OriNet: A Fully Convolutional Network for 3D Human Pose Estimation

TL;DR

This paper introduces OriNet, a fully convolutional network that estimates 3D human poses from monocular images using limb orientations and bounding boxes, achieving strong results and good generalization.

Contribution

The paper presents a novel approach combining limb orientations with bounding boxes for 3D pose estimation, improving accuracy and robustness without extra constraints.

Findings

Achieves good results on large-scale benchmarks.

Generalizes well to new scenes.

Robust to bounding box inaccuracies.

Abstract

In this paper, we propose a fully convolutional network for 3D human pose estimation from monocular images. We use limb orientations as a new way to represent 3D poses and bind the orientation together with the bounding box of each limb region to better associate images and predictions. The 3D orientations are modeled jointly with 2D keypoint detections. Without additional constraints, this simple method can achieve good results on several large-scale benchmarks. Further experiments show that our method can generalize well to novel scenes and is robust to inaccurate bounding boxes.