A carregar...
Publicação
Enhancing product variation generation with hyperparameter optimization of user-guided model
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 817.54 KB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Recent advances in generative AI have shown promising potential in creative tasks, yet challenges
persist in adapting these technologies for specialized industrial applications. This work project
presents an AI-powered product design assistant that bridges abstract concepts and visual
representations. Through model comparison and hyperparameter optimization, the system employs
Stable Diffusion XL for industrial product visualization and variations, enhanced by a dynamic
user preference analyzer. The system's structured database supports preference integration and
mood board functionality, enabling users to organize and visualize their design roadmap. Results
demonstrate improved image generation efficiency while maintaining industrial design
specifications and advancing AI-assisted design through personalization. In this paper, the
optimization of hyperparameter was mainly studied, achieving a reasonable balance of realism and
faithfulness on image outputs.
Descrição
Palavras-chave
Artificial Intelligence Image-to-image generation Stable Diffusion XL Product design Prompt engineering Industrial design Machine learning