Gun-Safe

Empower safer communities with Gun-Safe: a mobile application that makes gun security hassle free! From using nueral networks to detect gunshots, to live location tracking and alerting police.

Comment

Inspiration

Around 3 am one of our teammates' brother was going back in a metro where he suddenly heard loud sounds that he thought at the time were gunshots. Panicked and confused he didn't know what to do. The next day he received an email informing him that there was a gunshot at the same location where he was, but informed to him several hours later. Luckily nothing had happened that night, but if he was near the incident it could have been catastrophic.

What it does

  1. Using neural networks our app detects whether a certain audio is of a gunshot or not.
  2. Using Fourier transformation we were able to split a certain gun sound into its inndividuall frequency components which can be used to identify which type of gun is fired.
  3. Using a real time location feature to track the users live location and detecting the location of the gunshot we were able to send alerts to users within close proximity and simultaneously call the police with the location.
  4. We used video detection (which can be implemented via cctv cameras) to identify guns. This will also call the police and alert them about the following.

How we built it

We built the app using react native with node js. We used libraries like twilio, open cv. Our backend was coded in python and the data was stored on firebase to connect the entire project.

Challenges we ran into

Timezones were a huge issue for us as our timezone is in IST and we had to submit the project at 3 am. Additionally, integrating all of our codes (backend and frontend) was quite a tedious task. But nonethless that's the hackathon expiernce!

Accomplishments that we're proud of

We're extremely proud of our relatively accurate neural network and our Fourier transformation program which was able to detect the type of gun at a promising rate. We were also able to successfully integrate different code bases seamlessly, and we finished our project within the deadline (I hope).

What we learned

We used neural networks and audio dection with fourier transformation for the first time which was a great learning goal. We made our own data and worked with mobile apps for the first time as well using react native.

What's next for Gun-Safe

  1. We plan to improve user privacy in terms of working with their user location. We could achieve this by encrypting user data.
  2. We plan to work with a larger database and use more test data to develop an accurate model that has a higher frequency of detecting gunshots.
  3. To implement our project in the real world we would use audio from microphones on user devices (mobile phones).
  4. We plan to integrate the app on apple watches which would be far more convenient.
Share this project:

Updates