Abstract
This study investigates the application of machine learning in evaluating brand typicality during product line extension, taking inspiration from Dyson’s recent expansion into the headphone category. The research process comprised four main stages: extraction of visual features via Design Format Analysis (DFA), development of an image classification model on the WEKA platform, generation of Dyson-style speaker images using Stable Diffusion, and expert evaluation of the results. The classification model reached an accuracy of 87.21%, demonstrating an ability to distinguish between Dyson and non-Dyson styles. Expert feedback indicated that the model was effective in identifying surface-level features such as color and material, but suggested further refinement is needed to improve the recognition of form and structure. These findings suggest that machine learning has potential as a design support tool for assessing visual brand consistency and may contribute to greater efficiency in brand extension processes.
Keywords
Machine Learning Classification; Brand Typicality; Visual Design Language; AI-Generated Content
DOI
https://doi.org/10.21606/iasdr.2025.334
Citation
Huang, S.,and Wang, H.(2025) Is It Still Dyson? Evaluating Brand Identity in AI-Generated Designs Using Machine Learning, in Chang, C.-Y., and Hsu, Y. (eds.), IASDR 2025: Design Next, 02-05 December, Taiwan. https://doi.org/10.21606/iasdr.2025.334
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
Conference Track
Track 10 - Design Practices & Impacts
Is It Still Dyson? Evaluating Brand Identity in AI-Generated Designs Using Machine Learning
This study investigates the application of machine learning in evaluating brand typicality during product line extension, taking inspiration from Dyson’s recent expansion into the headphone category. The research process comprised four main stages: extraction of visual features via Design Format Analysis (DFA), development of an image classification model on the WEKA platform, generation of Dyson-style speaker images using Stable Diffusion, and expert evaluation of the results. The classification model reached an accuracy of 87.21%, demonstrating an ability to distinguish between Dyson and non-Dyson styles. Expert feedback indicated that the model was effective in identifying surface-level features such as color and material, but suggested further refinement is needed to improve the recognition of form and structure. These findings suggest that machine learning has potential as a design support tool for assessing visual brand consistency and may contribute to greater efficiency in brand extension processes.