Authors

Inspiration

Tackling the problem of traffic jams at the german border

What it does

Imagine you need to control every car at the borderline. We automatically analyze vehicle numbers and assign the detected car to the registered person.

How we built it

  • Using a gopro (or any other camera) as a camera.
  • Analyzing & Detecting vehicle numbers.
  • Mapping detected number to the "Grenzgänger"-Request.
  • Sending "OK" or "Not OK" to the officer's display. If the vehicle could be linked to a request, the picture and the passport of the vehicle driver is directly shown to the officer.

Challenges we ran into

  • Kafka deployment on the production server together with stunnel
  • Websocket notifications
  • Authorization: We use JWT offline tokens to authenticate and authorize the requests.

Accomplishments that we're proud of

  • Good results with error correction.
  • Almost production ready environment with deployment and 5 micro services.
  • No http calls at all (only kafka and websockets)
  • We're happy with the architecture.

What we learned

  • Used kafka the first time.
  • Machine learning error correction.

What's next for Border Crossing Analyzer

Depends on the feedback ;-)

Hackathons

Technologies

django, elasticsearch, googlevisionapi, gopro, kafka, openidconnect, python, vuejs, websockets

Devpost Software Identifier

257385