基于深度学习的服装版型栅格图重建  

Clothing Pattern Raster Graph Reconstruction Based on Deep Learning

作　　者：伍柳柳 王鑫 马颜雪 钟跃崎[1,2] WU Liu-liu;WANG Xin;MA Yan-xue;ZHONG Yue-qi(College of Textiles,Donghua University,Shanghai 201620;Key Laboratory of Textile Science&Technology,Ministry of Education,Donghua University,Shanghai 201620)

机构地区：[1]东华大学纺织学院,上海201620 [2]东华大学纺织面料技术教育部重点实验室,上海201620

出　　处：《纺织科学与工程学报》2023年第3期18-21,共4页Journal of Textile Science & Engineering

基　　金：上海自然科学基金项目(21ZR1403000)。

摘　　要：为了将服装版型栅格图自动转换为数字化服装版型,以深度学习为技术手段,以服装参数为驱动自建服装版型数据集,实现输入服装版型栅格图,自动输出与之对应的服装版型的关键点坐标,继而利用预测结果进一步绘制数字化服装版型。实验结果表明,模型针对本数据集的预测效果较好,其中MSE(5.28218)、RMSE(1.93216)、MAE(1.26072)、R~2(0.99848),由此可见MSE、MAE、RMSE数值较低,R~2接近于1。In order to automatically convert the clothing pattern raster graph into a digital clothing pattern,deep learning was taken as a technical means,the clothing parameters was adopted as the driving to self-build clothing pattern dataset,to realize the input of clothing pattern raster graph and automatically output the corresponding key point coordinates of the clothing pattern,and the digital clothing pattern was drawn by using the prediction results.The results showed that the prediction effects of the model for this dataset was good,and MSE(5.28218),RMSE(1.93216),MAE(1.26072),R2(0.99848),it could be seen that the MSE,MAE,RMSE values were relatively low,and R~2 was close to 1.

关 键 词：服装版型 数字化 栅格图 BEZIER曲线 神经网络 深度学习 

分 类 号：TS941.2[轻工技术与工程—服装设计与工程]