Forest Fire Detection Using Convolutional Neural Networks

Abstract

In this paper, we have used a Deep Learning approach called Convolutional Neural Networks to train our model. ResNet-34 is the architecture used to train the model to detect forest fires by taking satellite images as input. The dataset has been obtained online and has been subjected to Batch Normalization and Data Augmentation to achieve better results. The model has been trained using satellite images of forests with and without the occurrence of fires because satellites ensure a continuous supply of images for the model to keep testing. The result obtained has a high accuracy and is therefore able to detect forest fires easily with minimal errors.