Rantelinggi Parma Hadi, Shi Xintong, Bouazizi Mondher, Ohtsuki Tomoaki
Graduate School of Science and Technology, Keio University, Yokohama 223-8522, Japan.
Department of Computer Engineering, Universitas Papua, Manokwari 98314, Indonesia.
Sensors (Basel). 2025 Jun 23;25(13):3909. doi: 10.3390/s25133909.
Human skeleton estimation using Frequency-Modulated Continuous Wave (FMCW) radar is a promising approach for privacy-preserving motion analysis. However, the existing methods struggle with sparse radar point cloud data, leading to inaccuracies in joint localization. To address this challenge, we propose a novel deep learning framework integrating convolutional neural networks (CNNs), multi-head transformers, and Bi-LSTM networks to enhance spatiotemporal feature representations. Our approach introduces a frame concatenation strategy that improves data quality before processing through the neural network pipeline. Experimental evaluations on the MARS dataset demonstrate that our model outperforms conventional methods by significantly reducing estimation errors, achieving a mean absolute error (MAE) of 1.77 cm and a root mean squared error (RMSE) of 2.92 cm while maintaining computational efficiency.
使用调频连续波（FMCW）雷达进行人体骨骼估计是一种用于隐私保护运动分析的有前途的方法。然而，现有方法在稀疏雷达点云数据方面存在困难，导致关节定位不准确。为了应对这一挑战，我们提出了一种新颖的深度学习框架，该框架集成了卷积神经网络（CNN）、多头变换器和双向长短期记忆（Bi-LSTM）网络，以增强时空特征表示。我们的方法引入了一种帧拼接策略，在通过神经网络管道进行处理之前提高数据质量。在MARS数据集上的实验评估表明，我们的模型通过显著降低估计误差优于传统方法，在保持计算效率的同时，实现了1.77厘米的平均绝对误差（MAE）和2.92厘米的均方根误差（RMSE）。
