Within the past decade, many computational approaches have been developed to estimate gaze directions of persons based on their facial appearance. Most researchers used common face datasets with only a limited representation of different head poses to train and verify their algorithms. Moreover, in most datasets, faces have neither a defined gaze direction, nor do they incorporate different combinations of eye gaze and head pose. Therefore, we recorded an extended dataset of 20 subjects including faces in various combinations of head pose and eye gaze leading to a total amount of 2220 colour images (111 per subject). The images were produced under controlled conditions, i.e. we used a technique to make sure that the subjects adjust their head and eyes appropriately. Furthermore, all images are manually labelled with landmarks indicating important features of the face. Finally, we evaluate the dataset with two computational methods for head pose and gaze estimation respectively.