Reflection on Human and Automatic Image Description in Elearning Context

Abstract

Images have a strong presence in educational contexts, particularly in online environments. When images add vital information for the understanding of a given subject, this information must be perceived by all students, including those who have a visual impairment. Although most web tools and platforms have a field for the description or alternative text, most images do not contain this information. Since 2016, some platforms and web services have been providing features, based on Artificial Intelligence, which present a brief description of the images to screen readers. Are these descriptions effective for people with visual impairment, in an eLearning context? Are they enough for a blind person to create a mental image? How do they differ from the description performed by humans? To answer these questions, it is necessary to have an idea of how we perceive images and how they make sense in our brain, according to our values and culture. Thus, in this work we present a reflection related to these questions, using examples of descriptions obtained by the google chrome feature “Get image description” and descriptions made by people in the various editions of the MOOC “Image description in web context”.