Visonary

Inspiration

Caroline's sister was born with eyesight problems resulting in years of difficulty navigating her own home. While she was able to recover through glasses, millions of individuals continue living their life with visual impairments with no way of receiving eye care. According to the World Health Organization, nearly 90 million people lack access to corrective glasses. Thus, we present Visionary.

What it does

Visionary aims to assist blind and visually impaired (BVI) individuals with navigation, powered by OpenAI. difThe project is a no-cost option for BVI individuals to tremendously improve their navigational abilities. Although 75% of BVI people report difficulty navigating public spaces, Visionary provides the essential assistance they require to protect their safety.

Visonary runs on OpenAI's API to provide a thorough analysis of the user's surroundings, offering live feedback to provide the navigational assistance needed to power better human-environment interaction. After receiving the image analysis, ChatGPT speaks naturally in order to provide navigational info to the BVI individual.

How we built it

Visionary is built on NextJS, a powerful framework extended off ReactJS. Javascript, HTML, and CSS power our responsive and interactive website with smooth animations and aesthetic visuals.

Challenges we ran into

A major challenge we faced was the amount of time needed to analyze the surroundings. Originally, our program compressed the image and converted it to a base64 string before sending it to OpenAI for analysis, which averaged 20 seconds, far too slow for practical use. We solved this by converting the image to a URL via a Blob, drastically reducing the amount of data the program needed to process. As a result, the analysis now takes an average of just 3 to 5 seconds, restoring the real-time effectiveness of our program.