From Neck to Head: Bio-Impedance Sensing for Head Pose Estimation

2025-07-17

AI Summary

Visual head tracking is limited by line-of-sight requirements and susceptibility to occlusion. To address this, we propose NeckSense—a necklace-style wearable system leveraging multi-channel bioimpedance sensing to estimate head pose without visual constraints, by monitoring impedance variations in neck tissues. This work introduces bioimpedance sensing to head pose estimation for the first time and innovatively incorporates anatomical priors—specifically, cervical kinematic constraints—into the deep learning loss function to enhance model robustness and physiological plausibility. The system employs flexible dry electrodes, low-power multi-channel signal acquisition, and jointly optimized modeling. Evaluated on seven subjects via leave-one-subject-out cross-validation, NeckSense achieves a mean vertex error of 25.9 mm—comparable to state-of-the-art vision-based methods. It establishes a new paradigm for compact, privacy-preserving, and全天候 (all-weather) head tracking.

Technology Category

• Humans and AI: Brain-Sensing and Analysis
• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

We present NeckSense, a novel wearable system for head pose tracking that leverages multi-channel bio-impedance sensing with soft, dry electrodes embedded in a lightweight, necklace-style form factor. NeckSense captures dynamic changes in tissue impedance around the neck, which are modulated by head rotations and subtle muscle activations. To robustly estimate head pose, we propose a deep learning framework that integrates anatomical priors, including joint constraints and natural head rotation ranges, into the loss function design. We validate NeckSense on 7 participants using the current SOTA pose estimation model as ground truth. Our system achieves a mean per-vertex error of 25.9 mm across various head movements with a leave-one-person-out cross-validation method, demonstrating that a compact, line-of-sight-free bio-impedance wearable can deliver head-tracking performance comparable to SOTA vision-based methods.

Problem

• Develop wearable system for head pose tracking
• Estimate head pose using bio-impedance sensing
• Achieve performance comparable to vision-based methods

Innovation

• Multi-channel bio-impedance sensing with dry electrodes
• Deep learning framework with anatomical priors
• Compact wearable for line-of-sight-free head tracking