Our app📱 “Miranda” at Hack’s 20—was inspired by recent global law enforcement encounters, we wanted to build a seamless app that can help de-escalate an aggravating situation in case of getting pulled over. There are obvious psychological impacts, of driving among police cars— how do we remain calm while being fully discreet and cooperative? How do we rebuild trust in certain communities, where racial profiling has become a norm?

What our app does

This is an app that documents and analyses police encounters using Machine Learning to help mitigate negative interactions with the police. Additionally, 'Miranda' promotes community policing by alerting nearby users and family.

User Story

User: Innocent citizen being pulled over by the police

  • When I see the flashing red and blue police lights in my rear-view mirror, I ask my phone “Hey Siri, I’m being pulled over.” The App “The one stop for police stops” opens up and automatically starts recording audio and video of this scene, streaming it to the cloud for secure storage. While the phone is recording the scene, my Constitutional and Miranda Rights are presented clearly on the app’s screen.

  • I can tap a button (or the screen) to send a notification to my family and local concerned citizens that I’m being pulled over and may need help interacting with the police (e.g. recording the incident themselves).

  • My incident report (including a transcript, the officer’s name, and the officer’s license plate) is published securely. Optionally, this incident can be posted on Twitter (with location + hashtags) to solicit help from other folks in the community, especially when there’s racist or aggressive language involved.

How We built it

  • Using Google’s Speech Synchronous Recognition API, audio files longer than 80 minutes can be transcribed successfully.

  • It can also be translated to different languages to fight police brutality in other countries (HK, a notable one).

  • The NLP model analyses the sentiments, comes up with a list of words with saliences attached. Judging from the relevance/how negative or positive it is, a loved one is able to encapsulate the situation quicker.

  • Auto generates PDF. An actual log that went down during the whole interaction.

  • With Google’s Cloud Vision API, our dash cam was able to screen capture the vehicle’s license plate. Printing each digit carefully into the full report, as a copy for the victim’s attorney/representative.

  • Within Miranda, the app provides a comprehensive list of commandments the user is entitled to. IF the situation is aggravated, user can conveniently refer to it.

  • As a form of protection, the app is designed to be completely black on the exterior. In case of confiscation, the user’s data will be saved and a full report will be ready for review.

  • Our speech recognition can also detect screams/words highly categorized as “danger”, and using Twilio API, the user is able to send texts to friends and family if the user is in a dangerous situation

Challenges We ran into

  • Connect some of the endpoints together.
  • Figuring out which APIs to use from Google Cloud
  • Working remotely

Accomplishments that We're proud of

We were able to integrate a lot of functionality in a short amount of time.

What's next for MIRANDA

We plan on expanding and developing some of the functionalities of the app. For instance, we would like to integrate AR/VR in order to simulate a similar environment where you are pulled over by a cop and familiarize yourself with the actions you can take using this app. We also plan to use Twitter API to alert our community to flock to the location if the interaction is indicative of police brutality.

Built With

  • adobe-xd
  • android
  • google-cloud
  • google-cloud-natural-language-processing
  • google-cloud-speech-to-text
  • google-cloud-vision
  • google-cloud-vision-api
  • mobile-application
  • python
  • twilio
  • twilio-messaging-api
+ 4 more
Share this project: