How to See

Computer vision powers facial recognition, targeted ads, and surveillance. It's built to identify and extract people, objects, and behaviors in service of commerce and control.

“How to See” reappropriates these tools toward human, ephemeral concepts they weren’t designed to measure, made visible through patterned annotations.

  • Scroll and zoom to explore the archive
  • Tap Filters at the bottom to browse concepts and filter the archive.
  • Submit your own entry via the camera button on the top right, or click here

Info

How to See

Computer vision is built to measure, track, and extract, producing outputs treated as stable truth, optimized for commerce and surveillance. “How to See” redirects systems of detection toward unstable interpretation.

Each photograph is processed through shifting annotations that render machine perception visible, offering multiple provisional ways of seeing.

Tomás Carlson, 2026