This repository contains code and data used for the final project in the class "Computational Cognitive Science III" at the University of Copenhagen.
- Code to determine the dominant eye of a viewer.
- Code to analyse which fixation point (longest duration, first, last, etc.) usually lies on the object.
- Code to explore whether the fixation point with the longest duration yields better segmentation masks.
This folder contains the eye tracking data (fixation points) extracted from the POET dataset.
- Code to analyse the eye tracking data of the POET data set and store it as a .csv and .pkl file for further processing.
- Code to calculate the Dice coefficient between segmentation masks generated by passing SAM human fixation points and points generated by Ablation or GradCAM. Addiotnally, performs Wilcoxon signed-rank tests to see whether differences in Dice coefficients between human and model generated ppoints are significant.