FACE_OFF

Corrupt governments such as China are infecting its streets with surveillance facial recognition cameras to suppress its citizens. Protesting in these countries becomes increasingly dangerous as artificial intelligence develops. Our goal was to protect the identities and rights of protesters from A.I.

Jasper and I found these A.I. facial recognition bots can be tricked. By understanding how these cameras look at our faces we can generate the "convolutional filter" as masks and wear them on the streets to look like anyone in the bot's database. We used this trick to make masks that mirror the faces of anyone physically opposing the protest: the Hong Kong policemen.

So when the police tries to find the identities of the protesters using facial recognition, they'll only ever find themselves.

By Jasper Kamphuis & Lode Dijkers

Kind of work:
Research, concept,
art direction.

Techniques:
A.I. machine learning,
YOLO object detection,
mask making.

Made in 2020