Abstract

Many fundamental questions in ecology require the ability to recognize animals that have previously been encountered. Bat ecologists often mark individuals with uniquely numbered bands, also called rings, placed on the forearm. However, there are renewed concerns about the safety of this practice, leading some to look for alternative approaches, such as identifying individuals based on the pattern of collagen-elastin bundles in the wings. Visually identifying these bundles is a labor-intensive task. To help create an automated tool for identifying bats based on wing photographs, we developed a machine learning approach for recognizing collagen-elastin bundles, which appear as lines, in wing photographs. Future work will include training a machine learning model to identify individual bats based on a large dataset of photographs. Our approach holds promise in allowing bat ecologists, who often take pictures of wings in their research, to use these photographs to create their own databases of previously encountered animals for population studies.