For long time now, sea turtle researchers have been exploiting the unique scale pattern of sea turtle individuals to identify them from images and study several aspects of their ecology and biology. These patterns are unique to every individual and stay stable in time (like human fingerprints!). The process, called sea turtle photo-identification (photo-ID), is nowadays used in projects all over the world and examples of studies facilitated by it include in-water behaviour studies, occurrence of injuries, progress of diseases, population monitoring and measuring ecotouristic impact on individual turtles.

The way sea turtle photo-ID is traditionally done is via comparing the facial geometric scale patterns of the same sides of turtles’ heads. This is because these are different from side to side on any individual sea turtle. Therefore, any new image showing the profile of an individual (left or right) is only compared to images of previously identified individuals showing the same side (left or right).

A recent study used artificial intelligence to push the boundaries of the method. The study was led by Kostas Papafitsoros, long term member and researcher of ARCHELON and Lukáš Adam, a researcher from Czech Republic who had also been volunteer in ARCHELON Zakynthos project. ARCHELON’s scientific committee member, ALan F. Rees, also participated in the study.

One of the main findings was that the similarity of left and right profiles of the same turtle is higher than the similarities of profiles of different turtles.

Try to verify that yourself! Can you match the left side of "Viktoria" with her right side?

The same study showed that artificial intelligence algorithms, including one previously developed by the team, can detect this left-right similarity in three sea turtle species. The algorithms can match left and right profiles of the same individual sea turtles which have otherwise no spatial overlap. This results in higher photo-ID matching accuracies and in the future, it might be sufficient to photograph only one side of a turtle’s face when collecting photo-ID data, allowing for more flexible data collection protocols.

Read the research paper here