Research on Product Styling Design Evaluation by Genetic Algorithm Based Convolutional Neural Networks

In recent days, with the advancements of continuous development in industrial civilization, living standards of people as well as consumption habits had significantly evolved in product styling design. Traditional approaches for product styling design had faced several challenges such as high costs and personalization issues. Therefore, this research proposes Genetic Algorithm based Convolutional Neural Networks based (GA based CNN) for evaluating product design styling. Initially, input is taken from Computer-Aided Design (CAD) software, user surveys and feedback forms, design databases. After that, GA is utilized for selecting features which significantly influence product styling like such as geometric properties, symmetry, curvature, surface texture, and structural modifications like merging, stretching, and extrusion. Finally, CNN is used for evaluating the aesthetics, functionality, and user preferences of product styling by analysing spatial hierarchies and the selected features is processed through convolutional layers to extract visual attributes. The proposed GA based CNN had achieved better results in terms of accuracy (82.5<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>) when compared with existing approach like Multilayer Perceptron Genetic Algorithm Neural Network (GA-MLP-NN).