Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GesturePrint: Enabling User Identification for mmWave-based Gesture Recognition Systems

Lilin Xu, Keyi Wang, Chaojie Gu, Xiuzhen Guo, Shibo He, Jiming Chen

TL;DR

GesturePrint addresses the lack of user identification in mmWave gesture recognition by introducing a one-stop system that jointly performs gesture recognition and user identification on a commodity mmWave radar. It combines an efficient data preprocessing pipeline with GesIDNet, an attention-enhanced, multilevel feature fusion network tailored for sparse gesture point clouds, enabling both tasks from the same data. Across a self-collected ASL-gesture dataset and three public datasets, GesturePrint achieves gesture recognition accuracies near 98% and gesture-based user identification accuracies around 99%, with an average EER below 1% across datasets. The approach demonstrates strong distance robustness, speed-variation tolerance, and practical edge deployment potential, making mmWave-based gesture interactions more personalized and scalable for real-world smart devices and IoT systems.

Abstract

The millimeter-wave (mmWave) radar has been exploited for gesture recognition. However, existing mmWave-based gesture recognition methods cannot identify different users, which is important for ubiquitous gesture interaction in many applications. In this paper, we propose GesturePrint, which is the first to achieve gesture recognition and gesture-based user identification using a commodity mmWave radar sensor. GesturePrint features an effective pipeline that enables the gesture recognition system to identify users at a minor additional cost. By introducing an efficient signal preprocessing stage and a network architecture GesIDNet, which employs an attention-based multilevel feature fusion mechanism, GesturePrint effectively extracts unique gesture features for gesture recognition and personalized motion pattern features for user identification. We implement GesturePrint and collect data from 17 participants performing 15 gestures in a meeting room and an office, respectively. GesturePrint achieves a gesture recognition accuracy (GRA) of 98.87% with a user identification accuracy (UIA) of 99.78% in the meeting room, and 98.22% GRA with 99.26% UIA in the office. Extensive experiments on three public datasets and a new gesture dataset show GesturePrint's superior performance in enabling effective user identification for gesture recognition systems.

GesturePrint: Enabling User Identification for mmWave-based Gesture Recognition Systems

TL;DR

GesturePrint addresses the lack of user identification in mmWave gesture recognition by introducing a one-stop system that jointly performs gesture recognition and user identification on a commodity mmWave radar. It combines an efficient data preprocessing pipeline with GesIDNet, an attention-enhanced, multilevel feature fusion network tailored for sparse gesture point clouds, enabling both tasks from the same data. Across a self-collected ASL-gesture dataset and three public datasets, GesturePrint achieves gesture recognition accuracies near 98% and gesture-based user identification accuracies around 99%, with an average EER below 1% across datasets. The approach demonstrates strong distance robustness, speed-variation tolerance, and practical edge deployment potential, making mmWave-based gesture interactions more personalized and scalable for real-world smart devices and IoT systems.

Abstract

The millimeter-wave (mmWave) radar has been exploited for gesture recognition. However, existing mmWave-based gesture recognition methods cannot identify different users, which is important for ubiquitous gesture interaction in many applications. In this paper, we propose GesturePrint, which is the first to achieve gesture recognition and gesture-based user identification using a commodity mmWave radar sensor. GesturePrint features an effective pipeline that enables the gesture recognition system to identify users at a minor additional cost. By introducing an efficient signal preprocessing stage and a network architecture GesIDNet, which employs an attention-based multilevel feature fusion mechanism, GesturePrint effectively extracts unique gesture features for gesture recognition and personalized motion pattern features for user identification. We implement GesturePrint and collect data from 17 participants performing 15 gestures in a meeting room and an office, respectively. GesturePrint achieves a gesture recognition accuracy (GRA) of 98.87% with a user identification accuracy (UIA) of 99.78% in the meeting room, and 98.22% GRA with 99.26% UIA in the office. Extensive experiments on three public datasets and a new gesture dataset show GesturePrint's superior performance in enabling effective user identification for gesture recognition systems.
Paper Structure (25 sections, 3 equations, 15 figures, 2 tables)

This paper contains 25 sections, 3 equations, 15 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. mmWave-based Gesture Recognition
  4. mmWave-based User Identification
  5. Preliminaries
  6. GesturePrint Design
  7. System Overview
  8. Data Preprocessing Stage
  9. Classification Stage
  10. Implementation
  11. Performance Evaluation
  12. Datasets and Methodology
  13. Gesture Datasets
  14. Baseline
  15. Evaluation Metrics
  16. ...and 10 more sections

Figures (15)

  • Figure 1: Potential applications of mmWave-based gesture recognition systems with user identification capability. The identification capability can significantly improve the user experience in interacting with smart devices.
  • Figure 2: The visualization of gesture point clouds obtained from User A and User B when they perform ASL gestures.
  • Figure 3: Differences among point clouds are measured by three metrics: Hausdorff distance (HD), Chamfer distance (CD), and Jensen-Shannon divergence (JSD).
  • Figure 4: Overview of GesturePrint: a system that enables mmWave-based gesture recognition and gesture-based user identification, comprising the data preprocessing and classification stages.
  • Figure 5: The architecture of GesIDNet. MLP: Multilayer Perceptron, FC: Fully Connected Layer, RB: Resizing Block.
  • ...and 10 more figures