No More Stealing

Proof when person is gone the email is sent

Inspiration

I don't like having things not be where they are, so I thought of this to help warn me if someone moves something elsewhere.

What it does

So normally, it's supposed to email the user when items are taken or missing from the webcam feed. Sometimes spurious detections may cause a misfired email, so the project tries to prevent that from happening by raising the confidence threshold.

How we built it

Used yolov5 and pytorch. Pretrained models. Used the predictions to pull out the predicted objects with high confidence, and kept track of them

Challenges we ran into

I have never set up a raspberry pi project nor used YOLOv5 before, so that was interesting. Another challenge I ran into was the pretrained model was extremely limited and prone to some errors. For example, it only had 80 pretrained classes and was not very accurate. However,