基于DCGAN算法的服装效果图生成方法
Design method of garment effect drawing based on DCGAN algorithm
郭宇轩 1孙林1
作者信息
- 1. 大连工业大学 服装学院,辽宁 大连 116034
- 折叠
摘要
为了提高服装设计效率,适应时尚产品迭代加速的趋势,提出一种基于深度卷积对抗网络(DCGAN)的服装效果图生成方法.搭建适用于服装效果图生成任务的DCGAN模型,制作服装秀场数据集进行模型训练并生成服装效果图,设计师主观筛选具有设计参考价值的生成服装效果图,计算有效生成图像比例,评估该模型性能和生成图像质量,通过人机交互的方式优化部分生成图像并形成最终设计方案.结果表明:优化后的DCGAN模型可以快速提取流行趋势生成创意设计方案,辅助设计师高效完成设计效果表达,为服装设计的智能化提供有效途径和方法参考.
Abstract
In order to improve the efficiency of clothing design and adapt to the trend of accelerating fashion product iteration,a method of generating fashion illustration based on deep convolutional Adversarial network(DCGAN)was proposed.By building a DCGAN model suitable for the task of fashion illustration generating,making a clothing show data set for model training and fashion illustration generating,the designer subjectively selects the generated fashion illustration with design reference value,calculates the proportion of effectively generated images,and evaluates the performance of the model and the quality of generated images.Through human-computer interaction,the optimized part generates images and forms the final design scheme.The results show that the optimized DCGAN model can quickly extract fashion trends and generate creative design schemes,assist designers to efficiently complete the expression of design effects,and provide effective ways and methods for the intelligent fashion design.
关键词
卷积神经网络/DCGAN/服装效果图/交互设计/深度学习Key words
deep learning/convolutional neural network/DCGAN/effect drawing/interaction design引用本文复制引用
基金项目
辽宁省"兴辽英才计划"项目(XLYC2210025)
辽宁省教育厅科研项目(LJKFR20220220)
出版年
2024
NETL
NSTL
维普
万方数据