PedRecNet: Multi-task deep neural network for full 3D human pose and orientation estimation

TL;DR

PedRecNet is a multi-task deep learning model capable of estimating 2D and 3D human poses, as well as body and head orientations, using full body images, with performance comparable to current state-of-the-art methods.

Contribution

This paper introduces a simple yet effective multi-task neural network that jointly estimates human pose and orientation from full body images, reducing reliance on face recognition.

Findings

Performance on 3D pose and orientation estimation is comparable to state-of-the-art.

Simulation data enhances training for pose and orientation tasks.

The network architecture is adaptable and easy to extend.

Abstract

We present a multitask network that supports various deep neural network based pedestrian detection functions. Besides 2D and 3D human pose, it also supports body and head orientation estimation based on full body bounding box input. This eliminates the need for explicit face recognition. We show that the performance of 3D human pose estimation and orientation estimation is comparable to the state-of-the-art. Since very few data sets exist for 3D human pose and in particular body and head orientation estimation based on full body data, we further show the benefit of particular simulation data to train the network. The network architecture is relatively simple, yet powerful, and easily adaptable for further research and applications.