NASA and IBM released the NASA-IBM Lunar Foundation Model, an open-source AI system available on Hugging Face. In testing, it reduced errors by 23 percent compared to Microsoft's SwinV2-B in identifying potential ice on the lunar surface, and outperformed SwinV2-B by 19 percent at crater classification while using half the training data. The release includes a co-registered open-source dataset of more than two million data points from lunar orbiter missions for building future models.
No score is assigned. Sources and their independence are shown in the citation chain below.