Agirman, Ahmet K.Taşdemir, Kasím2025-09-252025-09-25201797815090649462165-0608https://doi.org/10.1109/SIU.2017.7960275https://hdl.handle.net/20.500.12573/2994Computer vision methods used for night-time fire detection are limited. Existing works are for detection of distant night fires recorded from watch towers. In this paper, detection of short to mid-range night fires from video cameras are aimed. Flames in short distance flicker, grow and move more rapidly compared to ones in long distance. Features obtained by taking advantage of these distinctions let us detect fire over 90% accuracy on average in videos containing deceptive light sources like common city lights and headlights of vehicles. © 2017 Elsevier B.V., All rights reserved.trinfo:eu-repo/semantics/closedAccessMajority VotingShort to Mid-Range Night FireSvmVideo Based Fire DetectionFire DetectorsLight SourcesSignal ProcessingVideo CamerasFire DetectionMajority VotingNight TimeFiresConference Object10.1109/SIU.2017.79602752-s2.0-85026320799