AI app helps scientists match dinosaur footprints across millions of years

A team of paleontologists has built an app that does something surprisingly tricky: it looks at a dinosaur footprint and finds others like it, without needing to know which dinosaur made it in the first place.

The app, called DinoTracker, uses artificial intelligence trained on 2,000 unlabeled footprint silhouettes. Instead of relying on human guesses about which footprints belonged to which dinosaurs—guesses that might be wrong—the system learned to spot patterns by itself. It measures eight key features: how spread the toes are, how much the foot pressed into the ground, where the heel landed, and more.

"When we find a dinosaur footprint, we try to do the Cinderella thing and find the foot that matches the slipper," said Prof Steve Brusatte from the University of Edinburgh. "But it's not so simple, because the shape of a dinosaur footprint depends not only on the shape of the dinosaur's foot but also the type of sand or mud it was walking through, and the motion of its foot."

The problem DinoTracker solves is real. Paleontologists have spent decades labeling footprints, assigning them to species based on educated guesses. But those guesses can compound into errors. Feed bad labels into an AI system, and it learns the wrong patterns. Dr Gregor Hartmann from Helmholtz-Zentrum in Germany puts it bluntly: "You never find a footprint and alongside it the dinosaur that had made this footprint. So, no offence to palaeontologists and such, but most likely some of these labels are wrong."

By training on unlabeled footprints instead, the researchers created something more honest. Upload a footprint silhouette to the free DinoTracker app, and it shows you the seven most similar prints in the system. You can also tweak those eight features and watch how the matches change—turning it into a kind of learning tool for understanding what actually matters when comparing ancient tracks.

What the Footprints Reveal

The system clusters footprints the way human experts would expect about 90% of the time, though paleontologists still need to verify details like the age and material of each print.

One striking finding: a set of footprints from the Triassic and early Jurassic periods look remarkably birdlike—despite being around 60 million years older than the oldest known bird fossils. This doesn't necessarily mean birds walked the earth that early. Brusatte suspects meat-eating dinosaurs with very birdlike feet made those tracks instead. Dr Jens Lallensack of Humboldt University offers another possibility: a regular theropod's foot might have simply left a birdlike impression when it sank into soft ground.

What matters is that DinoTracker gives paleontologists a clearer way to ask these questions. By removing human bias from the initial sorting, the app reveals patterns that might otherwise stay hidden in decades of accumulated assumptions. The next phase will be using it to untangle which footprints actually belong to which dinosaurs—a puzzle that's only gotten more solvable now that the sorting is honest.