As Python contiunes to gain popularity for numeric and scientific computing
analysis and big data analysis, more and more is published about the tools
available, numpy, scipy, pandas, etc. Blogs, tutorials, e-books and the
old fashioned print book, you won't have to consult many of these before you
meet Lena, or at least are presented with an image of her. She stares
provocatively at you over her right shoulder, a banded hat perched on her head.
The author of whatever content you are reading then proceeds to manipulate this
image to illustrate the usage of the package they are trying to explain. It
is time, long past time some might say, that we bid farewell to this picture of
Ms. Lena Söderberg and find better ways or at least different ways to illustrate
a concept.
Why? For the benefit of those: who don't know the back-story, here is a brief
summary. In the summer of 1973, Alexander Swachunk was an assistant professor
at the Universisty of Southern California, when he and some other colleagues
were desperate for a high qualtiy, glossy image to use in a paper about image
processing. A colleague happened by with an issue of Playboy in which Ms.
Söderberg was featured; the academics tore off the top third of the centerfold
and quickly scanned it, this image or remastered copies of it, live on today.[1, 2]
It is now a standard test is included in a sub-module of the scipy package.
As a community of developers who use, develop and evangelize Python, we often
trumpet our community as one which is open, friendly, inclusive, not
heirachical or cliquish. I have generally found this to be true, but we all
make mistakes and I think the continued use of the Lena image by our community
is one of them. We can argue endlessly around this issue but the fundamental
question is: Do we want to continue to use an 40+ year old image from a
magazine whose primary purpose is to objectify women? Is that consistent with
what we profess to value as a community of female and male Python developers?
All technical issues to the side for the moment, after all these are not
technical questions, let's definitively answer these two questions and let our
actions flow from the answers.
I recognize that the people who analyze images or write books about tools may
still want and need a stock test image, so I should suggest an alternative. I
am a terrible photographer but I think I could probably take a decent photo
with the same or better resolution than the Lena photo (512 x 512). Here are
three suggestions:
- How about a cat? YouTube seems to like cats a lot, I'm sure there are plenty
of high quality pictures of cats.
- I've heard that it is very comforting to watch the Washington DC panda-cam.
How about a panda? I sort of like this idea since we have nice Python package
of the same name, although not named after the animal.
- There might be an objection to using or exploiting animals with the above two
suggestions. I hear you, how about a nice flower? Here's a morning glory
that I particularly like.
I'm certainly not the first person to raise this issue and far from the most
eloquent. [3, 4] I don't think that eliminating the use of the Lena image in
the Python community will solve all problems, but it will solve this problem
for our community. But as community no one is going to impose a solution
we have to agree to solve it ourselves.
References
- Hutchison, Jamie (May–June 2001). "Culture, Communication, and an
Information Age Madonna" (PDF). IEEE Professional Communication Society
Newsletter (archive) 45 (3). image for image processing algorithms and
- Munson, David C, Jr (January 1996). "A Note on Lena". IEEE Transactions on
Image Processing (Archive) 5 (1)
- "A Centerfold Does Not Belong in the Classroom." Washington Post. The
Washington Post. Web. 3 May 2015.
http://www.washingtonpost.com/opinions/a-playboy-centerfold-does-not-belong-in-tj-classrooms/2015/04/24/76e87fa4-e47a-11e4-81ea-0649268f729e_story.html.
- Needell, Deanna; Ward, Rachel (February 29, 2012). "Stable image
reconstruction using total variation minimization" v1. Cornell University
Library / Arxiv.org.