Moose - An Early Traffic Warning System

• 

  Landing / Registration

• 

  Detected dangers's locations

• 

  Notifications

Inspiration

• 1,824 accidents involving members of the deer family (elk, deer, moose) took occurred in 2017. These collisions in Finland typically takes place on a two-lane motorway in the late evening or early morning.

• In urban area, drivers do not know early enough when emergency vehicles are approaching which might cost people’s lives

What it does

• Locate potential dangerous animals on the roads in real-time
• Provide real-time database for locations of potential threats
• Make use of a real-time database of car’s locations based on the plate number recognition

• Warn and notify drivers through mobile/car app clients

• Detect and notify users about emergency vehicles coming around

• Make use of infrared sensors to detect potential threats at night

• Provide API access as a subscription based service for self-driving cars

Business opportunities

• Provide services to city planers, map services
• Asssist self-driving technology or driving navigators with real-time database

How we built it

Real-time RGB camera => Tensorflow based risk detection => Backend server => Notification system

Challenges we ran into

Performance issues with tensorflow

Accomplishments that we're proud of

Complete system architecture.

What we learned

Nokia smart street lights, IOT, notification system, Tensorflow object detection

What's next for Moose - Early Traffic Warning System

Open API.

Presentation slides are available in git repo https://github.com/mooseetws/MooseETWS

Built With

• docker
• firebase
• flutter
• node.js
• python
• tensorflow