Deep Learning-enabled Smart Wearables to Assist People with Autism Spectrum Disorder in Dynamic Environments

Danowski, Przemyslaw, Ekpo, Sunday ORCID: https://orcid.org/0000-0001-9219-3759, Ijaz, Muhammad ORCID: https://orcid.org/0000-0002-0050-9435, Raza, Umar ORCID: https://orcid.org/0000-0002-9810-1285 and Carbrero, Raul (2023) Deep Learning-enabled Smart Wearables to Assist People with Autism Spectrum Disorder in Dynamic Environments. In: The Second International Conference on Adaptive and Sustainable Science, Engineering and Technology (ASSET) 2023, 18 July 2023 - 20 July 2023, Ikot Akpaden, Nigeria and Manchester, UK. (In Press)

Accepted Version File not available for download. Download (354kB)

Abstract

The purpose of this work is to explore the possibilities of using robotic emotion recognition model to help people with autism spectrum disorder improve their social skills. An OAK-D lite camera and Lenovo IdeaPad L340 Gaming were used to run a python algorithm that would estimate emotions in two stage Neural Network hybrid deep learning system. The data was gathered from 30 Manchester Metropolitan University engineering students. Results showed that the idea of robotic emotion recognition works, but the algorithm could be improved resulting in better accuracy with the hardware also due for enhancement. However, the project has the potential to expand on to other areas such as public security and mood monitoring services and combine it with other indicators such as heart rate sensors and7 rapid movement detection.