Research on 3D Digital Clothing Design Based on Optimized Capsule Network Using Enhanced Salp Swarm Optimization Algorithm

Xue Shen
DOI: https://doi.org/10.1109/ICDSNS62112.2024.10690948
2024-07-26
Abstract:The field of fashion design has been continuously transformed and developed as a result of the advancements in virtual reality technology. In prior research, a 3D digital clothing design model was developed by focussing on the key technologies and research challenges associated with the optimized Convolution Neural Network (CNN). Nevertheless, this approach yields an average performance in the field of 3D digital clothing design. Therefore, this investigation introduces an Enhanced Salp Swarm Optimization Algorithm (ESSOA) that is implemented in conjunction with the Capsule Network (CapsNet) model to identify 3D digital clothing designs. The ESSOA-CapsNet method involves mainly with the extraction of features from 2D images for design of 3D digital clothing. In order to achieve this, the ESSOA-CapsNet technique that has been introduced employs the Median Filtering (MF) technique to flatten the images. Subsequently, the hyperparameter of the CapsNet model is altered using ESSOA to enhance its efficacy in 3D digital clothing design recognition. In contrast to traditional CNN methods, such as 3D point cloud retrieval, 3D-CDM-CNN, and CNN The ESSOA-CapsNet surpasses the existing works and achieves a higher recognition accuracy of 98.95%.
Computer Science,Engineering
What problem does this paper attempt to address?