MIT300 is a dataset of 300 images of natural indoor and outdoor scenes, as well as corresponding recorded human eye movements. The dataset is intended for evaluation of visual saliency models.
Verified
@inproceedings{Judd_2012,
title = {A Benchmark of Computational Models of Saliency to Predict Human Fixations},
author = {Tilke Judd and Fr{\'e}do Durand and Antonio Torralba},
booktitle = {MIT Technical Report},
year = {2012}
}
