2025/11/06 by Valentin Meo · 1 voice
Computer Science · Engineering · #Advanced Image and Video Retrieval Techniques #Automated Road and Building Extraction #Infrared Target Detection Methodologies
paper · doi:10.1080/15230406.2025.2576496
openalex created_date 2025/11/06 · openalex publication_date 2025/11/06 · openalex updated_date 2026/06/24
This paper explores the field of deepfake geography, which involves the creation and detection of manipulated satellite images using advanced deep-learning techniques. It begins with a discussion of the increasing accessibility and realism of deepfake technology, as well as its potential impact on trust, public opinion, and the dissemination of disinformation. Then, the manipulation of maps and geographical information is examined, highlighting notable examples and recent advancements in generative AI for creating synthetic satellite imagery. While prior studies have explored the detection of synthetic satellite images, they do not address the more challenging task of identifying manipulated content within real geospatial data. To fill this gap, a new deep-learning-based detection method is introduced, and its performance is evaluated using a dataset of deepfake-geography images created with a state-of-the-art generative model. The results demonstrate the effectiveness of the proposed method in detecting fake areas in real satellite images. The paper concludes by discussing the implications of its findings and suggesting potential avenues for future research in deepfake-geography creation and detection.