Human presence in Antarctica has grown over the past century, raising concerns about how buildings, camps, and other structures may affect the continent's fragile environment. To better track and understand this impact, our research tested whether artificial intelligence (AI) could help map human-made structures using degraded aerial images. We used a powerful AI tool developed by Meta, called the Segment Anything Model, to find and label structures like buildings, tents, fuel tanks, and vehicles in high-resolution photos of Antarctic research sites. The model worked well for larger structures, like buildings, and moderately well for smaller ones such as research tents. We also tested how image quality affected the model's ability to detect features, finding that permanent stations can be mapped under the use of coarser images, while more temporary field camps require finer detail. This suggests that satellite images, already available at certain resolutions, could be used to semi-automatically monitor human activity across Antarctica. Our method could help scientists and policymakers track how humans are changing the landscape, and support efforts to protect the pristine environment of Antarctica.