The Feasibility of RGB-D Gaze Intention Measurement in Children with Autism Using Azure Kinect.

Area of Science:

• Computer Vision
• Developmental Psychology
• Biomedical Engineering

Background:

• Gaze interpretation is crucial for social communication, yet challenging for children with autism spectrum disorder (ASD).
• Existing gaze tracking technologies (RGB, RGB-D) face limitations in accurately measuring gaze direction.

Purpose of the Study:

• To propose and evaluate a novel approach using the Azure Kinect for improved gaze intention measurement in children with ASD.
• To provide accurate estimations of gaze direction and head position for enhanced social interaction analysis.

Main Methods:

• Utilized the Azure Kinect sensor for data acquisition.
• Conducted an experimental trial with eight participants of varying statures.
• Analyzed the intersection of estimated gaze with the target plane over 38-minute sessions.

Main Results:

• Achieved high accuracy in gaze measurement.
• Minimum angular error of 2.5° (pupil position) and 2.06° (head orientation).
• Average errors of 4.46° (pupil position) and 3.19° (head orientation).

Conclusions:

• The Azure Kinect-based approach offers a promising solution for accurate gaze tracking in children with ASD.
• This technology facilitates better assessment of social intent and interactions, leading to more precise clinical evaluations for autism.