Sensory Maze: A camera-based remote expression therapy interface
Korak Sengupta · 2023
In typical social settings, it can be difficult for autistic people, especially children, to express their feelings or recognize facial expressions. Even though modern communication technologies have made progress in making basic communication easier, they frequently lack the range of expression that is required for effective emotional communication. Emotion Expression Recognition (EER) is an arising field that means to peruse and decipher the feelings communicated by people continuously. In this paper, we present a user interface called Sensory Maze, that aims to teach how to express emotions, or how to identify emotions made by other people. Sensory Maze is a game that consolidates EER as a fundamental part, permitting users to obtain emotion/expression therapy in their home setting. Sensory Maze relies on a webcam that captures the facial images in real-time and uses faceapi.js to run facial expression classification models on the browser to give real-time feedback. The system has three distinct levels, starting from a virtual therapy session where the subjects are given feedback on the expression they are asked to make, to a level where a situation is presented to them, and the system verifies if they are reacting with the correct emotion. We also present a study of the performance of the system with respect to parameters such as illumination condition, presence of other subjects in the scene, and eyeglasses worn by the subject.