[1]吴攀超,范文博,王婷婷. 改进RT-DETR 的油田人员异常行为检测[J].机械与电子,2026,44(03):32-40.
　WU Panchao,FAN Wenbo,WANG Tingting. Improved RT-DETR for Abnormal Behavior Detection of Personnel in Oilfields[J].Machinery & Electronics,2026,44(03):32-40.
点击复制

 改进RT-DETR 的油田人员异常行为检测()
分享到：

《机械与电子》[ISSN:1001-2257/CN:52-1052/TH]

卷:
44
期数:
2026年03期
页码:
32-40
栏目:
智能检测
出版日期:
2026-03-25

文章信息/Info

Title:
 Improved RT-DETR for Abnormal Behavior Detection of Personnel in Oilfields
文章编号:
1001-2257(2026)03-0032-09
作者:
 吴攀超; 范文博; 王婷婷
 (东北石油大学电气信息工程学院,黑龙江 大庆 163319)
Author(s):
 WU Panchao; FAN Wenbo; WANG Tingting
 (School of Electrical Information Engineering,Northeast Petroleum University,Daqing 163319,China)
关键词:
RT-DETR; 小目标检测; 特征提取; 异常行为; 深度学习
Keywords:
 ; RT-DETR; small target detection; feature extraction; abnormal behavior; deep learning
分类号:
TP391.4
文献标志码:
A
摘要:
 为解决油田作业现场因操作不规范导致的安全事故频发问题,提出了一种高效的面向油田复杂场景下异常行为检测的RT DETR改进算法(HCH DETR)。首先,设计一种新型主干网络,结合双分支高频细节增强模块(HFERB)和CSP结构,提高模型高频细节特征提取能力,并有效减少模型计算量;其次,针对油田监控中目标尺度变化大、背景复杂等问题,提出一种基于上下文引导的空间特征重构特征金字塔网络(CGFRPN),通过矩形自校准注意力(RCA)增强多尺度特征融合,提高模型对多尺度目标的检测精度,并增强了其在复杂场景下的鲁棒性;最后,引入Haar小波下采样模块(HWD)优化传统下采样,提高模型对小目标的检测能力。在自建油田数据集上进行模型验证:mAP@0.5和mAP@0.5:0.95分别达到85.4%和55.1%,较原始RT DETR模型提升3.2百分点和2.4百分点,同时计算量减少7.1×109,参数量降低6.5×106;消融实验验证了各改进模块的有效性,泛化实验表明模型在VisDrone数据集上精度亦有提升。
Abstract:
 To address the frequent safety accidents caused by non standard operations at oilfield worksites,this study proposes an efficient improved RT DETR algorithm,named HCH DETR,for abnormal behavior detection in complex oilfield scenarios.Firstly,a novel backbone network is designed by integrating the dual branch High Frequency Enhancement Residual Block (HFERB) and the Cross Stage Partial (CSP) structure.This design enhances the model’s ability to extract high frequency detailed features while effectively reducing the model’s computational complexity.Secondly,to tackle the challenges such as large variations in target scale and complex backgrounds in oilfield monitoring,a Context Guided Feature Reconstruction Feature Pyramid Network (CGFRPN) is proposed.It employs Rectangular Self Calibration Attention (RCA) to strengthen multi scale feature fusion,improving the detection accuracy for multi scale targets and bolstering robustness in complex scenes.Finally,the Haar Wavelet Downsampling (HWD) module is introduced to optimize traditional downsampling,thereby improving the model’s capability for small target detection.Experimental validation on a self constructed oilfield dataset shows that the proposed model achieves 85.4% mAP@0.5 and 55.1% mAP@0.5:0.95,representing improvements
of 3.2 percentage points and 2.4 percentage points respectively over the original RT DETR model.Meanwhile,the computational complexity is reduced by 7.1×109,and the number of parameters is decreased by 6.5×106.Ablation experiments verify the effectiveness of each improved module,and generalization experiments demonstrate that the model also achieves improved accuracy on the VisDrone dataset.

