Perceiving Age Beyond Reality: Human and Machine Judgments of Virtual and AI-Generated Humans

• Ana Carolina de Oliveira Xavier PUCRS https://orcid.org/0009-0003-3103-692X
• Victor Flávio de Andrade Araujo PUCRS / Kunumi Institute https://orcid.org/0000-0002-7743-1365
• Paulo Ricardo Knob PUCRS / INCT SIMAI / CNPq https://orcid.org/0009-0001-9872-9355
• Soraia Raupp Musse PUCRS / Kunumi Institute / INCT SIMAI / CNPq https://orcid.org/0000-0002-3278-217X

DOI: https://doi.org/10.5753/semish.2026.22298

Resumo

This paper investigates how age is perceived and estimated in three categories of facial representations: Real Humans (photographs), Virtual Humans (3D MetaHuman renderings), and Digital Humans (photorealistic faces generated by a text-to-image model). We evaluate two age ranges (young: 20–30; older: 60–70) using a triangulated methodology: (i) four commercial age-estimation applications, (ii) a vision–language model (CLIP) under a zero-shot protocol, and (iii) an online user study (N = 112). Results indicate that commercial apps and CLIP exhibit their largest errors on 3D Virtual Humans (VHs), suggesting a domain generalization gap, while Artificial Intelligence (AI) generated 2D faces can be easier to classify than some real photographs. Human judgments show high accuracy for young AI-generated faces, but older AI-generated faces produce polarized responses. Overall, 3D VHs remain the most challenging domain for robust age perception.

Referências

Antipov, G., Baccouche, M., and Dugelay, J.-L. (2017). Face aging with conditional generative adversarial networks. arXiv preprint arXiv:1702.01983.

Buolamwini, J. and Gebru, T. (2018). Gender shades: Intersectional accuracy disparities in commercial gender classification. In Conference on fairness, accountability and transparency, pages 77–91. PMLR.

Dalvi, C. and Tuarob, S. (2021). A survey on the uncanny valley in virtual humans. IEEE Access, 9:53569–53587.

Grother, P., Ngan, M., and Hanaoka, K. (2019). Face recognition vendor test (frvt) part 3: Demographic effects. Technical Report NISTIR 8280, National Institute of Standards and Technology.

Hausladen, C. I., Knott, M., Camerer, C. F., and Perona, P. (2025). Social perception of faces in a vision-language model. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, pages 639–659.

Karkkainen, K. and Joo, J. (2021). Fairface: Face attribute dataset for balanced race, gender, and age. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 1548–1558.

Kim, T.-H., Kwon, H.-J., Choi, J.-Y., and Kweon, I. S. (2020). Face synthesis with identity preservation and perceived age calibration. In European Conference on Computer Vision (ECCV), pages 594–610. Springer.

Kumar, N., Berg, A. C., Belhumeur, P. N., and Nayar, S. K. (2009). Attribute and simile classifiers for face verification. In 2009 IEEE 12th International Conference on Computer Vision, pages 365–372. IEEE.

Levi, G. and Hassner, T. (2015). Age and gender classification using convolutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops, pages 34–42.

Liu, Z., Luo, P., Wang, X., and Tang, X. (2015). Deep learning face attributes in the wild. In Proceedings of the IEEE international conference on computer vision, pages 3730–3738.

Merler, M., Ratha, N., Ferrara, M., Zhang, J., and Smith, J. R. (2019). Diversity in faces. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 0–1. IEEE.

Pan, X., Han, Y., Shan, S., and Chen, X. (2018). Mean-variance loss for deep age estimation from a face. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5285–5294.

Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. (2021). Learning transferable visual models from natural language supervision. In International conference on machine learning, pages 8748–8763. PmLR.

Rothe, R., Timofte, R., and Van Gool, L. (2015). Dex: Deep expectation of apparent age from a single image. In Proceedings of the IEEE International Conference on Computer Vision Workshops (ICCVW), pages 10–15.

Thalmann, D. and Musse, S. R. (2000). The foundations to build a virtual human: The six-layer approach. In Proceedings of the workshop on virtual environments, pages 11–19.

Zhang, Z., Song, Y., and Qi, H. (2017). Age progression/regression by conditional adversarial autoencoder. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 5810–5818.