Related Experiment Video
Updated: Apr 25, 2026
06:17
Deep-Learning Based Multi-Joint Synchronous Tracking for Objective Quantification of Hindlimb Locomotor Kinematics in Rats
Published on: April 3, 2026
105
Development of vision based multiview gait recognition system with MMUGait database.
Hu Ng1, Wooi-Haw Tan1, Junaidi Abdullah1
1Faculty of Computing and Informatics, Multimedia University, 63100 Cyberjaya, Malaysia.
Thescientificworldjournal
|August 22, 2014
Summary
A new gait recognition system achieves over 90% accuracy using the MMUGait database. This multiview model-based approach effectively handles various walking conditions and occlusions, outperforming existing methods.
Area of Science:
- Biometrics
- Computer Vision
- Pattern Recognition
Background:
- Gait recognition is a challenging biometric modality due to variations in walking patterns and environmental factors.
- Existing gait recognition systems often struggle with occlusions and diverse walking trajectories.
Purpose of the Study:
- To introduce the MMUGait database, a new resource for gait recognition research.
- To propose and evaluate a novel multiview model-based gait recognition system with joint detection.
Main Methods:
- Developed the MMUGait database with 82 subjects under normal conditions and 19 under covariate factors, captured from two views.
- Proposed a multiview model-based system employing joint detection for gait recognition.
- Implemented silhouette enhancement, joint angular trajectory determination, feature extraction (crotch height, step-size), smoothing, normalization, and feature selection.
Main Results:
- The proposed system achieved a correct classification rate exceeding 90% on both the MMUGait and SOTON Small DB datasets.
- The system demonstrated robust performance across different walking trajectories and covariate factors, including occlusions.
- The approach outperformed other methods on the SOTON Small DB in most experimental cases.
Conclusions:
- The proposed multiview model-based gait recognition system is effective and robust, particularly under challenging conditions.
- The MMUGait database provides a valuable resource for advancing gait recognition research.
- The joint detection approach enhances performance in real-world scenarios with occlusions and trajectory variations.
