Triptych: 3 cameras and screens transform image recognition into a live collage of sound and images
Type
Generative Art, Installation
Technologies
Machine Learning, Python
Supported by
Goldsmiths, University of London
Project Description
Triptych 1 is an audiovisual installation where three cameras face their own screens.
Each screen shows a shifting sequence of images that functions as a score. As the cameras recognize objects in these images, they trigger sequencing scripts that generate a real-time collage of sound and visuals.
But there is one issue. Technology is neither neutral nor perfect. Image recognition, while powerful, can make mistakes. Wrong matches have led to people being misidentified and even wrongly convicted.
In this installation, errors are deliberately introduced: from time to time the system will misclassify, revealing images of people and situations where the consequences of such mistakes have been severe.
