Increasing geographical diversity of visual datasets

What is it about?

The study of fairness and diversity in visual datasets are heavily influenced by race. However, race does not take into view the immense diversity present across different cultures, regions and ethnicities. We propose a new point of view i.e. geography. We also propose a framework for analysing and studying the issue of geographical bias.

Featured Image

Why is it important?

Bias and fairness in datasets is heavily influenced by the western concept of race and race relations. Our research proposes a new way to view this issue, one that can encompass the diversity of humanity as whole.