Clarity

Team Members
Issues Page
Detailed report of issue
Settings Page (on web-app)
Customization Settings + Features for those with impairments
Customization Ctd.
More Permissions
Detection Model Identifying A Group of Chairs with visual proximity considered
Picture-In-Picture working on Instragram Reels
Detection Model Identifying People in Proximity (with respect to LiDAR)

Inspiration

Our surroundings and past experiences inspired us to build the next generation safety detection system that can prevent small trips to surgery trips. We are the victim of our society; everyday every second people are always staring down at their screen, they may be doing it on their way to class or on their way to work, but the one thing we can never let go is our phones. While trying to enjoy using our phones for whatever purpose on our short walks we have to deal with uneven pavement crosswalks, obstacles, and crowded streets. These quick walks make us more injury prone because we're too focused on whatever were doing that we don't see whats beneath or ahead of us. But that made us think, what about people with disabilities, they have a lifetime struggle of not being able to properly see, we can't forget about people with color blind people as well. The solution to all these issues is Clarity.

What it does

Clarity is easy to navigate and access for everyone to support their enjoyment and safety. A algorithmic detection system that detects the path ahead of you while you enjoy your phone time keeping you safe at all times. With our smart repeating vibration, text to speech, notification system, accurate distance measurement, increased visualization ability, automatic algorithmic issue reports and color blindness support, the chances of you running into something are less than 1%.

How we built it

For iOS App stack used is Swift for frontend, Xcode for app development, node.js + python training for backend. Used Reinforced Learning Human Feedback and Machine Learning for training. For Web App stack used is HTML frontend , CSS frontend, and java script frontend/backend. Used browser camera api's with simple algorithm detection system, web api's for sound, speech, and vibration notifications.

Challenges we ran into

LiDar scanning, detection bugs Cross platform IOS + android issues Camera not working in the background

Accomplishments that we're proud of

Learned to train an AI model Switch from web application to an IOS native app quickly

What we learned

Model Training Pivoting when errors come last minute

What's next for Clarity..

Options for Androids (LiDar limtation) Reduce latency delays More reliable hazard detection (crowds, bad lighting, etc) Accurate distance and direction Testing with blind / low vision individuals