An Intuitive and Unconstrained 2D Cube Representation for Simultaneous Head Detection and Pose Estimation

TL;DR

This paper introduces a novel 2D cube representation for head detection and pose estimation that is intuitive, unconstrained, and directly reflects head orientation, simplifying pose calculation and improving adaptability to full-view scenarios.

Contribution

The paper proposes a single-stage keypoint-based method using a 2D cube representation for head detection and pose estimation, avoiding complex algorithms like PnP and enhancing full-view applicability.

Findings

Achieves comparable accuracy on AFLW2000 and BIWI datasets.

Demonstrates seamless adaptation to unconstrained full-view HPE on CMU panoptic dataset.

Provides a closed-form solution for Euler angles from 2D cube predictions.

Abstract

Most recent head pose estimation (HPE) methods are dominated by the Euler angle representation. To avoid its inherent ambiguity problem of rotation labels, alternative quaternion-based and vector-based representations are introduced. However, they both are not visually intuitive, and often derived from equivocal Euler angle labels. In this paper, we present a novel single-stage keypoint-based method via an {\it intuitive} and {\it unconstrained} 2D cube representation for joint head detection and pose estimation. The 2D cube is an orthogonal projection of the 3D regular hexahedron label roughly surrounding one head, and itself contains the head location. It can reflect the head orientation straightforwardly and unambiguously in any rotation angle. Unlike the general 6-DoF object pose estimation, our 2D cube ignores the 3-DoF of head size but retains the 3-DoF of head pose. Based on the prior of equal side length, we can effortlessly obtain the closed-form solution of Euler angles from predicted 2D head cube instead of applying the error-prone PnP algorithm. In experiments, our proposed method achieves comparable results with other representative methods on the public AFLW2000 and BIWI datasets. Besides, a novel test on the CMU panoptic dataset shows that our method can be seamlessly adapted to the unconstrained full-view HPE task without modification.