期刊论文

中文

电脑与信息技术

1005-1228

2023

31

3

24-28

10.3969/j.issn.1005-1228.2023.03.007

教育部人文社会科学研究规划基金项目（项目编号：22YJA760038）；

本校为第一机构

针对依靠人的肉眼对纹样图像进行分类准确率不高以及工作效率低下的问题,本文提出一种基于改进的Res Net18网络模型中进行迁移学习的方式来对纹样图像按国家进行分类实验,从而提升该模型的训练效果。由于数据集纹样图像数量较少...展开更多 针对依靠人的肉眼对纹样图像进行分类准确率不高以及工作效率低下的问题,本文提出一种基于改进的Res Net18网络模型中进行迁移学习的方式来对纹样图像按国家进行分类实验,从而提升该模型的训练效果。由于数据集纹样图像数量较少,因此引入了数据增强技术对数据集纹样图像加以扩充,借此来进一步增强Res Net18网络模型的泛化能力。最终的实验结果表明在基于改进...

Aiming at the problem that the accuracy of classifying pattern images by human eye is not high and the work efficiency is low,this paper proposes a method of transfer learning based on the improved ResNet18 network model to classify pattern images by country,so as to im...MORE Aiming at the problem that the accuracy of classifying pattern images by human eye is not high and the work efficiency is low,this paper proposes a method of transfer learning based on the improved ResNet18 network model to classify pattern images by country,so as to improve the training effect of the model.Due to the sm...