How It Works

  • Reported crimes are geotagged and passed through a deep learning severity classification algorithm
  • Live heat maps are populated based on the severity and the frequency per location using triangulation
  • Messages are stored in a blockchain distributed network across user devices to prevent scaling database costs
  • Civilians are incentivized to report with ethereum tokens awarded for reports that help solve crime

The result is that civilians can avoid / navigate around danger, search for and be notified of crime that may potential involve them, and police can better allocate resources to tackle incidents by urgency at a minimal cost. The community is also transparent and closer than it ever has been with a "I scratch your back if you scratch mine" mentality. All of this is done with the automation provided by machine learning and the distributed blockchain system.

Tech Stack

  • Tensorflow classifier/spam filter was trained using a bag-of-words and stemming technique
  • Messages posted to the Ethereum network using Web3 & Solidity
  • GeoLocation is gathered using Google Maps API
  • Safe paths to navigate around crime calculated using the A star algorithm
  • Server side logic is hosted on AWS Lambda
  • The Police Portal is built on a Flask framework
  • The Civilian Mobile App is built on React-Native
  • Implemented crime searching using Elasticsearch

Dependencies to Run

  • tensorflow
  • tflearn
  • pandas
  • nltk
  • numpy
  • web3
  • flask
Share this project:
×

Updates