Epsilon Archive for Student Projects

Abstract

Recent technical advancements within the field of ecology have increased the ability to collect scientific data. This change in effectivization contrast the time consuming process of processing said data. This effectivization is exemplified in the use of camera traps for ecological research. The contrast between the relative ease of data collection and the time and resource intensive process of managing the data creates a bottleneck. Such a bottleneck in the flow of information warrants solutions. Automating data processing through pattern recognition software offers a potential solution to circumvent said bottleneck. Unpredictable performance of automating software warrants the implementation of smaller scale studies evaluating performance on a given sample of images. This is a necessity before large scale studies can utilize the methods on the whole dataset. The purpose of this study was to evaluate whether software used to automate the identification of individuals could be utilized on a collection of images from a camera trap project at the Ol Pejeta conservancy in Kenya. The image recognition programs Hotspotter and IBEIS were used to evaluate the performance on a set of images. Hotspotter and IBEIS produced a top-1 accuracy of 78,9% and 76,5% respectively. In the greater context of image recognition performance for ecological purposes these results were contextualized and shown to be adequate for the implementation of both Hotspotter and IBEIS for use in future projects. Future implications of automated identification of individuals within populations is explored through a lens of societal, sustainability-related and ethical implications. Such implications include replacing physical tags when possible, enabling a new frontier of citizen science and improving current research methods. The methodological significance of individual identification is vast and with this freedom comes the possibilities to pose a novel range of questions in the interest of broadening the field.