My grandmother lives by herself and has fallen down the steps on multiple occasions. She carries a cell-phone on a necklace that allows to get help but sometimes she forgets to wear it. What if there was a system that automatically detected that she has fallen over and can contact her family for her?
What it does
We set out to create a system that automatically detects if someone has fallen over using feedback from a video camera, accelerator, and sound. From the rapid changes in the image matrix we can see the speed someone has fallen over at. Using our system the remote server then can send a email message to a care-taker alerting them that someone may be in trouble.
How we built it
Android is our data gathering point. All information is fed back to a linode Ubuntu server.
Challenges we ran into
The original plan was to use a RPi2 to capture video and do the real-time processing on the same system. However, the hardware supply was limited in then number of webcams available so we instead replaced it with an android phone. We had to write our own app to capture and send a live video stream to a remote server. With our central processing unit removed (RPi2) we instead decided to look into a linode sever and have it do all computations for our system.
Accomplishments that we're proud of
It was are first time working with opencv and we made a working motion detector.
What we learned
Come to the hackaton with your own hardware and with all software pre-installed.
What's next for MonitorMe
We did not get a chance to implement a voice monitoring or accelerometer feedback system.