|本期目录/Table of Contents|

[1]蔡登江.基于计算机视觉的数据机房工业仪表识别研究[J].工业仪表与自动化装置,2025,(02):99-104.[doi:10.19950/j.cnki.CN61-1121/TH.2025.02.018]
　CAI Dengjiang.Research on industrial instrument recognition in data room based on computer vision[J].Industrial Instrumentation & Automation,2025,(02):99-104.[doi:10.19950/j.cnki.CN61-1121/TH.2025.02.018]
点击复制

基于计算机视觉的数据机房工业仪表识别研究(PDF)

《工业仪表与自动化装置》[ISSN:1000-0682/CN:61-1121/TH]

卷:
期数:
2025年02期
页码:
99-104
栏目:
出版日期:
2025-04-15

文章信息/Info

Title:
Research on industrial instrument recognition in data room based on computer vision
文章编号:
1000-0682（2025）02-0099-06
作者:
蔡登江
（中国海洋石油集团有限公司信息技术中心，北京 100010）
Author(s):
CAI Dengjiang
(Information Technology Center, China National Offshore Oil Corporation, Beijing 100010, China)
关键词:
计算机视觉; YOLOv5; 仪表识别; 工业机房
Keywords:
computer vision;  YOLOv5;  instrument recognition;  industrial computer room
分类号:
TP 998
DOI:
10.19950/j.cnki.CN61-1121/TH.2025.02.018
文献标志码:
A
摘要:
数据机房作为保障数据安全的重要设施，其内部的各种仪器仪表对于监控数据机房的安全至关重要。然而，传统的仪表识别依赖人工操作，易出错且效率低下。为了解决数据机房中工业仪表在复杂场景下准确率低的问题，研究提出了一种结合渐进式注意力机制与目标检测网络的仪表识别算法。此外，研究还引入了幽灵模块以增强特征表达能力，提高模型的识别准确率。实验结果表明，在不同复杂场景下的仪表识别任务中该模型均表现出较高的准确率。例如，在光照不均或有遮挡的场景A中，准确率达到了92.15%，召回率为89.47%，F1分数为90.78%，处理时间为34.21 ms；在背景复杂或多仪表密集的场景B中，准确率提升至93.24%，召回率为90.85%，F1分数达到92.02%，处理时间为35.17 ms。因此研究所提出的基于计算机视觉的数据机房工业仪表识别模型在复杂场景下的工业仪表识别任务中取得了显著成果，显示出较高的实用价值。
Abstract:
As an important facility for ensuring data security, the various instruments and meters inside the data center are crucial for monitoring the security of the data center. However, traditional instrument recognition relies on manual operation, which is prone to errors and inefficient. In order to solve the problem of low accuracy of industrial instruments in complex scenarios in data centers, a instrument recognition algorithm combining progressive attention mechanism and object detection network is proposed. In addition, the study also introduced ghost modules to enhance feature expression ability and improve the recognition accuracy of the model. The experimental results show that the model exhibits high accuracy in instrument recognition tasks in different complex scenarios. For example, in scene A with uneven lighting and occlusion, the accuracy reached 92.15%, the recall rate was 89.47%, the F1 score was 90.78%, and the processing time was 34.21 ms; In scene B with complex background and dense instruments, the accuracy is improved to 93.24%, the recall rate is 90.85%, the F1 score reaches 92.02%, and the processing time is 35.17 ms. Therefore, the computer vision based industrial instrument recognition model proposed by the research institute has achieved significant results in industrial instrument recognition tasks in complex scenarios, demonstrating high practical value.

