Multi-Person 3D Human Pose Estimation from Monocular Images

TL;DR

This paper introduces HG-RCNN, a modular and simple network for multi-person 3D human pose estimation from monocular images, achieving state-of-the-art results without requiring multi-person 3D datasets.

Contribution

The paper presents HG-RCNN, a novel two-stage, Mask-RCNN based architecture that estimates 2D keypoints and lifts them to 3D in camera coordinates without needing 3D annotated data.

Findings

Achieves state-of-the-art results on MuPoTS-3D dataset.

Does not require multi-person 3D pose datasets.

Provides accurate 3D pose estimation in camera coordinates.

Abstract

Multi-person 3D human pose estimation from a single image is a challenging problem, especially for in-the-wild settings due to the lack of 3D annotated data. We propose HG-RCNN, a Mask-RCNN based network that also leverages the benefits of the Hourglass architecture for multi-person 3D Human Pose Estimation. A two-staged approach is presented that first estimates the 2D keypoints in every Region of Interest (RoI) and then lifts the estimated keypoints to 3D. Finally, the estimated 3D poses are placed in camera-coordinates using weak-perspective projection assumption and joint optimization of focal length and root translations. The result is a simple and modular network for multi-person 3D human pose estimation that does not require any multi-person 3D pose dataset. Despite its simple formulation, HG-RCNN achieves the state-of-the-art results on MuPoTS-3D while also approximating the 3D pose in the camera-coordinate system.