• Pipeline for Preprocessing
  • Node Degree Centrality
  • Seahaven Paradigm
  • Landmark
  • Gaze Graph
  • Application on Gaze-Based-Paradigms

Outline

PARADIGM

Exploration in Seahaven

  • 30 minutes in 3 blocks
  • 22 young participants
  • Data Collection?

PARADIGM -- Data

  • 30 Hz Eye Tracking Data
  • Raycasting (like a Superman!)
  • Building ID/NoHouse/NoData
  • HMD Position Data

3. Check missing data ratio -- Threshold 30%

4. Interpolation (Walter et al. 2022)

5. Gaze-Noise Label (Gaze defined > 266 ms)

2. Condense Data -- Merge Same Gaze Seq.

1. Format Transform (JSON to CSV)

PIPELINE OVERVIEW

Gaze Graph

Gaze Graph

NODE DEGREE CENTRALITY

Def: # edges connecting to a node

Possible measures:

  • Mean + SD for each participant
  • Mean + SD for each node (building)

NODE DEGREE CENTRALITY

Graph of a participant

Mean + SD for each participant

Mean + SD for each building

LANDMARK

gaze-graph-based landmark

Mean Node Degree

across buildings and participants

:=

2 * std. dev. Node Degree

across buildings and participants

+

Node degree centrality

>=

APPLICATION

  • Not limited to navigation
  • ROI definition restricted to objects
  • Applicable to any task similar to seek & destroy
  • Task-oriented gaze definition (navigation > 250 ms)

THANKS!

Made with Slides.com