[1]龚德燕,赵晨萌,孙云洲,等.基于改进RT-DETR的遥感影像林火烟雾检测[J].南京师大学报(自然科学版),2026,49(01):115-124.[doi:10.3969/j.issn.1001-4616.2026.01.012]
　Gong Deyan,Zhao Chenmeng,Sun Yunzhou,et al.Improved RT-DETR for Forest Fire and Smoke Detection in Remote Sensing Images[J].Journal of Nanjing Normal University(Natural Science Edition),2026,49(01):115-124.[doi:10.3969/j.issn.1001-4616.2026.01.012]
点击复制

基于改进RT-DETR的遥感影像林火烟雾检测()

《南京师大学报（自然科学版）》[ISSN:1001-4616/CN:32-1239/N]

卷:
49
期数:
2026年01期
页码:
115-124
栏目:
计算机科学与技术
出版日期:
2026-02-10

文章信息/Info

Title:
Improved RT-DETR for Forest Fire and Smoke Detection in Remote Sensing Images
文章编号:
1001-4616(2026)01-0115-10
作者:
龚德燕1; 2; 3; 赵晨萌1; 2; 3; 孙云洲1; 2; 3; 袁淑婷1; 2; 3; 蒋雨婷1; 2; 3; 张 卡1; 2; 3; 4
(1.南京师范大学气候系统预测与变化应对全国重点实验室,江苏 南京 210023)
(2.南京师范大学地理科学学院,江苏 南京 210023)
(3.南京师范大学虚拟地理环境教育部重点实验室,江苏 南京 210023)
(4.江苏省地理信息资源开发与利用协同创新中心,江苏 南京 210023)
Author(s):
Gong Deyan1; 2; 3; Zhao Chenmeng1; 2; 3; Sun Yunzhou1; 2; 3; Yuan Shuting1; 2; 3; Jiang Yuting1; 2; 3; Zhang Ka1; 2; 3; 4
(1.State Key Laboratory of Climate System Prediction and Risk Management,Nanjing Normal University,Nanjing 210023,China)
(2.School of Geography,Nanjing Normal University,Nanjing 210023,China)
(3.Key Laboratory of Virtual Geographic Environment,Nanjing Normal University,Ministry of Education,Nanjing 210023,China)
(4.Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application,Nanjing 210023,China)
关键词:
遥感影像; 火焰烟雾检测; 多尺度特征; 分组重排卷积; 特征调制融合; RT-DETR
Keywords:
remote sensing images; smoke and fire detection; multi-scale features; group shuffle convolution; feature modulation fusion; RT-DETR
分类号:
P237; TP751
DOI:
10.3969/j.issn.1001-4616.2026.01.012
文献标志码:
A
摘要:
针对遥感影像中森林火焰烟雾检测任务存在的多尺度特征差异显著、复杂背景干扰严重以及小目标漏检等问题,本文提出一种基于改进RT-DETR的森林火焰烟雾目标检测方法. 该方法引入特征调制融合模块,强化多尺度跨层级特征的有效融合; 设计轻量化瓶颈结构,实现空间语义特征与局部细节特征之间的信息交互; 同时,添加P2小目标检测层,增强模型对小目标火焰图像局部特征信息的关注程度. 实验结果表明,本文算法参数量降低7.40%、精确率提升1.07%、召回率提升3.58%、平均精度均值mAP50、mAP50-95分别提升3.49%、1.12%,同时,F1分数从0.799 3提升至0.824 0,能更好满足森林火焰、烟雾等复杂场景下小目标的检测定位需求.
Abstract:
To address the issues in the task of forest fire and smoke detection in remote sensing images,such as significant multi-scale feature differences,severe interference from complex backgrounds,and missed detections of small targets,this paper proposes a forest fire and smoke object detection method based on an improved RT-DETR. This method introduces a feature modulation and fusion module to strengthen the effective fusion of multi-scale cross-level features; designs a lightweight bottleneck structure to realize the information interaction between spatial semantic features and local detail features; meanwhile,adds a P2 small target detection layer to enhance the model's attention to the local feature information of small-target fire images. Experimental results show that the algorithm proposed in this paper reduces the parameter amount by 7.40%,improves the precision by 1.07% and recall by 3.58%,and increases the mean average precision(mAP)mAP50 and mAP50-95 by 3.49% and 1.12%,respectively. Meanwhile,the F1-score rises from 0.799 3 to 0.824 0,which makes it more capable of meeting the requirements for detecting and locating small targets in complex scenarios such as forest fires and smoke.

