Argus

Inspiration

Following the news of a Tesla Model 3 colliding with a crashed tow truck in Moscow, Telsa claimed that there was a problem with the software in regards to detecting accidents.

What it does

A web-app that uses object detection to monitor for accidents in CCTV cameras and report them immediately to a user's phone and browser client.

How I built it

• Uses TensorFlow's object detection model for accident "detection".
• Google Maps API with custom interface. :0
• Server set up in Node.js. npm install nodemon server/server.js To run: python crash_detection_model.py
• Express.js and Socket.io used to update the client with accident details.
• Twillio API to send SMS to the user at the moment of an accident.
• Google Geocoding API used to convert the unique markers' longitude and latitude to human-readable addresses.

Challenges I ran into

Google does not like to be spammed with 100s of API calls in less than a second.

Accomplishments that I'm proud of

Getting this finished in time!

What I learned

To move from static webpages to client-server web-applications. This includes: Node.js, Express.js, and Socket.io

What's next for Argus

To use real live footage from CCTV cameras around the globe.

Built With

• css3
• express.js
• google-maps
• html5
• java
• javascript
• node.js
• opencv
• python
• socket.io
• twilio