Ghost Protocol

Ghost Protocol Logo

Inspiration

We were inspired by AI's capabilities for face recognition and wanted to find a way to implement Redis databases alongside it.

What it does

Ghost Protocol is a surveillance & antisurveillance defense system. Utilizing video input and AI facial recognition technology, Ghost Protocol manages a live Redis database of blacklisted and whitelisted users to provide various security functions. Decide who knows what through our credentials file, alert executives when blacklisted users are at a location, track and identify these face IDs in the log, and protect whitelisted users' privacy by blurring their faces.

How we built it

Initially, we experimented with the different categories of the competition and found out about the capabilities of OpenCV. While playing around with facial recognition, we developed a whitelist feature which lead us down the rabbit hole of biometric ID, authentication, and security. We then used a Redis database to track and store all of this information, expanding the utility of our blacklists and whitelists. Finally, we implemented a front-end through Flask to add new faces and users to the database. Throughout this process, multiple passes of optimizations were made to the live facial recognition to maintain as much performance as possible.

Challenges we ran into

Managing and expanding user information in our Redis database
Encoding and decoding faces through NumPy
Ensuring the speed of our software
Choosing a topic and deciding
Applying the blur filter
Sending warning emails of potential threats

Accomplishments that we're proud of

How cool our app looks! We've got blur, colored boxes, name/whitelist recognition, and we are all proud of how we developed our front end to show the quality of our project.

What we learned

Dellie: How to use Redis and manage/query databases, how to use OpenCV, and how to serialize and store important user data. Jace: How to use facial recognition technology and using ChatGPT for larger projects with a version control system. Dylan: How to use OpenCV to apply image filters, and Flask file management

What's next for Ghost Protocol

What's next:

Further optimizing the live facial recognition
Making a more user-friendly experience through Flask
Use a relational database on top of Redis of who knows/connects with who
When we source a better pre-trained model, better camera quality, and more user data our model could be even more accurate.