Approach based on heuristic optimization and homography estimation for the study of adaptive interfaces

Abstract

A system is proposed to determine the alterations in concentration and attention, as part of the study of the effects on user performance when using systems with adaptive interfaces. At the same time, solutions are proposed to avoid technical limitations that were present in previous implementations of the project.

Specifically, it is proposed to use a particle swarm optimization algorithm instead of a genetic algorithm, as well as to incorporate a surrogate model that allows the user’s response to be approximated, thereby reducing the need to directly evaluate each configuration of the interface with the user and allowing parallelism.

To detect the concentration of the user, a novel use to the homography estimation is proposed for the detection of the face and the orientation of the head, due to the robustness of its preliminary results compared to other techniques evaluated, as well as the need for a low-cost solution that prevented the use of specialized equipment in gaze tracking.

The results of the different system tests are presented and a series of modifications or suggestions are proposed for future iterations of the project.