GeoSET: first generalist SAR-to-EO translation model trained on 3M pairs
Jeonghyeok Do · hf · 2026-10-06
Researchers introduce GeoSET, the first generalist foundation model for SAR-to-EO image translation:
- Curated 3M+ high-quality SAR-EO pairs from over 10M SAR observations, spanning diverse sensors, resolutions, and ground sampling distances.
- Built on a single pretrained parent adapted to downstream datasets via LoRA, updating only 0.60% of generator parameters in about one hour per dataset.
- A speckle-robust SAR encoder bridges the modality gap to the pretrained image generator.
- Achieves state-of-the-art FID and DISTS across six downstream benchmarks with either full fine-tuning or LoRA.
More from Research
- AI formal verification hits reality: the halting problem blocks existing codebases — danbri · 2026-10-06
- GeoCR: one generalist model unifies cloud removal across 10 datasets, no fine-tuning needed — KAIST-VIClab · 2026-10-06
- Pi²: turning Wikipedia tables into long-context reasoning training data (COLM 2026) — tuvllms · 2026-10-06
- New RL simulation environment for port logistics built on real-world port data — yb2698 · 2026-10-06
- Interactive explainer makes TF-IDF, BM25 and NDCG click, BM25 by example — JnBrymn · 2026-10-06
- Poster on curating long-context reasoning data presented today at Poster Session 1 — tuvllms · 2026-10-06