CrossEarth-SAR: A SAR-Centric and Billion-Scale Geospatial Foundation Model for Domain Generalizable Semantic Segmentation

Export BriefConnect with Author
View PDF ↗
PDF Viewer
Open Full PDF

BUILDER'S SANDBOX

Build This Paper

Use an AI coding agent to implement this research.

OpenAI Codex
OpenAI CodexAI Agent

Lightweight coding agent in your terminal.

Claude Code
Claude CodeAI Agent

Agentic coding tool for terminal workflows.

AntiGravity IDE
AntiGravity IDEScaffolding

AI agent mindset installer and workflow scaffolder.

Cursor
CursorIDE

AI-first code editor built on VS Code.

VS Code
VS CodeIDE

Free, open-source editor by Microsoft.

MVP Investment

$9K - $12K
6-10 weeks
Engineering
$8,000
Cloud Hosting
$240
SaaS Stack
$300
Domain & Legal
$100

6mo ROI

2-4x

3yr ROI

10-20x

Lightweight AI tools can reach profitability quickly. At $500/mo average contract, 20 customers = $10K MRR by 6mo, 200+ by 3yr.

Talent Scout

Z

Ziqi Ye

Fudan University

Z

Ziyang Gong

Shanghai Jiao Tong University

N

Ning Liao

Shanghai Jiao Tong University

X

Xiaoxing Hu

Beijing Institute of Technology

Find Similar Experts

Geospatial experts on LinkedIn & GitHub

References (60)

[1]
DINOv3
2025Oriane Sim'eoni, Huy V. Vo et al.
[2]
A semantic-enhanced multi-modal remote sensing foundation model for Earth observation
2025Kang Wu, Yingying Zhang et al.
[3]
Adaptively Distilled ControlNet: Accelerated Training and Superior Sampling for Medical Image Synthesis
2025Kunpeng Qiu, Zhiying Zhou et al.
[4]
SkySense V2: A Unified Foundation Model for Multi-modal Remote Sensing
2025Yingying Zhang, Lixiang Ru et al.
[5]
Noise-Consistent Siamese-Diffusion for Medical Image Synthesis and Segmentation
2025Kunpeng Qiu, Zhiqiang Gao et al.
[6]
A Complex-Valued SAR Foundation Model Based on Physically Inspired Representation Learning
2025Mengyu Wang, Hanbo Bi et al.
[7]
Can Generative Geospatial Diffusion Models Excel as Discriminative Geospatial Foundation Models?
2025Yuru Jia, V. Marsocci et al.
[8]
Earth-Adapter: Bridge the Geospatial Domain Gaps with Mixture of Frequency Adaptation
2025Xiaoxing Hu, Ziyang Gong et al.
[9]
ASANet: Asymmetric Semantic Aligning Network for RGB and SAR image land cover classification
2024Pan Zhang, Baochai Peng et al.
[10]
MGFNet: An MLP-dominated gated fusion network for semantic segmentation of high-resolution multi-modal remote sensing images
2024Kan Wei, Jinkun Dai et al.
[11]
CrossEarth: Geospatial Vision Foundation Model for Domain Generalizable Remote Sensing Semantic Segmentation
2024Ziyang Gong, Zhixiang Wei et al.
[12]
Improved Coherent Processing of Synthetic Aperture Radar Data through Speckle Whitening of Single-Look Complex Images
2024L. Alparone, A. Arienzo et al.
[13]
Convolutional Modulated Scattering Feature Network for Aircraft Classification in SAR Images
2024Ziqi Ye, Xiayang Xiao et al.
[14]
HyperSIGMA: Hyperspectral Intelligence Comprehension Foundation Model
2024Di Wang, Meiqi Hu et al.
[15]
CoDA: Instructive Chain-of-Domain Adaptation with Severity-Aware Visual Prompt Tuning
2024Ziyang Gong, Fuhao Li et al.
[16]
Parsing All Adverse Scenes: Severity-Aware Semantic Segmentation with Mask-Enhanced Cross-Domain Consistency
2024Fuhao Li, Ziyang Gong et al.
[17]
MTP: Advancing Remote Sensing Foundation Model via Multitask Pretraining
2024Di Wang, Jing Zhang et al.
[18]
SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection
2024Yuxuan Li, Xiang Li et al.
[19]
Airborne Multi-Dimensional SAR Land Cover Dataset and Fusion Classification Method
2024Nai-Rong Zheng, Zi-An Yang et al.
[20]
Neural Plasticity-Inspired Foundation Model for Observing the Earth Crossing Modalities
2024Zhitong Xiong, Yi Wang et al.

Showing 20 of 60 references

Founder's Pitch

"SAR imaging for domain-generalizable semantic segmentation with billion-scale SAR foundation model."

Geospatial AIScore: 8View PDF ↗

Commercial Viability Breakdown

0-10 scale

High Potential

4/4 signals

10

Quick Build

1/4 signals

2.5

Series A Potential

4/4 signals

10

Sources used for this analysis

arXiv Paper

Full-text PDF analysis of the research paper

GitHub Repository

Code availability, stars, and contributor activity

Citation Network

Semantic Scholar citations and co-citation patterns

Community Predictions

Crowd-sourced unicorn probability assessments

Analysis model: GPT-4o · Last scored: 3/12/2026

🔭 Research Neighborhood

Generating constellation...

~3-8 seconds

Why It Matters

Semantic segmentation in SAR is essential for applications such as disaster mitigation and urban management, and improving its accuracy and robustness can serve critical infrastructural and humanitarian functions.

Product Angle

Bundle the CrossEarth-SAR model with APIs and visualization tools to offer a comprehensive package for SAR data analysis, targeting governmental and environmental agencies.

Disruption

This solution could replace existing SAR analysis tools by offering superior accuracy and domain generalization capabilities, particularly in environments with challenging conditions.

Product Opportunity

The demand for improved SAR data analysis is growing, driven by sectors like disaster management, defense, and environmental monitoring, potentially unlocking a billion-dollar market.

Use Case Idea

Develop a platform for governments and disaster response agencies that leverages SAR data for real-time monitoring and decision support during environmental crises.

Science

CrossEarth-SAR uses a sparse mixture-of-experts (MoE) architecture integrated into a DINOv2 backbone for SAR semantic segmentation, applying a physics-guided routing mechanism to manage the heterogeneity in SAR data.

Method & Eval

Evaluated on 22 sub-benchmarks, CrossEarth-SAR achieved state-of-the-art performance in 20, demonstrating over 10% improvement in mIoU compared to prior methods on several benchmarks.

Caveats

The model's reliance on extensive computational resources and domain-specific datasets may limit its accessibility and scalability in non-specialist settings.

Author Intelligence

Ziqi Ye

Fudan University

Ziyang Gong

Shanghai Jiao Tong University

Ning Liao

Shanghai Jiao Tong University

Xiaoxing Hu

Beijing Institute of Technology

Di Wang

Wuhan University

Hongruixuan Chen

The University of Tokyo

Chen Huang

Sun Yat-sen University

Yiguo He

Shanghai Jiao Tong University

Yuru Jia

KU Leuven

Xiaoxing Wang

Shanghai Jiao Tong University

Haipeng Wang

Fudan University

Xue Yang

Shanghai Jiao Tong University

Junchi Yan

Shanghai Jiao Tong University

Related Papers

Loading…