Skip to content

Latest commit

 

History

History
28 lines (15 loc) · 1.41 KB

README.md

File metadata and controls

28 lines (15 loc) · 1.41 KB
Raw

Saliency Maps

The saliency maps are generated using edge splatting technique. In splatting multiple textures with different alpha maps over a mesh. In our project we augmented the method devised by Krueger et al [1].

  • In the first step, a gaussian kernel of fixed standard deviation (10% of max-pixel width) and saccades amplitude as max gaussian value is applied to all the fixation points. The values at the overlapping pixels are added.

  • In the second step the density values are mapped to a color gradient.

Code

We use git gist to save the script for the saliency maps. Our python script to generate and compare saliency maps can be found at git gists

Scores (Performance Metrics)

  • Mean Squared Error (MSE)

where, (m, n) are dimentions of image I and K

MSE score distribution for all 170 pic_IDs

SCORES

References

[1] Krueger, Robert & Koch, Steffen & Ertl, Thomas. (2016). SaccadeLenses: Interactive Exploratory Fltering of Eye Tracking Trajectories. 31-34. 10.1109/ETVIS.2016.7851162.