首页 期刊 论文 学科刊群 资讯 加盟 关于
EN
image
Article(id=1319955325730538082, tenantId=1146029695717560320, journalId=1278651732997652489, issueId=1319955241370502145, articleNumber=null, orderNo=null, doi=10.12086/oee.2026.250152, pmid=null, cstr=32245.14.oee.2026.250152, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1748188800000, receivedDateStr=2025-05-26, revisedDate=1756915200000, revisedDateStr=2025-09-04, acceptedDate=1757260800000, acceptedDateStr=2025-09-08, onlineDate=1792631473238, onlineDateStr=2026-10-22, pubDate=1769270400000, pubDateStr=2026-01-25, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1792631473238, onlineIssueDateStr=2026-10-22, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1792631473238, creator=13701087609, updateTime=1792631473238, updator=13701087609, issue=Issue{id=1319955241370502145, tenantId=1146029695717560320, journalId=1278651732997652489, year='2026', volume='53', issue='1', pageStart='250064', pageEnd='1', issueExtLink='null', onlineDate='null', pubDate='1769270400000', pubDateStr='2026-01-25', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1792631453125, creator='13701087609', updateTime=1792631453125, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext=null, issueFiles=null, downloadFileDto=null}, startPage=250152, endPage=1, ext={EN=ArticleExt(id=1319955325940253283, articleId=1319955325730538082, tenantId=1146029695717560320, journalId=1278651732997652489, language=EN, title=Lightweight RT-DETR and multi-scale featurefusion algorithm for flame and smoke detection, columnId=1279511634116841602, journalTitle=Opto-Electronic Engineering, columnName=Article, runingTitle=null, highlight=null, articleAbstract=
Objective

With the advancement of modernization and the development of urbanization, fire safety issues in various public venues, commercial buildings, and residential homes have become increasingly prominent. Fire incidents not only directly threaten lives and property but also generate smoke that often causes irreversible pollution and damage to the ecological environment. Traditional fire detection methods primarily rely on physical devices such as sensors, yet these approaches have limitations—including insufficient sensitivity, frequent false alarms and missed detections, susceptibility to environmental interference, and difficulty in achieving early warning. Consequently, there is an urgent need for more efficient fire detection algorithms. In recent years, rapid advancements in artificial intelligence have driven breakthroughs in deep learning within computer vision, particularly demonstrating exceptional performance in object detection. This has opened new avenues and research perspectives for fire detection studies.

Methods

Addressing current limitations in flame and smoke detection technologies—such as insufficient accuracy in complex scenarios, susceptibility to false negatives and positives, and high computational costs—this research innovatively applies the cutting-edge RT-DETR model from computer vision to fire detection tasks. The aim is to thoroughly explore this model's capability to capture and recognize flame and smoke features in complex, dynamic environments, striving to develop a fire detection algorithm with high detection accuracy, low false alarm rates, and the ability to meet real-time early warning requirements. RT-DETR is an efficient real-time end-to-end detector based on the Transformer architecture. It organically combines the local feature extraction advantages of convolutional neural networks (CNNs) with the global modeling capabilities of Transformers through the introduction of a hybrid encoder. By employing an innovative IoU-aware query selection mechanism to precisely filter initial target queries, it achieves a good balance between detection accuracy and inference speed while significantly reducing computational latency. However, when handling complex, multi-scale, and morphologically variable targets like flames and smoke, this model still suffers from insufficient capture of local information and limited extraction of fine-grained features. This often leads to false negatives and false positives for small targets, limiting its application in scenarios demanding high accuracy. To address this, this paper proposes a lightweight RT-DETR-based flame and smoke detection algorithm with multi-scale feature fusion to enhance the model's perception and discrimination capabilities for such targets. We first adopt RT-DETR-r18 as the baseline model, introducing the lightweight EfficientNetV1 architecture into the backbone network. Through evaluation and experimentation with mainstream lightweight networks like GhostNet, RepViT, and EfficientViT, results demonstrate that EfficientNetV1 achieves the optimal balance between accuracy and efficiency via its composite scaling strategy. It significantly reduces parameter count and computational burden while maintaining robust feature extraction capabilities, providing the optimal feature foundation for subsequent detection tasks. Second, the VoVGSCSP module is designed at the model's neck. This architecture integrates the efficient feature reuse of VoVNet with the gradient splitting concept of CSPNet, effectively enhancing the model's expression capabilities for key features like flame-smoke texture and color while suppressing feature redundancy. Finally, addressing the core challenge of extreme scale variations in flame-smoke targets, this paper introduces the more efficient multi-scale feature fusion mechanism bidirectional feature pyramid network (BiFPN)-E. During the weighted fusion process in the BiFPN, it innovatively employs the ELU activation function for weight normalization. The non-zero saturation property of the ELU function when handling negative inputs generates smoother and more discriminative fusion weights, thereby strengthening the interaction and complementarity between feature maps at different scales. This significantly improves the model's perception and discrimination accuracy in complex scenarios such as large-scale flames, faint smoke, and occluded objects.

Results and Discussions

The final experimental results fully validate the comprehensive effectiveness of the proposed method. On the NEWFire dataset, the improved model achieved mAP50 and mAP50-95 scores of 84.5% and 57.0%, respectively, representing improvements of 1.1% and 1.7% over the baseline RT-DETR model. On the self-built E-fire dataset, the accuracy metrics mAP50 and mAP50-95 improved by 1.8% and 2.2%, respectively. More importantly, while achieving these accuracy gains, the model demonstrated significant lightweighting: model parameters were reduced by 32%, and floating-point operations were drastically reduced by 65%. In-depth ablation experiments further confirm that the BiFPN-E architecture is the most critical contributor to accuracy gains, while the EfficientNetV1 backbone serves as the core for achieving efficient model compression. Visual analysis also reveals that the improved model's feature heatmap responses are more concentrated on actual flame-smoke regions, demonstrating stronger suppression of fire-like distractors and complex backgrounds, higher recognition accuracy, and effectively reducing both false negatives and false positives.

Conclusions

Through systematic enhancements to the RT-DETR model, this research achieves a synergistic breakthrough in accuracy, efficiency, and feature discriminative power for fire detection. The EfficientNetV1 backbone enables efficient model compression, while the introduced VoVGSCSP module enhances the model's ability to capture irregular fire shapes and dynamic changes. The BiFPN-E architecture significantly improves detection accuracy through optimized multi-scale feature fusion. The synergistic effect of these three components enables the model to maintain high accuracy while reducing parameters by 32% and computational load by 65%. This breakthrough overcomes computational resource constraints while enhancing performance, providing a practical solution for achieving high-precision real-time fire alerts on resource-constrained edge devices.

, authors=Guangling Sun1, 2, Xueting Wang1, authorsList=Guangling Sun, Xueting Wang, authorCompany=null, correspAuthors=Guangling Sun, authorNote=null, correspAuthorsNote=
, copyrightStatement=Copyright © 2026 Opto-Electronic Engineering. All rights reserved., copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, fund=null), CN=ArticleExt(id=1319955331342516882, articleId=1319955325730538082, tenantId=1146029695717560320, journalId=1278651732997652489, language=CN, title=轻量化RT-DETR与多尺度特征融合的火焰烟雾检测算法, columnId=1319955120360636489, journalTitle=光电工程, columnName=研究论文, runingTitle=null, highlight=null, articleAbstract=

针对火焰烟雾检测复杂性高，精度偏低，并且经常会产生漏检误检等问题，提出了轻量化RT-DETR与多尺度特征融合的火焰烟雾检测算法。该方法首先以RT-DETR-r18为基础架构，在主干网络中引入了一种轻量化网络结构EfficientNetV1，降低了模型参数量；接着引入了VoVGSCSP模块，有效减少了特征冗余，并且增强了模型特征提取能力；最后结合双向特征融合金字塔网络BiFPN和ELU激活函数的优点，在BiFPN权重归一化过程中引入ELU激活函数，设计了BiFPN-E结构，提高了模型的跨尺度交互能力，有效提升了模型的检测精度。根据实验结果显示，改进的RT-DETR模型相较于原模型在公开数据集NEWFire和自建数据集E-fire上mAP50分别提高了1.1%和1.8%，mAP50-95分别提高了1.7%和2.2%，参数量下降32%，浮点计算量下降65%。说明改进模型在提高检测精度的同时保持了轻量化要求，满足了在火焰烟雾检测时的精度和效率的实际需求。

, authors=孙光灵1, 2, *, *, 王雪婷1, authorsList=孙光灵, 王雪婷, authorCompany=null, correspAuthors=孙光灵, authorNote=

作者贡献

孙光灵：指导论文选题及论文的修改；王雪婷：研究资料和背景分析，实验实施与分析，论文写作。

孙光灵(1976-)，男，硕士，教授，主要研究方向为机器学习、图像处理。E-mail：

王雪婷(1999-)，女，硕士研究生，主要研究方向为图像处理、目标检测。E-mail：

, correspAuthorsNote=
孙光灵，
, copyrightStatement=版权所有©《光电工程》编辑部 2026, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=rzq3Nf1wLgTpB+Cb+TO/zQ==, magXml=ulyF/KshH4yaZr7BJH/jTQ==, pdfUrl=null, pdf=uhwh/ThvSh2e5481xT7hVA==, pdfFileSize=6455189, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=QVP+Olnbutzv6EWVsY87Zw==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=c6Q0CRwxhcnSya6HA+1bLQ==, mapNumber=null, fund=null)}, authors=[Author(id=1319955331824861850, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=sunguangling@163.com, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1319955331912942237, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955331824861850, language=EN, stringName=Guangling Sun, firstName=Guangling, middleName=null, lastName=Sun, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, address=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China
2Anhui Interational Joint Research Center for Intelligent Perception and High-dimensional Modeling of Ancient Buildings, Anhui Jianzhu University, Hefei, Anhui 230601, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955331980051102, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955331824861850, language=CN, stringName=孙光灵, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, *, *, address=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601
2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601, bio={"img":"RelyF6Q49UAosq9MiMEGiw==","content":"

