Fei-Fei Li: ImageNet and the Foundation of Computer Vision

Fei-Fei Li is a computer scientist and professor at Stanford University, widely recognized for her contributions to computer vision and human-centered AI. She is the principal architect of ImageNet, a large-scale dataset that played a key role in advancing deep learning.

Background and Education

Fei-Fei Li was born in 1976 in Chengdu, China, and immigrated to Parsippany, New Jersey, in 1992. While pursuing her education, Li helped her parents run a local dry-cleaning business. She completed her undergraduate studies at Princeton University and earned a PhD from Caltech, balancing her academic research with managing her family's business on weekends.

The Development of ImageNet

In 2007, as an assistant professor at Princeton, Li recognized that the limited size of existing datasets was a significant bottleneck in computer vision research. She hypothesized that improving algorithms required exposing them to a vastly larger and more comprehensive dataset of real-world images.

Li initiated the ImageNet project to map the visual world, aiming to collect millions of images categorized across thousands of concepts. The project initially faced skepticism and struggled to secure funding, as some reviewers considered large-scale data collection to be outside the scope of traditional scientific research.

Amazon Mechanical Turk and the Deep Learning Boom

To label the millions of images required, Li utilized Amazon Mechanical Turk, crowdsourcing the task to thousands of global workers. Released in 2009, ImageNet contained 14 million annotated images categorized into 22,000 groups.

In 2012, a deep convolutional neural network named AlexNet, developed by Geoffrey Hinton's team at the University of Toronto, won the ImageNet Large Scale Visual Recognition Challenge. AlexNet achieved a significant reduction in error rates, demonstrating the effectiveness of deep learning when trained on large datasets and marking a major shift in artificial intelligence research.

Human-Centered AI

In 2017, Li took a leave from Stanford to serve as the Chief Scientist of AI/ML at Google Cloud. During this period, she observed the practical applications of AI and became concerned about issues related to bias and the ethical deployment of technology.

She returned to Stanford to co-direct the Stanford Institute for Human-Centered Artificial Intelligence (HAI). Her work focuses on ensuring AI technologies are developed and deployed ethically, advocating for the democratization of computing resources and advising policymakers on AI regulation.