SpaceFace

Inspiration

The inspiration for our project came about when our team was looking for a table at Worcester Dining Commons. However, we could not find a table for around 15 minutes. This is an irritating problem for many, and we would not have entered Worcester Dining Commons if we had known beforehand that there was no available seating.

What it does

SpaceFace uses image recognition to track people entering and exiting a building. This information is then displayed on a website, where users can check the data and check whether a building is full, the rate of change of people entering and exiting the building, what hours in the day the building is at its lowest and highest activity, etc.

How we built it

We use a raspberry-pi 4 with a camera module and feed its footage to an image recognition algorithm. The image recognition algorithm we used is called a Single Shot Multibox Detection (SSD) algorithm. An SSD algorithm is applied to the video feed using an OpenCV model. A bounding box will be applied to any person in frame, and then the centroid of the bounding box is calculated and tracked. The final position of the centroid tracker determines whether the person has entered or exited the building. If a person enters the building, the building total is incremented by 1. This increment (along with a decrement if a person exits the building) is then sent to a database. Multiple of these IP cameras can be set up to point at various entrances and exits within a building, and they don't even have to communicate with one another. All they each have to do is send their footage to the SSD algorithm.