In this page you can download the dataset described on the paper CrowdPupil: A crowdsourced, pupil-center annotated image dataset by David Gil de Gómez Pérez and Roman Bednarik.
The paper is part of the Communication by Gaze Interaction (COGAIN) 2017 symposium inside the European Conference on Eye Movements (ECEM) 2017.
Permission to use the dataset is granted for non-commercial research or educational purposes. For other types of uses please contact dgil[at]uef.fi
If you use it on a publication, please cite the paper mentioned above.
The images on the annotated dataset are named according to this convention:
Img_(Subject Number)_[L|R]_(Image Number).jpg
The L and R refer to images from the left (L) and right (R) eye. The results are provided in a comma separated values file (CSV) that contains, in order, the file name, the X coordinate of the resulting point and the Y coordinate of the resulting point separated by the semicolon (;) character. Both coordinates are given in pixels.