StealthProtecc

Inspiration

In our surroundings, we were very touched by terrorist attacks in schools and public spaces. Most of the time, it takes a while before everyone is notified of a school shooter when you are in a large establishment. Hate induced attacks including firearms has been on the rise this 21st century and we felt like we had to do something about it for our city to grow.

What it does

Run on server.io continusiously and send a HTTP request when possible video containing a subject with a firearm passes in front of cameras. Then, a short 5 sec clip of the event is sent on a typescript app which can be reviewed and act upon quickly as our api also sends the exact localisation of the event (level, building, alley, etc...).

How I built it

We started with Google Object Detection API and OpenCV for the video treatment, we trained it on 3000 pictures, labelled them and packaged them locally. Then we started working on a great UI for the phone app.

Challenges I ran into

We got stuck for quite a while at the sockets implementation to communicate between the server and the phone.

Accomplishments that I'm proud of

We kept going and didn't lost our end goal. We had a great time with the team.

What I learned

Apart from the technical amount of knowledge in 24hrs, we had the chance to work on something possibly life changing in the future. We learned about each other, how we can put our mind to a goal and manage to come through.

What's next for StealthProtecc

Learning about the limits of our implementation in a real time environment (IP cameras and thread management), train the model for knives also. But it's surely not our last goodbyes.