Aegis

Advanced AI-powered website that analyzes security camera feeds to identify weapons and depth charts, improving public safety and streamlining threat detection for schools and communities.

Comment

Inspiration

This past year, the high school that my partner (Aiden) and I (Kevin) attended had two bomb threats. Additionally, I had transferred school districts before attending my current high school, and the high school I was supposed to go to had a few knife stabbings that made the news. Many of my friends felt unsafe and chose to attend a different high school because of this. Overall, we believe safety should never interfere with education.

What It Does

Aegis inputs a video and returns an annotated video with gun and knife detections along with a depth map video. It also features a dashboard so users can track statistics.

How We Built It

We trained a custom YOLOv8 model and used the Depth Anything V2 model for our machine learning needs. We deployed everything using Docker to a cloud GPU on Paperspace. For the backend, we used a Python Flask REST API to integrate the components, and for the frontend, we used React, Tailwind CSS, and a few other CSS libraries, deploying it all to DigitalOcean. Firebase was used for authentication.

Challenges We Ran Into

While setting up our website on DigitalOcean, we encountered an issue where the site wasn’t correctly rendering with the necessary JS and CSS. This was because our website was only partially using React Router. The root cause was our uncertainty about how to implement React Routing with our Flask backend authorization. After a few hours of searching online and troubleshooting, we were able to fully implement React Routing into our website.

Accomplishments That We're Proud Of

We are very proud of hosting this as a public website on DigitalOcean and completing this project on time. A few other accomplishments, in no particular order, include:

  • Setting up Firebase authentication
  • Using Docker
  • Deploying the API to a cloud GPU

What We Learned

We learned a lot from this project, including React Routing, Firebase authentication, Docker, and the general process for deploying a website.

What's Next for Aegis

We want to expand this technology to make it more efficient. Currently, our cloud GPU is very costly, so we hope to migrate to a lower-cost solution, making the software more inclusive. Additionally, we aim to partner with nonprofits and startups to implement this solution in the real world.

Built With

Share this project:

Updates