Inspiration
In Marin county there are 2 primary public transity systems: SMART, and Marin Transit bus lines. To me, this used to be the wolrd of transit, but that changed very quickly. In 2025 I rode Bay Area Rapid Transit for the first time, travelling from El Cerrito Del Norte to Downtown Berkeley. Witnessing the speedy 80mph maximum and the ease of travel made me realize just how useful this type of transit was, yet when discussing with others they feared the danger of riding on transit. This inspired me to try and lower this concern with TransitGuard.
What it does
TransitGuard listens to bart trains and detects danger within the trains such as violence. It then reports this to authorities so they can assist the situation. This helps to motivate people not to commit violent acts on public transit.
How we built it
Transitguard uses a microcontroller and microphone to listen to trains and sends that data to a local machine learning model which evaluates how likely something violent is occurring. If so, it calls a flask server to update a website for drivers and law enforcement to observe.
Challenges we ran into
Synchronizing the microphone and the ML model posed a challenge due to the varying audio quality, however this was fixed by trying to raise the audio quality of the microphone.
Accomplishments that we're proud of
I am very proud of the ML model which during testing reached accuracy as high as 97%.
Log in or sign up for Devpost to join the conversation.