Predicting human eye fixations


Toronto Data set


Input Image

Superimposed eye fixations

Human density map

Itti et al. (1998)

Harel et al. (2007)

Torralba et al. (2006)

Hou and Zhang (2007)

Zhang et al. (2008)

Bruce and Tsotsos (2009)

Seo and Milanfar (2009)

Goferman et al. (2010)

Our approach (Model 1)

Our approach (Model 2)

Our approach (Model 3)

Our approach (Model 4)

Input Image

Superimposed eye fixations

Human density map

Itti et al. (1998)

Harel et al. (2007)

Torralba et al. (2006)

Hou and Zhang (2007)

Zhang et al. (2008)

Bruce and Tsotsos (2009)

Seo and Milanfar (2009)

Goferman et al. (2010)

Our approach (Model 1)

Our approach (Model 2)

Our approach (Model 3)

Our approach (Model 4)

Input Image

Superimposed eye fixations

Human density map

Itti et al. (1998)

Harel et al. (2007)

Torralba et al. (2006)

Hou and Zhang (2007)

Zhang et al. (2008)

Bruce and Tsotsos (2009)

Seo and Milanfar (2009)

Goferman et al. (2010)

Our approach (Model 1)

Our approach (Model 2)

Our approach (Model 3)

Our approach (Model 4)


MIT1003 Data set


Input Image

Superimposed eye fixations

Human density map

Itti et al. (1998)

Harel et al. (2007)

Torralba et al. (2006)

Hou and Zhang (2007)

Zhang et al. (2008)

Bruce and Tsotsos (2009)

Seo and Milanfar (2009)

Goferman et al. (2010)

Our approach (Model 1)

Our approach (Model 2)

Our approach (Model 3)

Our approach (Model 4)

Input Image

Superimposed eye fixations

Human density map

Itti et al. (1998)

Harel et al. (2007)

Torralba et al. (2006)

Hou and Zhang (2007)

Zhang et al. (2008)

Bruce and Tsotsos (2009)

Seo and Milanfar (2009)

Goferman et al. (2010)

Our approach (Model 1)

Our approach (Model 2)

Our approach (Model 3)

Our approach (Model 4)

Input Image

Superimposed eye fixations

Human density map

Itti et al. (1998)

Harel et al. (2007)

Torralba et al. (2006)

Hou and Zhang (2007)

Zhang et al. (2008)

Bruce and Tsotsos (2009)

Seo and Milanfar (2009)

Goferman et al. (2010)

Our approach (Model 1)

Our approach (Model 2)

Our approach (Model 3)

Our approach (Model 4)