Sinkcam is a social experiment on how clean (or dirty) a communal sink can get. At the beginning of the fall term at ITP a new student sink was installed.
It didn’t take long for it to get dirty. In the hopes of getting users to wash after them. At first I thought of sensors on the dish rack or a front facing camera to remind users to wash their dishes.
I eventually settled for an overhead camera to both see the sink and the user at the same time.
Because of the unpredictable nature of the sink I was only allowed to install a camera more than 10 feet above the sink. This meant that a simpler solution like the kinect would not be possible because of the depth limitation. So I used an Axis IP camera and using Blob Detection in Processing to analyze the images.
Initially I recorded stills before announcing the camera’s existence to the student population to monitor behavioral patterns.
Once I was satisfied that the camera was able to determine items in the sink from people and other external items.
I then installed the application on a public terminal on the floor. Some users became more conscious on what they left in sink. Eventually the sink was clear of dishes which was followed by a new problem. Keeping the dishes. But I’ll leave that for another project.
Built using Processing and code from Dan O’Sullivan using IP cameras in Processing and the Blob Detection library.
Developed for the Fun Theory class under Katherine Dillon at ITP.
Shown at the ITP Winter Show 2012.
Download the code from github.