Optimizing Deep Learning Models for Climate-Related Natural Disaster Detection from UAV images and Remote Sensing Data

This research project involves a comprehensive approach to utilizing artificial intelligence (AI), machine learning (ML), & deep learning (DL), for detecting climate change related natural disasters such as flooding/desertification from aerial imagery. By compiling images of numerous datasets from an open-access data site, this work offers an extensive novel dataset, the Climate Change Dataset. This dataset was utilized to train DL models including transfer learning models for the detection of climate related natural disasters. Four ML models trained on the Climate Change Dataset were compared including: a convolutional neural network (CNN), DenseNet201, VGG16, and ResNet50. Our DenseNet201 model was chosen for optimization leading to improved performance. The 4 ML models all performed well with DenseNet201 Optimized and ResNet50 yielding the highest accuracies of 99.37% and 99.21% respectively. By advancing our scientific knowledge in climate change impacts, desertification & flood detection, this research project has demonstrated the potential of AI to proactively address environmental challenges. Our study is intended for the use of AI for Climate Change and Environmental Sustainability.