Search Results

The rise of Artificial Intelligence (AI) has brought up both opportunities and challenges for today's evolving GIScience. Its ability in image classification, object detection and feature extraction has been frequently praised. However, it may also apply for falsifying geospatial data. To demonstrate the thrilling power of AI, this research explored the potentials of deep learning algorithms in capturing geographic features and creating fake satellite images according to the learned 'sense'. Specifically, Generative Adversarial Networks (GANs) is used to capture geographic features of a certain place from a group of web maps and satellite images, and transfer the features to another place. Corvallis is selected as the study area, and fake datasets with 'learned' style from three big cities (i.e. New York City, Seattle and Beijing) are generated through CycleGAN. The empirical results show that GANs can 'remember' a certain 'sense of place' and further apply that 'sense' to another place. With this paper, we would like to raise both public and GIScientists' awareness in the potential occurrence of fake satellite images, and its impacts on various geospatial applications, such as environmental monitoring, urban planning, and land use development.