PhaceSpace – AI-Generated Composite Sketches
DOI:
https://doi.org/10.52825/th-wildau-ensp.v3i.3531Keywords:
Human-Computer Interaction, Human-Centered Design, Phantom Image, StyleGAN, Facial Recognition, Generative AIAbstract
The PhaceSpace research project investigated how composite sketches can be created using generative AI (StyleGAN). To this end, a prototype was developed which extends StyleGAN with an iterative search process in face space through a task-specific human-AI dialogue. In two user studies, the user interface and interaction options, as well as the search and selection process, were experimentally investigated and further developed. The results showed that users were able to create suitable images of target individuals using the prototype. There is particular potential for further development in the targeted search for and selection of not only faces as a whole, but also individual facial features.
Downloads
References
[1] M. Roscher, “Visuelle Fahndungshilfe,” info110, vol. 25, no. 02, pp. 40–41, 2017.
[2] H. Mendelin and R. Wortmann, Phantombilder: Das Handbuch für Phantombildersteller und Zeugen. BoD–Books on Demand, 2017.
[3] T. Karras, T. Aila, S. Laine, and J. Lehtinen, “Progressive Growing of GANs for Improved Quality, Stability, and Variation,” ArXiv171010196 Cs Stat, Feb. 2018, Accessed: Aug. 18, 2020. [Online]. Available: http://arxiv.org/abs/1710.10196
[4] T. Karras, S. Laine, and T. Aila, “A Style-Based Generator Architecture for Generative Adversarial Networks,” presented at the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 4401–4410. Accessed: Aug. 18, 2020. [Online]. Available: https://openaccess.thecvf.com/content_CVPR_2019/html/Karras_A_Style-Based_Generator_Architecture_for_Generative_Adversarial_Networks_CVPR_2019_paper.html
[5] T. Karras, S. Laine, M. Aittala, J. Hellsten, J. Lehtinen, and T. Aila, “Analyzing and Improving the Image Quality of StyleGAN,” ArXiv191204958 Cs Eess Stat, Mar. 2020, Accessed: Jun. 17, 2020. [Online]. Available: http://arxiv.org/abs/1912.04958
[6] T. Karras et al., “Alias-Free Generative Adversarial Networks,” Oct. 18, 2021, arXiv: arXiv:2106.12423. Accessed: Oct. 17, 2023. [Online]. Available: http://arxiv.org/abs/2106.12423
[7] R. Stollhoff, “Modeling prosopagnosia: computational theory and experimental investigations of a deficit in face recognition.,” University of Leipzig, 2010.
[8] R. Stollhoff, I. Kennerknecht, T. Elze, and J. Jost, “A computational model of dysfunctional facial encoding in congenital prosopagnosia,” Neural Netw., vol. 24, no. 6, pp. 652–664, Aug. 2011, doi: 10.1016/j.neunet.2011.03.006.
[9] C. Frowd, “Facial composites and techniques to improve image recognizability,” Forensic Facial Identif. Theory Pract. Identif. Eyewitnesses Compos. CCTV, pp. 43–70, 2015.
[10] C. D. Frowd, P. J. Hancock, and D. Carson, “EvoFIT: A holistic, evolutionary facial imaging technique for creating composites,” ACM Trans. Appl. Percept. TAP, vol. 1, no. 1, pp. 19–39, 2004.
[11] DIN EN ISO 9241-210:2020-03, Ergonomie der Mensch-System-Interaktion_- Teil_210: Menschzentrierte Gestaltung interaktiver Systeme (ISO_9241-210:2019); Deutsche Fassung EN_ISO_9241-210:2019, 2019. doi: 10.31030/3104744.
[12] A. Radford et al., “Learning Transferable Visual Models From Natural Language Supervision,” Feb. 26, 2021, arXiv: arXiv:2103.00020. doi: 10.48550/arXiv.2103.00020.
[13] K. Karkkainen and J. Joo, “FairFace: Face Attribute Dataset for Balanced Race, Gender, and Age for Bias Measurement and Mitigation,” in 2021 IEEE Winter Conference on Applications of Computer Vision (WACV), Waikoloa, HI, USA: IEEE, Jan. 2021, pp. 1547–1557. doi: 10.1109/WACV48630.2021.00159.
[14] S. Serengil and A. Özpınar, “A Benchmark of Facial Recognition Pipelines and Co-Usability Performances of Modules,” Bilişim Teknol. Derg., vol. 17, no. 2, pp. 95–107, Apr. 2024, doi: 10.17671/gazibtd.1399077.
[15] DIN EN ISO 9241-110:2020-10, Ergonomie der Mensch-System-Interaktion_- Teil_110: Interaktionsprinzipien (ISO_9241-110:2020); Deutsche Fassung EN_ISO_9241-110:2020, 2020. doi: 10.31030/3147467.
[16] DIN EN ISO 9241-11:2018-11, Ergonomie der Mensch-System-Interaktion_- Teil_11: Gebrauchstauglichkeit: Begriffe und Konzepte (ISO_9241-11:2018); Deutsche Fassung EN_ISO_9241-11:2018. doi: 10.31030/2757945.
[17] C. D. Frowd et al., “Recovering faces from memory: The distracting influence of external facial features.,” J. Exp. Psychol. Appl., vol. 18, no. 2, p. 224, 2012.
[18] VoiceAI. KI-Servicezentrum für sensible und kritische Infrastrukturen. [Online]. Available: https://kisski.gwdg.de/leistungen/6-09-voice-ai/
[19] P. Mayring, Qualitative Inhaltsanalyse, 11th ed. Beltz Verlagsgruppe, 2010. [Online]. Available: https://content-select.com/de/portal/media/view/519cc17d-6158-4e6c-9944-253d5dbbeaba
[20] R Core Team, “R: a language and environment for statistical computing,” R Foundation for Statistical Computing, Vienna, Austria, manual, 2025. [Online]. Available: https://www.R-project.org/
[21] D. Bates, M. Mächler, B. Bolker, and S. Walker, “Fitting linear mixed-effects models using lme4,” J. Stat. Softw., vol. 67, no. 1, pp. 1–48, 2015, doi: 10.18637/jss.v067.i01.
[22] V. Goldité, H. Raz, B. Nguyen, and I. Dormoy, GAN Face Editing. (2022). [Online]. Available: https://github.com/valentingol/gan-face-editing
[23] F. Strohm, M. Bâce, and A. Bulling, “HAIFAI: H uman- AI Interaction for Mental F ace Reconstruct i on,” ACM Trans. Interact. Intell. Syst., p. 3725891, Apr. 2025, doi: 10.1145/3725891.
[24] O. Patashnik, Z. Wu, E. Shechtman, D. Cohen-Or, and D. Lischinski, “StyleCLIP: Text-driven manipulation of StyleGAN imagery,” in Proceedings of the IEEE/CVF international conference on computer vision (ICCV), Oct. 2021, pp. 2085–2094.
Downloads
Published
How to Cite
Conference Proceedings Volume
Section
License
Copyright (c) 2026 Rainer Stollhoff , Christin Buley , Stefan Laenger, Ashley Karongo
This work is licensed under a Creative Commons Attribution 4.0 International License.