Jiya Adama Enoch, Ilesanmi Banjo Oluwafemi, Olulope K. Paul, F. Ibikunle, Osaji Emmanuel, Ariba Folashade Olamide
{"title":"A Comparative Performance Study of Support Vector Machine, KNN, and Ensemble Classifiers on through-wall human detection Dataset","authors":"Jiya Adama Enoch, Ilesanmi Banjo Oluwafemi, Olulope K. Paul, F. Ibikunle, Osaji Emmanuel, Ariba Folashade Olamide","doi":"10.1109/SEB-SDG57117.2023.10124476","DOIUrl":null,"url":null,"abstract":"There is a chance that many injured people will be present in the disaster area when natural disasters happen. To lower the death rate, rescue workers must provide these injured people with assistance as soon as they can. In search and rescue (SAR) operations, distance estimation, position determination, and classification criteria are all equally crucial, the accuracy of each would be impacted by a decrease in the other. Rescue crews have used a variety of methods and approaches to anticipate and find victims in unsafe, collapsed building structures. Classification algorithms have shown a dynamic capacity to acquire key dataset properties by utilizing just a few sample sets, thanks to the recent development of machine learning methods. The detection of a human target's state behind a wall in diverse samples is the main topic of this work. Here, SVM, KNN, and Ensemble algorithms are selected in the machine learning model to categorize and recognize human targets behind walls. The classification algorithms can derive clear data-feature representations by automatically learning about the dataset's innate characteristics. The performance of the classifier was used to extract more effective feature representations. The classification and identification of the behind-the-wall human-target states were separately carried out, and then the results were compared with those of other classification algorithms. The outcome demonstrates that the use of KNN with 85% accuracy outperforms other classifiers (SVM 81% and Ensemble 81%) and is more effective for the prediction of human targets behind walls. This work may help with human identification during search and rescue efforts after a disaster.","PeriodicalId":185729,"journal":{"name":"2023 International Conference on Science, Engineering and Business for Sustainable Development Goals (SEB-SDG)","volume":"158 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 International Conference on Science, Engineering and Business for Sustainable Development Goals (SEB-SDG)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SEB-SDG57117.2023.10124476","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
There is a chance that many injured people will be present in the disaster area when natural disasters happen. To lower the death rate, rescue workers must provide these injured people with assistance as soon as they can. In search and rescue (SAR) operations, distance estimation, position determination, and classification criteria are all equally crucial, the accuracy of each would be impacted by a decrease in the other. Rescue crews have used a variety of methods and approaches to anticipate and find victims in unsafe, collapsed building structures. Classification algorithms have shown a dynamic capacity to acquire key dataset properties by utilizing just a few sample sets, thanks to the recent development of machine learning methods. The detection of a human target's state behind a wall in diverse samples is the main topic of this work. Here, SVM, KNN, and Ensemble algorithms are selected in the machine learning model to categorize and recognize human targets behind walls. The classification algorithms can derive clear data-feature representations by automatically learning about the dataset's innate characteristics. The performance of the classifier was used to extract more effective feature representations. The classification and identification of the behind-the-wall human-target states were separately carried out, and then the results were compared with those of other classification algorithms. The outcome demonstrates that the use of KNN with 85% accuracy outperforms other classifiers (SVM 81% and Ensemble 81%) and is more effective for the prediction of human targets behind walls. This work may help with human identification during search and rescue efforts after a disaster.