孙光灵(1976-)，男，硕士，教授，主要研究方向为机器学习、图像处理。E-mail：

"}, bioImg=RelyF6Q49UAosq9MiMEGiw==, bioContent=

孙光灵(1976-)，男，硕士，教授，主要研究方向为机器学习、图像处理。E-mail：

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955331564814995, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=1, ext=[AuthorCompanyExt(id=1319955331573203604, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331581592213, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601)]), AuthorCompany(id=1319955331699032726, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=2, ext=[AuthorCompanyExt(id=1319955331707421335, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331699032726, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Anhui Interational Joint Research Center for Intelligent Perception and High-dimensional Modeling of Ancient Buildings, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331711615640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331699032726, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601)])]), Author(id=1319955332097491616, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=18755443812@163.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955332227515042, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955332097491616, language=EN, stringName=Xueting Wang, firstName=Xueting, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955332332372643, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955332097491616, language=CN, stringName=王雪婷, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601, bio={"img":"H5CiDKmoIagGXm4zWwY6tA==","content":"

王雪婷(1999-)，女，硕士研究生，主要研究方向为图像处理、目标检测。E-mail：

"}, bioImg=H5CiDKmoIagGXm4zWwY6tA==, bioContent=

王雪婷(1999-)，女，硕士研究生，主要研究方向为图像处理、目标检测。E-mail：

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955331564814995, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=1, ext=[AuthorCompanyExt(id=1319955331573203604, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331581592213, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601)])])], keywords=[Keyword(id=1319955332466590372, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, orderNo=1, keyword=EfficientNetV1), Keyword(id=1319955332537893541, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, orderNo=2, keyword=VoVGSCSP), Keyword(id=1319955332617585318, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, orderNo=3, keyword=activation function), Keyword(id=1319955332676305575, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, orderNo=4, keyword=BiFPN), Keyword(id=1319955332743414440, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, orderNo=1, keyword=EfficientNetV1), Keyword(id=1319955334475661993, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, orderNo=2, keyword=VoVGSCSP), Keyword(id=1319955334588908202, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, orderNo=3, keyword=激活函数), Keyword(id=1319955334693765803, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, orderNo=4, keyword=BiFPN)], refs=[Reference(id=1319955339353637580, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=1, rfOrder=0, authorNames=null, journalName=null, refType=null, unstructuredReference=谢永超, 杨利, 严俊. 基于MQ-2型传感器的烟雾探测报警器的设计[J]. 计算机测量与控制, 2021, 29(8): 255−259., articleTitle=null, refAbstract=null), Reference(id=1319955339429135053, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=1, rfOrder=1, authorNames=null, journalName=null, refType=null, unstructuredReference=Xie Y C, Yang L, Yan J. Design of smoke detection alarm based on MQ-2 sensor[J]. Comput Meas Control, 2021, 29(8): 255−259., articleTitle=null, refAbstract=null), Reference(id=1319955339517215438, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=2, rfOrder=2, authorNames=null, journalName=null, refType=null, unstructuredReference=Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, 2014: 580–587.https://doi.org/10.1109/CVPR.2014.81., articleTitle=null, refAbstract=null), Reference(id=1319955339592712911, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=3, rfOrder=3, authorNames=null, journalName=null, refType=null, unstructuredReference=Ren S Q, He K M, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Trans Pattern Anal Mach Intell, 2017, 39(6): 1137−1149., articleTitle=null, refAbstract=null), Reference(id=1319955339664016080, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=4, rfOrder=4, authorNames=null, journalName=null, refType=null, unstructuredReference=Majid S, Alenezi F, Masood S, et al. Attention based CNN model for fire detection and localization in real-world images[J]. Expert Syst Appl, 2022, 189: 116114., articleTitle=null, refAbstract=null), Reference(id=1319955339777262289, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=5, rfOrder=5, authorNames=null, journalName=null, refType=null, unstructuredReference=Barmpoutis P, Dimitropoulos K, Kaza K, et al. Fire detection from images using faster R-CNN and multidimensional texture analysis[C]//2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, 2019: 8301–8305. https://doi.org/10.1109/ICASSP.2019.8682647., articleTitle=null, refAbstract=null), Reference(id=1319955339861148370, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=6, rfOrder=6, authorNames=null, journalName=null, refType=null, unstructuredReference=孙光灵, 王薪博. THI-YOLO: 改进YOLOv8的非机动车驾驶员头盔检测[J]. 光电工程, 2024, 51(12): 240244., articleTitle=null, refAbstract=null), Reference(id=1319955339940840147, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=6, rfOrder=7, authorNames=null, journalName=null, refType=null, unstructuredReference=Sun G L, Wang X B. THI-YOLO: improved non-motorized drivers helmet detection of YOLOv8[J]. Opto-Electron Eng, 2024, 51(12): 240244., articleTitle=null, refAbstract=null), Reference(id=1319955339999560404, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=7, rfOrder=8, authorNames=null, journalName=null, refType=null, unstructuredReference=宋华伟, 屈晓娟, 杨欣, 等. 基于改进YOLOv5的火焰烟雾检测[J]. 计算机工程, 2023, 49(6): 250−256., articleTitle=null, refAbstract=null), Reference(id=1319955340079252181, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=7, rfOrder=9, authorNames=null, journalName=null, refType=null, unstructuredReference=Song H W, Qu X J, Yang X, et al. Flame and smoke detection based on improved YOLOv5[J]. Comput Eng, 2023, 49(6): 250−256., articleTitle=null, refAbstract=null), Reference(id=1319955340175721174, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=8, rfOrder=10, authorNames=null, journalName=null, refType=null, unstructuredReference=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., articleTitle=null, refAbstract=null), Reference(id=1319955340234441431, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=8, rfOrder=11, authorNames=null, journalName=null, refType=null, unstructuredReference=Ma Y M, Zhang P F, Tan F S. Improved YOLOv8s model for smoke and flame detection in complex backgrounds[J]. Comput Eng Appl, 2025, 61(3): 121−130., articleTitle=null, refAbstract=null), Reference(id=1319955340301550296, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=9, rfOrder=12, authorNames=null, journalName=null, refType=null, unstructuredReference=张小雪, 王雨, 吴思远, 等. 基于级联稀疏查询机制的轻量化火灾检测算法[J]. 光电工程, 2023, 50(10): 230216., articleTitle=null, refAbstract=null), Reference(id=1319955340372853465, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=9, rfOrder=13, authorNames=null, journalName=null, refType=null, unstructuredReference=Zhang X X, Wang Y, Wu S Y, et al. An improved lightweight fire detection algorithm based on cascade sparse query[J]. Opto-Electron Eng, 2023, 50(10): 230216., articleTitle=null, refAbstract=null), Reference(id=1319955340444156634, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=10, rfOrder=14, authorNames=null, journalName=null, refType=null, unstructuredReference=Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//16th European Conference on Computer Vision, Glasgow, 2020: 213–229. https://doi.org/10.1007/978-3-030-58452-8_13., articleTitle=null, refAbstract=null), Reference(id=1319955340515459803, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=11, rfOrder=15, authorNames=null, journalName=null, refType=null, unstructuredReference=Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, 2017: 6000–6010., articleTitle=null, refAbstract=null), Reference(id=1319955340574180060, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=12, rfOrder=16, authorNames=null, journalName=null, refType=null, unstructuredReference=Bodla N, Singh B, Chellappa R, et al. Soft-NMS—improving object detection with one line of code[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision, Venice, 2017: 5562–5570. https://doi.org/10.1109/ICCV.2017.593., articleTitle=null, refAbstract=null), Reference(id=1319955340653871837, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=13, rfOrder=17, authorNames=null, journalName=null, refType=null, unstructuredReference=Meng D P, Chen X K, Fan Z J, et al. Conditional DETR for fast training convergence[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision, Montreal, 2021: 3631–3640. https://doi.org/10.1109/ICCV48922.2021.00363., articleTitle=null, refAbstract=null), Reference(id=1319955340712592094, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=14, rfOrder=18, authorNames=null, journalName=null, refType=null, unstructuredReference=Zhao Y A, Lv W Y, Xu S L, et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2024: 16965–16974. https://doi.org/10.1109/CVPR52733.2024.01605., articleTitle=null, refAbstract=null), Reference(id=1319955340800672479, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=15, rfOrder=19, authorNames=null, journalName=null, refType=null, unstructuredReference=Tan M X, Le Q. EfficientNet: rethinking model scaling for convolutional neural networks[C]//Proceedings of the 36th International Conference on Machine Learning, Long Beach, 2019: 6105–6114., articleTitle=null, refAbstract=null), Reference(id=1319955340901335776, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=16, rfOrder=20, authorNames=null, journalName=null, refType=null, unstructuredReference=Li H L, Li J, Wei H B, et al. Slim-neck by GSConv: A better design paradigm of detector architectures for autonomous vehicles[Z]. arXiv: 2206.02424, 2022. https://arxiv.org/abs/2206.02424v1., articleTitle=null, refAbstract=null), Reference(id=1319955341022970593, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=17, rfOrder=21, authorNames=null, journalName=null, refType=null, unstructuredReference=Tan M X, Pang R M, Le Q V. EfficientDet: scalable and efficient object detection[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2020: 10778–10787. https://doi.org/10.1109/CVPR42600.2020.01079., articleTitle=null, refAbstract=null), Reference(id=1319955341123633890, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=18, rfOrder=22, authorNames=null, journalName=null, refType=null, unstructuredReference=Clevert D A, Unterthiner T, Hochreiter S. Fast and accurate deep network learning by exponential linear units (ELUs)[C]//4th International Conference on Learning Representations, San Juan, 2016., articleTitle=null, refAbstract=null), Reference(id=1319955341190742755, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=19, rfOrder=23, authorNames=null, journalName=null, refType=null, unstructuredReference=He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 2016: 770–778. https://doi.org/10.1109/CVPR.2016.90., articleTitle=null, refAbstract=null), Reference(id=1319955341262045924, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=20, rfOrder=24, authorNames=null, journalName=null, refType=null, unstructuredReference=Sandler M, Howard A, Zhu M L, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 2018: 4510–4520. https://doi.org/10.1109/CVPR.2018.00474., articleTitle=null, refAbstract=null), Reference(id=1319955341345932005, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=21, rfOrder=25, authorNames=null, journalName=null, refType=null, unstructuredReference=Glorot X, Bordes A, Bengio Y. Deep sparse rectifier neural networks[C]//Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, Fort Lauderdale, 2011: 315–323., articleTitle=null, refAbstract=null), Reference(id=1319955341408846566, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=22, rfOrder=26, authorNames=null, journalName=null, refType=null, unstructuredReference=Chollet F. Xception: deep learning with depthwise separable convolutions[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, 2017: 1800–1807. https://doi.org/10.1109/CVPR.2017.195., articleTitle=null, refAbstract=null), Reference(id=1319955341480149735, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=23, rfOrder=27, authorNames=null, journalName=null, refType=null, unstructuredReference=Zhang X Y, Zhou X Y, Lin M X, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 2018: 6848–6856. https://doi.org/10.1109/CVPR.2018.00716., articleTitle=null, refAbstract=null), Reference(id=1319955341551452904, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=24, rfOrder=28, authorNames=null, journalName=null, refType=null, unstructuredReference=Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, 2017: 936–944. https://doi.org/10.1109/CVPR.2017.106., articleTitle=null, refAbstract=null), Reference(id=1319955341622756073, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=25, rfOrder=29, authorNames=null, journalName=null, refType=null, unstructuredReference=Liu X Y, Peng H W, Zheng N X, et al. EfficientViT: Memory efficient vision transformer with cascaded group attention[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, 2023: 14420–14430. https://doi.org/10.1109/CVPR52729.2023.01386., articleTitle=null, refAbstract=null), Reference(id=1319955341689864938, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=26, rfOrder=30, authorNames=null, journalName=null, refType=null, unstructuredReference=Han K, Wang Y H, Tian Q, et al. GhostNet: More features from cheap operations[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2020: 1577–1586. https://doi.org/10.1109/CVPR42600.2020.00165., articleTitle=null, refAbstract=null), Reference(id=1319955341761168107, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=27, rfOrder=31, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang A, Chen H, Lin Z J, et al. RepViT: revisiting mobile CNN from ViT perspective[Z]. arxiv: 2307.09283, 2023. https://doi.org/10.48550/arXiv.2307.09283., articleTitle=null, refAbstract=null)], funds=null, companyList=[AuthorCompany(id=1319955331564814995, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=1, ext=[AuthorCompanyExt(id=1319955331573203604, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331581592213, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601)]), AuthorCompany(id=1319955331699032726, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=2, ext=[AuthorCompanyExt(id=1319955331707421335, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331699032726, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Anhui Interational Joint Research Center for Intelligent Perception and High-dimensional Modeling of Ancient Buildings, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331711615640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331699032726, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601)])], figs=[ArticleFig(id=1319955334832177836, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.1, caption=RT-DETR model, figureFileSmall=VbIfotWO9SIOWcwvv6+PXw==, figureFileBig=QVP+Olnbutzv6EWVsY87Zw==, tableContent=null), ArticleFig(id=1319955334907675309, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图1, caption=RT-DETR模型, figureFileSmall=VbIfotWO9SIOWcwvv6+PXw==, figureFileBig=QVP+Olnbutzv6EWVsY87Zw==, tableContent=null), ArticleFig(id=1319955335104807598, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.2, caption=Improved RT-DETR model, figureFileSmall=KzKTlTflqPywA6lj4gm6Cw==, figureFileBig=Zi1PMgF4M6G26+bRvfzjcQ==, tableContent=null), ArticleFig(id=1319955335180305071, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图2, caption=改进的RT-DETR模型, figureFileSmall=KzKTlTflqPywA6lj4gm6Cw==, figureFileBig=Zi1PMgF4M6G26+bRvfzjcQ==, tableContent=null), ArticleFig(id=1319955335301939888, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.3, caption=Structure of MBConv module, figureFileSmall=4kBag7nydnskNWirPbNmeA==, figureFileBig=7R6uJLiIVnkQwDAvPH/RHg==, tableContent=null), ArticleFig(id=1319955335394214577, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图3, caption=MBConv模块结构, figureFileSmall=4kBag7nydnskNWirPbNmeA==, figureFileBig=7R6uJLiIVnkQwDAvPH/RHg==, tableContent=null), ArticleFig(id=1319955335461323442, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.4, caption=Structure of GSConv module, figureFileSmall=/1CqSXXspuEya7+Q3Bskkw==, figureFileBig=sW7C5wr80uJ9gHoAhM3x4w==, tableContent=null), ArticleFig(id=1319955335549403827, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图4, caption=GSConv模块结构, figureFileSmall=/1CqSXXspuEya7+Q3Bskkw==, figureFileBig=sW7C5wr80uJ9gHoAhM3x4w==, tableContent=null), ArticleFig(id=1319955335612318388, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.5, caption=Structure of GS bottleneck module, figureFileSmall=JqeZRO5/qQULlh/EPj4qAQ==, figureFileBig=+hUJ8jAa3E9nOJ5tPneUOQ==, tableContent=null), ArticleFig(id=1319955335683621557, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图5, caption=GS bottleneck模块结构, figureFileSmall=JqeZRO5/qQULlh/EPj4qAQ==, figureFileBig=+hUJ8jAa3E9nOJ5tPneUOQ==, tableContent=null), ArticleFig(id=1319955335763313334, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.6, caption=Structure of VoVGSCSP module, figureFileSmall=YJEl18oT0WWAVF2Ko/XyMA==, figureFileBig=kDBkme0Pzt7Qxkt4HGjXbg==, tableContent=null), ArticleFig(id=1319955335843005111, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图6, caption=VoVGSCSP模块结构图, figureFileSmall=YJEl18oT0WWAVF2Ko/XyMA==, figureFileBig=kDBkme0Pzt7Qxkt4HGjXbg==, tableContent=null), ArticleFig(id=1319955335931085496, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.7, caption=Structure of PANet module, figureFileSmall=zW7EPrpiNbZ3sGyDj1P46A==, figureFileBig=+TsUkwESVY4D1MG0B2/68w==, tableContent=null), ArticleFig(id=1319955336014971577, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图7, caption=PANet模块结构图, figureFileSmall=zW7EPrpiNbZ3sGyDj1P46A==, figureFileBig=+TsUkwESVY4D1MG0B2/68w==, tableContent=null), ArticleFig(id=1319955336136606394, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.8, caption=Structure of BiFPN module, figureFileSmall=6ZM1PlwDPgoIG3KAnfC1ag==, figureFileBig=E07vKf5KSs526qjFa5UzUA==, tableContent=null), ArticleFig(id=1319955336203715259, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图8, caption=BiFPN模块结构图, figureFileSmall=6ZM1PlwDPgoIG3KAnfC1ag==, figureFileBig=E07vKf5KSs526qjFa5UzUA==, tableContent=null), ArticleFig(id=1319955336316961468, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.9, caption=Comparison of confusion matrices on the NEWFire dataset, figureFileSmall=K62mUte9cQAyo6rja9EY7g==, figureFileBig=hPznD55sXNVLFeRT0fGM7Q==, tableContent=null), ArticleFig(id=1319955336409236157, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图9, caption=在NEWFire数据集上混淆矩阵对比结果, figureFileSmall=K62mUte9cQAyo6rja9EY7g==, figureFileBig=hPznD55sXNVLFeRT0fGM7Q==, tableContent=null), ArticleFig(id=1319955336547648190, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Fig.10, caption=Comparison of detection results, figureFileSmall=+5CF5nUP2lvT/LchUJWb/g==, figureFileBig=5y4OZNt6/MO0RsQPOdCzjg==, tableContent=null), ArticleFig(id=1319955336669283007, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=图10, caption=检测结果对比, figureFileSmall=+5CF5nUP2lvT/LchUJWb/g==, figureFileBig=5y4OZNt6/MO0RsQPOdCzjg==, tableContent=null), ArticleFig(id=1319955336753169088, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Tab.1, caption=

