应用深度学习的大姿态人脸对齐
作者:
作者简介:

姜悦卉(1995-),女,江苏东台人,硕士生,主要研究领域为视频编码,模式识别;张倩(1983-),女,博士,副教授,主要研究领域为视频和图像信息处理;王斌(1987-),女,博士,副教授,主要研究领域为图像处理,大数据;沈慧中(1995-),女,硕士生,主要研究领域为模式识别;黄继风(1963-),男,博士,教授,主要研究领域为模式识别,数字图像处理,生物信息学,视频图像识别;严涛(1981-),男,博士,主要研究领域为图像处理.

通讯作者:

姜悦卉,E-mail:1714861314@qq.com;张倩,E-mail:qianzhang@shnu.edu.cn

基金项目:

国家自然科学基金(61741111)


Large-pose Face Alignment Based on Deep Learning
Author:
Fund Project:

National Natural Science Foundation of China (61741111)

  • 摘要
  • | |
  • 访问统计
  • |
  • 参考文献 [18]
  • |
  • 相似文献 [20]
  • | | |
  • 文章评论
    摘要:

    针对大姿态人脸对齐算法中的精确度低的问题,设计并实现了一种新的分层并行和多尺度Inception-Resnet网络来实现大姿态人脸对齐.首先,构建了一个四阶级联沙漏网络模型.该模型通过端到端的方式直接输入图像进行人脸对齐.其次,网络内部使用预先设定的参数进行采样和特征提取.最后,直接输出对应的人脸特征点提取图像以及同等人脸大小的二维坐标点绘制图,并将所提出的方法在AFLW2000-3D数据集上进行测试.实验结果表明,对于任意无约束的二维人脸图像,该方法的归一化平均误差为4.41%.与传统方法相比,该方法输出的正脸姿态图像视觉质量高、保真度更强.

    Abstract:

    Aiming at the low accuracy of the large-pose face alignment algorithm, this paper designs and implements a new hierarchical parallel and multi-scale Inception-resnet network to achieve large-pose face alignment. Firstly, a four-class Hourglass network model is constructed. The model directly inputs images for face alignment in an end-to-end manner. Secondly, the network internally uses pre-set parameters for sampling and feature extraction. Finally, the corresponding face feature points are directly output. A two-dimensional coordinate point drawing of the image and the equivalent face size is extracted, and the proposed method is tested on the AFLW2000-3D data set. Experimental results show that the normalized average error of this method is 4.41% for any unconstrained two-dimensional face image. Compared with the traditional method, the positive face attitude image outputted in this paper has high visual quality and fidelity.

    参考文献
    [1] Kemelmacher-Shlizerman I, Basri R. 3D face reconstruction from a single image using a single reference face shape. IEEE Trans. on Software Engineering, 2011,33(2):394-405.
    [2] Jourabloo A, Liu X. Pose-invariant 3D face alignment. In:Proc. of the 2015 IEEE Int'l Conf. on Computer Vision (ICCV). IEEE, 2015. 3694-3702.
    [3] Zhang HW, Li Q, Sun Z, et al. Combining data-driven and model-driven methods for robust facial landmark detection. IEEE Trans. on Information Forensics and Security, 2018,13(10):2409-2422.
    [4] Sun Y, Wang X, Tang X. Deep convolutional network cascadefor facial point detection. In:Proc. of the 2013 IEEE Conf. on Computer Vision and Pattern Recognition (CVPR). IEEE, 2013. 3476-3483.
    [5] Zhang Z, Luo P, Loy CC, Tang X. Facial landmark detection bydeep multi-task learning. In:Proc. of the Computer Vision (ECCV) 2014. Springer-Verlag, 2014. 94-108.
    [6] Zhu X, Lei Z, Liu X, Shi H, Li SZ. Face alignment acrosslarge poses:A 3D solution. In:Proc. of the 2016 IEEE Conf. on Computer Vision and Pattern Recognition (CVPR). IEEE, 2016. 146-155.
    [7] Fan HJ, Zhou E. Approaching human level facial landmark localization by deep learning. Image & Vision Computing, 2016,47:27-35.
    [8] Hong R, Cheng WH, Yamasaki T, et al. Small object detection using deep feature pyramid networks. In:Advances in Multimedia Information Processing-PCM 2018. 2018. 554-564.
    [9] Jourabloo A, Liu X. Pose-invariant face alignment via CNN-based dense 3D model fitting. International Journal of Computer Vision, 2017,124(2):187-203.
    [10] Yuen K, Trivedi MM. An occluded stacked hourglass approach to facial landmark localization and occlusion estimation. IEEE Trans. on Intelligent Vehicles, 2017,2(4):321-331.
    [11] Deng JK, Zhou YX, Cheng SY, et al. Cascade multi-view hourglass model for robust 3D face alignment. In:Proc. of the 201813th IEEE Int'l Conf. on Automatic Face & Gesture Recognition (FG 2018). 2018. 399-403.
    [12] Newell A, Yang KY, Deng J. Stacked hourglass networks for human pose estimation. In:Proc. of the Computer Vision-ECCV 2016. Springer-Verlag, 2016. 483-499.
    [13] Lai HJ, Xiao SG, Yan P, et al. Deep recurrent regression for facial landmark detection. IEEE Trans. on Circuits & Systems for Video Technology, 2018,28(5):1144-1157.
    [14] Peng X, Feris RS, Wang XY, et al. RED-net:A recurrent encoder-decoder network for video-based face alignment. Int'l Journal of Computer Vision, 2018(2):1-17.
    [15] Liu F, Zeng D, Li J, et al. On 3D face reconstruction via cascaded regression in shape space. Frontiers of Information Technology & Electronic Engineering, 2017,18(12):1978-1990.
    [16] Barbu A, Lay N, Gramajo G. Face detection with a 3D model. Academic Press Library in Signal Processing, 2014,21(4):237-259.
    [17] Ranjan R, Sankaranarayanan S, Bansal A, et al. Deep learning for understanding faces:Machines may be just as good, or better, than humans. IEEE Signal Processing Magazine, 2018,35(1):66-83.
    [18] Chen FX, Liu F, Zhao QJ. Robust multi-view face alignment based on cascaded 2D/3D face shape regression. In:Proc. of the Chinese Conf. on Biometric Recognition. Springer-Verlag, 2016. 40-49.
    引证文献
    网友评论
    网友评论
    分享到微博
    发 布
引用本文

姜悦卉,张倩,王斌,沈慧中,黄继风,严涛.应用深度学习的大姿态人脸对齐.软件学报,2019,30(S2):1-8

复制
分享
文章指标
  • 点击次数:1961
  • 下载次数: 4326
  • HTML阅读次数: 0
  • 引用次数: 0
历史
  • 收稿日期:2019-08-17
  • 在线发布日期: 2020-01-02
文章二维码
您是第19863542位访问者
版权所有:中国科学院软件研究所 京ICP备05046678号-3
地址:北京市海淀区中关村南四街4号,邮政编码:100190
电话:010-62562563 传真:010-62562533 Email:jos@iscas.ac.cn
技术支持:北京勤云科技发展有限公司

京公网安备 11040202500063号