Artist and CSNI researcher Nicolas Malevé has written a computer script that cycles through ImageNet — a vast dataset of 14,197,122 photographs — at a speed of 90 milliseconds per image. To exhibit all images, this runs over a two month period (until 01 Sep 2019) as a live stream on the web and on the Media Wall at The Photographers Gallery. The script pauses at random points to enable the viewer to ‘see’ some of the images and how they are categorised, thus raising questions about the relation of scale between the overwhelming quantities of images needed to train algorithms and the human attention and labour required to annotate and categorise the images. Further notes on the project can be found here.
Related to this, TPG are hosting other events:
What Does the Dataset Want? Photography & Classification in the 21st Century Symposium followed by launch of Heather Dewey-Hagborg Commission, 14th Sept 13:30 – 18:00. A symposium exploring the crucial role of photographic datasets in the development of machine vision and artificial intelligence, in collaboration with the AI Now Institute. The day will conclude with opening drinks for a new commission by Heather Dewey-Hagborg for The Media Wall. Speakers include: Zach Blas, Heather Dewey-Hagborg, Nicolas Malevé, Daniel Rubinstein, Katrina Sluis, Jon Uriate, Anna Ridler.
10th Birthday Party for ImageNet, with a talk by Dr Fei-Fei Li, Sat 21st Sept 4-7pm. ImageNet has become one of the most influential photographic datasets in the fields of Deep Learning and AI. More than 14 million photographs were gathered through a benchmarking effort that propelled the outbreak of Computer Vision and its wide range of applications such as surveillance, phone filters, medical imaging, biometry and autonomous cars. ImageNet is organised through 21,000 categories that are still being used today to train computational models. On Saturday afternoon we will celebrate the 10th anniversary of ImageNet with a series of playful challenges, activities and artists’ presentations to showcase the impact of ImageNet, including a talk by its creator Dr Fei-Fei Li of Stanford University/AI4ALL.