Project Description
Different types of annotators vary by their task-specific expertise, their costs and availability. While domain experts are more likely to provide high-quality labels, they are more costly and less frequently available. In this study we compare the annotation quality of multiple profiles of annotators, ranging from domain experts to laypersons. In addition, we assess the capability of GPT vision models in image annotation. Substantively annotators are categorizing satellite images of industrial land in three classes, relevant to Fraunhofer IIS.
Contact Person
Jacob Beck