VGG Face facial recognition dataset

VGG Face is a dataset of 2.6 million facial images of 2,622 people that was created to provide researchers working on facial recognition systems with access to biometric data.  

The dataset mostly comprises celebrities, public figures, actors, and politicians whose names were chosen 'by extracting males and females, ranked by popularity, from the Internet Movie Data Base (IMDB) celebrity list.' Information about ethnicity, age, and kinship was also collected from IMDB.

Operator: ChaLearn; Chinese Academy of Sciences; Delft University of Technology; Simula Research Laboratory; University of Applied Sciences & Arts Western Switzerland; University of California, Berkeley; Universitat Autònoma de Barcelona
Developer: University of Oxford

Country: UK

Sector: Research/academia

Purpose: Develop facial recognition systems

Technology: Dataset; Facial recognition  
Issue: Privacy; Copyright

Transparency: Privacy

Risks and harms 🛑

At no point did any individual whose personal details were collected provide consent or information about how they were being used, according to Adam Harvey at exposing.ai, raising the question as to whether the images of public figures should be available for any organisation or person to use as they see fit.

The dataset has since been removed from Oxford University's website.

Investigations, assessments, audits 🧐

Page info
Type: Dataset
Published: January 2023