The project was inspired by the recent tragedy that was the Parkland shooting. As high school students and many people our age go out to protest we felt inspired to do our part with out technical skill and knowledge. As we read about school shooting we noticed that there was a surprisingly large amount of time between a shooter being seen and first responders arriving on scene.
What it does
The Anti Shooter Alert System (or ASAS for short) uses Google's Cloud Vision API to decide whether a picture contains a firearm. Using this information it will then use the SafeTrek API to inform emergency personnel of what is happening and where it is happening reducing the latency in a shooter being sighted and an officer getting on the scene.
How we built it
We used Google Cloud Projects so that our python script can upload images to make use of Google's Vision API. Once we tested this part of the application, we used SafeTrek's API to create and alarm instance that would alert the right personnel.
Challenges we ran into
We initially wanted to do this using the Raspberry Pi to show that this can be deploy-able on small embedded systems such as the current CCTV infrastructure that is currently implemented in schools. Unfortunatly the Pi's camera module broke causing us to migrate the script to a PC. Figuring out python's own Requests API also took us some time but we eventually managed to make it work.
Accomplishments that we're proud of
As a team we feel like this is definitely a project that has potential to impact a lot of young people, save lives, and embodies the safety aspect of this competition. We are also proud of the way we overcame the challenges we faced especially when out hardware failed.
What we learned
Although we had used similar APIs in the past such as Microsoft's Computer Vision API, this was our first time setting up a google project and using a part of Google Cloud's suite of APIs. Additionally, we learnt how to format requests in python and got a peek at SafeTrek's API.
What's next for A.S.A.S
Given the opportunity, we would definitely like to try this on a more mobile system such as a Pi or use actual CCTV footage to implement the idea. Furthermore, we would love to work with a more established company to develop such a system with a neural network more attenuated to the project specification and actually see it implemented to protect our friends and family.