RadProPoser: Probabilistic Radar Tensor Human Pose Estimation That Knows Its Limits

TL;DR

RadProPoser is a probabilistic radar-based human pose estimation framework that quantifies uncertainty, achieves high accuracy, and operates in real-time, suitable for privacy-preserving motion tracking.

Contribution

It introduces a novel end-to-end probabilistic model with spectral attention for radar tensor data, providing uncertainty estimates and extending easily to multi-radar setups.

Findings

Achieves 6.425 cm mean per-joint position error on a new benchmark.

Provides calibrated total uncertainty with an expected calibration error of 0.027.

Runs at 89 FPS on an NVIDIA RTX 3090, surpassing radar frame rate.

Abstract

Radar-based human pose estimation enables privacy-preserving motion tracking for ambient intelligence, yet the noisy nature of radar sensing makes uncertainty quantification essential. We present RadProPoser, an end-to-end probabilistic framework that predicts three-dimensional body joints with per-joint uncertainties from raw radar tensor data. Using a variational encoder-decoder with spectral attention that fuses real and imaginary radar components across temporal frames, we model aleatoric uncertainty through learnable Gaussian and Laplace distributions. Trained on a new benchmark dataset with optical motion-capture ground truth, our method achieves 6.425 cm mean per-joint position error. The model outputs per-joint aleatoric uncertainties, and isotonic recalibration yields calibrated total uncertainty with expected calibration error of 0.027. Since spectral attention operates on individual radar tensor components, extending to multi-radar configurations requires only concatenating additional input streams. On the HuPR benchmark with dual orthogonal radars, this achieves 5.042 cm MPJPE. The framework runs at 89 frames per second (FPS) on an NVIDIA RTX 3090, exceeding the 15 Hz radar frame rate.