Scientists Launch AI App to Identify Dinosaurs From Footprints

Scientists have launched DinoTracker, an AI app that identifies dinosaurs from ancient footprints with about 90% expert-level accuracy.
The team trained the system using 2,000 unlabelled footprint silhouettes instead of preclassified examples.
The AI grouped prints by shared shape features, including toe spread, heel position, and ground contact.
Researchers published the study in Proceedings of the National Academy of Sciences.
Steve Brusatte said footprint shapes reflect both anatomy and ground conditions.
Gregor Hartmann said incorrect labels often mislead earlier AI systems.
The app lets users upload footprints and compare them with similar examples.
The AI supports earlier findings that some Triassic tracks appear surprisingly birdlike.
Brusatte said the tracks likely came from birdlike meat-eating dinosaurs, not true birds.
Humboldt University of Berlin researcher Jens Lallensack urged caution when interpreting the results.