Fire spread analysis and Fire detection using CNN

Early detection of forest fires allows for swift intervention, preventing the flames from spreading uncontrollably and safeguarding both human lives and the delicate ecosystems that thrive within our forests. Our study attempts to identify fire, locate its precise position relative to its surroundings, and determine its rate of spread. Initially, we begin with pre-processing the images using OpenCV, followed by segmentation of the same. As the images generated by the drone cameras are trained, a convolutional neural network model is built using the Keras API. The photographs can be categorized as “Fire” or “Non-Fire” with the aid of this model. Additionally, the rate at which the fire spreads over the measured area is also determined using the recorded footage. The normalized FLAME dataset is utilized to derive the outcomes for the suggested technique.