Claire Mitchell
Primitives
An interactive sculpture that draws the patterns of its own perception.

Classes
Thesis
Thesis
Primitives is a sculptural illustration of the computational metaphors used to understand visual perception.
The sculpture learns to recognize basic shapes and uses these shapes as a vocabulary to physically draw patterns on a chalkboard, leaving a trail of its development on the wall over time.
Background
Tactile-Vision Device Paul Bach-Y-Rita
Vision by David Marr
Visual Intelligence by Donald Hoffman
I is an Other by James Geary
The Structure of Scientific Revolutions by Thomas Kuhn
illustrations by Fritz Kahn
illustrations of perception by Descartes, Johann Zahn, Isaac Newton
work by Chuck Close, Sol Lewitt, Cy Twombly
Implementation
Visitors are invited to submit hand-drawn primitive shapes on a small chalkboard slate with an overhead camera. The machine tries to detect the shape drawn by the user then draws the shape it recognizes on a wall - as if attempting to communicate or make sense of the world that it perceives. Each human interaction improves the machine's recognition abilities by contributing new data.
The sculpture learns to recognize basic shapes and uses these shapes as a vocabulary to physically draw patterns on a chalkboard, leaving a trail of its development on the wall over time.
Background
Tactile-Vision Device Paul Bach-Y-Rita
Vision by David Marr
Visual Intelligence by Donald Hoffman
I is an Other by James Geary
The Structure of Scientific Revolutions by Thomas Kuhn
illustrations by Fritz Kahn
illustrations of perception by Descartes, Johann Zahn, Isaac Newton
work by Chuck Close, Sol Lewitt, Cy Twombly
Implementation
Visitors are invited to submit hand-drawn primitive shapes on a small chalkboard slate with an overhead camera. The machine tries to detect the shape drawn by the user then draws the shape it recognizes on a wall - as if attempting to communicate or make sense of the world that it perceives. Each human interaction improves the machine's recognition abilities by contributing new data.