Related Experiment Video
Updated: Aug 25, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
4.0K
Identity and Gender Recognition Using a Capacitive Sensing Floor and Neural Networks
Daniel Konings1, Fakhrul Alam1, Nathaniel Faulkner1
1Department of Mechanical & Electrical Engineering (MEE), School of Food & Advanced Technology (SF&AT), Massey University, Auckland 0632, New Zealand.
Sensors (Basel, Switzerland)
|October 14, 2022
Summary
Capacitive sensing floors can identify individuals and their gender using walking patterns. Advanced deep learning models like Bi-directional Long Short-Term Memory (BLSTM) achieved 98.12% accuracy for subject recognition.
Area of Science:
- Computer Science
- Biomedical Engineering
- Human-Computer Interaction
Background:
- Capacitive sensing floors offer unobtrusive individual localization.
- Gait analysis from floor sensors is an emerging biometric identification method.
Purpose of the Study:
- To investigate the potential of capacitive sensing floors for subject and gender recognition based on walking characteristics.
- To develop and compare neural network-based machine learning algorithms for this task.
Main Methods:
- Development of several neural network algorithms, including Deep Neural Networks (DNNs), Bi-directional Long Short-Term Memory (BLSTM), and Convolutional Neural Networks (CNNs).
- Training and validation using a dataset of walking patterns from 23 subjects on a capacitive sensing floor.
- Benchmarking against Support Vector Machine (SVM) for performance comparison.
Main Results:
- A BLSTM network achieved the highest accuracy for identity recognition at 98.12%.
- A CNN model demonstrated superior performance for gender recognition with 93.3% accuracy.
- Most neural network models outperformed SVM in accuracy metrics for floor-based recognition tasks.
Conclusions:
- Capacitive sensing floors, combined with advanced machine learning, can effectively recognize individuals and their gender based on gait.
- Deep learning approaches, particularly BLSTM and CNNs, show significant promise for enhancing biometric identification systems using unobtrusive sensing floors.