EfficientNetV1 neural network structure

, figureFileSmall=null, figureFileBig=null, tableContent=
StageOperateResolutionChannelsLayers
1Conv 3×3224×224321
2MBConv 1, K 3×3112×112161
3MBConv 6, K 3×3112×112242
4MBConv 6, K 3×356×56402
5MBConv 6, K 3×328×28803
6MBConv 6, K 3×314×141123
7MBConv 6, K 3×314×141924
8MBConv 6, K 3×37×73201
9Conv 1×1 & Pooling & FC7×712801
), ArticleFig(id=1319955336832860865, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=表1, caption=

EfficientNetV1神经网络结构

, figureFileSmall=null, figureFileBig=null, tableContent=
StageOperateResolutionChannelsLayers
1Conv 3×3224×224321
2MBConv 1, K 3×3112×112161
3MBConv 6, K 3×3112×112242
4MBConv 6, K 3×356×56402
5MBConv 6, K 3×328×28803
6MBConv 6, K 3×314×141123
7MBConv 6, K 3×314×141924
8MBConv 6, K 3×37×73201
9Conv 1×1 & Pooling & FC7×712801
), ArticleFig(id=1319955336912552642, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Tab.2, caption=

Experimental environment configurations

, figureFileSmall=null, figureFileBig=null, tableContent=
ConfigurationEnvironmental parameter
CPUIntel(R) Core(TM) i7-12700F 2.10 GHz
GPUNVIDIA GeForce RTX 3090 24 G
Memory32 G
Python3.9.19
CUDA11.8
Pytorch2.0.0
), ArticleFig(id=1319955336988050115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=表2, caption=

实验环境配置

, figureFileSmall=null, figureFileBig=null, tableContent=
ConfigurationEnvironmental parameter
CPUIntel(R) Core(TM) i7-12700F 2.10 GHz
GPUNVIDIA GeForce RTX 3090 24 G
Memory32 G
Python3.9.19
CUDA11.8
Pytorch2.0.0
), ArticleFig(id=1319955337046770372, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Tab.3, caption=

Results of different lightweight backbones on NEWFire dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelsmAP50mAP50-95Params/MGFLOPs
Baseline0.8340.55320.058.0
EfficientViT0.8200.54010.928.2
GhostNetV30.8300.55025.841.5
RepViT0.8350.56015.242.2
EfficientNetV10.8390.56613.524.7
), ArticleFig(id=1319955337122267845, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=表3, caption=

主干使用不同轻量模型在NEWFire数据集上的实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelsmAP50mAP50-95Params/MGFLOPs
Baseline0.8340.55320.058.0
EfficientViT0.8200.54010.928.2
GhostNetV30.8300.55025.841.5
RepViT0.8350.56015.242.2
EfficientNetV10.8390.56613.524.7
), ArticleFig(id=1319955337201959622, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Tab.4, caption=

Experimental results of BiFPN combined with different activation functions on NEWFire dataset

, figureFileSmall=null, figureFileBig=null, tableContent=
FunctionmAP50mAP50-95Params/MGFLOPs
BiFPN0.8360.55419.856.6
BiFPN+ReLU0.8350.56019.856.6
BiFPN+ELU0.8390.56319.856.6
), ArticleFig(id=1319955337285845703, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=表4, caption=

BiFPN结合不同激活函数在NEWFire数据集上的实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
FunctionmAP50mAP50-95Params/MGFLOPs
BiFPN0.8360.55419.856.6
BiFPN+ReLU0.8350.56019.856.6
BiFPN+ELU0.8390.56319.856.6
), ArticleFig(id=1319955338950984392, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Tab.5, caption=

Ablation results on the NEWFire and E-fire datasets

, figureFileSmall=null, figureFileBig=null, tableContent=
EfficientNetV1VoVGSCSPBiFPN-ENEWFireE-fireParams/MGFLOPs
mAP50mAP50-95mAP50mAP50-95
0.8340.5530.7730.50920.058.0
√0.8390.5660.7840.51913.524.7
√0.8390.5630.7800.52219.856.6
√0.8360.5560.7780.51720.055.3
√√0.8330.5500.7700.50519.754.0
√√0.8400.5660.7860.52613.423.5
√√0.8380.5570.7870.52614.021.6
√√√0.8450.5700.7910.53113.720.3
), ArticleFig(id=1319955339055841993, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=表5, caption=

在NEWFire和E-fire数据集上的消融实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
EfficientNetV1VoVGSCSPBiFPN-ENEWFireE-fireParams/MGFLOPs
mAP50mAP50-95mAP50mAP50-95
0.8340.5530.7730.50920.058.0
√0.8390.5660.7840.51913.524.7
√0.8390.5630.7800.52219.856.6
√0.8360.5560.7780.51720.055.3
√√0.8330.5500.7700.50519.754.0
√√0.8400.5660.7860.52613.423.5
√√0.8380.5570.7870.52614.021.6
√√√0.8450.5700.7910.53113.720.3
), ArticleFig(id=1319955339160699594, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=EN, label=Tab.6, caption=

Comparative results on NEWFire and E-fire datasets

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelNEWFireE-fireParams/MGFLOPs
mAP50mAP50-95mAP50mAP50-95
Faster R-CNN0.7260.4250.7020.4541.7207.2
RT-DETR-l0.8400.5420.7850.51232.7107.4
YOLOv5s0.8300.5450.7700.5049.123.8
YOLOv8m0.8340.5590.7760.51425.878.7
YOLOv8s0.8320.5540.7780.51811.128.4
YOLOv9c0.8300.5520.7760.51225.3102.3
ours0.8450.5700.7910.53113.720.3
), ArticleFig(id=1319955339248779979, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, language=CN, label=表6, caption=

在NEWFire和E-fire数据集上的对比实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelNEWFireE-fireParams/MGFLOPs
mAP50mAP50-95mAP50mAP50-95
Faster R-CNN0.7260.4250.7020.4541.7207.2
RT-DETR-l0.8400.5420.7850.51232.7107.4
YOLOv5s0.8300.5450.7700.5049.123.8
YOLOv8m0.8340.5590.7760.51425.878.7
YOLOv8s0.8320.5540.7780.51811.128.4
YOLOv9c0.8300.5520.7760.51225.3102.3
ours0.8450.5700.7910.53113.720.3
)], attaches=null, journal=Journal(id=1278641367198941188, delFlag=0, nameCn=光电工程, nameEn=Opto-Electronic Engineering, nameHistory1=null, nameHistory2=null, issn=1003-501X, eissn=2097-4019, cn=51-1346/O4, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=4Vimkd+qXLWNxtdpr9mFNw==, journalPrice=null, startedYear=null, abbrevIsoEn=Opto-Electronic Engineering, journalRemark=null, publicationField=null, createdTime=1782781457950, updatedTime=1784021784694, createdBy=18614031015, updatedBy=13041195026, firstLetterCn=G, firstLetterEn=G, subjectCode=Engineering, subjectName=null, subjectCodeEn=Engineering, subjectNameEn=null, picCn=4Vimkd+qXLWNxtdpr9mFNw==, picEn=vAq9s20WLs1ODDfbWq+Gjg==, jcr=null, cjcr=null, exts=[JournalExt(id=1283843675721536154, language=CN, name=光电工程, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1784021784954, updatedTime=1784021784954, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.manuscripts.com.cn/gdgc, submissionEditorUrl=http://www.manuscripts.com.cn/gdgc, submissionReviewUrl=http://www.manuscripts.com.cn/gdgc, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1283843675771867803, language=EN, name=Opto-Electronic Engineering, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1784021784966, updatedTime=1784021784966, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.manuscripts.com.cn/gdgc, submissionEditorUrl=http://www.manuscripts.com.cn/gdgc, submissionReviewUrl=http://www.manuscripts.com.cn/gdgc, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1278651732997652489, websiteList=[Website(id=1278723867418018151, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1278651732997652489, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/oee/CN, language=CN, createTime=1782801127533, createBy=18614031015, updateTime=1782804494442, updateBy=18614031015, name=光电工程-中文, tplId=1146099689490845704, title=光电工程, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1278738091150128034, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=articleTextType, value=kx, createTime=1782804518735, updateTime=1782804518735, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091120767903, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=banner, value=null, createTime=1782804518728, updateTime=1782804518728, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091171099557, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=grayFlag, value=0, createTime=1782804518740, updateTime=1782804518740, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091108184990, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=logo, value=https://castjournals.cast.org.cn/joweb/oee/CN/file/pic?fileId=A1C6uwqtMazluiWkEpR0Mg==, createTime=1782804518725, updateTime=1782804518725, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091179488167, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=minRunFlag, value=0, createTime=1782804518742, updateTime=1782804518742, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091141739425, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/oee/CN/file/pic, createTime=1782804518733, updateTime=1782804518733, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091175293862, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=silenceFlag, value=0, createTime=1782804518741, updateTime=1782804518741, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091129156512, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1782804518730, updateTime=1782804518730, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091154322339, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=themeColor, value=null, createTime=1782804518736, updateTime=1782804518736, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091162710948, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=themeStyle, value=null, createTime=1782804518738, updateTime=1782804518738, creator=18614031015, updator=18614031015)]), Website(id=1278723867522875769, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1278651732997652489, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/oee/EN, language=EN, createTime=1782801127558, createBy=18614031015, updateTime=1782804490442, updateBy=18614031015, name=光电工程-英文, tplId=1146101810881728533, title=Opto-Electronic Engineering, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1278738063660659607, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=articleTextType, value=kx, createTime=1782804512181, updateTime=1782804512181, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063635493780, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=banner, value=null, createTime=1782804512175, updateTime=1782804512175, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063924900762, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=grayFlag, value=0, createTime=1782804512244, updateTime=1782804512244, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063606133651, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=logo, value=https://castjournals.cast.org.cn/joweb/oee/EN/file/pic?fileId=A1C6uwqtMazluiWkEpR0Mg==, createTime=1782804512168, updateTime=1782804512168, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063937483676, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=minRunFlag, value=0, createTime=1782804512247, updateTime=1782804512247, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063652270998, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/oee/EN/file/pic, createTime=1782804512179, updateTime=1782804512179, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063933289371, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=silenceFlag, value=0, createTime=1782804512246, updateTime=1782804512246, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063639688085, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1782804512176, updateTime=1782804512176, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063664853912, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=themeColor, value=null, createTime=1782804512182, updateTime=1782804512182, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063916512153, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=themeStyle, value=null, createTime=1782804512242, updateTime=1782804512242, creator=18614031015, updator=18614031015)])], journalTitle=光电工程, weixinUrl=null, journalUrl=https://www.oejournal.org/oee, iacademicId=null, status=1, seqNo=null, journalTitleEn=Opto-Electronic Engineering, journalPhotoCn=4Vimkd+qXLWNxtdpr9mFNw==, journalPhotoEn=vAq9s20WLs1ODDfbWq+Gjg==, journalFirstLetter=G, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false, interPubPlatform=, interPubPlatformUrl=null), detailUrlCn=https://castjournals.cast.org.cn/joweb/oee/CN/10.12086/oee.2026.250152, detailUrlEn=https://castjournals.cast.org.cn/joweb/oee/EN/10.12086/oee.2026.250152, pdfUrlCn=https://castjournals.cast.org.cn/joweb/oee/CN/PDF/10.12086/oee.2026.250152, pdfUrlEn=https://castjournals.cast.org.cn/joweb/oee/EN/PDF/10.12086/oee.2026.250152, aliStartDate=0, aliEndDate=0, collectionFlag=false, citedCount=null, citedUrl=null, previewStatus=0, delFlag=0, hasFullText=1, orderTime=1769270400000, fullTextJson=null, articleText=null, reference=null)
收藏切换
轻量化RT-DETR与多尺度特征融合的火焰烟雾检测算法
收藏切换
PDF下载
孙光灵 1, 2, *, * , 王雪婷 1
光电工程 | 研究论文 2026,53(1): 250152-1
收起
收藏切换
光电工程 |研究论文 2026 , 53 (1) : 250152 -1
轻量化RT-DETR与多尺度特征融合的火焰烟雾检测算法
全屏
[Author(id=1319955331824861850, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=sunguangling@163.com, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1319955331912942237, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955331824861850, language=EN, stringName=Guangling Sun, firstName=Guangling, middleName=null, lastName=Sun, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, address=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China
2Anhui Interational Joint Research Center for Intelligent Perception and High-dimensional Modeling of Ancient Buildings, Anhui Jianzhu University, Hefei, Anhui 230601, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955331980051102, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955331824861850, language=CN, stringName=孙光灵, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, *, *, address=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601
2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601, bio={"img":"RelyF6Q49UAosq9MiMEGiw==","content":"

