Fast and Scalable Human Pose Estimation using mmWave Point Cloud

TL;DR

This paper introduces FUSE, a fast and scalable framework for human pose estimation using mmWave radar point clouds, leveraging multi-frame data and meta-learning to improve adaptability and accuracy in low-data, high-sparsity scenarios.

Contribution

The paper presents a novel FUSE framework that combines multi-frame representation and meta-learning to enhance mmWave-based human pose estimation, achieving faster adaptation and competitive accuracy.

Findings

FUSE adapts to unseen scenarios 4 times faster than supervised methods.

Achieves approximately 7 cm mean absolute error in joint coordinate estimation.

Demonstrates effective handling of sparse mmWave point cloud data.

Abstract

Millimeter-Wave (mmWave) radar can enable high-resolution human pose estimation with low cost and computational requirements. However, mmWave data point cloud, the primary input to processing algorithms, is highly sparse and carries significantly less information than other alternatives such as video frames. Furthermore, the scarce labeled mmWave data impedes the development of machine learning (ML) models that can generalize to unseen scenarios. We propose a fast and scalable human pose estimation (FUSE) framework that combines multi-frame representation and meta-learning to address these challenges. Experimental evaluations show that FUSE adapts to the unseen scenarios 4$\times$ faster than current supervised learning approaches and estimates human joint coordinates with about 7 cm mean absolute error.