ImageNet is assembled and released
Fei-Fei Li's group began building a labelled image collection at a scale nobody had attempted, organised on the WordNet noun hierarchy and annotated largely through crowdsourced work, and released it from 2009. The bet was explicit and contrarian: that recognition was limited by the absence of data rather than by algorithm design. The annual challenge run on a subset of it from 2010 turned that bet into a public scoreboard, which is what allowed the 2012 result to be recognised immediately as a result.
Historical — it happened, and the record is settled.
Subjects
- ImageNet · technology · United States · not located
- Fei-Fei Li · person · United States · not located
Evidence
Partly verified — Core facts are sourced; some optional detail is deliberately absent.
ImageNet Classification with Deep Convolutional Neural Networks
An eight-layer convolutional network — five convolutional and three fully connected layers — trained on graphics processors with dropout regularisation achieved a top-5 error rate of 15.3% at ILSVRC 2012 against 26.2% for the second-placed entry, learning hierarchical features rather than using hand-designed ones. V55 verification basis: the paper was not fetched. The 15.3% and 26.2% figures and the architecture description were carried in multiple retrieved summaries and in a hosted copy of the paper itself, but the attribution in those summaries is to secondary explainers as often as to the paper, and a curator should confirm both error rates against the ILSVRC 2012 results table before publication.
Advances in Neural Information Processing Systems 25 (NIPS 2012) · 2012 · Partly verified
Krizhevsky, A., Sutskever, I. and Hinton, G. E., NIPS 2012; the ImageNet Large Scale Visual Recognition Challenge 2012 results
Recorded on ImageNet
- ImageNet is assembled and released
- A deep convolutional network wins the ImageNet challenge
- The AI Index records both fast benchmark movement and doubts about benchmarks
Related
Related because they share a subject in the Atlas — never because the text looks similar.