基于深度卷积神经网络的织物花型分类

Fabric pattern classification based on depth convolution neural network

针对织物纹理花型人工视觉分类效率低的问题,提出一种基于深度卷积神经网络(CNN)的条状、格子和波点纹理织物的识别分类方法.首先,建立由条状花型、格子花型和波点花型织物组成的图像样本集和标签数据集.然后,分别建立了基于GoogLeNet和AlexNet两种卷积神经网络的织物花型分类模型.最后,通过模型评价指标选择最优的训练迭代期.实验结果表明,利用深度卷积神经网络分类织物花型是可行有效的.

To solve low efficiency problems of fabric pattern artificial visual classification,a method to classify fabric strip,lattice and wave point patterns based on depth convolution neural network（CNN）is presented.First,image samples and tag data are constructed,which include fabric strip,lattice and wave point patterns.Then,fabric pattern classification models are developed based on GoogLeNet and AlexNet CNN,respectively.Finally,optimal training epoch periods are selected by the model evaluation method.Experimental results indicate that it is feasible and effective to classify fabric patterns by deep convolution neural networks.