<abstract xmlns="http://www.w3.org/1999/xhtml">

<p>The evolution of stage design is fundamentally linked to advancements in modern science and technology, which, in turn, catalyze further opportunities for innovation in this field. This study meticulously explores the potential of artificial intelligence in enhancing performance space design and introduces an advanced visual design system tailored for performance spaces underpinned by virtual reality technologies. To facilitate accurate three-dimensional modeling of performance spaces, this research adopts the generative adversarial network (GAN) training mechanism, incorporating point cloud data as a direct input to enhance network architecture. This approach innovatively integrates a multi-resolution point cloud completion network structure that leverages fused graph attention features. Furthermore, to streamline model complexity, an inverse residual network is employed. Alongside this, a novel semantic segmentation method tailored for 3D scenes utilizes a self-attention mechanism, demonstrating significant advancements in the field. Through comparative experiments and the construction of 3D scenes, the study evaluates the efficacy of the proposed design model. The findings reveal substantial reductions in the Chamfer Distance (CD) mean—ranging between 43.75% to 67.47% for the residual point cloud and 40.68% to 67.89% for the complete point cloud—significantly outperforming three alternative algorithms. The semantic segmentation method further showcases enhanced precision in building 3D scenes, achieving an average intersection ratio of 47.33% and a pixel accuracy of 76.05%. Collectively, the experiments corroborate that the model developed in this study not only surpasses traditional models in terms of innovation but also meets the stringent real-time and accuracy standards required for engineering applications.</p>
</abstract>

The evolution of stage design is fundamentally linked to advancements in modern science and technology, which, in turn, catalyze further opportunities for innovation in this field. This study meticulously explores the potential of artificial intelligence in enhancing performance space design and introduces an advanced visual design system tailored for performance spaces underpinned by virtual reality technologies. To facilitate accurate three-dimensional modeling of performance spaces, this research adopts the generative adversarial network (GAN) training mechanism, incorporating point cloud data as a direct input to enhance network architecture. This approach innovatively integrates a multi-resolution point cloud completion network structure that leverages fused graph attention features. Furthermore, to streamline model complexity, an inverse residual network is employed. Alongside this, a novel semantic segmentation method tailored for 3D scenes utilizes a self-attention mechanism, demonstrating significant advancements in the field. Through comparative experiments and the construction of 3D scenes, the study evaluates the efficacy of the proposed design model. The findings reveal substantial reductions in the Chamfer Distance (CD) mean—ranging between 43.75% to 67.47% for the residual point cloud and 40.68% to 67.89% for the complete point cloud—significantly outperforming three alternative algorithms. The semantic segmentation method further showcases enhanced precision in building 3D scenes, achieving an average intersection ratio of 47.33% and a pixel accuracy of 76.05%. Collectively, the experiments corroborate that the model developed in this study not only surpasses traditional models in terms of innovation but also meets the stringent real-time and accuracy standards required for engineering applications.

Artificial intelligence-driven performance space design and virtual reality interaction model

Applied Mathematics and Nonlinear Sciences

This work is licensed under the Creative Commons Attribution 4.0 International License.

The evolution of stage design is fundamentally linked to advancements in modern science and technology, which, in turn, catalyze further opportunities for...

Artificial intelligence-driven performance space design and virtual reality interaction model

Data publikacji: 05 sie 2024

Zakres stron: -

Otrzymano: 04 kwi 2024

Przyjęty: 02 lip 2024

DOI: https://doi.org/10.2478/amns-2024-1906

Słowa kluczowe
Generative adversarial networks, Graph attention, Self-attention mechanism, Semantic segmentation, Artificial intelligence, Performance space design

© 2024 Xinjing Wu et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Artificial intelligence-driven performance space design and virtual reality interaction model

Data publikacji: 05 sie 2024

Zakres stron: -

Otrzymano: 04 kwi 2024

Przyjęty: 02 lip 2024

DOI: https://doi.org/10.2478/amns-2024-1906

Słowa kluczoweGenerative adversarial networks, Graph attention, Self-attention mechanism, Semantic segmentation, Artificial intelligence, Performance space design

© 2024 Xinjing Wu et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Słowa kluczowe
Generative adversarial networks, Graph attention, Self-attention mechanism, Semantic segmentation, Artificial intelligence, Performance space design