Tripo AI 3D Generator
Mini toolsOlmoEarth-v1-Base
Model family: Earth
OlmoEarth-v1-Base is an 89M-parameter Vision Transformer (ViT-Base) pre-trained as a general-purpose encoder for Earth observation. It supports both single images and multi-temporal time series from Sentinel-1 (SAR radar), Sentinel-2 (multispectral optical), and Landsat satellite platforms. The model is designed for fine-tuning on downstream remote sensing tasks including semantic segmentation, object classification, and feature embedding extraction. Tutorials are provided for computing embeddings and fine-tuning workflows using the rslearn and olmoearth_run libraries. The model is also deployed via the OlmoEarth Platform, which enables organizations to apply Earth observation AI without requiring deep AI or engineering expertise.
Image
Gen 4
Overview
OlmoEarth-v1-Base is a ViT-Base (89M parameter) vision foundation model pre-trained for Earth observation tasks. It processes Sentinel-1, Sentinel-2, and Landsat satellite images and image time series, and can be fine-tuned for downstream remote sensing tasks such as segmentation and classification.
About The Allen Institute for Artificial Intelligence
Semantic Scholar uses groundbreaking AI and engineering to understand the semantics of scientific literature to help Scholars discover relevant research.
Location:
August, Washington, US
Website:
semanticscholar.org
Last updated: June 30, 2026
