Ngo Duong Ha, Nhu Y. Tran, Le Nhi Lam Thuy, Ikuko Shimizu, Pham The Bao
{"title":"视频中的暴力区域定位与校园暴力行为分类","authors":"Ngo Duong Ha, Nhu Y. Tran, Le Nhi Lam Thuy, Ikuko Shimizu, Pham The Bao","doi":"10.3389/fcomp.2023.1274928","DOIUrl":null,"url":null,"abstract":"Classification of school violence has been proven to be an effective solution for preventing violence within educational institutions. As a result, technical proposals aimed at enhancing the efficacy of violence classification are of considerable interest to researchers. This study explores the utilization of the SORT tracking method for localizing and tracking objects in videos related to school violence, coupled with the application of LSTM and GRU methods to enhance the accuracy of the violence classification model. Furthermore, we introduce the concept of a padding box to localize, identify actions, and recover tracked objects lost during video playback. The integration of these techniques offers a robust and efficient system for analyzing and preventing violence in educational environments. The results demonstrate that object localization and recovery algorithms yield improved violent classification outcomes compared to both the SORT tracking and violence classification algorithms alone, achieving an impressive accuracy rate of 72.13%. These experimental findings hold promise, especially in educational settings, where the assumption of camera stability is justifiable. This distinction is crucial due to the unique characteristics of violence in educational environments, setting it apart from other forms of violence.","PeriodicalId":52823,"journal":{"name":"Frontiers in Computer Science","volume":"29 1","pages":"0"},"PeriodicalIF":2.4000,"publicationDate":"2023-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Violence region localization in video and the school violent actions classification\",\"authors\":\"Ngo Duong Ha, Nhu Y. Tran, Le Nhi Lam Thuy, Ikuko Shimizu, Pham The Bao\",\"doi\":\"10.3389/fcomp.2023.1274928\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Classification of school violence has been proven to be an effective solution for preventing violence within educational institutions. As a result, technical proposals aimed at enhancing the efficacy of violence classification are of considerable interest to researchers. This study explores the utilization of the SORT tracking method for localizing and tracking objects in videos related to school violence, coupled with the application of LSTM and GRU methods to enhance the accuracy of the violence classification model. Furthermore, we introduce the concept of a padding box to localize, identify actions, and recover tracked objects lost during video playback. The integration of these techniques offers a robust and efficient system for analyzing and preventing violence in educational environments. The results demonstrate that object localization and recovery algorithms yield improved violent classification outcomes compared to both the SORT tracking and violence classification algorithms alone, achieving an impressive accuracy rate of 72.13%. These experimental findings hold promise, especially in educational settings, where the assumption of camera stability is justifiable. This distinction is crucial due to the unique characteristics of violence in educational environments, setting it apart from other forms of violence.\",\"PeriodicalId\":52823,\"journal\":{\"name\":\"Frontiers in Computer Science\",\"volume\":\"29 1\",\"pages\":\"0\"},\"PeriodicalIF\":2.4000,\"publicationDate\":\"2023-10-16\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Frontiers in Computer Science\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.3389/fcomp.2023.1274928\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Frontiers in Computer Science","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.3389/fcomp.2023.1274928","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"COMPUTER SCIENCE, INTERDISCIPLINARY APPLICATIONS","Score":null,"Total":0}
Violence region localization in video and the school violent actions classification
Classification of school violence has been proven to be an effective solution for preventing violence within educational institutions. As a result, technical proposals aimed at enhancing the efficacy of violence classification are of considerable interest to researchers. This study explores the utilization of the SORT tracking method for localizing and tracking objects in videos related to school violence, coupled with the application of LSTM and GRU methods to enhance the accuracy of the violence classification model. Furthermore, we introduce the concept of a padding box to localize, identify actions, and recover tracked objects lost during video playback. The integration of these techniques offers a robust and efficient system for analyzing and preventing violence in educational environments. The results demonstrate that object localization and recovery algorithms yield improved violent classification outcomes compared to both the SORT tracking and violence classification algorithms alone, achieving an impressive accuracy rate of 72.13%. These experimental findings hold promise, especially in educational settings, where the assumption of camera stability is justifiable. This distinction is crucial due to the unique characteristics of violence in educational environments, setting it apart from other forms of violence.