Aerial photographic surveys have long been used to monitor gray and harbor seal populations, but extracting data from those images is a time-intensive process. Just one image from a haul-out or pupping site may contain over one hundred seals, which must be manually counted and categorized by sex or age. This project aimed to automate this image analysis and obtain population estimates more efficiently through machine learning, specifically with the software VIAME (Video and Image Analytics for Marine Environments). Survey images from gray seal pupping sites on Monomoy and Muskeget Islands off of Cape Cod, Massachusetts were manually annotated to identify pups and adults, after which the annotations were used to generate a trained detector through deep learning. After 4 iterations of this process, the detector’s probability of correctly identifying a pup or adult of this species on sandy substrates is 0.87. Going forward, this detector will be further refined to provide accurate estimates of gray and harbor seal abundance in various environments in a fraction of the time it would take using traditional methods.