Automated Identification and Classification of Teeth Fossils

Deep-learning algorithms achieve up to 82% precision and 80% recall in classifying fossil teeth from Wadi Al-Hitan, outperforming traditional manual methods. The ResNet50 model demonstrated the highest accuracy (80%) among tested models, providing a reliable automated approach for identifying and extracting information about teeth fossils from sharks, elephants, hyrax, and Afradapis. This method streamlines taxonomic identification, enabling efficient research into past diets, trophic interactions, and environmental conditions. The study validated its performance on a dataset of over 50,000 images, offering potential applications in comparative anatomy and paleontology.

Elsayed, O. R., ElKot, Y. G., ElRefaai, D. A., Abdelfattah, H. M., ElSayed, M., & Hamdy, A. (2023). Automated Identification and Classification of Teeth Fossils. 2023 International Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC), 179–186. https://doi.org/10.1109/miucc58832.2023.10278368