VGG Face - dataset

Transparency, accountability 🙈

The VGG Face dataset is seen to suffer from several significant transparency limitations:

• Lack of consent. The dataset was created by scraping images of 2,622 individuals from the internet without obtaining their consent or informing them about how their biometric data would be used.

• Unclear data collection process. While some details are provided about using IMDB and Google Image Search to collect images, the full extent of the data collection and curation process is not entirely transparent,

• Limited demographic information. Although some information on ethnicity, age, and kinship was collected from IMDB, it is unclear how comprehensive or accurate this demographic data is.

• Potential biases. The dataset primarily consists of celebrities and public figures, which may not represent a diverse range of faces and could introduce biases in facial recognition technologies developed using this data. But the dataset lacks comprehensive documentation about potential biases, limitations, or ethical considerations that researchers and developers should be aware of when using the data.

• Lack of clear usage guidelines. There appears to be no clear guidelines or restrictions on how the dataset can be used, potentially leading to misuse or unethical applications of the biometric data.

Risks, harms 🛑

The VGG Face dataset has raised significant ethical concerns and potential harms by collecting and distributing biometric data of over 2,600 individuals without their consent, potentially enabling privacy and copyright violations, surveillance, and the development of biased facial recognition technologies.

Investigations, assessments, audits 👁️

• Harvey, A., LaPlace, J. (2019). Exposing.ai