Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 20, 2026SensorsOpen Access

Synthetic AI-Generated Satellite Imagery to Improve Earth Observation-Based Neural Networks

View Full Paper
Ask AI
Bookmark
Share

Authors

EAEnrique Albalate-PrietoNVNoelia VállezJEJosé Luís Espinosa-Aranda

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates the effectiveness of synthetic data in improving satellite imagery analysis, suggesting strong benefits for Earth observation.

Key Points

  • This research aims to improve Earth observation through the use of synthetic AI-generated satellite imagery.
  • Implemented Pix2Pix, CUT, and ControlNet models to synthesize satellite imagery.
  • Trained U-Net instances for building, road, and water segmentation tasks using synthetic datasets.
  • Tested performance on independent authentic imagery to analyze generalizability.
  • Increased maximum Dice scores by 0.9% (to 54.1%) for buildings.
  • Increased maximum Dice scores by 2.3% (to 38.6%) for roads.
  • Increased maximum Dice scores by 4.1% (to 46.5%) for waterbodies.

Cite This Study

Albalate-Prieto et al. (2026) studied this question.

synapsesocial.com/papers/6a363147db0793dc1a5382eahttps://doi.org/10.3390/s26123895
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Generating Synthetic Satellite Imagery for Rare Objects: An Empirical Comparison of Models and Metrics2024
  2. 2Generating Synthetic Satellite Imagery With Deep-Learning Text-to-Image Models -- Technical Challenges and Implications for Monitoring and Verification2024 · 1 citations
  3. 3Synthetic Data Matters: Re-training with Geo-typical Synthetic Labels for Building Detection2025
  4. 4Generative AI for Urban Planning: Synthesizing Satellite Imagery via Diffusion Models2025
  5. 5Synthetic Data for Sentinel-2 Semantic Segmentation2024 · 2 citations