Watchdog

Inspiration

During our plane ride to the venue, we ran into someone who had their stuff stolen at the airport. In that moment, we both came had a sudden urge to contruct a system that not only watches over your items with real-time camera processing, but also preventing thiefs from taking your belongings.

What it does

The premise of the program is to watch over belongings and to actively prevent thieves from stealing items.

How we built it

We utilized Raspberry Pi 4B to create a portable camera system using a USB Logitech to stream real-time video. In addition to this, we coded YOLO and OpenCV to process and determine belongings. The info is then referenced to a website which allows for the user to activate and send information to the SSH server.

Challenges we ran into

Essentially, majority of the group had never used ML learning or Raspberry Pi, so a lot of learning on the fly was done. Whether it be consulting mentors for hours, or terrible internet connection which affected our stream bitrate, we managed to conquer every obstacle that we encountered.

Accomplishments that we're proud of

Being able to present the judges a finished product
Had a better creation experience than all of our previous hackathons
Met life-long connections and friends
A severe caffeine addiction

What we learned

We each learned a little more about ML, Raspberry Pi, the communication between servers, SSH communication, and connecting the Backend with the entire system

What's next for Watchdog

In the future, we plan to scale our product to be more affordable and also train our own AI model to determine thief facial expressions, suspicious movement patterns and have a more reactable program toward the thief rather than the belongings. Additionally, the effects of the Raspberry Pi are effectively wasted and we plan to integrate a cheaper system to work around Wifi and cost issues.