Gender classification based on gait features for healthy children / Nur Khalidah Zakaria
This thesis describes a representation of gait analysis for the purpose of gender classification of healthy children using the most significant gait features. In this study, the gait parameters were measured using 3D motion analysis system. In order to obtain the most significant gait features, nume...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | https://ir.uitm.edu.my/id/eprint/98985/1/98985.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This thesis describes a representation of gait analysis for the purpose of gender classification of healthy children using the most significant gait features. In this study, the gait parameters were measured using 3D motion analysis system. In order to obtain the most significant gait features, numerical analysis using statistical method was performed. The three gait parameters which were spatiotemporal, kinematics and kinetics parameters were tested. From the result, it is noted that the most significant difference between genders only existed at kinematic parameter. Out of 36 gait features in kinematic parameter, only four were found to be the most significant gait features. These features were added into the ANN as an input to classify gender of healthy children. The ANN networks were optimized by adjusting the number of hidden neurons and thresholds. Since the size of the original gait features was too small due to small sample size of healthy children, the generation of synthetic data was performed based on the original gait features data. The result from the performance measures show that synthetic data obtained better accuracy compared to original data, since ANN performed better with large amount of data. The ANN network for this study is 4-6-1 with 0.3 thresholds. The performance measures show that this network achieved 76%, 87% and 91.3% of accuracy for training, validation and testing respectivel |
|---|