Mass media is one of the most effective ways to reach the general public. Over lately there has been a lot of unrest over the idea of AI being able to use deep generative models to create artificial media (Fake News). The idea is to develop a tool that can combat these tactics so everyone can enjoy cat videos without any political bias or subliminal messages.

What it does

It analysis bias levels on videos to help the user determine if it's real or fake.

How we built it

Our designer created wireframes and later high-fidelity mockups for the front-end developer to work with. Meanwhile, back-end and full-stack developers focused on the video analyzing aspect by developing a SaaS application deployed to the Azure platform, that leverage a variety of Microsoft Services our application's functionality:

  • MongoDB Atlas on Azure, Azure Deploycenter for Continuous integration and deployment from our GitHub repository.
  • AMS Video Indexer API, to extract metadata that explains what is contained in the video (brands, celebrities, emotions, etc).
  • Azure Web App and Kubernetes services to make applications scalable and performance.

Challenges we ran into

Using the Video indexer API was challenging as the documentation was not straightforward.

Accomplishments that we're proud of

Good teamwork.

What we learned

We all grew in our own way based on our skill sets. For instance, the backend developer, Derrick, learned more about using MongoDB in a live application. And the designer, Mais, learned about collaborating with developers, using wireframes as a communication tool with the developers, and creating a developer hand-off of the final user interface.

What's next for Properganda

Making a custom CNN and training it to scan a wider array of videos and other media

Built With

Share this project: