An EARTH initiative
Peer-reviewed papers from the earth observation field, summarised for researchers, practitioners, and decision-makers.
Papers · updated weekly
Featured paper
A deep learning model detects nearly 500,000 wildebeest and zebra across the Serengeti-Mara from 38 to 50 cm satellite imagery, with an F1-score of 84.75%. It is the first demonstration that very large wildlife populations can be censused automatically from space.
Key finding
A deep learning pipeline detects nearly 500,000 wildebeest and zebra in the Serengeti-Mara from 38 to 50 cm satellite imagery with an F1-score of 84.75%, showing wildlife populations can be censused from space.
All research
Sort by
Showing
0
of
0
papers