Automated classification of natural habitats using ground-level imagery
Mahdis Tourian, Remy Vandaele, Sareh Rowlands, Max Fancourt, Rebecca Mein, Bethany Hadfield, Alison Saunders, Sophie Potter · 2026
Read Automated classification of natural habitats using ground-level imagery by Mahdis Tourian, Remy Vandaele, Sareh Rowlands, Max Fancourt, Rebecca Mein, Bethany Hadfield, Alison Saunders, Sophie Potter (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0351335.txt.
Description
Knowing which habitat fills a landscape helps protect wildlife, and Mahdis Tourian and colleagues taught a computer to name habitats from ordinary ground-level photos. Their 2026 PLOS One study, built with Natural England, sorts each picture into one of sixteen classes under an established English habitat framework.
The team adjusted and balanced thousands of photos, then fine-tuned a deep learning model first made for outlining objects in images. Repeated cross-checks measured how often the model agreed with human labels. Because the method needs only photos, it could tap the huge supply of pictures from volunteers and field visits, widening habitat mapping beyond satellite views.
Catalog record
| CREATOR | Mahdis Tourian, Remy Vandaele, Sareh Rowlands, Max Fancourt, Rebecca Mein, Bethany Hadfield, Alison Saunders, Sophie Potter |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55175 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |