Any Model, Any Place, Any Time: Get Remote Sensing Foundation Model Embeddings On Demand

Dingqi Ye; Daniel Kiv; Wei Hu; Jimeng Shi; Shaowen Wang

cs.CVFeb 27, 2026

Any Model, Any Place, Any Time: Get Remote Sensing Foundation Model Embeddings On Demand

Authors: Dingqi Ye, Daniel Kiv, Wei Hu, Jimeng Shi, Shaowen Wang

Organizations: University of Illinois Urbana-Champaign, Urbana, Illinois, USA

Abstract

The remote sensing community is witnessing a rapid growth of foundation models, which provide powerful embeddings for a wide range of downstream tasks. However, practical adoption and fair comparison remain challenging due to substantial heterogeneity in model release formats, platforms and interfaces, and input data specifications. These inconsistencies significantly increase the cost of obtaining, using, and benchmarking embeddings across models. To address this issue, we propose rs-embed, a Python library that offers a unified, region of interst (ROI) centric interface: with a single line of code, users can retrieve embeddings from any supported model for any location and any time range. The library also provides efficient batch processing to enable large-scale embedding generation and evaluation. The code is available at: https://github.com/cybergis/rs-embed

Figures & tables

Figure 1. Comparison of a complex current workflow vs. rs-embed, which generates standardized embeddings (plus metadata) from location, time, and model ID in a single line of code.

Figure 2. rs-embed architecture overview.

Figure 3. Parallel batch export pipeline (orchestration + I/O prefetch + inference + async export) that decouples network, compute, and disk writes to sustain high throughput reliably.

Figure 4. Performance comparison of different rsfm embeddings for maize yield regression in Illinois. The top row shows bar plots of test-set $R^{2}$ and RMSE (mt/ha) for each model. The bottom row presents predicted (Pred) vs. true (True) scatter plots for the top three models, with the dashed line indicating the ideal 1:1 reference.

Figure 5. Spatial maps of test predictions and residuals in Illinois(Argrifm). Left: predicted maize yield (mt/ha). Right: residuals (pred - true), where red indicates overestimation and blue indicates underestimation.

Figure 6. Visualization of embeddings from different models ( OutputSpec.grid() , input_prep=’resize’ ). The numbers in parentheses denote the channel, height, and width. For visualization, we apply PCA and use the top three principal components as pseudo-RGB channels.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 7. Dataset overview (n = 991). Left: yield label distribution (mt/ha). Middle: maize field area distribution (ha). Right: spatial distribution of sampling points colored by yield label within the study region.

Figure 8. Sample input and spatial context.

Remote Sensing Benchmarks

Remote sensing benchmarks measure how well models interpret Earth-observation data across sensors, regions, time, tasks, and operating conditions.

See all 194 papers

Earth Embeddings
Aug 4, 2026Adam J. Stewart, Heng Fang, Isaac A. Corley +1Geospatial Representation Learning Land Cover Classification
Chair of Data Science in Earth Observation, Technical University of Munich, Munich, Germany · Munich Center for Machine Learning, Munich, Germany · KTH Royal Institute of Technology, Stockholm, Sweden +1
OlmoEarth v1.3: Embeddings
Oct 23, 2026Gabriel Tseng, Favyen Bastani, Patrick Alan Johnson +13Geospatial Representation Learning Multimodal Embedding
Allen Institute for AI
VLM2GeoVec: Toward Universal Multimodal Embeddings for Remote Sensing
Dec 12, 2025Emanuel Sánchez Aimar, Gulnaz Zhambulova, Fahad Shahbaz Khan +2Multimodal Understanding Cross-Modal Learning
Linköping University, Sweden

Aug 4, 2026J/K move · Enter open · S save

Earth Embeddings

Adam J. Stewart, Heng Fang, Isaac A. Corley, Xiao Xiang Zhu

Chair of Data Science in Earth Observation, Technical University of Munich, Munich, Germany · Munich Center for Machine Learning, Munich, Germany · KTH Royal Institute of Technology, Stockholm, Sweden+1

Geospatial Representation Learning Land Cover Classification Remote Sensing Earth Science

Open paper AskPDF arXiv

Earth observation is moving from foundation models that users must run themselves toward embedding products that package model feature outputs as reusable data without needing to download and process the imagery used to generate them. Earth embeddings are vectors that summarize locations, image patches, or pixels, letting users analyze compact features instead of repeatedly training or running large models on raw satellite imagery. This chapter explains the main types of Earth embeddings, from implicit location encoders to explicit patch and pixel products, and compares their coverage, resolution, dimensionality, storage cost, licenses, and reproducibility. We review their use in land cover and crop mapping, ecological and hazard modeling, socioeconomic prediction, and semantic search, with evidence on when embeddings improve on conventional features and when pooling, fusion, or spatial transfer limit performance. Two case studies show practical workflows for similarity search and land cover mapping. We close with guidance for choosing, evaluating, storing, compressing, and publishing embeddings, and with open problems in oceanic and atmospheric coverage, uncertainty, and benchmarking.

Abstract

Remote Sensing Benchmarks

Explore similar work

Earth Embeddings

OlmoEarth v1.3: Embeddings

VLM2GeoVec: Toward Universal Multimodal Embeddings for Remote Sensing