With increasing awesome application where are our information is stored and shared, multiple attack attack vectors are created for those with a malicious intent.
What it does
It allows you to set a list of sites that you would like to protect. It watches for activity on those sites and blocks anyone trying to access the site that isn't you. It does so by using the camera to implement facial recognition via Microsoft's Oxford Cognition Project.
How we built it
- Trained a Machine learning model for a user which took a few hours
- Overwatch chrome extension detects inactivity and restricted site and automatically engages the locking mechanism
- Overwatch api that handles image processing, facial recognition, and user authentication for the client
Challenges we ran into
- encoding and decoding incoming images to process in a manner that would work nicely with our solution
- We quickly exceed our request limit on our inital implementaion with an existing facial recognition api (Kairos), at one point our request would simply hang and we thought it was something about out logic, when in fact it was the api
- Ultilizing the ML Project in a smooth manner.
Accomplishments that we're proud of
- Finishing what we set out to build with a working solution.
- A clean and maintainable codebase, allowing for further development of the project
- Implemented improvements to Microsoft's node package for Oxford-Project
- Accurately being able to detect the desktops owner and authenticating that user to their configured sites
- Once again our solution WORKS!!!! AND SECURES OUR SITES!!!!!! :D FIU
What we learned
Machine Learning and what it takes to train models for accurate results
What's next for Overwatch
Thanks alot for reading and considering us a place!!!! :D