Cheesehacks Track: SECURITY AND PRIVACY

Inspiration

Explaining and Harnessing Adversarial Examples (Ian J. Goodfellow, Jonathon Shlens, Christian Szegedy, 2015)

What it does

From a single video feed source, we can classify objects through 3 different architectures, while maintaining integrity from attackers using noise attacks or intentional changes on the bitwise level of frames.

How we built it

Python for our encryption/decription algoritms, React for the frontend, and Docker cluster for the AI object detection nodes. Kafka/RedPanda for sending frames upstream so all nodes can fetch frames from 1 source, rather than 1 source pushing to many nodes.

Challenges we ran into

Ensuring that all nodes were in sync with each other during frame ingestion.

Accomplishments that we're proud of

Implementing a functional solution to a real world problem that is only growing in its attack vector.

Built With

Share this project:

Updates