Rethinking Common Assumptions to Mitigate Racial Bias in Face Recognition Datasets

Many existing works have made great strides towards reducing racial bias in face recognition. However, most of these methods attempt to rectify bias that manifests in models during training instead of directly addressing a major source of the bias, the dataset itself. Exceptions to this are BUPT-Balancedface/RFW and Fairface, but these works assume that primarily training on a single race or not racially balancing the dataset are inherently disadvantageous. We demonstrate that these assumptions are not necessarily valid. In our experiments, training on only African faces induced less bias than training on a balanced distribution of faces and distributions skewed to include more African faces produced more equitable models. We additionally notice that adding more images of existing identities to a dataset in place of adding new identities can lead to accuracy boosts across racial categories. Our code is available at github.com. Accepted as an oral paper to the Human-centric Trustworthy Computer Vision (HTCV 2021) workshop at ICCV 2021; 17 pages, 9 figures, 6 tables