Generating an open AI training data set for Moon craters with crowd science via a videogame

This study investigated the integration of citizen science tasks into a fully realized gaming environment as an innovative solution to challenges in participant retention and data generation. The contractor developed a moon-base building game that incorporates the annotation of small lunar craters (10–100 m) using high-resolution Lunar Reconnaissance Orbiter Camera (LROC) images. Unlike traditional gamification approaches, the design embeds the annotation tasks directly within the core gameplay mechanics, creating a more immersive and engaging user experience.
The results demonstrated that while annotation precision and recall were comparable to those achieved in other citizen science projects, participants in the game-based environment marked significantly more craters. Additionally, the game fostered considerably higher long-term engagement, with users remaining active for extended periods compared to those in conventional, non-gamified setups.