Details

Autor(en) / Beteiligte

• Riche, Nicolas

Titel

Databases for Saliency Model Evaluation

Ist Teil von

• From Human Attention to Computational Attention, p.191-207

Ort / Verlag

New York, NY: Springer New York

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• The comparison between saliency algorithms needs two prerequisites: a dataset of stimuli with a ground truth on which the algorithms can be compared and a metric which measures in an objective way how close an algorithm and the ground truth are. This chapter focuses on the stimuli datasets and the ground truths. In computer vision, the databases for the modeling of visual attention contain two kinds of ground truth: eye movement recording and salient region labeling. The stimuli are still images or videos. A review of the main databases is achieved and more details are given for the most used among them. Stimuli differences are also shown.