Robust Head-Pose Estimation Based on Partially-Latent Mixture of Linear Regressions

  title={Robust Head-Pose Estimation Based on Partially-Latent Mixture of Linear Regressions},
  author={Vincent Drouard and Radu Horaud and Antoine Deleforge and Sil{\`e}ye O. Ba and Georgios Evangelidis},
  journal={IEEE Transactions on Image Processing},
  • Vincent DrouardR. Horaud Georgios Evangelidis
  • Published 31 March 2016
  • Computer Science
  • IEEE Transactions on Image Processing
Head-pose estimation has many applications, such as social event analysis, human-robot and human-computer interaction, driving assistance, and so forth. Head-pose estimation is challenging, because it must cope with changing illumination conditions, variabilities in face orientation and in appearance, partial occlusions of facial landmarks, as well as bounding-box-to-face alignment errors. We propose to use a mixture of linear regressions with partially-latent output. This regression method… 

Figures and Tables from this paper

Deep Mixture of Linear Inverse Regressions Applied to Head-Pose Estimation

This paper proposes the coupling of a Gaussian mixture of linear inverse regressions with a ConvNet and describes the methodological foundations and the associated algorithm to jointly train the deep network and the regression function, and is the first to incorporate inverse regression into deep learning for computer vision applications.

Continuous Head Pose Estimation Using Manifold Subspace Embedding and Multivariate Regression

A continuous head pose estimation system is proposed to estimate yaw and pitch head angles from raw facial images based on manifold learning-based methods, due to their promising generalization properties shown for face modeling from images.

Learning toward practical head pose estimation

This paper investigates the impact of the failure of these assumptions, i.e., misalignment of face images, uncertainty and undersampling of training data, on head pose estimation accuracy of state-of-the-art methods.

Coupled cascade regression from real and synthesized faces for simultaneous landmark detection and head pose estimation

A unified framework, named coupled cascade regression (CCR), is introduced, for simultaneous facial landmark detection and head pose estimation, based on the cascade regression framework, to learn two separate regressors to update the landmark locations and three-dimensional face model parameters at each cascade level.

Hybrid coarse-fine classification for head pose estimation

This paper combines the coarse and fine regression output together for a deep network to do head pose estimation without facial landmarks, and achieves the state-of-the-art on AFLW2000, BIWI and performs favorably on AFLw.

Near Real-Time Three Axis Head Pose Estimation Without Training

A novel pose estimation method that exploits a quad-tree-based representation of facial features is described, providing accurate yaw, pitch and roll axis estimates almost in real-time, without need for any training or previous knowledge about the subject.

An Improved Single Shot Multibox for Video-Rate Head Pose Prediction

This work proposes an efficient and robust method to monitor the driver’s attention by binning continuous angles into specific classes, the 3D angle regression problem is converted into angle classification and face box regression, and the model directly outputs Euler angles without detecting face landmarks.

FDN: Feature Decoupling Network for Head Pose Estimation

A novel three-branch network architecture, termed as Feature Decoupling Network (FDN), a more powerful architecture for landmark-free head pose estimation from a single RGB image that outperforms other state-of-the-art methods based on a singleRGB image and behaves on par with approaches based on multimodal input resources.



Head pose estimation via probabilistic high-dimensional regression

This paper proposes a method that maps HOG-based descriptors, extracted from face bounding boxes, to corresponding head poses using a relatively small number of locally-linear regression functions to deal with the non-linear mapping problem at hand.

Robust Head Pose Estimation Using Supervised Manifold Learning

This paper presents a taxonomy of methods for incorporating continuous pose angle information into one or more stages of the manifold learning process, and discusses its implementation for Neighborhood Preserving Embedding and Locality Preserving Projection.

A Two-Layer Framework for Piecewise Linear Manifold-Based Head Pose Estimation

  • J. FoytikV. Asari
  • Computer Science, Mathematics
    International Journal of Computer Vision
  • 2012
A pose estimation framework that seeks to describe the global nonlinear relationship in terms of localized linear functions and is tested using widely accepted pose-varying face databases and shown to perform fine head pose estimation with competitive accuracy when compared with state of the art nonlinear manifold methods.

On partial least squares in head pose estimation: How to simultaneously deal with misalignment

This paper shows that the kernel version of PLS (kPLS) achieves better than state-of-the-art results on the estimation problem and develops a technique to reduce misalignment based on the learned PLS factors.

Synergistic Face Detection and Pose Estimation with Energy-Based Models

We describe a novel method for real-time, simultaneous multi-view face detection and facial pose estimation. The method employs a convolutional network to map face images to points on a manifold,

Face detection, pose estimation, and landmark localization in the wild

It is shown that tree-structured models are surprisingly effective at capturing global elastic deformation, while being easy to optimize unlike dense graph structures, in real-world, cluttered images.

Query Driven Localized Linear Discriminant Models for Head Pose Estimation

This work develops a query point driven, localized linear subspace learning method that approximates the non-linearity of the head pose manifold structure with piece-wise linear discriminating subspaces/metrics.

Head pose estimation in the wild using approximate view manifolds

Pose estimation using the proposed approach shows significant improvement in accuracy and reduced Mean Angular Error compared to other methods on the challenging AFLW, McGill, McGill and YouTube Faces datasets.