four stages: detect => align => represent => classify. We revisit both the alignment step and the representation step by employing explicit 3D face modeling in order to apply a piecewise affine transformation, and derive a fa...

Full description

Bibliographic Details

Published in:	2014 IEEE Conference on Computer Vision and Pattern Recognition pp. 1701 - 1708
Main Authors:	Taigman, Yaniv, Ming Yang, Ranzato, Marc'Aurelio, Wolf, Lior
Format:	Conference Proceeding Journal Article
Language:	English
Published:	IEEE 01.06.2014
Subjects:	Agriculture Alignment Computer vision Face Face recognition Facial Networks Neural networks Pattern recognition Representations Shape Solid modeling State of the art Three-dimensional displays Training
ISSN:	1063-6919, 1063-6919
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!