[U-12] Future of Earth and Planetary Sciences Boosted by Artificial Intelligence

[U12-01] Beyond the Horizon: Advancing Earth Observation with Deep Learning

★Invited Papers

*Naoto Yokoya¹ (1.The University of Tokyo)

Keywords:Earth observation, deep learning, image processing

Deep learning has brought significant innovation to the analysis of Earth observation data. It not only improves the accuracy of analysis results for existing tasks but also realizes tasks that were previously difficult with existing technologies through generative models. On the other hand, there are many tasks in Earth observation where it is difficult to collect sufficient training data on a large scale. In this talk, we will introduce recent efforts to address the lack of training data in the understanding and acquisition of Earth observation images. In particular, we will present deep learning techniques that serve as the basis for intelligent information processing, such as automatic extraction of land cover maps, digital elevation models, and their changes from Earth observation data, and computational imaging that overcomes the limitations of spatial, temporal, and spectral resolutions of imaging systems.