参考文献/References:

[1]魏书精,罗斯生,罗碧珍,等. 气候变化背景下森林火灾发生规律研究[J]. 林业与环境科学,2020,36(2):133-143.
[2]Son R,Kim H,Wang S Y,et al. Changes in fire weather climatology under 1.5 ℃ and 2.0 ℃ warming[J]. Environmental Research Letters,2021,16(3):034058.
[3]World Resources Institute(WRI). Global forest review[R]. Washington,DC:World Resource Institute,2024.
[4]Ullah F,Khan S I,Munawar H S,et al. UAV based spatiotemporal analysis of the 2019-2020 New South Wales bushfires[J]. Sustainability,2021,13(18):10207.
[5]Forest Survey of India. India state of forest report(ISFR)2021[R]. Dehradun:Forest Survey of India,2021.
[6]Xu X,Li F,Lin Z,et al. Holocene fire history in China:responses to climate change and human activities[J]. Science of the Total Environment,2021,753:142019.
[7]Yuan D,Chang X,Huang P Y,et al. Self-supervised deep correlation tracking[J]. IEEE Transactions on Image Processing,2020,30:976-985.
[8]覃先林,李晓彤,刘树超,等. 中国林火卫星遥感预警监测技术研究进展[J]. 遥感学报,2020,24(5):511-520.
[9]周恩泽,胡思雨,张录军,等. 电网山火灾害特征及风险预警技术[J]. 电力工程技术,2020,39(3):58-64.
[10]梁新宇,刘凯,王笑松,等. 基于视频的林火烟雾检测算法研究[J]. 林业科技,2023,48(6):45-48.
[11]朱弥雪,刘志强,张旭,等. 林火视频烟雾检测算法综述[J]. 计算机工程与应用,2022,58(14):16-26.
[12]徐海文,张贵,谭三清,等. 基于卷积神经网络的林火烟雾检测[J]. 中南林业科技大学学报,2023,43(7):23-31,64.
[13]Jin L,Yu Y,Zhou J,et al. SWVR:a lightweight deep learning algorithm for forest fire detection and recognition[J]. Forests,2024,15(1):204.
[14]马耀名,张鹏飞,谭福生. 面向复杂背景下烟雾火焰检测的改进YOLOv8s算法[J]. 计算机工程与应用,2025,61(3):121-130.
[15]付燕,杨旭,叶鸥. 基于CNN和Transformer特征融合的烟雾识别方法[J]. 计算机工程与科学,2024,46(11):2045-2052.
[16]Qian J,Lin J,Bai D,et al. Omni-dimensional dynamic convolution meets bottleneck transformer:a novel improved high accuracy forest fire smoke detection model[J]. Forests,2023,14(4):838.
[17]Wang S,Chen T,Lv X,et al. Forest fire detection based on lightweight YOLO[C]//Proceedings of the 33rd Chinese Control and Decision Conference(CCDC). Kunming,China:IEEE,2021:1560-1565.
[18]况瑞杰,李祥,刘宇,等. 基LEM-YOLO的轻量化多尺度无人机林火烟雾检测[J]. 激光与光电子学进展,2025,62(14):425-436.
[19]Li G,Li Y,Jiang S. Real-time smoke detection network based on multi-scale feature recognition and lightweight architecture design[C]//Proceedings of the 2024 IEEE International Conference on Systems,Man,and Cybernetics(SMC). Kuching,Malaysia:IEEE,2024:262-267.
[20]Zhao L,Zhi L,Zhao C,et al. Fire-YOLO:a small target object detection method for fire inspection[J]. Sustainability,2022,14(9):4930.
[21]Huang J,Zhou J,Yang H,et al. A small-target forest fire smoke detection model based on deformable transformer for end-to-end object detection[J]. Forests,2023,14(1):162.
[22]颜谨,肖满生,王瑶瑶,等. 一种改进DETR的森林火灾烟雾识别模型[J]. 计算机技术与发展,2025,35(2):24-32.
[23]Ren S,He K,Girshick R,et al. Faster R-CNN:Towards real-time object detection with region proposal networks[C]//Advances in Neural Information Processing Systems. Montreal,Canada:IEEE,2015,39(6):91-99.
[24]Redmon J,Divvala S,Girshick R,et al. You only look once:unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Las Vegas,NV,USA:IEEE,2016:779-788.
[25]Zhao Y,Lv W,Xu S,et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Seattle,WA,USA:IEEE,2024:16965-16974.
[26]Zhang Y,Zhou S,Li H. Depth information assisted collaborative mutual promotion network for single image dehazing[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR),Seattle,WA,USA:IEEE,2024:2846-2855.
[27]Li H,Li J,Wei H,et al. Slim-neck by GSConv:a lightweight-design for real-time detector architectures[J]. Journal of Real-Time Image Processing,2024,21:62.
[28]Wang G,Li H,Li P,et al. M4SFWD:a multi-faceted synthetic dataset for remote sensing forest wildfires detection[J]. Expert Systems with Applications,2024,248:123489.
[29]吴栋,范永胜,桑彬彬.基于改进RT-DETR的轻量级火焰检测算法[J]. 计算机与现代化,2025(7):106-111,118.
[30]张彼德,邱杰,娄广鑫,等. 基于CNN和Transformer的轻量化电能质量扰动识别模型[J]. 电力工程技术,2025,44(1):69-78.
[31]Varghese R M S. YOLOv8:a novel object detection algorithm with enhanced performance and robustness[C]//Proceedings of the International Conference on Advances in Data Engineering and Intelligent Computing Systems(ADICS),Chennai,India:IEEE,2024:1-6.
[32]Wang A,Chen H,Liu L,et al. YOLOv10:real-time end-to-end object detection[C]//Proceedings of the 38th Conference on Neural Information Processing Systems(NeurIPS),Vancouver,Canada:ACM,2024:3429-3456.
[33]Liu W,Anguelov D,Erhan D,et al. SSD:single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision(ECCV),Amsterdam,Netherlands:Springer,Cham,2016:21-37.
[34]倪一华,闫胜业.基于改进CenterNet的轻量级目标检测算法[J]. 计算机应用与软件,2025,42(4):135-141,149.
[35]朱雨,郝晓丽,牛保宁,等. RetinaNet目标检测算法改进[J]. 计算机工程与设计,2024,45(11):3271-3278.
[36]赵晨萌,张卡,盛业华,等. 多场景烟雾火焰目标检测数据集(S-FireData)[DS/OL]. 国家地球系统科学数据中心--长江三角洲分中心,2025. http://geodata.nnu.edu.cn.
[37]Zhao C M,Zhao L K,Zhang K,et al. Smoke and fire-you only look once:a lightweight deep learning model for video smoke and flame detection in natural scenes[J]. Fire,2025,8(3):104.