孙光灵(1976-)，男，硕士，教授，主要研究方向为机器学习、图像处理。E-mail：

"}, bioImg=RelyF6Q49UAosq9MiMEGiw==, bioContent=

孙光灵(1976-)，男，硕士，教授，主要研究方向为机器学习、图像处理。E-mail：

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955331564814995, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=1, ext=[AuthorCompanyExt(id=1319955331573203604, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331581592213, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601)]), AuthorCompany(id=1319955331699032726, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=2, ext=[AuthorCompanyExt(id=1319955331707421335, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331699032726, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2Anhui Interational Joint Research Center for Intelligent Perception and High-dimensional Modeling of Ancient Buildings, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331711615640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331699032726, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601)])]), Author(id=1319955332097491616, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=18755443812@163.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955332227515042, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955332097491616, language=EN, stringName=Xueting Wang, firstName=Xueting, middleName=null, lastName=Wang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955332332372643, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, authorId=1319955332097491616, language=CN, stringName=王雪婷, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601, bio={"img":"H5CiDKmoIagGXm4zWwY6tA==","content":"

王雪婷(1999-)，女，硕士研究生，主要研究方向为图像处理、目标检测。E-mail：

"}, bioImg=H5CiDKmoIagGXm4zWwY6tA==, bioContent=

王雪婷(1999-)，女，硕士研究生，主要研究方向为图像处理、目标检测。E-mail：

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955331564814995, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, xref=1, ext=[AuthorCompanyExt(id=1319955331573203604, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China), AuthorCompanyExt(id=1319955331581592213, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955325730538082, companyId=1319955331564814995, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1安徽建筑大学电子与信息工程学院，安徽 合肥 230601)])])]
孙光灵1, 2, *, * , 王雪婷1
作者信息
  • 1安徽建筑大学电子与信息工程学院，安徽 合肥 230601
  • 2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601
通讯作者:
作者简介:

作者贡献

孙光灵：指导论文选题及论文的修改；王雪婷：研究资料和背景分析，实验实施与分析，论文写作。

孙光灵(1976-)，男，硕士，教授，主要研究方向为机器学习、图像处理。E-mail：

王雪婷(1999-)，女，硕士研究生，主要研究方向为图像处理、目标检测。E-mail：

Lightweight RT-DETR and multi-scale featurefusion algorithm for flame and smoke detection
Guangling Sun1, 2 , Xueting Wang1
Affiliations
  • 1School of Electronic and Information Engineering, Anhui Jianzhu University, Hefei, Anhui 230601, China
  • 2Anhui Interational Joint Research Center for Intelligent Perception and High-dimensional Modeling of Ancient Buildings, Anhui Jianzhu University, Hefei, Anhui 230601, China
    • Corresponding Author:
出版时间: 2026-01-25 doi: 10.12086/oee.2026.250152
文章导航
收藏切换
摘要
收起

针对火焰烟雾检测复杂性高，精度偏低，并且经常会产生漏检误检等问题，提出了轻量化RT-DETR与多尺度特征融合的火焰烟雾检测算法。该方法首先以RT-DETR-r18为基础架构，在主干网络中引入了一种轻量化网络结构EfficientNetV1，降低了模型参数量；接着引入了VoVGSCSP模块，有效减少了特征冗余，并且增强了模型特征提取能力；最后结合双向特征融合金字塔网络BiFPN和ELU激活函数的优点，在BiFPN权重归一化过程中引入ELU激活函数，设计了BiFPN-E结构，提高了模型的跨尺度交互能力，有效提升了模型的检测精度。根据实验结果显示，改进的RT-DETR模型相较于原模型在公开数据集NEWFire和自建数据集E-fire上mAP50分别提高了1.1%和1.8%，mAP50-95分别提高了1.7%和2.2%，参数量下降32%，浮点计算量下降65%。说明改进模型在提高检测精度的同时保持了轻量化要求，满足了在火焰烟雾检测时的精度和效率的实际需求。

EfficientNetV1  /  VoVGSCSP  /  激活函数  /  BiFPN
Objective

With the advancement of modernization and the development of urbanization, fire safety issues in various public venues, commercial buildings, and residential homes have become increasingly prominent. Fire incidents not only directly threaten lives and property but also generate smoke that often causes irreversible pollution and damage to the ecological environment. Traditional fire detection methods primarily rely on physical devices such as sensors, yet these approaches have limitations—including insufficient sensitivity, frequent false alarms and missed detections, susceptibility to environmental interference, and difficulty in achieving early warning. Consequently, there is an urgent need for more efficient fire detection algorithms. In recent years, rapid advancements in artificial intelligence have driven breakthroughs in deep learning within computer vision, particularly demonstrating exceptional performance in object detection. This has opened new avenues and research perspectives for fire detection studies.

Methods

Addressing current limitations in flame and smoke detection technologies—such as insufficient accuracy in complex scenarios, susceptibility to false negatives and positives, and high computational costs—this research innovatively applies the cutting-edge RT-DETR model from computer vision to fire detection tasks. The aim is to thoroughly explore this model's capability to capture and recognize flame and smoke features in complex, dynamic environments, striving to develop a fire detection algorithm with high detection accuracy, low false alarm rates, and the ability to meet real-time early warning requirements. RT-DETR is an efficient real-time end-to-end detector based on the Transformer architecture. It organically combines the local feature extraction advantages of convolutional neural networks (CNNs) with the global modeling capabilities of Transformers through the introduction of a hybrid encoder. By employing an innovative IoU-aware query selection mechanism to precisely filter initial target queries, it achieves a good balance between detection accuracy and inference speed while significantly reducing computational latency. However, when handling complex, multi-scale, and morphologically variable targets like flames and smoke, this model still suffers from insufficient capture of local information and limited extraction of fine-grained features. This often leads to false negatives and false positives for small targets, limiting its application in scenarios demanding high accuracy. To address this, this paper proposes a lightweight RT-DETR-based flame and smoke detection algorithm with multi-scale feature fusion to enhance the model's perception and discrimination capabilities for such targets. We first adopt RT-DETR-r18 as the baseline model, introducing the lightweight EfficientNetV1 architecture into the backbone network. Through evaluation and experimentation with mainstream lightweight networks like GhostNet, RepViT, and EfficientViT, results demonstrate that EfficientNetV1 achieves the optimal balance between accuracy and efficiency via its composite scaling strategy. It significantly reduces parameter count and computational burden while maintaining robust feature extraction capabilities, providing the optimal feature foundation for subsequent detection tasks. Second, the VoVGSCSP module is designed at the model's neck. This architecture integrates the efficient feature reuse of VoVNet with the gradient splitting concept of CSPNet, effectively enhancing the model's expression capabilities for key features like flame-smoke texture and color while suppressing feature redundancy. Finally, addressing the core challenge of extreme scale variations in flame-smoke targets, this paper introduces the more efficient multi-scale feature fusion mechanism bidirectional feature pyramid network (BiFPN)-E. During the weighted fusion process in the BiFPN, it innovatively employs the ELU activation function for weight normalization. The non-zero saturation property of the ELU function when handling negative inputs generates smoother and more discriminative fusion weights, thereby strengthening the interaction and complementarity between feature maps at different scales. This significantly improves the model's perception and discrimination accuracy in complex scenarios such as large-scale flames, faint smoke, and occluded objects.

