GEONN for Satellite Image-to-Map Translation via Enhanced Geospatial Processing

View at publisher Summarise with VIRAT
Authors: Satya Sandeep, Rama B Bhargavi, J Varsha, Bhavana K Mythili, V Varshini
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Satellite image-based automatic map generation is crucial for urban planning, environmental surveillance, and disaster response. Current methods like Convolutional Neural Networks (CNNs), Variational Autoencoders (VAEs), and Conditional GANs (C-GANs) have major drawbacks: CNNs are challenged by geographical heterogeneity, VAEs sacrifice reconstruction quality, and C-GANs are data-diversity-sensitive, lowering accuracy and recall. To overcome these issues, we introduce GeoNN (Geographical Neural Network), a deep learning architecture specifically designed to improve spatial relationship learning and geospatial feature extraction. GeoNN uses multi-scale processing to successfully capture local fine-grained features and large-scale geographic patterns, allowing for better and high-resolution map generation. Experimental results demonstrate that GeoNN surpasses other models with accuracy >0.95, precision up to 0.97, and recall of 0.96, with loss converging to around 0.05. In addition, GeoNN enhances explainability and computation speed, becoming a strong contender for large-scale geospatial tasks. Having broken the bar of conventional practices, GeoNN provides a new benchmark for satellite image-to-map translation, bringing more trustworthy and actionable geographic intelligence for real-world use cases.

Fields of Study & Subjects

Remote-Sensing Image Classification Satellite Image Processing and Photogrammetry Generative Adversarial Networks and Image Synthesis

Publication details

Journal / source
CRC Press eBooks
Pages
397–405
Publication date
25 March 2026
Language
EN
Document type
Book chapter