Gait: a new fingerprint ?
This repository houses a project centered around gait analysis and recognition, a fascinating area of biometric recognition. Gait refers to a person’s manner of walking, and is as individual as a fingerprint. Our project uses data collected from accelerometer sensors and a variety of computational techniques to recognize individuals based on their unique gait patterns.
Project Overview
This project began by exploring the importance of gait. Delving into the history of gait research, we focused on distinct methods of biometric gait recognition, including machine vision (MV) based, floor sensor (FS) based, and wearable sensor (WS) based techniques. Our project underscores the advantages of WS-based methods, which offer the benefit of being unobtrusive and flexible in terms of sensor placement.
Our work heavily relies on the Actibelt device, a high-tech 3D-accelerometer tucked away in a belt buckle. This device has been previously employed to examine changes in real-world walking speeds in patients with multiple sclerosis. We collected and analyzed data from the accelerometer, yielding visual plots that highlight the unique aspects of an individual's gait.
Repository Structure
Presentation/: Houses the final presentation of our project.Reading/: Stores a collection of research materials referenced throughout the project.Scripts and Plots/: Contains all source codes, documentation, and figures related to the project.
Future Work
Gait recognition holds vast potential in several areas:
- Biometric User Identification: Gait recognition could serve as a biometric identification method on smartphones.
- Security and Surveillance: As a non-invasive biometric recognition method, gait recognition could offer an alternative to face recognition techniques.
- Medical Applications: Identifying and monitoring gait abnormalities could assist in the diagnosis