Digital Creative Product Design based on Hyper Parameter Tuned Deep Convolutional Neural Network using Artificial Intelligence Images

Artificial Intelligence (AI) image generation techniques have proved to be potential techniques in the development of the digital creative product design. However, integrating digital culture with modernity while classifying AI image generation technology was a major concern. In this research, the hyper parameter tuned Deep Convolutional Neural Network with Stochastic Gradient (DCNN-SGD) is proposed to overcome these drawbacks, by classifying $A I$ images of product design. The open source AI images are collected and pre-processed to achieve noise free images by applying Gaussian filter. The pre-processed images are then fed to the CNN to extract features. The creative product design process is analysed which includes research and analysis, design expression and the final evaluation. The DCNN with convolutional, pooling and fully connected layers are used for the classification and identification of cultural symbols. For the effective identification of cultural heritage, the hyperparameters of the DCNN are finetuned with SGD. The experimental results showed that the proposed DCNN-SGD achieved better performance with classification accuracy of $\mathbf{9 8. 6 7 \%}$, precision of $\mathbf{9 5. 7 7 \%}$, and recall of $94.26 \%$ when compared to the existing methods improved Backpropagation Neural Network (BPNN) and Artificial Neural Network (ANN).