Intelligent Clothing Design and Embodied Cognitive System Based on Convolutional Neural Network and Generative Adversarial Network

With the popularization of virtual reality networks, monitoring intelligent clothing equipped with various flexible sensors is developing in the direction of combining clothing with science and technology, becoming an important niche market in the clothing industry. From the perspective of art and design, traditional clothing has three core attributes: beauty, comfort, and economy. However, as a specific type of clothing, smart clothing should not only meet daily wear requirements, but also consider functionality and safety. This requires designers to integrate the materials and electronic elements in smart clothing with the overall design of the garment. In addition, in smart clothing, the use of soft, elastic, and easily washable materials to replace some electronic components to enhance their wearing comfort and safety is also a concern for designers and users. This study is based on convolutional neural network (CNN) and generating adversarial networks (GAN) technologies, combined with the dataset extended by Bok intelligent CAD software, for intelligent clothing design. In the experiment, using a detailed body size dataset of 1100 children, the training accuracy of the CNN model increased from the initial 75% to 92%, and the F1 score increased from 73% to 91%. In addition, the quality score of the images generated by the GAN model increased from 6.8 to 8.0. The results showed that the designed smart clothing scored above 8 points in terms of comfort, functionality, and safety, demonstrating the superiority of smart clothing in practical applications. This study not only provides effective methods for the design of smart clothing, but also demonstrates its potential in improving user wearing experience and safety.