基于深度学习的卷积神经网络在民族图案语义标签中的应用研究

Application Research of Convolutional Neural Network Based on Deep Learning in National Pattern Semantic Labels

各个民族的服饰图案内容丰富,种类繁多,解读各个服饰图案的内容含义具有重要的现实意义。传统图案语义标签大多是提取了图案纹理、颜色等特征进行分类标注,无法准确地表达出图案的真实内容。笔者基于深度卷积神经网络对图案进行深度特征提取,并进行准确分类,构建了基于卷积神经网络的民族图案语义标签结构,与传统民族图案语义标签相比具有一定的先进性。

The costumes of various ethnic groups are rich in content and various in variety. It is of practical significance to interpret the meaning of the content of each costume pattern. Most of the traditional pattern semantic labels are extracted with features such as pattern texture and color, and cannot accurately express the true content of the pattern. This paper extracts the deep features of the pattern based on the deep convolutional neural network, and accurately classifies and constructs The semantic label structure of ethnic patterns based on convolutional neural networks has certain advancement compared with traditional national pattern semantic labels.