Prediction of gaze direction using Convolutional Neural Networks for Autism diagnosis

TL;DR

This paper presents a CNN-based algorithm that predicts gaze direction in real-time, offering a fast and accurate method for autism diagnosis by analyzing children's gaze preferences during video viewing.

Contribution

The study introduces a novel CNN approach for gaze prediction tailored for autism diagnosis, demonstrating real-time performance and high accuracy.

Findings

Achieved real-time gaze prediction response.

Demonstrated high accuracy in gaze direction classification.

Potential for rapid autism screening using gaze analysis.

Abstract

Autism is a developmental disorder that affects social interaction and communication of children. The gold standard diagnostic tools are very difficult to use and time consuming. However, diagnostic could be deduced from child gaze preferences by looking a video with social and abstract scenes. In this work, we propose an algorithm based on convolutional neural networks to predict gaze direction for a fast and effective autism diagnosis. Early results show that our algorithm achieves real-time response and robust high accuracy for prediction of gaze direction.