A product-form aesthetic evaluation system

An artificial neural network-based aesthetic evaluation system for product-form is presented in this paper. Fuzzy set theory and semantic difference (SD) method are applied to set up an aesthetic evaluation experiment, in which 400 product images were evaluated. First, a database of normalized images and quantitative aesthetic descriptions was constructed. Second, a back-propagation (BP) network model was constructed to learn the relationship between image spaces and quantified aesthetic words. Last, an intelligent evaluation system for product design was then developed based on the learned BP model. Applications demonstrated that the system evaluate product form-aesthetics accurately. Designers may promptly modify their design with the guidance of evaluation