面向电力场景图像的动态自适应增量目标检测DA-IOD算法

陈健华; 袁浩亮; 陈海淋; 许方园

doi:10.12052/gdutxb.250089

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

面向电力场景图像的动态自适应增量目标检测DA-IOD算法

电气工程 | 更新时间：2026-05-27

- 面向电力场景图像的动态自适应增量目标检测DA-IOD算法
- Dynamic Adaptive Enhancement for Power Scene Images DA-IOD Algorithm for Quantitative Target Detection
- 广东工业大学学报 2025年42卷第5期页码：86-95
- 作者机构：
  
  广东工业大学自动化学院,广东,广州,510006
- 作者简介：
  
  陈健华(1997–)，男，硕士研究生，主要研究方向为深度学习，E-mail：3064421739@qq.com
  袁浩亮(1986–)，男，副教授，博士，硕士生导师，主要研究方向为人工智能，E-mail：haoliangyuan@gdut.edu.cn
- 基金信息：
  
  广东省自然科学基金资助面上项目(2023A1515011041);南方电网公司科技项目(030100KK52230001)
- DOI：10.12052/gdutxb.250089
  中图分类号： TM08
- 纸质出版：2025
- 稿件说明：
移动端阅览
陈健华, 袁浩亮, 陈海淋, 等. 面向电力场景图像的动态自适应增量目标检测DA-IOD算法[J]. 广东工业大学学报, 2025, 42(5): 86–95.

Chen Jianhua, Yuan Haoliang, Chen Hailin, et al. Dynamic Adaptive Enhancement for Power Scene Images DA-IOD Algorithm for Quantitative Target Detection[J]. 2025, 42(5): 86-95.
陈健华, 袁浩亮, 陈海淋, 等. 面向电力场景图像的动态自适应增量目标检测DA-IOD算法[J]. 广东工业大学学报, 2025, 42(5): 86–95. DOI： 10.12052/gdutxb.250089

Chen Jianhua, Yuan Haoliang, Chen Hailin, et al. Dynamic Adaptive Enhancement for Power Scene Images DA-IOD Algorithm for Quantitative Target Detection[J]. 2025, 42(5): 86-95. DOI： 10.12052/gdutxb.250089.

摘要

针对传统的目标检测技术在电力场景中对新目标存在适应性不足、无法增量检测的问题，本文提出面向电力场景图像的动态自适应增量目标检测算法(Dynamic Adaptive Incremental Object Detection

DA-IOD)。首先，采用任务对齐的自适应特征解耦模块(Task-Aligned Adaptive Feature Decoupling

TAFD)增强特征表示能力；其次，引入轻量和高效的动态上采样器(Ultra-lightweight and Effective Dynamic Upsampler

DySample)降低计算负担和延迟；最后，采用完全交并比(Complete Intersection over Union

CIoU) 回归损失函数对基线增量Efficient-IOD算法的损失函数进行修正，进一步提高了算法的检测精度。在广东电网智慧现场作业数据集的3+3单步增量场景中，本文方法与基线算法相比，平均精度均值(mean Average Precision

mAP) 提高了2.4个百分点。

Abstract

Aiming to address the problems of insufficient adaptability to new targets and inability to perform incremental detection in power scenarios using traditional target detection technologies

this paper proposes a Dynamic Adaptive Incremental Object Detection algorithm for power scene images (DA-IOD). Firstly

a Task-Aligned Adaptive Feature Decoupling (TAFD) module is adopted to enhance the feature representation capability. Secondly

an Ultra-lightweight and Effective Dynamic Upsampler (DySample) is introduced to reduce computational burden and latency. Finally

a Complete Intersection over Union (CIoU) regression loss function is designed to modify the loss function of the baseline incremental Efficient-IOD algorithm

which further improves the detection accuracy of the proposed algorithm. In the 3+3 single-step incremental scenario of the Guangdong Power Grid Smart On-site Operation Dataset

out proposed method achieves an improvement of 2.4 percentage points in the mean Average Precision (mAP) when compared with the baseline algorithm.