Results and Discussions

The final experimental results fully validate the comprehensive effectiveness of the proposed method. On the NEWFire dataset, the improved model achieved mAP50 and mAP50-95 scores of 84.5% and 57.0%, respectively, representing improvements of 1.1% and 1.7% over the baseline RT-DETR model. On the self-built E-fire dataset, the accuracy metrics mAP50 and mAP50-95 improved by 1.8% and 2.2%, respectively. More importantly, while achieving these accuracy gains, the model demonstrated significant lightweighting: model parameters were reduced by 32%, and floating-point operations were drastically reduced by 65%. In-depth ablation experiments further confirm that the BiFPN-E architecture is the most critical contributor to accuracy gains, while the EfficientNetV1 backbone serves as the core for achieving efficient model compression. Visual analysis also reveals that the improved model's feature heatmap responses are more concentrated on actual flame-smoke regions, demonstrating stronger suppression of fire-like distractors and complex backgrounds, higher recognition accuracy, and effectively reducing both false negatives and false positives.

Conclusions

Through systematic enhancements to the RT-DETR model, this research achieves a synergistic breakthrough in accuracy, efficiency, and feature discriminative power for fire detection. The EfficientNetV1 backbone enables efficient model compression, while the introduced VoVGSCSP module enhances the model's ability to capture irregular fire shapes and dynamic changes. The BiFPN-E architecture significantly improves detection accuracy through optimized multi-scale feature fusion. The synergistic effect of these three components enables the model to maintain high accuracy while reducing parameters by 32% and computational load by 65%. This breakthrough overcomes computational resource constraints while enhancing performance, providing a practical solution for achieving high-precision real-time fire alerts on resource-constrained edge devices.

