This project started as a real-time installation. On the opening day, the system scanned people in front of the camera and tried to estimate “how Russian” they are — based on clothing, hair, tattoos, posture, and other visible features.
The system doesn’t actually understand who it is looking at. It only detects simple visual elements and combines them into a single score. When several people appear in the frame, their features merge. Five people can easily turn into one “large Russian.” Individual identity disappears, replaced by a set of overlapping signals.
The interface looks clean and convincing — like a real computer vision tool. At the same time, it assigns slightly strange labels to what it sees: things like Slavic stare, migration aura, or oversized look. This gap between technical language and subjective interpretation is the core of the work.
The project comes from a very basic observation: in migration, you start recognizing “your own” almost instantly — by style, posture, small visual details. This recognition feels intuitive, but it is built on stereotypes. Here, this process is pushed a bit further and turned into a system that looks objective, measurable, and reliable.
The installation was presented at KC Grad (Belgrade). What is shown here is a recording from the opening day, when the system was running live.

Collaboration with Amfeat

Tools: computer vision, real-time processing, custom interface design
Year: 2026
Format: video, interactive installation