Accidents happen on the road. Drivers use their phones to alert police/medical care. We want to decrease delay for information to pass to the important individuals.
What it does
Sound after a crash is recorded. Processed for loudness. Sends que to dashboard cam to take a picture. The picture goes through a machine model to get a prediction on whether it was an accident or not.
How we built it
Python scripts to take snapshots of sounds and pictures. Google cloud storage and AutoML vision to predict accident.
Challenges we ran into
Accomplishments that we're proud of
Running prototype, although not that accurate due to insufficient training data for visuals.
What we learned
ML is amazing. Period.
What's next for AutoSight