Page 105 - 无损检测2024年第十一期
P. 105
试验研究
试验研究
DOI:10.11973/wsjc240128
基于 YOLO 网络模型的多类别标签缺陷检测
孟令波 ,杨程午 ,李亚彬 2
1
2
(1. 兖煤蓝天清洁能源有限公司,邹城 273500;2. 天津中科智能识别有限公司,天津 300457)
摘 要:由于热转印滚筒温度的变化,热转印标签在转印的过程中会出现热转印标签褶皱、
脱模不完全等问题。针对某些较大标签存在缺陷种类较多及未知缺陷的问题,提出了一种基于
YOLO网络模型的多类别标签缺陷检测方法,将自适应匹配缺陷检测方法与改进的YOLO网络模
型相结合,增加注意力机制模块以提高小目标缺陷的检测能力。在处理过程中,首先,对不同区域
内的标签进行快速定位及预处理;然后,针对不同区域使用不同的检测方法进行检测;最后,将不
同区域结果融合,判断检测结果。试验结果表明,基于YOLO网络模型的多类别标签缺陷的检测方
法能够有效进行热转印标签的缺陷检测,检测准确率达98%,能够满足实际的生产要求。
关键词:多类别;热转印标签;传统图像处理;YOLO网络模型;注意力机制
中图分类号:TB92;TG115.28 文献标志码:A 文章编号:1000-6656(2024)11-0067-06
Detection of multi-class label defects based on YOLO network model
MENG Lingbo , YANG Chengwu , LI Yabin 2
1
2
(1. Yanmei Clean Energy Co., Ltd.,Zoucheng 273500,China;
2. Tianjin Academy for Intelligent Recognition Co. Ltd.,Tianjin 300457,China)
Abstract: During the use of heat transfer labels, due to changes in the temperature of the heat transfer drum, problems
such as wrinkles and incomplete demolding of the heat transfer labels may occur during the transfer process. In the problem
of defect detection in heat transfer printing, some larger labels have many types of defects and some unknown defects.
Therefore, a multi-class label defect detection method based on YOLO network model was proposed. The self-adaptive
matching defect detection method was combined with the improved YOLO network model for detection, and the YOLO
network model was added with an attention mechanism to improve the model's ability to detect small target defects. During the
processing, first, the labels in different regions were quickly located for preprocessing; Then, different regions were detected
using different detection methods; Finally, the results of different regions were fused to determine the detection results. The
results showed that the multi-class label defect detection method based on the YOLO network model can effectively detect
defects in heat transfer labels, with an overall detection accuracy of 98%, which can meet practical production requirements.
Key words: multi-category; thermal transfer label; traditional image processing; YOLO network model; attention
mechanism
表面质量已成为产品的重要竞争指标之一,对 产品标签在转印时会出现褶皱、划伤、脱模不完全等
产品表面的质量控制在工业生产中的作用日趋显 缺陷。表面缺陷检测已经成为工业生产过程中不可
著 。热转印标签作为产品最外层的包装,其转印 或缺的组成部分,而目前产品标签类型较多,检测方
[1]
质量直接影响到产品的质量。在实际的生产过程中, 式还是以人工检测为主,人工检测不仅检测效率低、
受到工艺流程、生产设备和工厂环境等因素的影响, 误检率高,而且成本较高,易受人工经验和主观因素
影响。因此,将机器视觉检测技术应用到标签缺陷
检测中已成为重要的研究方向之一。
收稿日期:2024-03-26
目前,众多学者提出了多种机器视觉检测缺陷
作者简介:孟令波,男,高级工程师,主要研究方向为自动化控制
通信作者:李亚彬,1286873614@qq.com 的算法,主要包括基于传统图像处理的检测方法与
67
2024 年 第 46 卷 第 11 期
无损检测

