[1]迪力夏提·多力昆,张太红,冯向萍.LabelMe 标注核对系统的设计与实现[J].计算机技术与发展,2022,32(03):214-220.[doi:10. 3969 / j. issn. 1673-629X. 2022. 03. 036]
　Duolikun DILIXIATI,ZHANG Tai-hong,FENG Xiang-ping.Design and Implementation of LabelMe Label Checking System[J].,2022,32(03):214-220.[doi:10. 3969 / j. issn. 1673-629X. 2022. 03. 036]
点击复制

LabelMe 标注核对系统的设计与实现()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年03期
页码:
214-220
栏目:
应用前沿与综合
出版日期:
2022-03-10

文章信息/Info

Title:
Design and Implementation of LabelMe Label Checking System
文章编号:
1673-629X(2022)03-0214-07
作者:
迪力夏提·多力昆; 张太红; 冯向萍
新疆农业大学 计算机与信息工程学院,新疆 乌鲁木齐 830052
Author(s):
Duolikun DILIXIATI; ZHANG Tai-hong; FENG Xiang-ping
School of Computer and Information Engineering,Xinjiang Agricultural University,Urumqi 830052,China
关键词:
核对标签; Django; OpenCV; LabelMe; SQLite
Keywords:
label checking; Django; OpenCV; LabelMe; SQLite
分类号:
TP311. 52
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 03. 036
摘要:
随着计算机技术的发展,深度学习领域对图像数据集的要求越来越高。 当无法构建大规模图像数据集时,一个高质量的数据集往往能够提高模型训练的质量和预测的准确率。 在对图像数据集进行标注时,往往需要多次核对数据集中的标签名称是否正确,且标签名称与标注物是否一一对应,并需要花费大量时间对图像数据集进行标注核对,费时费力。因此,该文基于 Django 框架设计并开发了 LabelMe 标注核对系统。 结合 OpenCV 图像处理技术,采用 Django 网页开发框架、Bootstrap 响应式网页开发技术、Python 语言和 MTV 的设计模式实现了项目管理、原图记录、导入 JSON、统一标签名、批量生成标注图、核对标注图、导出 JSON 等功能。 研究表明该系统能够简化对图像数据集的核对工作,提高核对工作的效率,有助于提升图像数据集质量。
Abstract:
With the development of computer technology,the field of deep learning has higher requirements for image data sets. When large-scale image data sets cannot be constructed,a high-quality data set can often improve the quality of model training and the accuracy of prediction. When labeling the image data set,it is often necessary to check whether the label name in the data set is correct,and whether the label name corresponds to the labeled object one by one,and it takes a lot of time to label and check the image data set,which is time-consuming and laborious. Therefore,we design and develop a LabelMe annotation checking system based on Django framework. Combined with OpenCV image processing technology,using Django web development framework,Bootstrap responsive web development technology,Python language and MTV design pattern,the functions of project management,original drawing recording,importing JSON, unifying tag name,batch generating mask drawing,checking mask drawing, exporting JSON are realized. Research shows that the system can simplify the checking work of image data set,improve the efficiency of checking work,and help to improve the quality of image data set.

相似文献/References:

[1]杨君,陈春玲,余瀚.基于 Scrapy 技术的数据采集系统的设计与实现[J].计算机技术与发展,2018,28(10):177.[doi:10.3969/ j. issn.1673-629X.2018.10.037]
　YANG Jun,CHEN Chun-ling,YU Han.Design and Implementation of Data Acquisition System Based on Scrapy Technology[J].,2018,28(03):177.[doi:10.3969/ j. issn.1673-629X.2018.10.037]
[2]张翠丽,孟小艳,杨 抒.基于 Django 框架的管理系统的设计与开发[J].计算机技术与发展,2019,29(10):63.[doi:10. 3969 / j. issn. 1673-629X. 2019. 10. 014]
　ZHANG Cui-li,MENG Xiao-yan,YANG Shu.Design and Development of Django Framework-based Management System[J].,2019,29(03):63.[doi:10. 3969 / j. issn. 1673-629X. 2019. 10. 014]

更新日期/Last Update: 2022-03-10