Skip to main content
For the Digital Public Good
Home » Innovation Program » Benchmarks of the World

Benchmarks of The World

A shared, community-governed system for evaluating geospatial AI models — measuring real-world performance, failure modes, compute costs, and generalization across geographies and data types.

Network graph with interconnected nodes of varying sizes, colored in blue, red, and a few yellow. Thin lines represent connections, creating a dense web against a dark background.

A rigorous, open benchmarking system gives the geospatial AI community — researchers, governments, and industry alike — a trusted, shared basis for comparing models and making informed decisions about which tools to use for which problems. It raises the bar for transparency and reproducibility, reduces duplicated effort across organizations, and helps ensure that advances in the field translate into reliable, real-world impact.

BotW creates simple accuracy scores to assess real-world performance, failure modes, compute costs, and generalization across geographies and data types.

Research Partners
Center for Geospatial Analytics, Clark University
Kerner Lab, Arizona State University
Evan Shelhamer, University of British Columbia
Ai2
Radiant Earth

Research Focus

Real-World Evaluation Methods

Evaluation beyond standard metrics — showing how models behave at deployment scale, where they fail, and how performance varies across geographic contexts.

Diagnostic Tools

New tools allow organizations to assess model strengths and blind spots without running full evaluations on every task — lowering the barrier to rigorous testing.

Pre-Task Performance Prediction

Methods for predicting model performance on a given task before running it — providing an early, low-cost signal for model selection.

Compute Transparency

Cost and compute requirements are tracked and reported as standard outputs alongside accuracy, making efficiency a first-class evaluation criterion.

Related News

A person wearing a scarf and jacket bends over a cart filled with apples at an outdoor market, with other people and produce visible in the background under a clear blue sky.
July 15, 2026

How geospatial AI is helping aid groups predict famine and deliver food

Devex — A St. Louis nonprofit is investing in AI to help WFP, FEWS NET, and other groups build and improve open-source tools for humanitarians to deliver food aid much more quickly and efficiently to people in need.
August 3, 2026

Introducing the Fields of The World User Advisory Group

Fields of The World is growing beyond its global launch, and we’re inviting practitioners to help shape what comes next. Applications for our first User Advisory Group are open through August 14.
An ornate white stone building with detailed carvings stands in front of trees and a domed structure, possibly a cathedral, under a blue sky with a small moon visible.
July 27, 2026

Satellites for Peace: Why the Vatican Cares About Earth Observation

International experts and humanitarian leaders recently convened at a Vatican workshop to explore how democratizing access to satellite imagery can create a transparent, community-governed public infrastructure for global humanitarian efforts.
Hannah Kerner shakes hands with attendees at the APEC ASPIRE ceremony.
June 25, 2026

Hannah Kerner Named U.S. Nominee for the APEC ASPIRE Prize

Taylor Geospatial research advisor Hannah Kerner recognized for advancing scientific research in AI and data science to promote industrial innovation and economic resilience.
June 25, 2026

Co-investment with AWS expands GIFS program

Geospatial Innovation for Food Security (GIFS) program adds 10 pilot projects
June 23, 2026

AI project aims to help humanitarian organizations identify food security risks in conflict zones

A new initiative at Arizona State University seeks to provide earlier, more accessible insights into emerging food security risks by combining artificial intelligence, or AI, satellite observations and natural-language tools.
June 18, 2026

SatSummit Comes to St. Louis

Taylor Geospatial to co-host conference for leaders in the satellite industry and experts in global development November 18-19, 2026, at The Post Building.
June 11, 2026

Mapping The World at Taylor Geospatial

Robin Cole from the Satellite Image Deep Learning podcast interviews Jennifer Marcus, VP for Strategic Innovation Programs and Isaac Corley, Director of AI/ML Research, about Fields of The World.
June 10, 2026

Mapping Every Field on Earth

Spatial Stack’s Matt Forrest interviews Jennifer Marcus, VP for Strategic Innovation Programs and Isaac Corley, Director of AI/ML Research, about Fields of The World.
Tractor moving across a freshly plowed field at sunset, dust clouds rising behind it.
June 8, 2026

Convening a National Roadmap on Agricultural Nitrous Oxide MMRV Research

Taylor Geospatial was honored to help guide this effort alongside colleagues at the Natural Resources Defense Council, bringing together researchers from universities, federal agencies, industry, and the nonprofit sector to reach consensus on the research needed to develop fit-for-purpose tools and methods to measure, model, and ultimately reduce the nitrous oxide emissions that come from the ways we grow our food.

Other Inititatives

Aerial view of a rural landscape with colorful patchwork fields, a winding road, scattered trees, and distant hills under a partly cloudy sky.

Fields of the World

Expanding the world’s first open ecosystem for global agricultural field boundary detection from satellite imagery — combining a global benchmark dataset, baseline ML models, inference tools, and web applications.

Aerial view of a densely packed neighborhood with red-roofed houses, green trees, and narrow roads winding through the area.

Features of the World

Applying the field-boundary approach to a broader set of physical features visible from space — training models capable of detecting roads, buildings, solar panels, trees, and more at global scale.