Publication

Abstract : The main objective of the project is to enhance the existing system for fire prediction and protection systems. Deep Learning approaches have shown the ability to provide better results for the prediction of wildfires. Previous systems are more complicated and sort of complete black box in image analysis. Faster recognition and passing the data to the concerned wildfire authorities in more important. The warning system needs to be automated and should send the information at a set of intervals for proper image analysis. Hence, we suggest including an automatic warning system to the system to avoid late actions taken against the wildfire and its damages. We propose different types of image analysis for categorizing the fires occurred. DWT is one of the image analysis methods that has been implemented for better result. This will help any people with less knowledge about the wildfire to understand the fire nature and enables them to take appropriate actions. The neural network model in the Restnet50 was faster in training with large datasets compared to other neural network models.