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

11/03/2020
by   Rafael Berral-Soler, et al.
1

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 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

READ FULL TEXT

page 2

page 4

page 8

page 14

page 15

page 16

research
11/02/2021

HHP-Net: A light Heteroscedastic neural network for Head Pose estimation with uncertainty

In this paper we introduce a novel method to estimate the head pose of p...
research
06/12/2019

Towards Real-Time Head Pose Estimation: Exploring Parameter-Reduced Residual Networks on In-the-wild Datasets

Head poses are a key component of human bodily communication and thus a ...
research
12/03/2018

Nose, eyes and ears: Head pose estimation by locating facial keypoints

Monocular head pose estimation requires learning a model that computes t...
research
04/27/2022

Ollivier-Ricci Curvature For Head Pose Estimation From a Single Image

Head pose estimation is a crucial challenge for many real-world applicat...
research
05/22/2020

RankPose: Learning Generalised Feature with Rank Supervision for Head Pose Estimation

We address the challenging problem of RGB image-based head pose estimati...
research
10/21/2022

MEEV: Body Mesh Estimation On Egocentric Video

This technical report introduces our solution, MEEV, proposed to the Ego...
research
07/20/2022

VirtualPose: Learning Generalizable 3D Human Pose Models from Virtual Data

While monocular 3D pose estimation seems to have achieved very accurate ...

Please sign up or login with your details

Forgot password? Click here to reset