参考文献/References:

[1] BOKAI ZHANG, DARRICK STURGEON, ARJUN RAVI SHANKAR, et al. Surgical instrument recognition for instrument usage documentation and surgical video library indexing[J]. Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023, 11(4): 1064-1072.
[2] 陈开峰,俞伟聪,唐雁文,等.基于轻量化深度学习网络的数字仪表识别[J].科学技术与工程,2023,23(2):674-680.
[3] GEORGII, KOSTIUCHIK, LALITH, SHARAN, BENEDIKT, MAYER, et al. Surgical phase and instrument recognition: how to identify appropriate dataset splits[J]. International journal of computer assisted radiology and surgery.,2024,19(4):699-711.
[4] 孙辉,李晓峰,程远方,等.基于改进联合滤波与CNN模型的仪表识别算法[J].制造业自动化,2023,45(10):166-172.
[5] 杨典,李小燕,刘培焱,等.基于OpenCV的变电站仪表识别方法研究[J].自动化与仪表,2022,37(4):75-80.
[6] WENTING ZHAO, SHIGANG WANG, YAN ZHAO, et al. A study on attention-based fine-grained image recognition: Towards musical instrument performing hand shape assessment[J]. Engineering Applications of Artificial Intelligence: The International Journal of Intelligent Real-Time Automation,2024,129(Mar.):107512.1-107512.9.
[7] 郝琨,韩冰,李志圣,等.基于投影阈值分割和数字序列校正的高噪声数字仪表图像识别方法[J].电子科技大学学报,2023,52(5):728-738.
[8] 张森,万吉林,王慧芳,等.基于注意力机制的卷积神经网络指针式仪表图像读数识别方法[J].电力自动化设备,2022,42(4):218-224.
[9] 钱玉宝,王紫涵,邱腾煌.指针式仪表读数识别的研究现状与发展[J].电子测量技术,2024,47(8):110-119.
[10] 张琳.基于传感器的工业仪表信号采集与处理技术研究[J].工业仪表与自动化装置,2024(3):111-116.
[11] 赵伟达,陈海文,郭陆阳,等.基于YOLO-E与改进OCRNet图像分割的变电站仪表读数自适应识别方法[J].电力建设,2023,44(11):75-85.
[12] 高志国,袁牧,童旸,倪修峰,程汪刘.基于PCBAM-YOLOv5网络的变电站表计缺陷检测[J].电子设计工程,2023,31(4):77-81.
[13] PENG G., DU B., LI Z., et al. Machine vision-based, digital display instrument positioning and recognition[J]. 2022,29(2):230-243.
[14] 金爱萍,袁亮,周德勤,等.基于YOLOv5和U-net的指针式仪表读数识别方法[J].仪表技术与传感器,2022(11):29-33.
[15] 孙顺远,魏志涛.基于刻度轮廓拟合的指针式仪表自动识别方法[J].仪表技术与传感器,2022,2(8):51-57.
[16] 刘培焱,叶尔扎提·努尔德别克,孙敏,等.基于智能机器人的变电站仪表识别系统[J].自动化与仪表,2022,37(12):28-33.

相似文献/References:

[1]阎 翔,谌海云,蒋 钰,等.基于计算机视觉的车道线检测与识别[J].工业仪表与自动化装置,2020,(01):118.
　YAN Xiang,SHEN Haiyun,JIANG Yu,et al.Lane line detection and recognition based on computer vision[J].Industrial Instrumentation & Automation,2020,(02):118.
[2]康朝海,刘 杨,任伟建,等.基于Mask R-CNN的油田井场指针仪表识别方法研究[J].工业仪表与自动化装置,2024,(05):76.[doi:DOI:10.19950/j.cnki.cn61-1121/th.2024.05.015]
　KANG Chaohai,LIU Yang,REN Weijian,et al.Research on Mask R-CNN-based pointer instrument recognition method for oilfield wellsite[J].Industrial Instrumentation & Automation,2024,(02):76.[doi:DOI:10.19950/j.cnki.cn61-1121/th.2024.05.015]

备注/Memo

备注/Memo:
收稿日期：2024-08-06第一作者：蔡登江（1980—），男，湖北孝感人，本科，信息通信工程高级工程师，研究方向为计算机网络传输及交换，数据中心基础设施建设等。E-mail：15533915062@163.com
更新日期/Last Update: 1900-01-01