Samuel Arnold, Dickinson College


Class Year:



Helen Bailey, Ph.D.

Project Title:

Analysis and Categorization of Green Turtle, Chelonia mydas, Carapace using Deep Machine Learning


Recent reports suggest that several green turtle (Chelonia mydas) subpopulations may be recovering after decades of decline. However, these reports are difficult to verify. The current practice of identifying individuals, determining their genetic lineage, and assigning them to a subpopulation requires a large team of scientists capturing individuals to obtain DNA samples, and analyzing specific genetic markers. While DNA sampling is a proven technique, it is time-consuming and expensive, therefore the sample size is typically small and may not be representative of the whole subpopulation. Recent studies have found a relationship between the genetic markers and the shape of an individual’s carapace, suggesting that individuals can be assigned to subpopulations by careful analysis of carapace images. In this study, this technique was further improved by using an image segmentation model and computer vision to automatically isolate the carapace from the rest of an image and prepare it for further carapace analysis. This significantly decreased the time needed to analyze a large collection of images, making the technique a practical alternative to DNA sampling. This technique can be used to distinguish among green turtle subpopulations and verify the recovery of populations.



REU Update

Geographic Information System Analyst with Inteliquest

The Blue Crab: Callinectes Sapidus

An essential resource for researchers, students, and managers.  Get your copy today!

pile of cooked crabs