EfficientNetV1  /  VoVGSCSP  /  activation function  /  BiFPN
孙光灵, 王雪婷. 轻量化RT-DETR与多尺度特征融合的火焰烟雾检测算法. 光电工程, 2026 , 53 (1) : 250152 -1 . DOI: 10.12086/oee.2026.250152
Guangling Sun, Xueting Wang. Lightweight RT-DETR and multi-scale featurefusion algorithm for flame and smoke detection[J]. Opto-Electronic Engineering, 2026 , 53 (1) : 250152 -1 . DOI: 10.12086/oee.2026.250152
正文
收起
1　引　言
收起
随着现代化的推进和城市化的发展，各类公共场所、商业建筑以及居民住宅的火灾安全问题日益凸显。火灾的发生，不仅会直接危及生命和财产安全，其产生的烟雾还会对生态环境造成不可逆转的污染和破坏。传统的火灾探测方法主要依赖传感器[1]等物理设备，这些设备存在一定局限性，比如灵敏度不足、误报和漏报频繁、易受环境干扰等，难以实现早期预警，因此迫切需要更加高效的火灾检测算法。近年来，随着人工智能技术的快速发展，深度学习在计算机视觉领域取得了突破性进展，特别是在目标检测方面展现出卓越的性能优势，这为火灾检测研究提供了新的方法和研究思路。
基于深度学习目标检测领域，检测算法主要分为两种，一种是双阶段目标检测算法，另外一种是单阶段目标检测算法。双阶段目标检测算法需同时识别和定位两种不同类型的目标，并且还需要考虑两者之间的依赖关系，检测任务通常涉及多类别协同处理；而单阶段目标检测算法则专注于单一类别目标的快速识别与定位，通过直接在特征图上回归目标框坐标实现高效检测。双阶段目标检测算法主要有CNN (convolutional neural networks)[2]、R-CNN (regions with convolutional neural networks)和Faster R-CNN (faster region-based convolutional neural networks)[3]。Majid等[4]提出了一种基于注意力机制的CNN模型，通过嵌入通道和空间注意力模块，动态增强火灾特征并抑制背景干扰，同时结合多尺度特征融合和双分支结构，有效提高了火灾检测的精度；Barmpoutis等[5]提出了一种结合Faster R-CNN目标检测框架和多维纹理特征分析的火灾检测方法，旨在提高复杂场景下的火灾检测精度。虽然此类方法对于火焰和烟雾的检测精度有了一定的提升，但是复杂的网络结构导致检测速度较慢，难以满足对实时性要求较高的火灾检测场景。单阶段目标检测算法比较典型的就是YOLO (you only look once) [6]系列算法。宋华伟等[7]通过将YOLOv5中的路径聚合网络(PAN)替换为双向交叉尺度融合模块，并且在其检测头加入了协调注意力机制 (coordinate attention)，有效提升了小目标的捕捉能力；马耀名等[8]提出一种面向复杂背景下火焰烟雾检测的 YOLOv8s 改进模型。设计了轻量级特征提取模块和多尺度上下文融合模块，有效提升了模型识别能力；最后使用Inner-SIoU 损失函数，有效改善了边界框不匹配的问题。张小雪等[9]通过在YOLOv5s网络中嵌入轻量化的级联稀疏查询模块，同时在主干部分融合高效通道注意力模块 (ECA, efficient channel attention)，并结合Inner-SIoU损失函数，以提升检测精度与鲁棒性。YOLO算法虽然有着计算效率高、参数较低、易于部署等优点，但其轻量化导致特征提取能力较弱，并且YOLO主要基于卷积结构，导致全局建模能力有限。而DETR (detection Transformer)[10]是基于 Transformer[11]架构的网络，其自注意力机制能够实现全局上下文建模，更擅长捕捉图像中远距离目标之间的关系，适用于复杂场景下的目标检测；同时DETR摒弃了传统的手工设计锚框与非极大值抑制 (non-maximum suppression, NMS)[12]流程，后者在处理密集目标时易导致误抑制，而DETR通过端到端的匹配机制有效规避了该问题。虽然DETR在目标检测领域中有一定优势，但也因其Transformer模块参数多，推理速度较慢，难以在资源受限设备上进行部署。并且其缺失特征金字塔，对局部信息提取能力有限，导致对小目标检测效果不如预期。为解决上述问题，相关研究者们在此基础上相继推出了一些改进版本。 Conditional DETR[13]通过引入空间条件查询机制，将参考点坐标作为位置条件动态生成解码器查询，使解码器的交叉注意力能够动态聚焦于图像特征的局部相关区域，从而显著提升检测精度与训练收敛速度，而RT-DETR (real-time detection Transformer)[14]则引入了更高效的特征编码方式和Query Selection策略，相比其他改进的DETR模型而言，在保持高精度检测性能的同时更加轻量化，使其在实际应用中更具优势。
RT-DETR作为一种端到端的实时目标检测框架，摒弃了传统检测模型中复杂的候选框生成与后处理流程，极大地优化了检测速度与系统响应能力，非常适用于火灾这类需要迅速反应的场景。但是由于局部信息捕获能力不足，细节提取能力有限，对小目标检测精度较低，容易出现漏检和误检问题，限制了其在高精度需求场景下的应用。因此本文提出改进的RT-DETR火焰烟雾检测模型。主要内容如下：
1) 在RT-DETR的主干网络中存在多个卷积和残差块的反复叠加，存在大量计算冗余，导致模型参数量偏大，因此本文重新设计了主干网络，引入了更轻量化的模型EfficientNetV1[15]。
2) 由于原模型的RepC3结构仅通过卷积堆叠提取特征，缺乏显式的多尺度交互机制，在特征提取能力上存在一定的局限性，因此本文用VoVGSCSP[16]结构代替了原模型的RepC3结构，进一步优化该检测模型。
3) 引入双向特征金字塔网络(BiFPN)[17]，并且在权重归一化时引入ELU[18]激活函数，增强跨尺度交互能力的同时提高梯度的稳定性，最终提升火焰烟雾的检测精度。
2　模型介绍
收起
2.1　RT-DETR模型
RT-DETR主要针对DETR收敛速度慢、计算资源需求高等问题进行了优化，并且推出了多个不同版本的模型架构如ResNet18[19]、ResNet50等，以适应多样化的应用场景，大幅度提高了模型的处理能力和资源利用效率，本实验基于研究背景的要求和条件，选择 RT-DETR-r18作为基准模型。
RT-DETR模型如图1所示。RT-DETR模型创新性地融合了CNN和Transformer架构的双重优势，主要由主干网络、编码器与解码器组成。主干网络主要用于特征提取。RT-DETR的编码器是一种高效的混合编码器，它通过解耦尺度内特征交互(AIFI)强化同层级特征的语义关联和跨尺度特征融合(CCFM)来实现不同分辨率特征的有效整合。这种设计能够在保持较高检测精度的同时，降低模型的计算复杂度。解码器则采用迭代优化机制，通过动态融合编码器输出的高级语义特征与空间位置信息，逐步优化检测结果。
2.2　改进的RT-DETR模型
改进的主干模型采用了轻量级的设计 EfficientNetV1来替换传统骨干网络，并且在深层次用VoVGSCSP模块替换了RepC3结构，进一步增强模型的特征融合能力，同时降低了检测模型的复杂度，最后使用融入了ELU激活函数的双向特征融合金字塔结构BiFPN，增强模型在复杂场景下的适应性，提高了检测的精度。改进后的RT-DETR如图2所示。
2.2.1　改进的主干网络
随着网络深度增加，模型会积累大量冗余信息，不仅显著增加了计算复杂度和参数量，还削弱了模型在复杂火灾场景下的泛化能力，因此本实验引入轻量化EfficientNetV1模型替代原ResNet-18主干网络，通过其高效的复合缩放机制和深度可分离卷积，在保持特征提取能力的同时，有效降低模型参数量和浮点计算量，提升检测效率。
EfficientNet是由Google Research团队提出的革命性神经网络架构，鉴于传统的神经网络架构通常采用单独调整深度、宽度或分辨率的方法进行缩放，单独增加一个维度可能会在提升网络性能的同时增加网络复杂性和计算量，而EfficientNet创新性地采用“复合缩放” (compound scaling)方法，通过协同调节网络的这三个关键维度来实现计算效率与模型性能的平衡，同时还采用了NAS (neural architecture search)技术来自动调整这三者的比例，来建立最优的复合缩放系数。EfficientNetV1神经网络结构如表1所示。
MBConv (Mobile inverted bottleneck convolution)模块是EfficientNetV1的核心构建单元，该模块是来自MobileNetV2[20]中的倒残差模块。首先，标准卷积用于增加输入特征图通道数，然后是深度可分离卷积用于空间特征处理，同时也能够减少模型复杂度，最后是逐点卷积用于融合扩增后的特征。在此过程中，MBConv采用了Swish激活函数，它结合了ReLU[21]的线性特性与Sigmoid的平滑性，具备两者的共同优点，从而提升了非线性表达能力。此外，为了增强通道维度上的特征表达能力，还在逐点卷积前引入了SE注意力机制。如图3所示，K×K表示卷积核尺寸，s1/s2表示步距。
2.2.2　重构RepC3结构
RepC3结构通过重复堆叠卷积模块提升特征表达，但增加了计算复杂度，且缺乏有效的信息交互与多尺度融合，限制了检测性能。为此，本文引入VoVGSCSP模块，它采用高效卷积操作并强化多尺度特征融合，在更广泛的尺度范围内提升火灾检测的精度与鲁棒性。
GSConv是构成VoVGSCSP模块的核心组件。它通过将卷积操作拆解成更小的部分(组卷积)，从而减少了计算复杂度，同时保持了良好的性能，是一种轻量级卷积操作。首先，输入特征图会经过一次标准卷积进行下采样处理，接着应用深度卷积 (DWConv) [22]，在此过程中，每个输入通道都使用独立的卷积核进行处理；接下来，标准卷积与深度卷积的结果会被拼接在一起；最终通过shuffle[23]操作进行均匀混合处理。这样标准卷积的信息就能完全混入深度可分离卷积的输出，在不同的通道上均匀地交换局部的特征信息，从而让模型更好地学习到更多的多尺度信息。GSConv模块结构如图4所示。
