2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features
| Format: | General Document |
|---|
| _version_ | 1860798287640002560 |
|---|---|
| building | INTELEK Repository |
| collection | Online Access |
| collectionurl | https://intelek.unisza.edu.my/intelek/pages/search.php?search=!collection3 |
| copyright | Copyright©PWB2025 |
| country | Malaysia |
| date | 2024-10-29 10:34 |
| format | General Document |
| id | 16818 |
| institution | UniSZA |
| originalfilename | 16818_9544daa5fe4740e.pdf |
| person | Muhammad Shazmil Bin Mohd Sabilan |
| recordtype | oai_dc |
| resourceurl | https://intelek.unisza.edu.my/intelek/pages/view.php?ref=16818 |
| sourcemedia | Server storage Scanned document |
| spelling | 16818 https://intelek.unisza.edu.my/intelek/pages/view.php?ref=16818 https://intelek.unisza.edu.my/intelek/pages/search.php?search=!collection3 General Document Malaysia Library Staff (Top Management) Library Staff (Management) Library Staff (Support) Terengganu Faculty of Informatics & Computing English application/pdf 1.5 Microsoft® Word 2016 125 Server storage Scanned document Universiti Sultan Zainal Abidin UniSZA Private Access Universiti Sultan Zainal Abidin Copyright©PWB2025 Pattern Recognition Pattern recognition systems Feature Extraction Dissertations, Academic Muhammad Shazmil Bin Mohd Sabilan Nowadays, the gender classification system has been spoofed by individuals who change and pretend to be of the opposite gender. Gender spoofing can cause various problems, particularly in terms of security, such as criminal activity. Moreover, existing identification methods such as facial recognition and handwriting analysis have been called into question for their reliability and lack of accuracy. Therefore, a Gender Spoofing Detection Using Gait Energy Image Spatial Based Features is proposed to detect the original gender. Since there is no gait spoof dataset, a new dataset called Spoof Gait Dataset (SpooGa) has been developed. The SpooGa dataset contains data on the individual's walking style in accordance with the study's requirements, and the data is recorded as a depth image.The proposed method consists of three phases: preprocessing, feature extraction, and classification. Data in SpooGa undergoes the preprocessing phase first, and then the Gait Energy Image (GEI) is computed. Feature extraction begins by dividing the GEI into three parts, of which only two parts are selected, i.e., the body and lower body since the feature extraction involves Hand, Leg, and Feet distances. After that, the PCA algorithm is applied to the GEI, and features are extracted using fusion features of leg, toe, and hand swing distances (LETH) formula from the reconstructed image.Linear Support Vector Machine (Linear SVM), Fine Tree, and Weighted K-Nearest Neighbor Classifier (Weighted KNN) were used separately as the classification method. The features are divided into two parts which are training and testing. 70% of the feature data is used for training, and 30% is used for testing purposes. Evaluation is measured by seeing if the proposed method can identify the original gender of an individual after disguise.The proposed method achieves high accuracy rates in identifying an individual's original gender even after disguise. The accuracy rates achieved by the Linear SVM, Weighted KNN, and Fine Tree classifiers were 92.30%, 96.15%, and 92.30%, respectively. Receiver Operating Characteristic (ROC) curves were plotted to evaluate the performance of each classifier, with the Area Under the Curve (AUC) indicating high discrimination ability, as observed with AUC values of 0.97 for Linear SVM, 0.98 for Weighted KNN, and 0.89 for Fine Tree classifiers.The proposed method is capable of detecting any attempt of an individual who tries to imitate someone else's way of walking and addressing security risks associated with it. The development of the SpooGa dataset and the use of the LETH formula contribute to the novelty of the study. Further research is needed to validate the results and explore the potential applications of this method in real-world settings. 2024-10-29 10:34 uuid:70d4bf7b-1dcd-41a8-83e1-b9d261c432a6 16818_9544daa5fe4740e.pdf Principal Component Analysis (PCA) 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features Gender Spoofing Gait Energy Image (GEI) Spoof Gait Dataset (SpooGa) Biometric Security Linear Support Vector Machine (SVM) Weighted K-Nearest Neighbor (KNN) Fine Tree Classifier Gait Analysis Gender Classification Biometry — Data processing Machine learning — Classification Gait in humans — Analysis Thesis |
| spellingShingle | 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features |
| state | Terengganu |
| subject | Pattern recognition systems Dissertations, Academic Biometry — Data processing Machine learning — Classification Gait in humans — Analysis |
| summary | Nowadays, the gender classification system has been spoofed by individuals who change and pretend to be of the opposite gender. Gender spoofing can cause various problems, particularly in terms of security, such as criminal activity. Moreover, existing identification methods such as facial recognition and handwriting analysis have been called into question for their reliability and lack of accuracy. Therefore, a Gender Spoofing Detection Using Gait Energy Image Spatial Based Features is proposed to detect the original gender. Since there is no gait spoof dataset, a new dataset called Spoof Gait Dataset (SpooGa) has been developed. The SpooGa dataset contains data on the individual's walking style in accordance with the study's requirements, and the data is recorded as a depth image.The proposed method consists of three phases: preprocessing, feature extraction, and classification. Data in SpooGa undergoes the preprocessing phase first, and then the Gait Energy Image (GEI) is computed. Feature extraction begins by dividing the GEI into three parts, of which only two parts are selected, i.e., the body and lower body since the feature extraction involves Hand, Leg, and Feet distances. After that, the PCA algorithm is applied to the GEI, and features are extracted using fusion features of leg, toe, and hand swing distances (LETH) formula from the reconstructed image.Linear Support Vector Machine (Linear SVM), Fine Tree, and Weighted K-Nearest Neighbor Classifier (Weighted KNN) were used separately as the classification method. The features are divided into two parts which are training and testing. 70% of the feature data is used for training, and 30% is used for testing purposes. Evaluation is measured by seeing if the proposed method can identify the original gender of an individual after disguise.The proposed method achieves high accuracy rates in identifying an individual's original gender even after disguise. The accuracy rates achieved by the Linear SVM, Weighted KNN, and Fine Tree classifiers were 92.30%, 96.15%, and 92.30%, respectively. Receiver Operating Characteristic (ROC) curves were plotted to evaluate the performance of each classifier, with the Area Under the Curve (AUC) indicating high discrimination ability, as observed with AUC values of 0.97 for Linear SVM, 0.98 for Weighted KNN, and 0.89 for Fine Tree classifiers.The proposed method is capable of detecting any attempt of an individual who tries to imitate someone else's way of walking and addressing security risks associated with it. The development of the SpooGa dataset and the use of the LETH formula contribute to the novelty of the study. Further research is needed to validate the results and explore the potential applications of this method in real-world settings. |
| title | 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features |
| title_full | 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features |
| title_fullStr | 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features |
| title_full_unstemmed | 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features |
| title_short | 2024_Gender Spoofing Detection Using Gait Energy Image Spatial Based Features |
| title_sort | 2024_gender spoofing detection using gait energy image spatial based features |