----------empyre- soft-skinned space----------------------
It’s been interesting to see what everyone’s been working on lately. We wanted
to announce a new project we will be debuting at Science Gallery Detroit as
part of the Tracked and Traced exhibition opening early this September [1].
The project, Boogaloo Bias, uses absurdity to reveal some of the problematic
uses of facial recognition software by law enforcement, such as the practice of
‘brute forcing’ where, in the absence of high-quality images of a suspect,
agents substitute hand drawn forensic sketches, proxy images generated from
artist sketches, and images of celebrities the suspect is reported to resemble.
In Boogaloo Bias, a facial recognition algorithm searches for members of the
anti-law enforcement militia the Boogaloo Bois by surveilling protest footage
and live video feeds using a facial recognition algorithm trained on faces from
the campy B-movie sequel Breakin’ 2: Electric Boogaloo, which is also the
namesake for the group. All matches made by the system will be false positives.
A study by the Georgetown Law Center on Privacy and Technology found that
because “[t]here are no rules when it comes to what images police can submit to
face recognition algorithms to generate investigative leads,” agents have been
known to substitute not only low-quality images from CCTVs, but hand-drawn
forensic sketches, proxy images generated from artist sketches, and images of
celebrities thought to resemble a suspect [2]. As one might suspect, this has
resulted in numerous false-positives. Through a juxtaposition of real video
footage and constructed movie characters, Boogaloo Bias reveals some of the
problems inherent in how facial recognition and algorithmic classification
technologies are used by law enforcement.
The use of Breakin’ 2: Electric Boogaloo is a reference to the Boogaloo Bois,
an anti-law enforcement militia that derives their name from the 1984 film. The
group emerged from 4chan meme culture and, since January 2020, have been
present at right and left-wing protests against coronavirus lockdowns and
police brutality. In the project, a facial recognition algorithm trained on
faces from Breakin’ 2 searches for those characters in a live video feed and
videos of protest footage in the United States from 2020 and 2021. When a match
is found, the face in the video is identified by a box displaying the name of
the character, an accuracy score, and a still image of the character is shown
to the right of the video feed. All matches produced by our system are
false-positives. The gallery installation will feature videos of protest
footage and a live-feed surveillance camera for viewers to interact with the
Boogaloo Bias facial recognition system. An online app that will allow visitors
to the project website to use their webcam to interact with the program is also
in development and will be available this summer.
Links:
[1] Science Gallery Detroit: https://detroit.sciencegallery.com/
[2] Angelyn, Clare. “Garbage In. Garbage Out. Face Recognition on Flawed Data.”
Georgetown Law Center on Privacy and Technology, May 16, 2019.
https://www.flawedfacedata.com/
Our current bios are available on our personal websites:
https://derekcurry.com/
http://jennifergradecki.com/
Please let us know if you have any questions about the project!
All our best,
Derek & Jennifer
--
Derek Curry, PhD.
Assistant Professor Art + Design
Office: 211 Lake Hall
http://derekcurry.com/
_______________________________________________
empyre forum
empyre@lists.artdesign.unsw.edu.au
http://empyre.library.cornell.edu
