The penguin is now being asked to keep its own census
the penguin is now being asked to keep its own census: computer vision reads guano stains in satellite images to count antarctic colonies, a weakly supervised model learns to segment adelie colonies from 2,044 images across 193 sites, and chilean researchers run thousands of camera trap photos of chinstraps in hours instead of weeks.
the colony now files its own head count.
Context
NASA's Earth Observatory describes how scientists find penguin colonies in remote parts of Antarctica by searching satellite images for guano stains. A CVPR 2019 workshop paper presents a deep learning model for semantic segmentation of Adelie penguin colonies in high-resolution satellite imagery, trained with weak labels from the Penguin Colony Dataset of 2,044 georeferenced cropped images from 193 Adelie penguin colonies.
The Polar Journal, in a report dated May 3, 2026, describes a project of the Chilean Antarctic Institute (INACH) led by marine biologist Magdalena Marquez Diaz that uses AI on camera trap images of chinstrap penguins, and says the AI processes thousands of images that previously took weeks of manual work in a few hours.
The 2,044 images and 193 sites match the CVPR paper. The NASA guano page is from 2017, so using guano in satellite images for counting is an older method, and the note's list of three studies spans 2017, 2019 and 2026.
The weeks-to-hours claim is the Polar Journal's caption, and the Chilean source describes the AI as differentiating adults from chicks in colony images. The note says 'thousands of photos in hours instead of weeks', which follows the Polar Journal wording.
The paper's model segments colony areas, and the note says it learns to segment, which is correct. Whether the segmentation yields a count was not confirmed in the pages read, so 'count' is the author's word for it. 'The colony now files its own head count' is the author's line.
Related work
- Penguin Droppings Are Fertile Ground for Science (NASA Science, June 13, 2017) ↗Source for the guano-in-satellite-imagery method.
- Weakly Labeling the Antarctic: The Penguin Colony Case (CVPR Workshops 2019) ↗Source for the weakly supervised segmentation and the 2,044 images across 193 colonies.
- When AI Counts Penguins: High-Tech in Antarctica (Polar Journal, May 3, 2026) ↗Source for the INACH camera trap project and the weeks-to-hours statement.
Watch next
- Read the INACH project's own publication for its accuracy. Check how the 2019 model's colony areas were turned into counts.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 3 October 2026 at 19:55 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →