3D Clothing Design Modeling Based on 3D Measurement and Generative Adversarial Networks

According to China's 12th Five-Year Development Program for the Disabled, the total number of disabled people in China is about 83 million, which means that on average, there is one disabled person with a certain degree of disability in every 16 people. Adaptive clothing is designed to meet the needs of people with disabilities or special needs. But so far, only a few apparel companies have developed such adaptive clothing. Helping people with disabilities to design diverse virtual clothing styles is an emerging field of research and innovation that utilizes the power of technology to help and enhance the aesthetics and self-presentation of clothing for people with disabilities. This field requires cutting-edge knowledge in a variety of fields such as computer vision, AIGC, augmented reality, etc., as well as a deep understanding of the different needs and challenges of people with different types of disabilities. This paper proposes a diverse clothing design model that specifically serves people with disabilities. In this paper, I focus my research on feature recognition of human body and digital design of 3D clothing. This paper proposes 3D clothing design generation based on generative adversarial network, applying deep learning neural network model for 3D human morphology feature recognition, recognizing the key parts of 3D human body such as neck, shoulder, armpit, chest, waist and hip, and extracting five measurement dimensions such as height, weight, chest circumference, waist circumference and hip circumference. On this basis, I propose an algorithm to construct a 3D clothing prototype by applying the clothing human body feature points to generate a natural clothing image from the clothing silhouette map based on the 3D features of the human body. I decompose the generation of garment images into two tasks: texture generation and garment shadow generation. Therefore, the whole system contains two modules, the image generator module and the shadow generator module. Finally, the feasibility and effectiveness of the designed garment design modeling algorithm are verified by experiments.