Intelligent Cognitive Artistic Style Recognition Model for Mechanical Design

The aesthetic design of mechanical products is an indispensable aspect of mechanical manufacturing, as it significantly influences the overall visual appeal. Central to achieving aesthetically pleasing designs is the recognition of artistic style, which predominantly reflects the underlying mechanisms of human cognition. However, existing algorithms for artistic style recognition primarily concentrate on quantifying them, thereby neglecting the pivotal role of human perception in effectively categorizing design artistic styles. Consequently, these algorithms encounter several challenges, including difficulties in feature screening and inadequate explanation of their internal mechanisms. To address these limitations, this study introduces a novel approach for intelligent artistic style recognition that leverages the human cognitive mechanism. The framework of this approach involves constructing the Artificial Residual Neural Network (ARNN) by integrating deep convolutional and residual neural networks. Design simples undergo feature extraction, and subsequently, the distinctive features of each convolutional layer are visualized using three visualization methods. Moreover, the network parameters are fine-tuned by integrating human perception of the feature maps at each layer. Furthermore, the inclusion of a human cognitive auxiliary structure facilitates network guidance based on human cognitive mechanism, this enables the network to concentrate on effectively screening pertinent features. Finally, the effectiveness of the proposed method is assessed through four classification tasks and generating corresponding design suggestions pertaining to artistic styles. This research makes a valuable contribution to the field of mechanical design artistic style recognition and holds the potential to offer significant aesthetic design guidance for mechanical products.