关键词

Keywords

references

XIONG W H, MAO X H, LI J L, et al . Influence of small resistance grounding mode on personal safety and research of intelligent resistance grounding mode[J ] . Power System Protection and Control, 2019, 47(14): 166-172.

MA F Q, WANG B, DONG X Z, et al . Safety image interpretation of power industry: basic concepts and technical framework[J ] . Proceedings of the CSEE, 2022, 42(2): 458-475.

YAN Y F, CHEN X, JIN H Y, et al . Research status and development of computer-vision-based power workers' behavior analysis[J ] . High Voltage Engineering, 2024, 50(5): 1842-1854.

HE M, QIN L, ZHAO F, et al . Intelligent detection algorithm of security risk management and control for power system on-site operation[J ] . High Voltage Technology, 2023, 49(6): 2442-2457.

CAO J, GUO Z B, PAN L Z, et al . Detection of safety belt wearing in aerial work[J ] . Journal of Hunan University of Science Technology (Natural Science Edition) , 2022, 37(1): 92-99.

HU C L, PEI S T, LIU Y P, et al . Real-time edge end defect detection method for transmission line based on LEE-YOLOv7[J ] . High Voltage Engineering, 2024, 50(11): 5047-5057.

GUO P T, LI L, WANG Y, et al . An improved DETR-based method for recognizing safety behaviors at electric power work sites[J ] . Electric Power Information and Communication Technology, 2024, 22(8): 63-70.

JIANG X J, WANG R T, MA Y H. Insulator defect detection based on lightweight improved RT-DETR edge deployment algorithm[J]. Transactions of China Electrotechnical Society, 2024, 40(3): 842-854.

GUO J, HAN K, WANG Y, et al . Distilling object detectors via decoupled features[C ] //Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 2154-2164.

JOSEPH K J, KHAN S, KHAN F S, et al . Towards open world object detection[C ] //Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 5830-5840.

DE L M, ALJUNDI R, MASANA M, et al . A continual learning survey: defying forgetting in classification tasks[J ] . IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 44(7): 3366-3385.

SHMELKOV K, SCHMID C, ALAHARI K. Incremental learning of object detectors without catastrophic forgetting[C]//Proceedings of the IEEE International Conference on Computer Vision. Venice: IEEE, 2017: 3400-3409.

FENG T, WANG M, YUAN H. Overcoming catastrophic forgetting in incremental object detection via elastic response distillation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022: 9427-9436.

FENG C, ZHONG Y, GAO Y, et al . Tood: task-aligned one-stage object detection[C ] //2021 IEEE/CVF International Conference on Computer Vision (ICCV) . Montreal: IEEE Computer Society, 2021: 3490-3499.

CHEN Z, YANG C, LI Q, et al . Disentangle your dense object detector[C ] //Proceedings of the 29th ACM International Conference on Multimedia. New York: ACM, 2021: 4939-4948.

BAI L, SONG H, FENG T, et al . Revisiting class-incremental object detection: an efficient approach via intrinsic characteristics alignment and task decoupling[J ] . Expert Systems with Applications, 2024, 257: 125057.

LIU W, LU H, FU H, et al . Learning to upsample by learning to sample[C ] //Proceedings of the IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2023: 6027-6037.

KIM K, LEE H S. Probabilistic anchor assignment with iou prediction for object detection[C]//Computer Vision–ECCV 2020: 16th European Conference. Glasgow: Springer, 2020: 355-371.

WANG J, CHEN K, XU R, et al . Carafe: content-aware reassembly of features[C ] //Proceedings of the IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2019: 3007-3016.

LU H, LIU W, FU H, et al . FADE: fusing the assets of decoder and encoder for task-agnostic upsampling[C ] //European Conference on Computer Vision. Cham: Springer Nature, 2022: 231-247.

LU H, LIU W, YE Z, et al . SAPA: similarity-aware point affiliation for feature upsampling[J ] . Advances in Neural Information Processing Systems, 2022, 35: 20889-20901.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于域不变判别特征学习的深子域自适应方法

基于主动学习和视觉状态空间模型的热点检测器

SCA-YOLO: 基于可变形卷积与上下文感知注意力的实时目标检测

面向灾后救援的多无人机通感算一体化资源分配优化方法

储能变流器电能质量治理策略研究