Utilizing Ai to Support Conversations in Individuals With Asd

This project explores the feasibility of integrating Computer Vision and Natural Language Processing to support conversations in individuals with autism spectrum disorder (ASD). Our goal was to develop the framework and prototype of a system capable of generating meaningful narratives from both images and videos. However, due to technical and resource constraints, we refined the scope to focus solely on image-based inputs. We implement vision-language models to generate simplified, contextually relevant captions that can be adapted to the user's communication level through a most-to-least prompting strategy. The end product is a web-based prototype using Streamlit. The prototype features an accessible interface designed for users with ASD, text-to-speech functionality, and adjustable prompt fading controls. This system establishes a foundation for future research aimed at improving the quality of narrative generation, incorporating user feedback, and implementing a scalable platform that supports expressive communication and social participation for individuals with ASD.