{"title":"一种利用基于熔合变压器的深度特征和深度神经网络的森林火灾探测的创新混合方法","authors":"Kemal Akyol","doi":"10.1016/j.asr.2025.04.020","DOIUrl":null,"url":null,"abstract":"<div><div>Forest fires, one of the most pernicious and devastating disasters, cause deforestation, wildlife extinction, global warming, and climate change. Early fire detection is critical before it reaches catastrophic dimensions. Artificial intelligence-based systems that detect forest fires accurately and quickly are needed for early intervention. Delayed extinguishing efforts without such systems cause tremendous damage and losses. An effective monitoring system allows for the reduction of fire damage and hence prevents forest loss. This study aims to develop a successful artificial intelligence model to detect fire from forest landscape images. In this context, this work is new as it provides new insights into robust and scientific modeling for forest fire detection by analyzing feature maps based on fused transformer architectures using Deep Neural Networks. The experimental models were validated using accuracy, sensitivity, precision, and area under the receiver operating characteristic curve measures. The validation findings reveal that the proposed hybrid model performs the best while all models yield reasonable results. To summarize, satisfactory accuracy values of 99.58% and 96.79% for both datasets, respectively, strongly support the proposed hybrid model’s fire detection achievement with the 5-fold cross-validation. Furthermore, the high sensitivity and high precision measures imply that the model has few false negatives and false positives. Considering the obtained accuracies, the proposed hybrid model could be used for comprehensive fire detection modeling. To the author’s best knowledge, this study is the first to use transformer architectures and Deep Neural Networks for forest fire detection and is therefore important for the relevant literature. In this context, this study presents a new approach to distinguishing landscape images of forest fires and further developing fire detection strategies by the role of transformer architectures in feature extraction. It is thought that by executing the proposed model in an unmanned aerial vehicle equipped with a real-time system, fire detection will provide decision support to field professionals in reducing damage and managing forest fires.</div></div>","PeriodicalId":50850,"journal":{"name":"Advances in Space Research","volume":"75 12","pages":"Pages 8583-8598"},"PeriodicalIF":2.8000,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"An innovative hybrid method utilizing fused transformer-based deep features and deep neural networks for detecting forest fires\",\"authors\":\"Kemal Akyol\",\"doi\":\"10.1016/j.asr.2025.04.020\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Forest fires, one of the most pernicious and devastating disasters, cause deforestation, wildlife extinction, global warming, and climate change. Early fire detection is critical before it reaches catastrophic dimensions. Artificial intelligence-based systems that detect forest fires accurately and quickly are needed for early intervention. Delayed extinguishing efforts without such systems cause tremendous damage and losses. An effective monitoring system allows for the reduction of fire damage and hence prevents forest loss. This study aims to develop a successful artificial intelligence model to detect fire from forest landscape images. In this context, this work is new as it provides new insights into robust and scientific modeling for forest fire detection by analyzing feature maps based on fused transformer architectures using Deep Neural Networks. The experimental models were validated using accuracy, sensitivity, precision, and area under the receiver operating characteristic curve measures. The validation findings reveal that the proposed hybrid model performs the best while all models yield reasonable results. To summarize, satisfactory accuracy values of 99.58% and 96.79% for both datasets, respectively, strongly support the proposed hybrid model’s fire detection achievement with the 5-fold cross-validation. Furthermore, the high sensitivity and high precision measures imply that the model has few false negatives and false positives. Considering the obtained accuracies, the proposed hybrid model could be used for comprehensive fire detection modeling. To the author’s best knowledge, this study is the first to use transformer architectures and Deep Neural Networks for forest fire detection and is therefore important for the relevant literature. In this context, this study presents a new approach to distinguishing landscape images of forest fires and further developing fire detection strategies by the role of transformer architectures in feature extraction. It is thought that by executing the proposed model in an unmanned aerial vehicle equipped with a real-time system, fire detection will provide decision support to field professionals in reducing damage and managing forest fires.</div></div>\",\"PeriodicalId\":50850,\"journal\":{\"name\":\"Advances in Space Research\",\"volume\":\"75 12\",\"pages\":\"Pages 8583-8598\"},\"PeriodicalIF\":2.8000,\"publicationDate\":\"2025-04-11\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Advances in Space Research\",\"FirstCategoryId\":\"89\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0273117725003564\",\"RegionNum\":3,\"RegionCategory\":\"地球科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ASTRONOMY & ASTROPHYSICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Advances in Space Research","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0273117725003564","RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ASTRONOMY & ASTROPHYSICS","Score":null,"Total":0}
An innovative hybrid method utilizing fused transformer-based deep features and deep neural networks for detecting forest fires
Forest fires, one of the most pernicious and devastating disasters, cause deforestation, wildlife extinction, global warming, and climate change. Early fire detection is critical before it reaches catastrophic dimensions. Artificial intelligence-based systems that detect forest fires accurately and quickly are needed for early intervention. Delayed extinguishing efforts without such systems cause tremendous damage and losses. An effective monitoring system allows for the reduction of fire damage and hence prevents forest loss. This study aims to develop a successful artificial intelligence model to detect fire from forest landscape images. In this context, this work is new as it provides new insights into robust and scientific modeling for forest fire detection by analyzing feature maps based on fused transformer architectures using Deep Neural Networks. The experimental models were validated using accuracy, sensitivity, precision, and area under the receiver operating characteristic curve measures. The validation findings reveal that the proposed hybrid model performs the best while all models yield reasonable results. To summarize, satisfactory accuracy values of 99.58% and 96.79% for both datasets, respectively, strongly support the proposed hybrid model’s fire detection achievement with the 5-fold cross-validation. Furthermore, the high sensitivity and high precision measures imply that the model has few false negatives and false positives. Considering the obtained accuracies, the proposed hybrid model could be used for comprehensive fire detection modeling. To the author’s best knowledge, this study is the first to use transformer architectures and Deep Neural Networks for forest fire detection and is therefore important for the relevant literature. In this context, this study presents a new approach to distinguishing landscape images of forest fires and further developing fire detection strategies by the role of transformer architectures in feature extraction. It is thought that by executing the proposed model in an unmanned aerial vehicle equipped with a real-time system, fire detection will provide decision support to field professionals in reducing damage and managing forest fires.
期刊介绍:
The COSPAR publication Advances in Space Research (ASR) is an open journal covering all areas of space research including: space studies of the Earth''s surface, meteorology, climate, the Earth-Moon system, planets and small bodies of the solar system, upper atmospheres, ionospheres and magnetospheres of the Earth and planets including reference atmospheres, space plasmas in the solar system, astrophysics from space, materials sciences in space, fundamental physics in space, space debris, space weather, Earth observations of space phenomena, etc.
NB: Please note that manuscripts related to life sciences as related to space are no more accepted for submission to Advances in Space Research. Such manuscripts should now be submitted to the new COSPAR Journal Life Sciences in Space Research (LSSR).
All submissions are reviewed by two scientists in the field. COSPAR is an interdisciplinary scientific organization concerned with the progress of space research on an international scale. Operating under the rules of ICSU, COSPAR ignores political considerations and considers all questions solely from the scientific viewpoint.