Fei-Fei Li

Stanford Professor, ImageNet Creator, World Labs Founder

Fei-Fei Li

Home > People > Fei-Fei Li

Profile

FieldDetails
Current PositionProfessor, Stanford University
RoleCo-Director, Stanford HAI (Human-Centered AI)
CompanyFounder, World Labs (2024)
PreviousGoogle Cloud AI/ML Chief Scientist (2017-2018)
PhDCaltech
BirthplaceBeijing, China

Key Contributions

  • ImageNet: Catalyst for the modern deep learning revolution, paradigm shift in computer vision
  • Stanford Vision Lab: Center for computer vision research
  • HAI: Research on AI ethics, policy, and human-centered AI
  • World Labs: Founded Spatial Intelligence company for 3D world understanding AI (2024)

Research Timeline

PhD & Early Career (2000-2007)

Caltech -> Princeton -> Stanford

YearWorkImpact
2005One-Shot LearningLearning from limited data
2006Joined Princeton as Professor
2007Moved to StanfordFounded Stanford Vision Lab

ImageNet Era (2007-2015)

ImageNet & Deep Learning Revolution

YearWorkImpact
2009ImageNet Released14 million images, 20,000+ categories
2010ImageNet Challenge BeginsAnnual competition
2012AlexNet WinsDeep learning revolution begins
2015ResNet Surpasses HumansExceeds human performance in image classification

Google & HAI (2016-2023)

Industry Experience & AI Ethics

YearWorkImpact
2017-18Google Cloud AI Chief ScientistIndustry experience
2019HAI Co-foundedHuman-Centered AI Institute
2021HAI Co-DirectorAI policy, ethics research

World Labs (2024-present)

Founded Spatial Intelligence Company

YearWorkImpact
2024World Labs Founded$1B+ valuation
2024Spatial Intelligence Research3D world understanding AI

Major Publications

ImageNet & Visual Recognition

  • ImageNet: A Large-Scale Hierarchical Image Database (CVPR 2009)
  • ImageNet Large Scale Visual Recognition Challenge (IJCV 2015)

One-Shot & Few-Shot Learning

  • One-Shot Learning of Object Categories (PAMI 2006)

Visual Understanding

  • Dense Captioning (CVPR 2017)
  • Visual Genome (IJCV 2017)

AI Ethics & Policy

  • Numerous papers and op-eds on AI ethics and policy

Key Ideas

ImageNet (2009)

Core: Large-scale, high-quality datasets are key to AI advancement

Before: Thousands to tens of thousands of images
ImageNet: 14 million images, 20,000+ categories, human-labeled

Impact:

  • 2012 AlexNet -> Triggered deep learning revolution
  • Established data-centric AI paradigm
  • Model for all subsequent large-scale datasets

World Labs & Spatial Intelligence (2024)

Core: AI that understands the 3D world beyond 2D images

Goals:
- 3D world reconstruction from images/videos
- Physical world simulation
- Applications in robotics, AR/VR, autonomous driving

Connection to Physical AI:

  • 3D world understanding -> Essential for robot manipulation
  • Spatial reasoning -> Predicting physical interactions

Philosophy & Direction

Research Philosophy

“True AI advancement requires not just technology but a human-centered approach”

Research Direction Evolution

  1. 2000-2007: One-shot learning, object recognition
  2. 2007-2015: Large-scale visual recognition, ImageNet
  3. 2015-2019: Visual understanding, dense captioning
  4. 2019-2023: AI ethics, policy, human-centered AI
  5. 2024-present: Spatial Intelligence, 3D world understanding

World Labs

Company Overview

  • Founded: 2024
  • Mission: Spatial Intelligence - AI that understands the 3D world
  • Valuation: $1B+ (2024)
  • Investors: a16z, Radical Ventures, and others

Technology Direction

  • Image/video to 3D reconstruction
  • Physics simulation
  • Applications in Embodied AI

Awards & Recognition

  • Member, National Academy of Engineering
  • Member, National Academy of Medicine
  • Member, American Academy of Arts and Sciences
  • ACM Fellow
  • Time 100 Most Influential People (2015)

Books & Media

  • “The Worlds I See” (2023) - Memoir
  • TED Talks (multiple)
  • Numerous AI-related interviews and publications

See Also