Trusted Receivables

Inspiration

Have you ever had your packages stolen? One of our team members had a package stolen not too long ago and the feeling sucks, so we designed a system to reduce some of the heartache!

What it does

We currently have it set up where we first check to see if a package has been delivered by checking to see if a package has been dropped off or not. When it has been dropped off, the user gets sent a notification that the package has been delivered. Then we constantly check to see if there is a box. If there isn't a box, then that means someone either stole it or picked it up. If this happens, the user is then notified that the package has been picked up. There is also an alarm that goes off when the package is removed.

How we built it

We built everything with python. We used opencv for the video feed and taking the photos from our laptop webcams. We used twilio to send SMS messages to the user, and then we used the esp8266 and micropython to make the alarm system. We used Microsoft Azure's Cognitive Services for Computer Vision to detect if a box has been dropped off or not and if it has been moved after. We used flask to deploy a local version of our project and then we deployed it on Google Cloud.

Challenges we ran into

We were switching between ideas for what we should do. Also, we tried to implement some custom superpixel color detection using computer vision and machine learning to classify boxes before opting to use Microsoft Cognitive Services. The main problem was that the centroid detection we were trying to use a list of properties that didn't actually record the centroids of the objects were were finding the segments for. We were deciding if we should pivot to a different idea 12 hours into the hackathon. We also have work done for a web scraping application to fill a database and a html front-end, but we were prioritizing a workable demo. Some of us slept because we were too tired and we all had work to do that was due the following week.