Former Lab Assistant Colin Murphy and Post-Doctoral Fellow Melissa Collier published an article in the Science of the Total Environment regarding the utilization of machine learning in assessing body conditions and health indicators of dolphins in the Potomac River and Chesapeake Bay.
You can read their article by clicking this link!
“Photographic health assessments are one of the best ways to monitor dolphin health, because it doesn't require taking any samples from the dolphins, which is an invasive and costly tactic. Skin lesions and scars captured in the photos we collect every field season can tell us a lot about stress and disease, but it takes a long time (hundreds of hours) to search for them in every single dolphin photograph. So we built an object-detection algorithm that automatically finds these small skin patterns. You can think of it like the iPhone photo search feature: type in “dog” and it gathers all your dog photos. Using this algorithm speeds up the manual processing of our 100,000+ photographs for skin lesions and scars, and gives us an outlook into the dolphin's health and stressors without having to use invasive methods,” says Dr. Collier when describing their paper.
