TY - GEN
T1 - Frontal to profile face verification in the wild
AU - Sengupta, Soumyadip
AU - Chen, Jun Cheng
AU - Castillo, Carlos
AU - Patel, Vishal M.
AU - Chellappa, Rama
AU - Jacobs, David W.
N1 - Publisher Copyright:
© 2016 IEEE.
PY - 2016/5/23
Y1 - 2016/5/23
N2 - We have collected a new face data set that will facilitate research in the problem of frontal to profile face verification 'in the wild'. The aim of this data set is to isolate the factor of pose variation in terms of extreme poses like profile, where many features are occluded, along with other 'in the wild' variations. We call this data set the Celebrities in Frontal-Profile (CFP) data set. We find that human performance on Frontal-Profile verification in this data set is only slightly worse (94.57% accuracy) than that on Frontal-Frontal verification (96.24% accuracy). However we evaluated many state-of-the-art algorithms, including Fisher Vector, Sub-SML and a Deep learning algorithm. We observe that all of them degrade more than 10% from Frontal-Frontal to Frontal-Profile verification. The Deep learning implementation, which performs comparable to humans on Frontal-Frontal, performs significantly worse (84.91% accuracy) on Frontal-Profile. This suggests that there is a gap between human performance and automatic face recognition methods for large pose variation in unconstrained images.
AB - We have collected a new face data set that will facilitate research in the problem of frontal to profile face verification 'in the wild'. The aim of this data set is to isolate the factor of pose variation in terms of extreme poses like profile, where many features are occluded, along with other 'in the wild' variations. We call this data set the Celebrities in Frontal-Profile (CFP) data set. We find that human performance on Frontal-Profile verification in this data set is only slightly worse (94.57% accuracy) than that on Frontal-Frontal verification (96.24% accuracy). However we evaluated many state-of-the-art algorithms, including Fisher Vector, Sub-SML and a Deep learning algorithm. We observe that all of them degrade more than 10% from Frontal-Frontal to Frontal-Profile verification. The Deep learning implementation, which performs comparable to humans on Frontal-Frontal, performs significantly worse (84.91% accuracy) on Frontal-Profile. This suggests that there is a gap between human performance and automatic face recognition methods for large pose variation in unconstrained images.
UR - http://www.scopus.com/inward/record.url?scp=84977647422&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84977647422&partnerID=8YFLogxK
U2 - 10.1109/WACV.2016.7477558
DO - 10.1109/WACV.2016.7477558
M3 - Conference contribution
AN - SCOPUS:84977647422
T3 - 2016 IEEE Winter Conference on Applications of Computer Vision, WACV 2016
BT - 2016 IEEE Winter Conference on Applications of Computer Vision, WACV 2016
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - IEEE Winter Conference on Applications of Computer Vision, WACV 2016
Y2 - 7 March 2016 through 10 March 2016
ER -