In this paper we present a dataset of visual saliency segments 
associated to social images, collected through a mobile game 
in conjunction with a crowdsourcing campaign. Our data 
are collected in the game itself, where players guess what 
is depicted in a gradually uncovered image. The game mechanics 
allow us to collect information about which image 
segments are required by the user to correctly recover the 
image content. The provided dataset can be applied to both, 
computer vision and image retrieval algorithms, and it aims 
to contribute to the understanding of human visual perception 
and attention. Moreover, the end objective is to test the 
game as a potential substitute to professional eye tracking 
systems.