AI Identifies Dinosaur Footprints With 90% Accuracy – New App Launched

For over a century, dinosaur footprints have offered a tantalizing glimpse into the past, but interpreting them has always been a challenge. These aren’t simple impressions; they’re records shaped by squishing mud, sliding toes, collapsing edges, and centuries of erosion. Now, a new tool leveraging artificial intelligence is aiming to bring order to this complex field, and it’s already yielding unexpected insights – potentially rewriting our understanding of when birds first emerged.

Researchers have developed DinoTracker, a mobile app that allows users to upload a photo or even a sketch of a dinosaur footprint and receive an instant analysis of the dinosaur species that likely created it. The app isn’t intended to replace expert paleontological analysis, but rather to act as a consistent, data-driven second opinion and to accelerate the process of identifying and classifying these ancient tracks.

AI Was Trained to “See” Variation

Traditional footprint research relies heavily on expert judgment and comparison with known examples. Older computer-based methods often required researchers to manually compile datasets, assigning tracks to specific dinosaurs – a process that can introduce bias. The team behind DinoTracker, led by researchers at the Helmholtz Research Center in Berlin, working with colleagues at the University of Edinburgh, took a different approach. Instead of forcing footprints into rigid categories, they trained their algorithms to recognize the natural variations that occur in tracks.

The AI was trained on nearly 2,000 real fossil footprints, but crucially, it also learned from millions of simulated variations. These simulations replicated the effects of compression, edge displacement, and other distortions that naturally occur in fossilization. This allowed the system to focus on key traits that distinguish trackmakers even when the print isn’t perfect, such as toe spread, heel position, contact area, and weight distribution. The research, published in the journal PNAS, details this method.

When AI Agrees with Experts – and Challenges Assumptions

After training, the model was tested by asking it to predict the dinosaur responsible for existing fossil tracks. The algorithm achieved approximately 90 percent agreement with classifications made by human experts, even in cases that are typically debated. This level of accuracy doesn’t equate to absolute certainty – footprints can be ambiguous, and paleontology often deals with the most plausible interpretations – but it provides a valuable, consistent perspective.

One of the most intriguing findings emerged from the analysis of very old footprints, dating back more than 200 million years. The AI flagged several tracks exhibiting unusually bird-like features, resembling prints associated with both extinct and modern birds. This observation suggests two possibilities: birds may have originated tens of millions of years earlier than currently believed, or some early dinosaurs possessed feet that coincidentally resembled those of birds. This doesn’t resolve the debate, but it highlights the potential for footprints to contain previously overlooked signals about evolutionary history.

The implications extend beyond simply refining timelines. Understanding the subtle variations in dinosaur footprints can reveal details about their gait, speed, and even social behavior. As Earth.com reported previously, the search for the origins of dinosaurs is ongoing, and new discoveries continually reshape our understanding of these ancient creatures.

Re-examining Scotland’s Puzzling Tracks

DinoTracker was also applied to puzzling footprints discovered on the Isle of Skye in Scotland. These tracks, made around 170 million years ago on a muddy lagoon shore, have been difficult to definitively assign to a specific dinosaur group. The AI analysis suggests that some of the oldest known relatives of duck-billed dinosaurs may have been the trackmakers. If confirmed, this could alter our understanding of when and where this lineage first began to spread.

DinoTracker: From Research Tool to Public Resource

DinoTracker isn’t solely a research demonstration; it’s designed for broader use. Footprints are among the most common types of dinosaur evidence encountered by the public, and an accessible tool could benefit both scientists and amateur enthusiasts. In research, it can quickly screen large numbers of tracks and identify patterns across sites. In education, it transforms footprints into an interactive learning experience. And in the field, it offers a rapid way to test hypotheses, particularly in locations where track interpretation traditionally relies on the expertise of whoever is present.

“This study is an exciting contribution for paleontology and an objective, data-driven way to classify dinosaur footprints,” said paleontologist Steve Brusatte from the University of Edinburgh. “It opens up exciting new possibilities for understanding how these incredible animals lived and moved, and when major groups like birds first evolved.”

The Future of Footprint Analysis

Dinosaur footprint research will likely never be fully settled by an app. Tracks are inherently messy, and the past doesn’t offer definitive answers. However, DinoTracker represents a valuable step forward: a tool that treats variation as information rather than noise. By reliably recognizing how real footprints warp and still connecting them to likely trackmakers, it could accelerate research, broaden participation, and ground debates in firmer evidence.

analyzing footprints allows us to connect with these ancient animals on a more intimate level. A footprint is a fleeting moment of contact between an organism and the earth beneath it. By more accurately reading these moments, we move closer to understanding how dinosaurs truly lived, moved, and evolved. Further research will focus on expanding the AI’s training dataset and refining its ability to account for different sediment types and preservation conditions. The team also plans to integrate the app with existing geological databases to provide more contextual information about track sites.

You can learn more about dinosaur footprints and their preservation at the University of California Museum of Paleontology.