Enhancing Forest Fire Detection: Utilizing VGG16 for High-Accuracy Image Classification and Machine Learning Integration

Forest fires cause significant problems for nature, communities. They damage forests, animals’ homes, and places where people live. The bad effects of forest fires go beyond just hurting the air, water, and animals. They also make climate change worse and mess up the balance of nature for a long time. In our efforts to address this significant issue, we utilize trained Convolutional Neural Networks (CNNs) algorithm VGG16 model to classify images containing fire and smoke versus non fire images. To enhance the training performance and the model’s ability to generalize to unseen data, we curated a dataset comprising approximately 12,000 images from each category. We proposed using the features extracted the VGG16 model for training a simple neural network and multiple machine learning algorithms. Following the training process of the neural network model, our model achieved a training accuracy of 98.4231% and a validation accuracy of 99.0717%. Subsequently, to assess the robustness of our model, we subjected the output layer of our trained model to 13 different machine learning algorithms. Remarkably, all 6 models surpassed a performance threshold of 99% across all metrics.