“Inferring Gaze Shifts from Captured Body Motion” by Rakita, Pejsa, Mutlu and Gleicher
Conference:
Type:
Entry Number: 77
Title:
- Inferring Gaze Shifts from Captured Body Motion
Presenter(s)/Author(s):
Abstract:
Motion-captured performances seldom include eye gaze, because capturing this motion requires eye tracking technology that is not typically part of a motion capture setup. Yet having eye gaze information is important, as it tells us what the actor was attending to during capture and it adds to the expressivity of their performance.
In this paper, we propose an approach for automated inference of gaze shifts and fixations by analyzing the kinematic properties of a captured body motion. Our focus is on gaze shifts, the intentional movement of the eyes, head, and upper body toward specific targets. Our system accurately infers gaze shift timings and gaze fixation targets . Applications of the inferred information include hand-authoring and automated synthesis of the missing gaze behavior, as well as multimodal analysis of the captured behavior.
Keyword(s):
Acknowledgements:
This work was supported in part by NSF award NRI-1208632.