参考文献/References:

 [1] 张千,梁鸿,童彦淇,等.基于深度学习的油田在线视频目标检测[J].计算机与数字工程,2024,52(3):864 872.
[2] 杨嘉如,秦忆南,李天旭,等.基于改进YOLOv8s的井下安全帽检测算法[J].煤矿安全,2025,56(5):221 228.
[3] 王贞,邱杭,吴斌,等.基于CCG YOLOv8的施工场景下安全帽佩戴检测[J].武汉理工大学学报,2024,46(6):73 80.
[4] 赖惠木,李俊平,江朝金,等.基于改进YOLOv10的监控区域人员工装穿着检测方法[J].电脑编程技巧与维护,2025(6):124 126.
[5] ZHAO Y,LV W,XU S L,et al.Detrs beat yolos on real time object detection[C]∥2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition,2024:16965 16974.
[6] 武慧荣,张敬宜,郭春敏.基于深度学习的公交客流检测算法[J].控制与决策,2025,40(6):1827 1837.
[7] 王东城,段伯伟,邢佳文,等.基于改进RT DETR 的铜带表面缺陷轻量化检测方法[J].中国有色金属学报,2025,35(10):3527 3538.
[8] 李鹏,余珺泽,于涛,等.基于轻量化RT DETR 的PCB缺陷检测算法[J].计算机工程与设计,2025,46(9):2714 2721.
[9] 耿嘉雯,严云洋,朱妍,等.基于改进RT DETR 的轻量化交通标志检测方法[J].南京师范大学学报(工程技术版),2025,25(2):69 78.
[10] 娄文,郭杜杜,张杰等.基于YOLOv7的驾驶人使用手机与抽烟行为识别方法[J].电子测量技术,2023,46(21):123 131.
[11] LI A,ZHANG L,LIU Y,et al.Feature modulation transformer:cross refinement of global representation via high frequency prior for image super resolution[C]∥2023 IEEE/CVF International Conference on Computer Vision (ICCV),2023:12480 12490.
[12] WANG C Y,LIAO H Y M,WU Y H,et al.CSPNet:a new backbone that can enhance learning capability of CNN[C]∥2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Worksh (CVPRW),2020:1571 1580.
[13] ELFWING S,UCHIBE E,DOYA K.Sigmoid weighted linear units for neural network function approximation in reinforcement learning[J].Neural networks,2018,107:3 11.
[14] XU G P,LIAO W T,ZHANG X,et al.Haar wavelet downsampling:a simple but effective downsampling module for semantic segmentation[J].Pattern recognition,2023,143:109819.
[15] CHEN J R,KAO S H,HE H,et al.Run,don’t walk:chasing higher FLOPS for faster neural networks [C]∥2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition,2023:12021 12031.
[16] LIU X,PENG H,ZHENG N,et al.EfficientVit:memory efficient vision transformer with cascaded group attention[C]∥Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2023:14420 14430.
[17] WANG A,CHEN H,LlU L H,et al.YOLOvl0:Reatime end to end object detection[J].Computer vision and pattern recognition,2024:14458.
[18] 秦建华,陈振伦,万保雄,等.基于RT DETR的复杂果园环境下青橘检测方法[J].电子测量技术,2025,48(11):175 186.
[19] 解浩龙,张孝龙,魏培旭,等.基于RT DETR的轻量化交通标志检测[J].应用光学,2025,46(2):300 308.
[20] 孙海青,杨传颖,敖乐根.基于改进RT DETR 的公路路面交通标识检测算法[J].电子测量技术,2025,48(8):187 195.

相似文献/References:

[1]吴攀超,王佳硕,王婷婷.基于 FACV-YOLOv10 模型的油田作业人员异常行为检测[J].机械与电子,2025,(10):18.
　WU Panchao,WANG Jiashuo,WANG Tingting.Abnormal Behavior Detection of Oilfield Operators Based on FACV-YOLOv10 Model[J].Machinery & Electronics,2025,(03):18.
[2]潘学华,梁炜皓,李金瑾,等.面向绝缘子微小缺陷的 YOLOv8s 的改进算法研究[J].机械与电子,2025,(12):10.
　PAN Xuehua,LIANG Weihao,LI Jinjin,et al.Research on Improved YOLOv8s Algorithm for Micro-defect Detection in Insulators[J].Machinery & Electronics,2025,(03):10.

备注/Memo

备注/Memo:
 收稿日期:2025-11-07
基金项目:国家自然科学基金资助项目(52474036)
作者简介:吴攀超 (1981-),男,黑龙江大庆人,博士,讲师,研究方向为图像处理与计算机视觉;范文博 (2002-),男,黑龙江哈尔滨人,硕士研究生,研究方向为模式识别应用与图像处理,通信作者,E-mail:978605215@qq.com;王婷婷 (1982-),女,黑龙江大庆人,博士,教授,博士研究生导师,研究方向为油气田岩石脆性分析。
更新日期/Last Update: 2026-04-29