Thousands of accidents happen on the road every day, but rarely is video footage available. As a result, victims have to endure drawn out court proceedings, authorities have to spend resources investigating, and insurance companies often falsely assign blame. With the DashOwl dashboard camera and cloud processing suite, vehicle crash footage is uploaded by the onboard camera, processed in the cloud, and archived for law enforcement and insurance companies to access.
The camera unit provides the same functionality of a conventional dashcam along with a GPS and cellular connectivity module.
For each crash event detected, the dashcam collects important metadata such as GPS location, date, and time.
Video streams are batch classified using our custom Convolutional Neural Network. Classifier runs on Google Compute Engine.
Videos classified as crashes are stored with their metadata in Firebase Datastore and Real-Time Databases.
As more DashOwl devices collect footage, our collection of crash events will expand and provide more pinpointed information for relevant authorities.
Using Firebase Authentication, we can limit access to the portal to authorized insurance providers and law enforcement officials.
Karthik Kalyanaraman, Akash Singhal, William Wong, and Harshayu Girase (Left to Right). We are second year students studying CS at UC Berkeley. Our love for cool projects, fast cars, and dumb CS jokes brought us together.