With marijuana legalization inevitably becoming the norm, we anticipate cases of THC-impaired driving will as well. We decided to tackle the issue of THC-impaired driving before it becomes a widespread issue like drunk-driving, and for once, be proactive rather than reactive. We were also inspired to find an innovative and effective method for doctors to diagnose Convergence Insufficiency (CI) Disorder.
What it does
Through computer vision, machine learning, and a nearly-automated version of a field sobriety test conducted by authoritative officials in the modern day, we determine whether the subject is impaired/fit for driving based on THC substance exposure and their ability to converge their eyes. Three subsequent failures of the test will result in the application preventing the subject from starting their vehicle, simulating an IID (Ignition interlock device), locating them and calling a cab based on their geographic location. In terms of diagnosing CI Disorder, failure of a patient to converge their eyes with accordance to the application will provide a Doctor or a Researcher with critical information that will help in the diagnosis of the disorder, and streamline the process.
How we built it
- Front-end in HTML/CSS/Vanilla js
- Back-end in Node.js, MongoDB
- Machine learning/computer vision in Python OpenCV
- Google Cloud Platform and Twilio API
Challenges we ran into
Simulating a stimulus that is gradually getting closer to the subject's eyes/bridge of the nose, formatting and scraping every single cab company based on state and major city, and general Computer Vision and ML work, such as accurate gaze tracking with head tilt compensation.
Accomplishments that we're proud of
Implementing complex algorithms and APIs in a short amount of time and perfectly integrating them to create something that can have a positive impact on the lives of thousands of people around the world.
What we learned
How to combine APIs in interesting and exciting ways, and make batch requests to cloud servers to make data transfer faster and smoother
What's next for WeeDontDrive
Converting the web application into a mobile application for Android and iOS that will allow car companies (like GM) to integrate our application during the manufacturing process of their vehicles, such that the ignition can be locked.