1000 Synsets (Vinyl Edition)
Two vinyl records, 33 RPM. Volume I: 38'23''. Volume II: 37'15'', 2019.
Featuring the voices of Sara Hamadeh & Joseph Hughes. Excerpt from Volume II:
Created in 1985, Wordnet is a hierarchical taxonomy that describes the world. It was inspired by theories of human semantic memory developed in the late 1960s. Nouns, verbs, adjectives and adverbs are grouped into synonym sets also known as "synsets." Each express a different concept. ImageNet is an image dataset based on the WordNet 3.0 nouns hierarchy. Each synset was populated with thousands of images.
From 2010 until 2017, the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) was a key benchmark in object category classification and localization for images, having a major impact on software for photography, image searches and image recognition.
In the first year of the the ILSVRC Object localization challenge, the 1000 synsets were selected randomly from the ones used by ImageNet. In the following years there were some changes and manual filtering applied, but since 2012 the selection of synsets has remained the same.
1000 synsets (Vinyl Edition) contains these 1000 object categories recorded at the high sound quality that this analog format for audiophiles and collectors allows.
This work highlights the impact of the datasets used to train artificial intelligence (AI) models that run on the systems and devices that we use on a daily basis. It invites us to reflect on them, and listen to them carefully.
Each copy of the vinyl records displays a different synset and picture used by researchers participating in the ILSVRC Challenge to train their models.
Special thanks to Michaela Lakova for her support in the production and sound edit of the vinyl records. Sound mastering by Any Colour Post Production Studios.