Eyes Lazy and Colorblind

This has appeared around the ‘net in the last couple days:

[youtube=http://www.youtube.com/watch?v=t4DT3tQqgRM&w=475]If the video does not appear, you may need to download the latest version of Adobe Flash Player.

HP’s response partially explains what’s going on: “The technology we use is built on standard algorithms that measure the difference in intensity of contrast between the eyes and the upper cheek and nose. We believe that the camera might have difficulty “seeing” contrast in conditions where there is insufficient foreground lighting.”

So is HP racist?

No. They didn’t purposefully exclude black people from using their tech and they seem to be making an honest, good-faith effort to fix things now that they know about the problem. But they should have known about the problem, and they should have fixed it before releasing the tech.

They were shortsighted, and they messed up.

They wrote their facial-recognition software; they coded it into the camera; and they tested it. But they probably only tested it on light-skinned people. They got it working with their testers and released it. It probably didn’t even occur to them that all their testing left something out—as it clearly did. They were colorblind, and as a result they failed to provide for all of their customers.

2 thoughts on “Eyes Lazy and Colorblind”

  1. What I find ironic is that HP felt they needed to reinvent the wheel. Several years ago, similar technology was employed by a Playstation component named “Eye Toy” You control the avatar on the television monitor through body movements tracked by an eyetoy camera mounted above it.

    The FIRST calibration test for the camera is it creates an oval, you position yourself so your face is within it, then it moves the oval and dings when it “sees” your head move to that spot.

    From then on, the camera tracks your head, body and arm movements. If you need to roll, you swing your arms in an arc. Need to jump over something – you jump.

    That was Playstation 2 (for those needing a date). It worked well with every dark brown person in my household (big and small).

    So HP has the same problem so many other companies have – which is to test a product on a limited scale and assume it works for everyone.

    Only this isn’t new technology – so HP doesn’t get a pass for the omission. More likely (as occurred at a former employer) someone in management decided ethnic people weren’t a large enough audience to worry about and screened them out of the focus group. The good news is I’m betting HP won’t make the mistake on future product offerings given the viral nature of the internet for spreading consumer reactions – good and not so good.

  2. I’m a photographer on the side–I do weddings, engagement shots, that kind of thing. Darker complexions photograph differently than lighter complexions, and as the HP site says, it’s because of contrast and lighting. But as Christine notes above, plenty of other facial recognition software takes that into account, so why didn’t HP’s? It just seemed pretty odd to me.

Comments are closed.