Inspiration
The whole inspiration behind this project was mostly me being impatient when packages were coming in, and that's about all. But not everyone comes home to a package they ordered, so I thought I could make it detect porch piracy too.
What it does
As I said above, the camera is connected to a raspberry pi or laptop which then looks for packages that have come and go
How I built it
Since Tensorflow was being a bit problematic for me, I had to take another approach. I decided to look for huge changes in the video stream, and after those were done the app would wait 15 seconds and then check again so it would know if it was just a cat looking for food or an intimate object. If it decides it's a box, it will e-mail the recipient specified on launch using the -r flag, which can either be a normal e-mail address, or in my case, an e-mail to sms service from your mobile carrier.
Challenges I ran into
Trying to train the tensorflow model didn't quite work out- it was using my CPU instead of GPU so the training would take ages, not overnight. So I've had to take a different approach for the time being, but I'll be sure to revisit this project. I've also tried to use a color mask to detect boxes because just looking for changes would result in funky results, but that didn't work out well.
Accomplishments that I'm proud of
They'll come later
What I learned
I've learned how to use OpenCV and SMTPlib to capture video and send e-mails and sms messages, and many ways python is different than c# (why is there no null)
What's next for Anti Porch-Piracy Camera
-Custom trained tensorflow model for object detection -Proper SMS sending without an e-mail to sms server -Actual testing on a raspberry pi -Maybe a GUI to launch it but idk about that
Log in or sign up for Devpost to join the conversation.