基于深度残差网络的多损失头部姿态估计

doi:10.19678/j.issn.1000-3428.0056176

摘要/Abstract

摘要： 为提高真实场景下头部姿态估计的准确性，提出一种采用深度残差网络的头部姿态估计方法。将深度残差网络RestNet101作为主干网络，引入优化器提高深层卷积网络训练时的梯度稳定性，使用RGB图像并采用分类器计算交叉熵损失，同时结合回归损失预测欧拉角表示头部姿态。实验结果表明，与FAN地标检测方法和无关键点细粒度方法相比，该方法在AFLW2000数据集和BIWI数据集上的平均绝对误差值更小，分别达到5.396和2.922，在300W_LP数据集上测试精度超过95%，在真实场景下具有较好的鲁棒性。

关键词: 深度残差网络, 欧拉角, 梯度优化, 回归损失, 姿态估计

Abstract: In order to improve the accuracy of headposture estimation in real scenes,this paper proposes a head posture estimation method using deep residual network.The deep residual network RestNet101 is used as the backbone network,and the optimizer is introduced to improve the gradient stability of the deep convolution network training.The cross entropy loss is calculated by using the RGB image and the classifier.At the same time,the Euler angle is predicted combined with the regression loss to represent the head posture.Experimental results show that,compared with the FAN landmark detection method and non key point fine-grained method,the proposed method has a smaller average absolute error on AFLW2000 dataset and BIWI dataset,which reaches 5.396 and 2.922,respectively.The accuracy of the method is over 95% on 300W_LP dataset,and the method has good robustness in real scenes.

Key words: deep residual network, Euler angle, gradient optimization, regression loss, posture estimation

中图分类号:

TP391.4

齐永锋, 马中玉. 基于深度残差网络的多损失头部姿态估计[J]. 计算机工程, 2020, 46(12): 247-253.

QI Yongfeng, MA Zhongyu. Multi-Loss Head Posture Estimation Based on Deep Residual Network[J]. Computer Engineering, 2020, 46(12): 247-253.

