Inspiration
While taking an online exam I thought to myself there is no way for the proctor to see if students are lying or not. So, we tried to solve that problem.
What it does
figures out if a person is lying or not based upon the dilation of their pupil
How we built it
We created a truth or dare game (published on Google Play Store) which detects if a person lied. We use the phone's camera to take a video of the person while they answer a question. A frame by frame image is sent to Google Cloud Platform where the algorithm analysis these images to measures the ratio of pupil to iris, in order to detect for dilation. An answer is sent back to the phone application which lets the user know if the person answering the question told the truth or not.
Challenges we ran into
Splitting a video frame by frame and communicating with GCP. Data collection approach (how close does someone need to hold the phone camera to the eye). Figuring out how to account for the dilation of pupil. Distinguishing between pupil & iris. No Dataset available on the internet to train our classifier.
Accomplishments that we're proud of
Implemented a model that actually works and can be improved upon. Solving a real-world problem.
What we learned
Determination to solve a problem gets things done.
What's next for Eye See Your Lies
Improve accuracy by thoroughly training our classifier. Better UI/UX Multiplayer feature
Log in or sign up for Devpost to join the conversation.