Level of Confidence is an art project to commemorate the mass kidnapping of 43 students from the Ayotzinapa normalista school in Iguala, Guerrero, Mexico. It was released on March 26, 2015, exactly six months after the kidnapping took place. The project consists of a face-recognition camera that has been trained to tirelessly look for the faces of the disappeared students. As you stand in front of the camera, the system uses algorithms to find which student’s facial features look most like yours and gives a “level of confidence” on how accurate the match is, in percent.
The biometric surveillance algorithms used, -Eigen, Fisher and LBPH-, are typically used by military and police forces to look for suspicious individuals whereas in this project they are used to search for victims instead. The piece will always fail to make a positive match, as we know that the students were likely murdered and burnt in a massacre where government, police forces and drug cartels were involved, but the commemorative side of the project is the relentless search for the students and the overlap of their image with the public’s own facial features.
The project software is available for free download so that any university, cultural centre, gallery or museum can set-up the piece and exhibit it. To show the work the institution must download the project software and provide a computer, screen and webcam. The full instructions and specifications are in this PDF document: lozano-hemmer.com/texts/manuals/levelofconfidence_specs.pdf
The project also exists as an open source software, which can be modified by any programmer with knowledge of OpenFrameworks so that he or she can make their own version, with different content. An example may be someone who trains the algorithms with images from missing aboriginal women in Canada. To download the source code please visit GitHub:
The piece can be acquired for art collections, but all proceeds are directed to a fund to help the affected community, for example in scholarships for new students at the normalista school: github.com/antimodular/Level-of-Confidence