LUVLi Face Alignment: Estimating Landmarks' Location, Uncertainty, and Visibility Likelihood

被引:104
作者
Kumar, Abhinav [1 ]
Marks, Tim K. [2 ]
Mou, Wenxuan [3 ]
Wang, Ye [2 ]
Jones, Michael [2 ]
Cherian, Anoop [2 ]
Koike-Akino, Toshiaki [2 ]
Liu, Xiaoming [4 ]
Feng, Chen [5 ]
机构
[1] Univ Utah, Salt Lake City, UT 84112 USA
[2] Mitsubishi Elect Res Labs MERL, Cambridge, MA USA
[3] Univ Manchester, Manchester, Lancs, England
[4] Michigan State Univ, E Lansing, MI 48824 USA
[5] NYU, New York, NY USA
来源
2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2020) | 2020年
关键词
D O I
10.1109/CVPR42600.2020.00826
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Modern face alignment methods have become quite accurate at predicting the locations of facial landmarks, but they do not typically estimate the uncertainty of their predicted locations nor predict whether landmarks are visible. In this paper, we present a novel framework for jointly predicting landmark locations, associated uncertainties of these predicted locations, and landmark visibilities. We model these as mixed random variables and estimate them using a deep network trained with our proposed Location, Uncertainty, and Visibility Likelihood (LUVLi) loss. In addition, we release an entirely new labeling of a large face alignment dataset with over 19,000 face images in a full range of head poses. Each face is manually labeled with the ground-truth locations of 68 landmarks, with the additional information of whether each landmark is unoccluded, self-occluded (due to extreme head poses), or externally occluded. Not only does our joint estimation yield accurate estimates of the uncertainty of predicted landmark locations, but it also yields state-of-the-art estimates for the landmark locations themselves on multiple standard face alignment datasets. Our method's estimates of the uncertainty of predicted landmark locations could be used to automatically identify input images on which face alignment fails, which can be critical for downstream tasks.
引用
收藏
页码:8233 / 8243
页数:11
相关论文
共 91 条
[1]  
[Anonymous], CVPR
[2]  
[Anonymous], 1995, COMPUTER VISION IMAG
[3]  
[Anonymous], 2019, ICCV, DOI DOI 10.1109/ICCV.2019.01025
[4]  
[Anonymous], 2016, CVPR, DOI DOI 10.1109/CVPR.2016.371
[5]  
Asthana A., 2014, CVPR
[6]  
Asthana A, 2011, IEEE I CONF COMP VIS, P937, DOI 10.1109/ICCV.2011.6126336
[7]  
Atoum Yousef, 2017, ICCV
[8]  
Bao Y, 2014, AAAI CONF ARTIF INTE, P2
[9]  
Belagiannis V., 2017, FG
[10]  
Bertoni L., 2019, ICCV