Deep learning model generates lunar elevation maps from single satellite images

Researchers have developed LunarDepthNet, a novel deep learning model designed to generate detailed Digital Elevation Models (DEMs) of the lunar surface using monocular satellite images. The model employs a UNet architecture with an EfficientNet encoder and custom layers to accurately interpret shadow information and estimate elevation. LunarDepthNet achieved a mean nRMSE of 0.437 and an MAE of 4.5m, demonstrating its capability to produce reliable elevation maps, particularly for lunar regions lacking stereo imagery. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Enables creation of lunar elevation maps from single images, aiding lunar exploration and mission planning in data-scarce regions.

RANK_REASON This is a research paper detailing a new deep learning model for generating lunar elevation maps.

Read on arXiv cs.CV →

• LunarDepthNet
• UNet
• EfficientNet
• Chandrayaan-2
• DEM
• DTM
• arXiv

• paper
• other