geoHD

Sentinel-2 imagery is free and revisits frequently, but at 10 m a small building is a single pixel and narrow roads, waterways and vegetation patches blur together; buying native high-resolution commercial imagery for large areas or past dates is costly or impossible. geoHD, delivered by Geoimage and developed by its partner SI Analytics, is an AI super-resolution product that uses a deep-learning model trained on paired low- and high-resolution imagery to reconstruct finer spatial detail from existing satellite data, with no new tasking or re-acquisition. Standard geoHD halves pixel size: Sentinel-2 from 10 m to 5 m, and commercial imagery from Pléiades Neo, WorldView-3, SPOT and KOMPSAT-3/3A (for example 30 cm to 15 cm). The geoHD+ tier, launched in March 2026, takes Sentinel-2 from 10 m to 2.5 m. It can be applied to archive imagery as well as recent captures.

Geoimage reports validation metrics for the output (MTF50 improvement above 250%, SNR increase above 40%, PLCC 0.984) and states that NDVI and other derived products remain virtually unchanged, so the imagery can feed quantitative and AI/ML workflows as well as visual interpretation. Geoimage also cautions that AI-reconstructed detail is an enhanced representation, not directly observed ground information; the original imagery remains the authoritative record. Target users named by the provider include government and environmental agencies, mining and resources, emergency management, urban planning, agriculture and AI/ML developers. Geoimage runs a demonstration on a buyer's area of interest before purchase.