Inspiration
Mass shootings across the nation have increased, and as they approach closer to home, we noticed that it was a much larger problem than we originally thought. Thousands of people have either been injured or found dead due to mass shootings, and the efficiency and flexibility of security today isn't increasing to keep up with the rate of mass violence.
What it does
We used machine learning to create QuickPoint - a faster, less invasive and automated security checkpoint. QuickPoint beings once the user scans their ID card to verify their identity. Then, it scans a group of objects after the individual empties their pockets and belongings, and speeds up the process by detecting any potential threats with computer vision. If no threats are detected, the user is prompted to continue. If a threat is detected, staff and security are immediately notified, and an alarm goes off.
How we built it
The scanning is done by using an RFID sensor, and much of the hardware was created using Arduino. The software utilized Xcode and was essentially an iOS app that could take in the current screen and output a probability of whether or not an item is dangerous.
Challenges we ran into
A challenge that we ended up running into was that we were unable to process frames from a continuously streaming live video. For the most part, we overcame this by taking individual photos and then processing them.
Accomplishments that we're proud of
One thing we're proud of is that we both met for the first time at this hackathon as fairly basic hackers, and were able to create our product by working together and collaborating. We're also proud of the fact that we were able to create a working model although there were no great datasets for this purpose, nor were there excellent, easy to use, trained models. We had to search for, organize, split, and add to our initial data to create a solid amount of data that our custom model could reliably train on.
What we learned
We learned how to create a custom model for CoreML that we could use in our iOS app. We also learned how to interface the camera with the app, and take pictures and videos in an iOS app.
What's next for Quickpoint
In the future, we hope to achieve concealed weapon detection, better automation, as well as a more accurate model.
Log in or sign up for Devpost to join the conversation.