VoVGSCSP是一种高效的轻量级模块，其核心是由GSConv和标准卷积组成的GS bottleneck基本单元，采用一次性聚合策略实现跨阶段特征融合。该设计通过GS bottleneck模块的堆叠增强特征提取能力，在降低计算复杂度的同时保持了较高的检测精度，实现了轻量化和性能的良好平衡。GS bottleneck模块结构和VoVGSCSP模块结构分别如图5和图6所示。
2.2.3　BiFPN-E结构
在RT-DETR模型中，由于缺乏专门设计的多尺度特征融合模块，模型在处理小目标、细粒度信息时表现不足，同时原有的特征流动方式较为单一，浅层与深层特征之间的交互不充分，限制了特征表达能力的提升。因此，为了进一步增强局部细节感知和跨尺度信息交互能力，同时保持实时推理的优势，本文引入了更加高效灵活的双向特征融合金字塔网络 (BiFPN)。
PANet模块结构如图7所示。BiFPN模块结构如图8所示。BiFPN是一种高效的多尺度特征融合网络，它在传统特征金字塔网络(PANet)[24]的基础上进行了优化。双向特征融合在BiFPN中指的是一种机制，它允许特征网络层中的信息在自顶向下和自底向上两个方向上流动和融合，使不同层级间的特征融合更高效，而且去除了冗余连接和计算节点，使得结构更加精简，计算量更少，火灾检测精度更高。BiFPN通过对不同尺度的特征加权融合，确保在融合过程中，不同尺度的特征能根据其贡献程度被合理利用，这种方式能够减少信息丢失和冗余情况，从而增强了特征表达的有效性，特别是在火焰烟雾检测任务中表现最佳。
虽然BiFPN在特征融合时有不少优点，但在进行权重归一化时只是简单的加权平均归一化，这种方法通常难以捕捉到特征间相对复杂的非线性关系，导致融合的特征图不能充分利用某些尺度上的信息，特别是在比较复杂的背景下，会影响检测精度和鲁棒性。因此本实验在BiFPN进行权重归一化时结合了ELU激活函数，设计了BiFPN-E模块，对于不同尺度的特征，它能根据特征的响应强度进行灵活的调整，从而增强权重归一化的表达能力。其表达式为
$ \widehat{\omega_i}=\dfrac{E L U\left(\omega_i\right)}{\displaystyle\sum_{j=1}^n E L U\left(\omega_j\right)}\;, $
式中：$ \widehat {{\omega _i}} $表示归一化后的第i个特征图的权重；$ {\omega _i} $表示第i个融合分支的可学习权重，用于衡量输入特征的重要性；j表示求和过程中的索引，用于遍历所有n个特征图；ELU(·)表示数线性单元的激活函数，用于增强梯度传播的非线性处理。ELU激活函数公式为
$ E L U(x)= \begin{cases}x, & x \gt 0 \\ \alpha(\exp (x)-1), & x \leqslant 0\end{cases} \;,$
式中：$ \alpha $表示设置为1的超参数，用来控制负区域的斜率，可以发现ELU对所有负值输入均能产生非零梯度，有效避免了神经元死亡问题。作为非饱和函数，它也不会遇到梯度爆炸或者消失的问题。除此之外，该函数还保持了很好的非线性特性，不仅能够拟合更复杂的数据，还能增强模型的表达能力。
3　实验与结果分析
收起
3.1　实验环境介绍
训练过程中使用批处理大小为8，图片尺寸为640 pixel×640 pixel，学习率设置为0.0001，优化器选择的是AdamW，迭代次数为200轮。实验在pytorch框架下进行训练，实验环境配置如表2所示。
3.2　数据集
本实验采用的是来自Roboflow网站的公开数据集，即NEWFire数据集。该数据集包含各种不同场景下的环境，该数据集被划分成5923张训练集、1681张验证集和839张测试集(用于评估模型性能)。同时，为了进一步验证模型的有效性，从网上搜集了相关火焰烟雾图像，去掉一些重复和不相关的图像，然后对没有标注的图片利用脚本工具进行标注，最终保留了6000张火焰烟雾数据集，最后数据集(E-fire)被划分成4000张训练集、1000张验证集和1000张测试集。
3.3　评价指标
图像目标检测常用的评价指标主要有参数量(parameters)、计算量(GFLOPs)、精准率(precision, P)、召回率(recall, R)、平均精度(average precision, AP)、平均精度均值(mean average precision, mAP)。精准率和召回率的计算公式分别为
$ P = \frac{{TP}}{{TP + FP}} \;,$
$ R = \frac{{TP}}{{TP + FN}} \;,$
式中：TP (true positive)代表真实为真样本，预测为真样本；FP (false positive)代表真实为负样本，预测为真样本；FN (false negative)代表真实为真样本，预测为负样本。
$ AP=\displaystyle\int_0^1P(R)\mathrm{d}R. $
使用不同的置信度阈值可以获得一系列对应的Precision和Recall坐标点，这些点连接形成的连续曲线即为P-R曲线。而AP就表示为P-R曲线与坐标轴围成的面积，它衡量的是模型在不同召回率下的平均精度表现，AP值越大，模型精确度越高。
$ mAP = \frac{1}{N}\sum\limits_{i = 1}^N {A{P_i}}\;, $
式中：N表示目标类别数量；$ A{P_i} $表示第i个类别的AP值。mAP50代表将阈值IoU设为0.50时，模型的平均精度；mAP50-95代表阈值IoU从0.50到0.95时，以步长为0.05计算得出的平均精度。
3.4　实验结果分析
3.4.1　RT-DETR-r18模型在主干部分使用不同轻量化模型结果分析
为了初步减少模型参数，本实验在主干部分分别使用EfficientViT[25]、GhostNetV3[26]、 RepViT[27]和本实验采用的EfficientNetV1，其中EfficientViT和RepViT都是最小模型类别。其实验结果如表3所示，可以发现，EfficientViT虽然参数和浮点数都很低，但检测精度下降不少；GhostNetV3只有浮点数有所下降，其参数较于原模型还有一定上涨，并且精度也有所下降；RepViT虽然相比原模型参数和浮点数都有下降，但相比 EfficientNetV1，效果还是不够突出。EfficientNetV1很好地兼顾了精度和轻量化这两个方面。因此本研究选择EfficientNetV1作为特征提取的主干网络。
3.4.2　BiFPN使用不同激活函数分析
在RT-DETR-r18模型中对BiFPN权重归一化时引入不同激活函数进行实验，根据表4数据可以发现，本实验方法效果最好。
3.4.3　消融实验
根据表5的消融实验结果，可以确定改进点的有效性。在引入EfficientNetV1后发现模型在NEWFire和E-fire数据集上mAP50分别提升了0.5%和1.1%，mAP50-95分别提升1.3%和1.0%，参数减少了6.5 M，浮点计算量下降了33.3 G；在引入BiFPN-E后发现模型在NEWFire和E-fire数据集上mAP50分别提升了0.5%和0.7%，mAP50-95分别提升1.0%和1.3%，参数减少了0.2 M，浮点计算量下降了0.4 G；在加入VoVGSCSP后可以发现相较于基准模型在NEWFire和E-fire数据集上mAP50分别提升0.2%和0.5%，mAP50-95分别提升0.3%和0.8%，参数无明显变化，浮点计算量下降了2.7 G；在只引入VoVGSCSP和BiFPN-E后，模型虽然在浮点计算量和参数量上都有所下降，但是精度指标mAP50和mAp50-95也都下降了， 这是由于VoVGSCSP主要提取局部多尺度特征，而BiFPN则通过双向跨层级连接进行全局多尺度特征融合，说明两个模块对同一特征进行了重复处理，存在信息冗余的冲突；而且过度融合导致特征细节变得模糊和丢失，进而导致检测精度的下降。而在引入EfficientNetV1和BiFPN-E后，模型在NEWFire和E-fire数据集上mAP50分别提升了0.6%和1.3%，mAP50-95分别提升1.3%和1.7%，参数减少了6.6 M，浮点计算量下降了34.5 G；在引入EfficientNetV1和VoVGSCSP后，模型在NEWFire和E-fire数据集上mAP50分别提升了0.4%和1.4%，mAP50-95分别提升0.4%和1.7%，参数减少了6.0 M，浮点计算量下降了36.4 G；而在引入这三个模块后的改进模型在NeWFire和E-fire数据集上mAP50分别提高了1.1%和1.8%，mAP50-95分别提升1.7%和2.2%，参数下降了6.3 M，浮点计算量下降了37.7 G。结果说明改进模型在精度和轻量化方面均有不错表现。
混淆矩阵中每个数值代表真实类别为某行的样本，被预测为某列类别的具体数量。对角线上的数值表示正确分类的样本数，其高低直接反映对应类别的识别准确率；非对角线数值则揭示了类别间的混淆情况。改进前后的混淆矩阵对比如图9所示。通过对比改进前后的混淆矩阵可以发现，优化后的模型在性能上取得了显著提升。
3.4.4　对比实验
为了进一步验证模型的优越性，将本实验结果与Faster R-CNN、RT-DETR-l、YOLOv5s、YOLOv8m、YOLOv8s、YOLOv9c模型对比可以发现改进模型在精度和轻量化方面都有着一定的优越性。其结果如表6所示。
3.4.5　检测效果对比
改进前后检测结果对比如图10所示。为了直观验证改进模型的优越性，本文对比了改进前后模型在NEWFire测试集上的检测效果图，选取了在不同背景下的火焰烟雾检测情况。可以发现本实验的改进模型检测出的火焰和烟雾精度都比原始模型有所提高，表明了改进模型的优越性。
4　结　论
收起
随着计算机科技的发展，基于深度学习的火焰烟雾检测已越来越成熟，本文将使用比较广泛的检测模型RT-DETR与轻量化模型EfficientNeTV1相结合，然后引入了VoVGSCSP模块，增强模型特征提取能力，再在双向特征金字塔网络BiFPN的基础之上融入ELU激活函数，进一步提高模型对火灾目标在不同尺度和复杂场景下的检测能力。火焰烟雾的检测相较于其他目标检测来说，对精度和轻量化方面都有更高的要求，后续可以在此基础上再朝着这两个方面继续优化，使得模型更加全面，提高其实际应用价值。

