Eyelid and Iris Annotations for Three Public RGB Datasets

RGB cameras are widely spread and present in nearly any mobile phone. For a utilization of those devices for medical applications like supportive diagnosis systems, a rich database of facial features is required. Most modern datasets are either very small or do not contain annotated eye features like the eyelid or the iris. We propose annotations for three huge public datasets containing the eyelid and the iris. With those features it is possible to train DNNs for the extraction and use them in real world applications. We hope that this contribution will push the limits of application areas of eye tracking further.