https://www.ecice06.com/CN/Y2020/V46/I12/247

图/表 12

20201216142511

20201216142515

20201216142518

20201216142523

20201216142527

20201216142530

20201216142533

20201216142538

20201216142541

20201216142546

20201216142549

20201216142552

参考文献

[1] MURPHY C E,DOSHI A,TRIVEDI M M.Head pose estimation for driver assistance systems:a robust algorithm and experimental evaluation[C]//Proceedings of 2007 IEEE Intelligent Transportation Systems Conference.Washington D.C.,USA:IEEE Press,2007:709-714.
[2] CHEN Jingying,LUO Nan,LIU Yuanyuan,et al.A hybrid intelligence-aided approach to affect-sensitive e-learning[J].Computing,2016,98(1):215-233.
[3] ZHAO Minghao.Research on computer posture detection based on computer vision[D].Chengdu:University of Electronic Science and Technology of China,2018.(in Chinese)赵明皓.基于计算机视觉的头部姿态检测方法研究[D].成都:电子科技大学,2018.
[4] MURPHY C E,TRIVEDI M M.Head pose estimation in computer vision:a survey[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(4):607-626.
[5] PATACCHIOLA M,CANGELOSI A.Head pose estimation in the wild using convolutional neural networks and adaptive gradient methods[J].Pattern Recognition,2017,71(11):132-143.
[6] WANG Chao,SONG Xubo.Robust head pose estimation via supervised manifold learning[J].Neural Networks,2014,53(5):15-25.
[7] WU J W,TRIVEDI M M.A two-stage head pose estimation framework and evaluation[J].Pattern Recognition,2008,41(3):1138-1158.
[8] LI D Q,PEDRYCZ W.A central profile-based 3D face pose estimation[J].Pattern Recognition,2014,47(2):525-534.
[9] HAGGERTY S E,WU A R,SIENKO K H,et al.A shared neural integrator for human posture control[J].Journal of Neurophysiology,2017,118(2):894-903.
[10] LIU Qingshan,YANG Jing,DENG Jiankang,et al.Robust facial landmark tracking via cascade regression[J].Pattern Recognition,2017,66(6):53-62.
[11] LI Ya,ZHANG Yunan,PENG Cheng,et al.Face attributes recognition method based on multi-task learning[J].Computer Engineering,2020,46(3):229-236.(in Chinese)李亚,张雨楠,彭程,等.基于多任务学习的人脸属性识别方法[J].计算机工程,2020,46(3):229-236.
[12] MIN Qiusha,LIU Neng,CHEN Yating,et al.Head pose estimation based on facial feature point localization[J].Computer Engineering,2018,44(6):263-269.(in Chinese)闵秋莎,刘能,陈雅婷,等.基于面部特征点定位的头部姿态估计[J].计算机工程,2018,44(6):263-269.
[13] ZHU Xiangyu,LIU Xiaoming,LEI Zhen,et al.Face alignment in full pose range:a 3D total solution[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2019,41(1):78-92.
[14] KUMAR A,ALAVI A,CHELLAPPA R.KEPLER:simultaneous estimation of keypoints and 3D pose of unconstrained faces in a unified framework by learning efficient H-CNN regressors[J].Image and Vision Computing,2018,79(9):49-62.
[15] SZEGEDY C,LIU W,JIA Y Q,et al.Going deeper with convolutions[C]//Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2015:1-9.
[16] RANJAN R,PATEL V M,CHELLAPPA R.HyperFace:a deep multi-task learning framework for face detection,landmark localization,pose estimation,and gender recognition[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2019,41(1):121-135.
[17] RANJAN R,SANKARANARAYANAN S,CASTILLO C D,et al.An all-in-one convolutional neural network for face analysis[C]//Proceedings of 2017 IEEE International Conference on Automatic Face and Gesture Recognition.Washington D.C.,USA:IEEE Press,2017:17-24.
[18] GUPTA A,THAKKAR K,GANDHI V,et al.Nose,eyes and ears:head pose estimation by locating facial keypoints[C]//Proceedings of ICASSP'19.Washington D.C.,USA:IEEE Press,2019:1-7.
[19] WANG Haofan,CHEN Zhenghua,ZHOU Yi.Hybrid coarse-fine classification for head pose estimation[EB/OL].[2019-08-25].https://www.researchgate.net/publication/330553261_Hybrid_coarse-fine_classification_for_head_pose_estimation.
[20] CHANG F J,TRAN A T,HASSNER T,et al.FacePoseNet:making a case for landmark-free face alignment[C]//Proceedings of 2017 IEEE International Conference on Computer Vision Workshops.Washington D.C.,USA:IEEE Press,2017:23-28.
[21] RUIZ N,CHONG E,REHG J M.Fine-grained head pose estimation without keypoints[EB/OL].[2019-08-25].https://arxiv.org/abs/1710.00925v1.
[22] LATHUILIERE S,JUGE R M,MESEJO P,et al.Deep mixture of linear inverse regressions applied to head-pose estimation[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C.,USA:IEEE Press,2017:31-37.
[23] DROUARD V,HORAUD R,DELEFORGE A,et al.Robust head-pose estimation based on partially-latent mixture of linear regressions[J].IEEE Transactions on Image Processing,2017,26(3):1428-1440.
[24] YIN X,YU X,SOHN K,et al.Towards large-pose face frontalization in the wild[C]//Proceedings of 2017 IEEE International Conference on Computer Vision.Washington D.C.,USA:IEEE Press,2017:3990-3999.
[25] LUO Liangchen,XIONG Yuanhao,LIU Yan,et al.Adaptive gradient methods with dynamic bound of learning rate[C]//Proceedings of 2019 International Conference on Learning Representations.New York,USA:ACM Press,2019:27-35.
[26] FANELLI G,DANTONE M,GALL J,et al.Random forests for real time 3D face analysis[J].International Journal of Computer Vision,2013,101(3):437-458.
[27] ADRIAN B,TZIMIROPOULOS,G.How far are we from solving the 2D & 3D face alignment problem?[EB/OL].[2019-08-25].https://www.researchgate.net/publication/315489658_How_Far_are_We_from_Solving_the_2D_3D_Face_Alignment_Problem_and_a_Dataset_of_230000_3D_Facial_Landmarks.

选择文件类型/文献管理软件名称

选择包含的内容