Application of convolution neural networks‐based hierarchical perception technology in 3D clothing designs. Issue 6 (6th September 2022)

• Record Type: Journal Article
• Authors: Hong, Yu
• Is Part Of: IET networks. Volume 11:Issue 6(2022)
• Page Start: 257
• Page End: 263
• Publication Date: 2022-09-06
• Subjects: Convolutional neural network -- Digital dress -- Deep learning -- 3D clothing design; Computer network architectures -- Periodicals; Computer network protocols -- Periodicals; Information networks -- Periodicals; Telecommunication systems -- Periodicals; 004.605
• DOI: 10.1049/ntw2.12047
• Languages: English
• ISSNs: 2047-4954

Abstract

Abstract: The purpose of this research is to solve the problems in the modern life, such as the inconvenience caused by fast life pace, the need to save time, the unsatisfactory products for consumers, the low brand awareness in the clothing industry, serious product homogenisation and low life quality. The study of the application of hierarchical perception technology in 3D (three‐dimensional) clothing designs is carried out based on CNN (convolutional neural networks). The study includes the training of CNN, the selection of 3D clothing models and 2D (two‐dimensional) projection of 3D human body models. A simulation experiment is conducted by taking 49‐year‐old males, 22‐year‐old males and 50‐year‐old females as research objects. The findings show that the training network can effectively identify the typical lines of the human's hindneck, chest, shoulders, and waist, and the accuracy of the front and side recognition of the human body can be reached as high as 93% on average. The effective identification of human characteristics is achieved. The study has an important impact on the technical field and clothing industry and also has profound theoretical and practical significance.