Pebble Dataset: Interpolated
A poetic machine-learning dataset made of 5,000 images of found pebbles. Mimicking the datasets used to train computer vision systems, this project is an intervention that highlights the things we ignore when recording and classifying the world through data