In 2012, AlexNet software revolutionized the AI industry by successfully classifying images, a breakthrough developed by Alex Krizhevsky, a doctoral student from the University of Toronto, alongside fellow student Ilya Sutskever (later co-founder of OpenAI) and their professor, Geoffrey Hinton. This milestone project is now open-source and accessible on GitHub.
The following year, in 2013, the trio continued their groundbreaking work at the University of Toronto, introducing a novel technique at the time: the convolutional neural network (CNN). AlexNet’s practical application of training on gaming graphics cards (using two GeForce GTX 580 cards) allowed it to triumph in the ImageNet Large Scale Visual Recognition Challenge of 2012, achieving a remarkable accuracy of 84.7%, a significant improvement from the 74.2% benchmark of the previous year’s winner. As a result, AlexNet set a new standard in image classification.
After founding DNNResearch, the three developers sold the company to Google, thereafter following separate career paths. The original 2012 version of AlexNet’s source code became Google’s property, with Alex Krizhevsky later introducing Geoffrey Hinton to the Computer History Museum (CHM) team in 2020. Working closely, they successfully located and released the original open-source code of AlexNet in 2025, a compact file sized at just 200KB.
Sources: Computer History Museum, University of Toronto, ZDNet
TLDR: In 2012, AlexNet, developed by Alex Krizhevsky and team, paved the way for image classification in AI. The trio’s groundbreaking work led to the development and open-sourcing of AlexNet, setting a new standard for the industry.
Leave a Comment