A Caltech Library Service

Robust face landmark estimation under occlusion

Burgos-Artizzu, Xavier P. and Perona, Pietro and Dollár, Piotr (2013) Robust face landmark estimation under occlusion. In: Proceedings of the 2013 IEEE International Conference on Computer Vision (ICCV 2013). IEEE Computer Society , Washington, DC, pp. 1513-1520. ISBN 978-1-4799-2840-8.

PDF - Submitted Version
See Usage Policy.


Use this Persistent URL to link to this item:


Human faces captured in real-world conditions present large variations in shape and occlusions due to differences in pose, expression, use of accessories such as sunglasses and hats and interactions with objects (e.g. food). Current face landmark estimation approaches struggle under such conditions since they fail to provide a principled way of handling outliers. We propose a novel method, called Robust Cascaded Pose Regression (RCPR) which reduces exposure to outliers by detecting occlusions explicitly and using robust shape-indexed features. We show that RCPR improves on previous landmark estimation methods on three popular face datasets (LFPW, LFW and HELEN). We further explore RCPR's performance by introducing a novel face dataset focused on occlusion, composed of 1,007 faces presenting a wide range of occlusion patterns. RCPR reduces failure cases by half on all four datasets, at the same time as it detects face occlusions with a 80/40% precision/recall.

Item Type:Book Section
Related URLs:
URLURL TypeDescription
Perona, Pietro0000-0002-7583-5809
Contact Email
Additional Information:© Copyright 2013 IEEE. This work is funded by the Gordon and Betty Moore Foundation and ONR MURI Grant N00014-10-1-0933.
Funding AgencyGrant Number
Gordon and Betty Moore FoundationUNSPECIFIED
Record Number:CaltechAUTHORS:20140530-015153032
Persistent URL:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:45988
Deposited By: SWORD User
Deposited On:30 May 2014 16:23
Last Modified:03 Oct 2019 06:39

Repository Staff Only: item control page