AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

OpenEarthAgent: A Unified Framework for Tool-Augmented Geospatial Agents

Akashah Shabbir, Muhammad Umer Sheikh, Muhammad Akhtar Munir, Hiyam Debary, Mustansar Fiaz, Muhammad Zaigham Zaheer, Paolo Fraccaro, Fahad Shahbaz Khan, Muhammad Haris Khan, Xiao Xiang Zhu, Salman Khan

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs · 16 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
SFT on validated tool-use trajectoriesQwen3-4B-Instruct-2507 · OpenEarthAgent benchmark end-to-endanswer accuracy: 13.72 -> 45.26Improved
SFT on validated tool-use trajectoriesQwen3-4B-Instruct-2507 · OpenEarthAgent benchmark step-by-stepargument value accuracy: 33.55 -> 62.10Improved
SFT on trajectoriesQwen3-4B-Instruct-2507 · OpenEarthAgent benchmark step-by-stepsummarization accuracy: 83.28 -> 83.64No clear effect
SFT on trajectoriesQwen3-4B-Instruct-2507 · ThinkGeoAns: 12.53 -> 16.35Improved
SFT on trajectoriesQwen3-4B-Instruct-2507 · Earth-Agent benchmark step-by-steptool selection accuracy: 60.61 -> 70.33Improved
SFT on trajectories vs frontier gpt-4oOpenEarthAgent · Earth-Agent benchmark step-by-steptool selection accuracy: 71.65 -> 70.33Hurt
combine easy and hard trajectories in training (ablation)OpenEarthAgent · OpenEarthAgent benchmark end-to-endanswer accuracy: 43.40 (easy) / 35.79 (hard) -> 45.26Improved
add in-context examples (k=15) to untuned model (ablation)Qwen3-4B-Instruct-2507 · OpenEarthAgent benchmark end-to-endanswer accuracy: 13.72 -> 33.06Improved
increase in-context examples from 10 to 15 (ablation)Qwen3-4B-Instruct-2507 · OpenEarthAgent benchmark end-to-endimage generation accuracy: 51.72 -> 46.90Hurt
provide full tool schema with JSON examples (Sch.5) vs minimal (ablation)OpenEarthAgent · OpenEarthAgent benchmark end-to-endanswer accuracy: 8.44 -> 45.26Improved
provide executable-constraint schema (Sch.4) vs signature schema (Sch.3) (ablation)OpenEarthAgent · OpenEarthAgent benchmark end-to-endtool order AnyOrder: 26.17 -> 26.55No clear effect
use optimized tool order with order rule vs shuffled (ablation)OpenEarthAgent · OpenEarthAgent benchmark end-to-endanswer accuracy: 43.35 -> 45.26Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy