Real-time 2D Multi-Person Pose Estimation on CPU: Lightweight OpenPose

TL;DR

This paper presents a lightweight, real-time multi-person pose estimation method optimized for CPU edge devices, achieving high frame rates with significantly reduced model size and computational complexity.

Contribution

It introduces a compact, efficient network design based on OpenPose that runs in real-time on CPU hardware, with optimized post-processing and integration into the OpenVINO toolkit.

Findings

Runs at 28 fps on Intel NUC and 26 fps on i7 CPU

Model has 4.1 million parameters and 9 GFLOPs complexity

Achieves similar quality to baseline OpenPose with much fewer resources

Abstract

In this work we adapt multi-person pose estimation architecture to use it on edge devices. We follow the bottom-up approach from OpenPose, the winner of COCO 2016 Keypoints Challenge, because of its decent quality and robustness to number of people inside the frame. With proposed network design and optimized post-processing code the full solution runs at 28 frames per second (fps) on Intel$\unicode{xAE}$ NUC 6i7KYB mini PC and 26 fps on Core$^{TM}$ i7-6850K CPU. The network model has 4.1M parameters and 9 billions floating-point operations (GFLOPs) complexity, which is just ~15% of the baseline 2-stage OpenPose with almost the same quality. The code and model are available as a part of Intel$\unicode{xAE}$ OpenVINO$^{TM}$ Toolkit.