Leveraging Generative Adversarial Networks for Dynamic UI Icon Generation: Addressing Style Consistency Challenges in Collaborative Design

DOI

10.2991/978-94-6463-823-3_90How to use a DOI?

Keywords

GANs; Semantic Style Transfer; Dynamic UI Icon Generation

Abstract

This paper investigates the application of Generative Adversarial Networks (GANs) in conjunction with semantic style transfer to enable dynamic UI icon generation for collaborative design environments. Recent advances in deep learning—particularly the use of Convolutional Neural Networks and GANs—have transformed image synthesis, but challenges remain in generating icons that preserve fine details, exhibit diverse styles, and maintain overall style consistency. The traditional GAN framework is prone to issues such as training instability, mode collapse, and limited style variety, especially when training data is scarce. To address these shortcomings, the proposed approach integrates several innovations. First, spectral normalization and Wasserstein GAN techniques are employed to enhance training stability and convergence, reducing the risk of mode collapse while producing higher-quality outputs. Second, the introduction of semantic segmentation and attention mechanisms directs the style transfer process; this ensures that key design elements and subtle details are maintained even as styles are transferred. Third, the architecture is optimized using lightweight network designs to lower computational demands without sacrificing performance or fidelity. Experimental results demonstrate that the integrated strategy yields realistic and high-quality icons with fewer artifacts and better overall style coherence, thus substantially easing the workload of designers by automating repetitive aspects of icon creation.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Hefan Chen
PY  - 2025
DA  - 2025/08/31
TI  - Leveraging Generative Adversarial Networks for Dynamic UI Icon Generation: Addressing Style Consistency Challenges in Collaborative Design
BT  - Proceedings of the 2025 3rd International Conference on Image, Algorithms, and Artificial Intelligence (ICIAAI 2025)
PB  - Atlantis Press
SP  - 917
EP  - 924
SN  - 2352-538X
UR  - https://doi.org/10.2991/978-94-6463-823-3_90
DO  - 10.2991/978-94-6463-823-3_90
ID  - Chen2025
ER  -