Hierarchical Clustering in Face Similarity Score Space
This work provides insights into the structure of face similarity data, which could aid in improving face recognition systems or understanding biases, but it is incremental as it applies standard clustering methods to a specific domain.
The paper tackled the problem of analyzing face recognition similarity scores by applying hierarchical clustering to group images based on subjects, gender, ethnicity, and illumination conditions, with evidence supporting the existence of these clusters.
Similarity scores in face recognition represent the proximity between pairs of images as computed by a matching algorithm. Given a large set of images and the proximities between all pairs, a similarity score space is defined. Cluster analysis was applied to the similarity score space to develop various taxonomies. Given the number of subjects in the dataset, we used hierarchical methods to aggregate images of the same subject. We also explored the hierarchy above and below the subject level, including clusters that reflect gender and ethnicity. Evidence supports the existence of clustering by race, gender, subject, and illumination condition.