Personalized skeuomorphic UI generation for industrial interfaces

To solve the issue of the digital transformation of Chinese manufacturing in terms of the bottleneck between industrial interfaces not being able to adapt to heterogeneous operators and the high cognitive load imposed on them, the authors propose the SkeuoUI-Gen framework based on the adaptation of skeuomorphic design principles and the use of conditional diffusion models to produce personalized industrial interfaces in the context of Chinese manufacturing. In this regard, the experiment used a within-subjects design involving 250 manufacturing industry operators (diverse in age, experience, and industry sectors) to evaluate three interface types: traditional flat interfaces, fixed skeuomorphic interfaces, and personalized adaptation interfaces. The experiment used objective evaluations (FID and PSNR) and subjective evaluations (SUS score and cognitive load), and trained the model on multiple sources: 50,000 interaction logs from operators and 50,000 screenshots of industrial user interfaces. The experiment found that the personalized adaptation interface resulted in a 78.6% SUS score (an increase of 15.4% compared to the traditional baseline), improved efficiency by 24.7%, and reduced serious safety-related errors by 52% and 67%. The network achieved a lower FID (21.5) than GAN-based approaches and required only 2.3 seconds per generation. In addition, the network presented robustness through multi-dimensional validation. This framework expands the cognitive load theory and the technology acceptance model.