Efficient Gait-Based Person Identification Using a Lightweight Deep Learning Model on Edge Devices

Core Concepts

A highly efficient and accurate deep learning model for real-time gait-based person identification that can be deployed on low-power edge devices.

Abstract

This paper presents a lightweight convolutional neural network (CNN) model for real-time person identification using gait analysis. The key highlights are:

Dataset: The authors used the public whuGAIT dataset and augmented it with data collected from the four team members, resulting in a total of 24 classes.

Feature Extraction: The authors used spectral analysis to extract features from the raw accelerometer and gyroscope data, leveraging the richer information in the gyroscope data compared to the accelerometer.

Model Architecture: The authors designed a 4-layer CNN model that achieves 96.7% accuracy on the test set, while consuming only 5KB of RAM and 70ms inference time on an Arduino Nano 33 BLE Sense board.

Deployment: The authors successfully demonstrated real-time person identification using the model running on the Arduino board, as well as on a smartphone. They also converted the CNN model to a spiking neural network (SNN) and deployed it on the BrainChip Akida neuromorphic processor.

Challenges and Iterations: The authors faced challenges in hyperparameter tuning, data collection, and deployment, requiring numerous iterations to arrive at the final solution.

Advantages: The edge-based approach offers several benefits, including reduced bandwidth, low latency, cost-effectiveness, reliability, and privacy preservation.

Ethical Considerations: The authors discuss the importance of addressing privacy, informed consent, and bias in the development and deployment of such gait-based identification systems.

Overall, this work presents a highly efficient and practical solution for real-time person identification using gait analysis, with the potential for deployment in various applications, such as surveillance, security, and healthcare.

Stats

Each person has a unique gait, influenced by factors like anatomy, musculoskeletal structure, and personal habits.
The final model achieves 96.7% accuracy and consumes only 5KB RAM with an inferencing time of 70 ms and 125mW power, while running continuous inference on Arduino Nano 33 BLE Sense.

Quotes

"Every person has a distinct gait, influenced by factors like anatomy, musculoskeletal structure, and personal habits. This distinctiveness makes gait analysis an intriguing and effective tool for identifying individuals in diverse settings, ranging from surveillance and security to healthcare and rehabilitation."
"Our model achieves 96.7% accuracy and consumes only 5KB RAM with an inferencing time of 70 ms and 125mW power, while running continuous inference on Arduino Nano 33 BLE Sense."

Key Insights Distilled From

Realtime Person Identification via Gait Analysis

by Shanmuga Ven... at arxiv.org 04-25-2024

https://arxiv.org/pdf/2404.15312.pdf

Deeper Inquiries

How can the proposed gait-based identification system be extended to accommodate individuals with physical disabilities or atypical gaits?

To extend the gait-based identification system to accommodate individuals with physical disabilities or atypical gaits, several considerations and adaptations can be implemented:

Diverse Dataset Inclusion: Collecting data from individuals with physical disabilities or atypical gaits is crucial to ensure the model's robustness and inclusivity. By incorporating a diverse range of gait patterns, the model can learn to recognize and differentiate between various walking styles.

Customized Feature Extraction: Implementing customized feature extraction techniques that focus on specific characteristics of atypical gaits can enhance the model's ability to identify individuals with physical disabilities. This may involve capturing unique movement patterns or anomalies in the gait cycle that are specific to certain conditions.

Adaptive Model Training: Training the model with data augmentation techniques that simulate variations in gait patterns can help improve its adaptability to different walking styles. By exposing the model to a wide range of gait variations, including those from individuals with physical disabilities, it can learn to generalize better and make accurate identifications.

Incorporating Sensor Fusion: Utilizing multiple sensors, such as pressure sensors or wearable cameras, in addition to IMU sensors, can provide a more comprehensive view of an individual's gait. This multi-modal approach can capture additional information about gait characteristics, aiding in the identification of individuals with physical disabilities or atypical gaits.

Continuous Model Evaluation and Feedback: Implementing a feedback loop mechanism that allows the model to continuously learn and adapt based on real-world data from individuals with physical disabilities can enhance its performance over time. Regular model evaluation and updates based on feedback from users with diverse gait patterns are essential for maintaining accuracy and reliability.

What are the potential limitations and vulnerabilities of using gait analysis for person identification, and how can they be addressed?

While gait analysis offers unique advantages for person identification, it also presents certain limitations and vulnerabilities that need to be addressed:

Environmental Factors: Gait analysis can be influenced by environmental conditions such as terrain, lighting, and obstacles, leading to variations in gait patterns. To address this, the system should be designed to account for environmental factors and adapt to different walking conditions to ensure accurate identification.

Privacy Concerns: Gait analysis raises privacy concerns as it involves collecting biometric data that can be sensitive. Implementing robust data encryption, anonymization techniques, and secure storage protocols can help mitigate privacy risks and ensure the confidentiality of user information.

Impersonation and Spoofing: Gait patterns can potentially be mimicked or spoofed, posing a risk of impersonation. To address this vulnerability, incorporating multi-factor authentication methods, such as combining gait analysis with other biometric modalities or behavioral traits, can enhance security and prevent unauthorized access.

Biometric Variability: Gait patterns can vary due to factors like fatigue, injury, or changes in walking speed, leading to inconsistencies in identification. Regular recalibration of the model, continuous monitoring of gait data, and adaptive learning algorithms can help account for biometric variability and improve the system's accuracy.

Bias and Fairness: Biases in the training data or algorithmic decisions can result in unfair treatment or inaccurate identifications, particularly for individuals with diverse gait patterns. To address bias and ensure fairness, it is essential to use representative and diverse datasets, conduct bias assessments, and implement bias mitigation strategies during model development and deployment.

How can the integration of gait-based identification with other biometric modalities, such as facial recognition or voice analysis, enhance the overall security and reliability of authentication systems?

Integrating gait-based identification with other biometric modalities like facial recognition or voice analysis can offer several benefits in terms of security and reliability:

Multi-Factor Authentication: Combining gait analysis with facial recognition or voice analysis creates a multi-factor authentication system that enhances security by requiring multiple biometric identifiers for user verification. This multi-modal approach increases the complexity of authentication, making it more robust against unauthorized access.

Improved Accuracy and Confidence: Integrating multiple biometric modalities allows for cross-validation of user identity, increasing the overall accuracy and confidence in authentication decisions. By corroborating gait patterns with facial features or voice characteristics, the system can achieve higher levels of reliability in identifying individuals.

Anti-Spoofing Measures: Leveraging different biometric modalities helps in implementing anti-spoofing measures to detect and prevent fraudulent attempts to deceive the system. By combining gait analysis with facial or voice recognition, the system can detect inconsistencies or anomalies that may indicate spoofing attacks, enhancing security.

Enhanced User Experience: Integrating diverse biometric modalities provides flexibility for users to choose the most convenient and accessible authentication method based on the context. This enhances the user experience by offering multiple options for identity verification, catering to individual preferences and accessibility needs.

Resilience to Biometric Variability: Combining gait-based identification with other biometric modalities can improve the system's resilience to biometric variability and environmental factors. In cases where gait analysis alone may be challenging, the integration with facial recognition or voice analysis can provide alternative means of authentication, ensuring reliable identification across different scenarios.