Region-based Annotation Data of Fire Images for Intelligent Surveillance System

<p>This data presents&nbsp;fire segmentation&nbsp;annotation&nbsp;data on 12&nbsp;commonly used&nbsp;and&nbsp;publicly available&nbsp;&ldquo;VisiFire&nbsp;Dataset&rdquo;&nbsp;videos&nbsp;from&nbsp;<a href="http://signal.ee.bilkent.edu.tr/VisiFire/">http://signal.ee.bilkent.edu.tr/VisiFire/</a>.&nbsp;This annotations dataset was obtained by per-frame, manual hand annotation over the fire region with 2,684&nbsp;total&nbsp;annotated frames.&nbsp;Since this&nbsp;annotation provides per-frame&nbsp;segmentation data,&nbsp;it offers a new&nbsp;and&nbsp;unique fire motion feature to the existing video, unlike other fire segmentation data that are collected from different still&nbsp;images.&nbsp;The&nbsp;annotations&nbsp;dataset&nbsp;also&nbsp;provides&nbsp;ground truth for segmentation task&nbsp;on videos.&nbsp;With segmentation task, it&nbsp;offers&nbsp;better insight on how well a machine learning model understood, not only detecting whether a fire is present, but also&nbsp;its&nbsp;exact location by calculating metrics such as Intersection over Union (IoU) with this annotations data.&nbsp;This&nbsp;annotations&nbsp;data&nbsp;is&nbsp;a&nbsp;tremendously useful&nbsp;addition&nbsp;to&nbsp;train, develop,&nbsp;and&nbsp;create&nbsp;a&nbsp;much better&nbsp;smart surveillance system&nbsp;for early detection&nbsp;in&nbsp;high-risk&nbsp;fire hotspots&nbsp;area.&nbsp;</p>