ABSTRACT
Despite the rapid development of detecting facial landmarks on 2D images in the past decade, detecting the landmarks on 3D faces under varying poses is still challenging. Existing methods rely either on pose-invariant feature descriptors, which are vulnerable to missing data (e.g., due to self-occlusion at large poses), or on multi-view landmark models, which are complicated to apply. In this paper, instead, we propose to first normalize the 3D face to frontal pose and then use a single deep model to regress the facial landmarks. We adapt a pose estimation network to estimate the yaw and pitch angles of the input 3D face based on its depth image, and then rotate the face to frontal pose and encode it as a position map. We employ another deep network to predict the location of landmarks on the position map, which is projected back onto the input 3D face to get the final detection results. Evaluation experiments using three public datasets (FRGCv2, UND and Bosphorus) prove the superiority of the proposed method for pose-invariant 3D facial landmark detection.
- Mauricio P. Segundo, Chau C. Queirolo, Olga Regina Pereira Bellon, and Luciano Silva, “Automatic 3d facial segmentation and landmark detection,” in ICIAP. IEEE, 2007, pp. 431–436.Google Scholar
- Marcelo Romero and Nick E. Pears, “Landmark localization in 3d face data,” inAVSS.IEEE,2009,pp.73–78.Google Scholar
- Nese Alyüz, Berk Gökberk, and Lale Akarun, “Regional registration for expression resistant 3-d face recognition,” in Trans. Information Forensics and Security. IEEE, 2010, pp. 425–440.Google ScholarDigital Library
- Xi Zhao, Emmanuel Dellandréa, Liming Chen, and Ioannis A. Kakadiaris, “Accurate landmarking of three dimensional facial data in the presence of facial expressions and occlusions using a three-dimensional statistical facial feature model,” in Trans. Systems, Man, and Cybernetics, Part. IEEE, 2011, pp. 1417–1428.Google ScholarDigital Library
- Xin Fan, Qi Jia, Kang Huyan, Xianfeng Gu, and Zhongxuan Luo, “3d facial landmark localization using texture regression via conformal mapping,” in Pattern Recognition Letters. ELSEVIER, 2016, pp. 395–402.Google ScholarDigital Library
- Jinwen Xu, Qijun Zhao, Xiaofeng Li, and Yang Wang, “2.5d cascaded regression for robust facial landmark detection,” in IJCB. IEEE, 2017, pp. 124–132.Google Scholar
- Clement Creusot, Nick E. Pears, and Jim Austin, “A machine-learning approach to keypoint detection and landmarking on 3d meshes,” in IJCV. Springer, 2013, pp. 146–179.Google Scholar
- G. Passalis, P. Perakis, T. Theoharis, and I. A. Kakadiaris, “Using facial symmetry to handle pose variations in real-world 3d face recognition,” inTPAMI.IEEE,2013, pp. 1938–951.Google Scholar
- P. Perakis, G. Passalis, T. Theoharis, and I. Kakadiaris, “3d facial landmark detection under large yaw and expression variations,” in TPAMI. IEEE, 2013, pp. 1552– 564.Google ScholarDigital Library
- Janez Krizaj, Ziga Emersicy, Simon Dobrisek, Peter Peery, and Vitomir Struc, “Localization of facial landmarks in depth images using gated multiple ridge descent,” in IWOBI. IEEE, 2018, pp. 1–8.Google Scholar
- Marek Kowalski, Jacek Naruniec, and Tomasz Trzcinski, “Deep alignment network: A convolutional neural network for robust face alignment,” in CVPR. IEEE, 2017, pp. 2034–2043.Google Scholar
- Xu Luo, Pengfei Li, Fuxuan Chen, and Qijun Zhao, “Improving large pose face alignment by regressing 2d and 3d landmarks simultaneously and visibility refinement,” in CCBR, 2018, pp. 349–357.Google ScholarCross Ref
- Nataniel Ruiz, Eunji Chong, and James M.Rehg, “Fine grained head pose estimation without keypoints,” in CVPR. IEEE, 2018, pp. 2074–2083.Google Scholar
- Kangkang Gao, Shan-Ming Yang, Keren Fu, and Peng Cheng, “Deep 3d facial landmark detection on position maps,” in IScIDE, 2019, pp. 299–311.Google Scholar
- Xiangyu Zhu, Zhen Lei, Xiaoming Liu, Hailin Shi, and Stan Z. Li, “Face alignment across large poses: A 3d solution,” in CVPR. IEEE, 2016, pp. 146–155.Google Scholar
- P. Jonathon Phillips, Patrick J. Flynn, W. Todd Scruggs, Kevin W.Bowyer, JinChang, Kevin Hoffman, Joe Marques, Jaesik Min, and William J. Worek, “Overview of the face recognition grand challenge,” in CVPR. IEEE, 2005, pp. 947–954.Google Scholar
- Ping Yan and Kevin W. Bowyer, “Empirical evaluation of advanced ear biometrics,” in CVPR.IEEE,2005, p. 41.Google Scholar
- Arman Savran, Nese Alyüz, Hamdi Dibeklioglu, Oya Çeliktutan, Berk Gökberk, Bülent Sankur, and Lale Akarun, “Bosphorus database for 3d face analysis,” in BIOID, 2008, pp. 47–56.Google Scholar
- Federico M. Sukno, John L. Waddington, and Paul F. Whelan, “3-d facial landmark localization with asymmetry patterns and shape regression from incomplete local features,” in Trans. Cybernetics. IEEE, 2015, pp. 1717–1730.Google ScholarCross Ref
- Kai Wang, Xi Zhao, Wanshun Gao, and Jianhua Zou, “Acoarse-to-fineapproachfor3dfaciallandmarkingby using deep feature fusion,” in Symmetry, 2018, p. 308.Google ScholarCross Ref
- M. P. Segundo, C. Queirolo, O. R. P. Bellon, and L. Silva, “Automatic 3d facial segmentation and landmark detection,” in ICIAP. IEEE, 2007, pp. 431–36.Google Scholar
Recommendations
Dynamic Weighting of Facial Features for Automatic Pose-Invariant Face Recognition
CRV '12: Proceedings of the 2012 Ninth Conference on Computer and Robot VisionThis paper proposes an automatic pose-invariant face recognition system. In our approach, we consider the texture information around the facial features to compute the similarity measure between the probe and gallery images. The weight of each facial ...
Rough set based pose invariant face recognition with mug shot images
This paper presents a pose invariant face recognition system to annihilate pose problem by using a modified log Gabor algorithm and the concepts of rough sets together. The proposed system consists of four separate modules; face detection, pose ...
Fully automatic pose-invariant face recognition via 3D pose normalization
ICCV '11: Proceedings of the 2011 International Conference on Computer VisionAn ideal approach to the problem of pose-invariant face recognition would handle continuous pose variations, would not be database specific, and would achieve high accuracy without any manual intervention. Most of the existing approaches fail to match ...
Comments