Generation, Detection and Localization of Synthetic Remote Sensing Imagery

Throughout the past several years for conspicuous rationale, there has been a hype about generative models and deepfakes. However, most of the applications target standard RGB images with a special focus on faces. In this thesis, we explore some generative models in the scope of remote sensing images, generating manipulated remote sensing images or plausible remote sensing images that do not exist. After building several generated remote sensing datasets, the detectability of those images among genuine ones was inspected.