preferred_colleague_dataset

This dataset provides a basis for analyzing human colleague preferences. It contains 2,200 proband face images, including 2,000 real faces taken from the SCUT-FBP5500-Database balanced by gender and ethnicity (Asian/Caucasian) and 200 synthetically generated faces. Preference judgments were collected from 451 diverse annotators. The repository includes the face images, raw data matrix, and annotator demographics, supporting research on bias and the development of fairer AI models.