相似文献/References:

[1]丁凯孟,朱长青,罗 文,等.基于自适应PCNN与PCA的遥感影像感知哈希认证算法[J].南京师大学报(自然科学版),2019,42(02):17.[doi:10.3969/j.issn.1001-4616.2019.02.003]
　Ding Kaimeng,Zhu Changqing,Luo Wen,et al.Perceptual Hash Algorithm Based on Adaptive PCNNand PCA for Remote Sensing Image Authentication[J].Journal of Nanjing Normal University(Natural Science Edition),2019,42(01):17.[doi:10.3969/j.issn.1001-4616.2019.02.003]
[2]汪 晨,张辉辉,乐继旺,等.基于深度学习和遥感影像的松材线虫病疫松树目标检测[J].南京师大学报(自然科学版),2021,44(03):84.[doi:10.3969/j.issn.1001-4616.2021.03.013]
　Wang Chen,Zhang Huihui,Le Jiwang,et al.Object Detection to the Pine Trees Affected by Pine Wilt Diseasein Remote Sensing Images Using Deep Learning[J].Journal of Nanjing Normal University(Natural Science Edition),2021,44(01):84.[doi:10.3969/j.issn.1001-4616.2021.03.013]

备注/Memo

备注/Memo:
收稿日期:2025-10-28.
基金项目:国家自然科学基金项目(42271342).
通讯作者:张卡,博士,教授,博士生导师,研究方向:摄影测量与遥感信息提取研究. E-mail:zhangka81@126.com
更新日期/Last Update: 2026-02-10