AI-Faces by Illinois is a database of AI-generated, photorealistic images of children’s faces designed for use in social and behavioral research. Developed by Drs. Michael T. Rizzo (psychology), Brenda Straka (psychology), and psychology PhD student Tim Chao, this CSBS-funded resource is available at no cost to researchers, providing a controlled, flexible alternative to traditional face databases, making it easier to conduct studies that may require diverse and standardized faces.
The AI Faces by Illinois dataset includes over 1,100 images of children at different ages (3, 6, 10, and 15 years old), gender presentations (female, male), racial backgrounds (Black, East Asian, Latine, Middle Eastern/North African, Native American, Pacific Islander, South Asian, White), and emotions (smiling, angry, sad). A researcher might use this database to source free, open-access images of, for example, a 6-year-old Latina girl smiling, a 10-year-old White boy looking angry, or a 15-year-old Native American girl frowning. The images are high-resolution, standardized, and background-free, making them easy to include a range of research contexts.
The images have been normed on CloudResearch by over 1,250 adult participants on their perceived (1) realness, (2) wealth, (3) niceness, (4) age, (5) emotion (happy, angry, sad), (6) gender (feminine, masculine), and (7) race (Black, East Asian, Latine, Middle Eastern/North African, Native American, Pacific Islander, South Asian, White), allowing researchers to carefully match images based on relevant criteria. Norming data from child participants, and images of adult faces, will be available in the future.
AI-generated faces provide a valuable tool for researchers who need high-quality images of human faces without privacy or copyright concerns. AI-Faces by Illinois could be particularly useful for researchers who want to avoid ethical concerns regarding using images of real children, publish images used as stimuli in journals or conference presentations, or use images that more closely align with the needs of the study. Researchers can also edit the images to suit their needs, adjusting features like clothing or background. The database offers flexibility, allowing for greater control over experimental variables.
Researchers can download the full set of images and accompanying data through OSF at https://osf.io/vurm5/. For further inquiries, contact Dr. Michael T. Rizzo (mtrizzo@illinois.edu) or Tim Chao (timchao1@illinois.edu).