What is it about?
We used AI to identify the key sources of variation in dinosaur footprints in a way that humans can easily interpret. The analysis shows that just eight features are sufficient to capture the essential structure of these tracks. To make our findings accessible to everyone, we also provide a public app called DinoTracker.
Featured Image
Photo by Adam Custer on Unsplash
Why is it important?
This unsupervised technique provides a purely mathematical perspective on footprint data, revealing patterns that are independent of human expectations. By acting as an additional, unbiased expert, the method helps researchers test hypotheses more rigorously and evaluate how reliable their interpretations truly are.
Perspectives
This approach can be expanded to other fossil traces and integrated with more complex models. Originally developed for photon science, the method has now been successfully transferred to paleontology, showing its potential to bridge disciplines. AI‑based analyses may become a key tool for testing hypotheses and improving the reliability of ichnological interpretations.
Gregor Hartmann
Hermann von Helmholtz-Gemeinschaft Deutscher Forschungszentren
Read the Original
This page is a summary of: Identifying variation in dinosaur footprints and classifying problematic specimens via unbiased unsupervised machine learning, Proceedings of the National Academy of Sciences, January 2026, Proceedings of the National Academy of Sciences,
DOI: 10.1073/pnas.2527222122.
You can read the full text:
Resources
Contributors
The following have contributed to this page
