What can large collections of user-uploaded photos reveal about places, events, and environmental perception? We analyze social media imagery to connect visual content with geography, land use, and human experience of landscapes. Recent work includes probabilistic embeddings for multi-scale zero-shot soundscape mapping from social and satellite data (PSM), mapping fine-grained textual descriptions from satellite images (Sat2Cap), and revisiting image geolocalization in the deep learning era. Social media provides ground-level diversity that complements structured earth observation in our geospatial AI and ecology programs.
Spotlight publications tagged for this research area.