RealHePoNet: a robust single-stage ConvNet for head pose estimation in the wild

被引:7
|
作者
Berral-Soler, Rafael [1 ]
Madrid-Cuevas, Francisco J. [1 ]
Munoz-Salinas, Rafael [1 ]
Marin-Jimenez, Manuel J. [1 ]
机构
[1] Univ Cordoba, Dept Comp & Numer Anal, Cordoba, Spain
来源
NEURAL COMPUTING & APPLICATIONS | 2021年 / 33卷 / 13期
关键词
Human head pose estimation; ConvNets; Human-computer interaction; Deep Learning; FRAMEWORK;
D O I
10.1007/s00521-020-05511-4
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Human head pose estimation in images has applications in many fields such as human-computer interaction or video surveillance tasks. In this work, we address this problem, defined here as the estimation of both vertical (tilt/pitch) and horizontal (pan/yaw) angles, through the use of a single Convolutional Neural Network (ConvNet) model, trying to balance precision and inference speed in order to maximize its usability in real-world applications. Our model is trained over the combination of two datasets: 'Pointing'04' (aiming at covering a wide range of poses) and 'Annotated Facial Landmarks in the Wild' (in order to improve robustness of our model for its use on real-world images). Three different partitions of the combined dataset are defined and used for training, validation and testing purposes. As a result of this work, we have obtained a trained ConvNet model, coined RealHePoNet, that given a low-resolution grayscale input image, and without the need of using facial landmarks, is able to estimate with low error both tilt and pan angles (4.4 degrees average error on the test partition). Also, given its low inference time (6 ms per head), we consider our model usable even when paired with medium-spec hardware (i.e. GTX 1060 GPU). Code available at: https://github.com/rafabs97/headpose_final Demo video at: https://www.youtube.com/watch?v=2UeuXh5DjAE.
引用
收藏
页码:7673 / 7689
页数:17
相关论文
共 50 条
  • [1] RealHePoNet: a robust single-stage ConvNet for head pose estimation in the wild
    Rafael Berral-Soler
    Francisco J. Madrid-Cuevas
    Rafael Muñoz-Salinas
    Manuel J. Marín-Jiménez
    Neural Computing and Applications, 2021, 33 : 7673 - 7689
  • [2] RetinaHand: Towards Accurate Single-Stage Hand Pose Estimation
    Xiao, Zilong
    Lin, Luojun
    Yang, Yuanxi
    Yu, Yuanlong
    ADVANCES IN NATURAL COMPUTATION, FUZZY SYSTEMS AND KNOWLEDGE DISCOVERY, ICNC-FSKD 2022, 2023, 153 : 639 - 647
  • [3] Single-Stage 6D Object Pose Estimation
    Hu, Yinlin
    Fua, Pascal
    Wang, Wei
    Salzmann, Mathieu
    2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2020, : 2927 - 2936
  • [4] FDNet: Feature decoupling for single-stage pose estimation in complex scenes
    Wang, Qianqian
    Liu, Qiong
    JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION, 2024, 98
  • [5] Evaluation of single-stage vision models for pose estimation of surgical instruments
    Burton, William
    Myers, Casey
    Rutherford, Matthew
    Rullkoetter, Paul
    INTERNATIONAL JOURNAL OF COMPUTER ASSISTED RADIOLOGY AND SURGERY, 2023, 18 (12) : 2125 - 2142
  • [6] Evaluation of single-stage vision models for pose estimation of surgical instruments
    William Burton
    Casey Myers
    Matthew Rutherford
    Paul Rullkoetter
    International Journal of Computer Assisted Radiology and Surgery, 2023, 18 : 2125 - 2142
  • [7] An integrated two-stage framework for robust head pose estimation
    Wu, JW
    Trivedi, MM
    ANALYSIS AND MODELLING OF FACES AND GESTURES, PROCEEDINGS, 2005, 3723 : 321 - 335
  • [8] Spacecraft Homography Pose Estimation with Single-Stage Deep Convolutional Neural Network
    Chen, Shengpeng
    Yang, Wenyi
    Wang, Wei
    Mai, Jianting
    Liang, Jian
    Zhang, Xiaohu
    SENSORS, 2024, 24 (06)
  • [9] A Compact and Powerful Single-Stage Network for Multi-Person Pose Estimation
    Xiao, Yabo
    Wang, Xiaojuan
    He, Mingshu
    Jin, Lei
    Song, Mei
    Zhao, Jian
    ELECTRONICS, 2023, 12 (04)
  • [10] Rethinking the Person Localization for Single-Stage Multi-Person Pose Estimation
    Jin, Lei
    Wang, Xiaojuan
    Nie, Xuecheng
    Wang, Wendong
    Guo, Yandong
    Yan, Shuicheng
    Zhao, Jian
    IEEE TRANSACTIONS ON MULTIMEDIA, 2024, 26 : 1436 - 1447