Iván Gonzáles Díaz: Using gaze and visual attention to decode intention and predict human actions

Datum / čas
Date(s) - 01.07.
13:30 - 14:30

Místo konání
CIIRC ČVUT 102 A

Kategorie


Lecture: „Using gaze and visual attention to decode intention and predict human actions“

Iván Gonzáles Díaz
Universidad Carlo III
Madrid, Spain

you do not need previous registration

Abstract: In this talk I will present our last research to decode intention and predict human actions. I will focus on the use of human visual attention and gaze, and their connections to perception, intentionality and planning. Among the different sensory signals that elaborate perception-anticipation, vision occupies a key position as visual searches can be considered experiments generating sensory data necessary for scene perception during a task. Based on this assumption, I will present a multi-task approach that learns interpretable models of how humans use their gaze to generate experimental data supporting a given task, and demonstrate that eye-movements, together with other physiological signals and the analysis of the visual field, can be used to decode human intention and predict human actions. Our target application is assistance to upper-limb amputees wearing neuroprostheses, by detecting the human intention of grasping a particular object in complex and cluttered environments with many distractors.