Related Experiment Video
Updated: Jun 23, 2025
05:41
A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
Published on: February 6, 2020
9.4K
A Novel Multi-Scaled Deep Convolutional Structure for Punctilious Human Gait Authentication.
Reem N Yousef1, Mohamed Maher Ata2,3, Amr E Eldin Rashed4
1Delta Higher Institute for Engineering and Technology, Mansoura 35681, Egypt.
Biomimetics (Basel, Switzerland)
|June 26, 2024
Summary
A new deep learning model, Multi-Scaled Deep Convolutional Structure for Punctilious Human Gait Authentication (MSDCS-PHGA), enhances human gait recognition accuracy. This system offers a contactless solution for secure identification, achieving over 99.8% accuracy.
Area of Science:
- Computer Science
- Biometrics
- Artificial Intelligence
Background:
- The COVID-19 pandemic highlighted the need for non-interactive human recognition systems.
- Contactless biometric solutions are crucial for maintaining user safety and equipment isolation.
Purpose of the Study:
- To introduce a novel Multi-Scaled Deep Convolutional Structure for Punctilious Human Gait Authentication (MSDCS-PHGA).
- To enhance the accuracy and efficiency of human gait recognition using a deep convolutional neural network (CNN).
Main Methods:
- The MSDCS-PHGA method segments, preprocesses, and resizes silhouette images into three scales.
- Gait features are extracted using custom convolutional layers and fused into an integrated feature set.
- A deep CNN architecture is employed for gait authentication.
Main Results:
- The proposed MSDCS-PHGA model achieved high accuracy across three benchmark datasets: CASIA (99.9%), OU-ISIR (99.9%), and OU-MVLP (99.8%).
- The model demonstrated superior performance compared to existing pre-trained models in gait recognition.
- Training time was minimal, around 3 minutes, indicating computational efficiency.
Conclusions:
- The multi-scaled deep convolutional approach significantly improves gait recognition accuracy.
- MSDCS-PHGA offers a highly accurate and efficient solution for contactless human authentication.
- The developed CNN model effectively addresses the demand for non-interactive biometric systems.
