"Capturing Attention," is a piece of networked art that observes the observer and records the first moment at which they contemplate the artwork.

Capturing Attention

It’s fascinating what we choose to pay attention to, and what really constitutes attention. This project, titled “Capturing Attention,” is a piece of networked art that observes the observer and records the first moment at which they contemplate the artwork.

In my first attempt, I tried to embed a camera inside a picture frame. When the visitor directs his attention at the picture for more than a five seconds (to indicate that it is actually attention and not just a passing glance), the camera takes a picture of the viewer and then uploads it to a website where people can view a gallery of people in that first moment of observation.

Through this process, perhaps we can gain a better understanding of how we consider the world around us. For those visiting the website, perhaps they too can gain insight into how people react to art, what faces they make when the view a piece, and how long a piece can sustain their attention.

I started off by creating a Processing sketch that could conduct the face detection. The OpenCV library has an excellent sketch example to build from. There was a bunch of tweaking to the code so that it only detected faces after a given period of time elapsed, and only from a certain distance. The camera is a hacked PSEye.

I then implemented the PDF2Web library to save the captured images to a file on the server. Finally, I wrote a PHP script to pull the images from the server and display them on a page that uses Javascript to dynamically load the images and keep checking to see if any new images are added to display.

Unfortunately, I couldn’t find a satisfactory method for disguising the camera. Any one instance of a lens in a frame is obvious, and multiple dummy lenses embedded to distract from the lens capturing the image still raises awareness about the presence of a camera which changes the interaction. There’s also the issue of hiding the cord. This may be easier depending on the environment, but in this particular case, it was impossible.

But the first camera in the first frame is only the first part of the project. What really becomes interesting to me is when you have multiple frames, all similar, housing different types of art. It would be cool to be able to capture reactions to each piece and create a new work that aggregates people’s reactions.

My next thought was that I could hide the camera inside the artwork itself. I mocked up a quick version at home with the materials at hand:

This is just a box containing the camera and a few LEDs. But once I covered it all up with an old 3M privacy screen, the camera was completely hidden. For the next iteration, I drew inspiration from Eyal’s piece outside the JRoom. With some foamcore and black masking tape, I created the latest version:

It’s actually only the black box, and not the stuff that it’s sitting on. Once I had a piece of art that I thought would attract attention, I started playing with the images. At first, the images were the same as those you see with face detection: gray with a red rectangle around the face:

They work for face detection, but they’re not very pretty. However, this method did help me to see that I needed to play with the settings to avoid false-positives:

After I started tinkering with the housing for the camera, I stumbled upon a Hipstomatic type look that I really like, entirely by accident:

The last part that I should touch on is the web interface. This is an important element of the project for me because it takes this moment of attention that might otherwise be forgotten, captures it, and re-contextualizes that time for a totally different audience. On one hand, I think that everyone that looks at a piece of art does so with the same expression, one that is both inquisitive and perplexed. Then a moment of recognition or understanding settles in and an opinion begins to form. This moment that I’m capturing occurs just before that.

Visitors to the website will see a pixelated box on the left hand side of the screen with squares of varying colors. Each square represents an hour in the day (there are 24 boxes). The color is determined with some degree of randomness based on the number of pictures captured in that hour. A white box indicates no images. Clicking on the box will reveal all of the images from that hour on the right side of the page, and scrolling over the images enlarges them.

If I were to work on the site further, I would probably add more ways to view the pixels on the left (by minute, by second) and perhaps some more information about each image on the right.

I’ve learned many lessons from this first go round. The camera height is important for face detection as well as for making sure people “notice” the work. Someone gave me the idea of putting a card with some small text in front of the display to entice people to get close, which is not a bad idea. I also need to consider whether I want people to know their picture is being captured and displayed on the Web, or whether that’s a totally different part of the experience.

Traceroutes Map

TraceroutesFor our study of traceroutes in Understanding Networks, I attempted to ping 25 African government websites. I wanted to see where the packets traveled and how many hops they took moving from New York across the Atlantic. In actuality, many government sites are hosted here in the US or in Europe, which means they are quicker to access and take less time to resolve. I used a program called WhatRoute for my research.

Among my findings:

  • Equatorial Guinea, Comoros, Mauritania, Eritrea and Niger did not resolve, meaning that along the way some of the routers would not respond to traceroute pings;
  • Central African Republic (hosted in the US) was the fewest hops (10);
  • Madagascar took the most hops (22);
  • Since my ISP is Time Warner, all of my pings began with Road Runner, a peering company with Level3. They often stayed on the Level3 network for the majority of the trace and branched off if (and when) they crossed the Atlantic closer to the website’s host server.

You can download the full map here. It says the image is broken, but you can right-click and view it on your computer without a problem.