Forest Fire Detection Using Deep Learning With Neural Network

Authors

  • V.Mariselvam, T.Aarthi, M.C.Oviya, S.Selva Meena,

Abstract

Forest fire is real threat to our lives, environment and surroundings. It is predicted that forest fire could destroy more than half of world’s forest bytheyear2030.Theonlyefficient wayt o minimize the forest fire damage is adopt early fire detection mechanisms. Thus, forest firedetectionmechanismshavingalotoffocusonseveralresearchcentersanduniversitiesaroundtheworld. Currently, there exists many simple fire detection sensor and live monitoring systems, buttheyaredifficulttoapplyinvastareaslikeforests,duetodelayinresponse,continuousmaintenance, low accuracy, high cost and other issues.[1-5] Here, image processing based has beenused due to several reasons such as quick development of digital cameras technology, the cameracan cover vast areas with maximum distance with best results, the response time and accuracy ofimage processing methods is better than that of the existingsystems, and the total cost of theimageprocessingwithdeeplearningislowerthansensorsystem.Stillaccurateforestfiredetectionalgorithmremainacomplicatedissue,because,fewobjectshavethesimilarfeaturesoffire,whichgives false message or alarm. In this paper a new image processing forest fires detection method,which is of four stages is used. Here the first step is, to detect moving regions by backgroundsubtractionalgorithm.Then,fire locations can be found using RGB colorspace. Thirdly, aoriginal fire and fire like objects can be easily differentiated by feature extraction. At last, convolutionalneural network algorithm is used to differentiate either real fire or non-fire. The final experimentresultverifies that theproposed method works effectivelyand identifies theforest fire.[6-10]

 

Downloads

Published

2021-05-01

How to Cite

V.Mariselvam, T.Aarthi, M.C.Oviya, S.Selva Meena,. (2021). Forest Fire Detection Using Deep Learning With Neural Network. International Journal of Modern Agriculture, 10(2), 3183 - 3189. Retrieved from http://modern-journals.com/index.php/ijma/article/view/1141

Issue

Section

Articles