文献
收起
参考文献 引证文献
排序方式：
1
谢永超, 杨利, 严俊. 基于MQ-2型传感器的烟雾探测报警器的设计[J]. 计算机测量与控制, 2021, 29(8): 255−259.
Xie Y C, Yang L, Yan J. Design of smoke detection alarm based on MQ-2 sensor[J]. Comput Meas Control, 2021, 29(8): 255−259.
2
Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, 2014: 580–587.https://doi.org/10.1109/CVPR.2014.81.
3
Ren S Q, He K M, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Trans Pattern Anal Mach Intell, 2017, 39(6): 1137−1149.
4
Majid S, Alenezi F, Masood S, et al. Attention based CNN model for fire detection and localization in real-world images[J]. Expert Syst Appl, 2022, 189: 116114.
5
Barmpoutis P, Dimitropoulos K, Kaza K, et al. Fire detection from images using faster R-CNN and multidimensional texture analysis[C]//2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, 2019: 8301–8305. https://doi.org/10.1109/ICASSP.2019.8682647.
6
孙光灵, 王薪博. THI-YOLO: 改进YOLOv8的非机动车驾驶员头盔检测[J]. 光电工程, 2024, 51(12): 240244.
Sun G L, Wang X B. THI-YOLO: improved non-motorized drivers helmet detection of YOLOv8[J]. Opto-Electron Eng, 2024, 51(12): 240244.
7
宋华伟, 屈晓娟, 杨欣, 等. 基于改进YOLOv5的火焰烟雾检测[J]. 计算机工程, 2023, 49(6): 250−256.
Song H W, Qu X J, Yang X, et al. Flame and smoke detection based on improved YOLOv5[J]. Comput Eng, 2023, 49(6): 250−256.
8
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.
Ma Y M, Zhang P F, Tan F S. Improved YOLOv8s model for smoke and flame detection in complex backgrounds[J]. Comput Eng Appl, 2025, 61(3): 121−130.
9
张小雪, 王雨, 吴思远, 等. 基于级联稀疏查询机制的轻量化火灾检测算法[J]. 光电工程, 2023, 50(10): 230216.
Zhang X X, Wang Y, Wu S Y, et al. An improved lightweight fire detection algorithm based on cascade sparse query[J]. Opto-Electron Eng, 2023, 50(10): 230216.
10
Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//16th European Conference on Computer Vision, Glasgow, 2020: 213–229. https://doi.org/10.1007/978-3-030-58452-8_13.
11
Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, 2017: 6000–6010.
12
Bodla N, Singh B, Chellappa R, et al. Soft-NMS—improving object detection with one line of code[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision, Venice, 2017: 5562–5570. https://doi.org/10.1109/ICCV.2017.593.
13
Meng D P, Chen X K, Fan Z J, et al. Conditional DETR for fast training convergence[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision, Montreal, 2021: 3631–3640. https://doi.org/10.1109/ICCV48922.2021.00363.
14
Zhao Y A, Lv W Y, Xu S L, et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2024: 16965–16974. https://doi.org/10.1109/CVPR52733.2024.01605.
15
Tan M X, Le Q. EfficientNet: rethinking model scaling for convolutional neural networks[C]//Proceedings of the 36th International Conference on Machine Learning, Long Beach, 2019: 6105–6114.
16
Li H L, Li J, Wei H B, et al. Slim-neck by GSConv: A better design paradigm of detector architectures for autonomous vehicles[Z]. arXiv: 2206.02424, 2022. https://arxiv.org/abs/2206.02424v1.
17
Tan M X, Pang R M, Le Q V. EfficientDet: scalable and efficient object detection[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2020: 10778–10787. https://doi.org/10.1109/CVPR42600.2020.01079.
18
Clevert D A, Unterthiner T, Hochreiter S. Fast and accurate deep network learning by exponential linear units (ELUs)[C]//4th International Conference on Learning Representations, San Juan, 2016.
19
He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 2016: 770–778. https://doi.org/10.1109/CVPR.2016.90.
20
Sandler M, Howard A, Zhu M L, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 2018: 4510–4520. https://doi.org/10.1109/CVPR.2018.00474.
21
Glorot X, Bordes A, Bengio Y. Deep sparse rectifier neural networks[C]//Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, Fort Lauderdale, 2011: 315–323.
22
Chollet F. Xception: deep learning with depthwise separable convolutions[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, 2017: 1800–1807. https://doi.org/10.1109/CVPR.2017.195.
23
Zhang X Y, Zhou X Y, Lin M X, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 2018: 6848–6856. https://doi.org/10.1109/CVPR.2018.00716.
24
Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, 2017: 936–944. https://doi.org/10.1109/CVPR.2017.106.
25
Liu X Y, Peng H W, Zheng N X, et al. EfficientViT: Memory efficient vision transformer with cascaded group attention[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, 2023: 14420–14430. https://doi.org/10.1109/CVPR52729.2023.01386.
26
Han K, Wang Y H, Tian Q, et al. GhostNet: More features from cheap operations[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, 2020: 1577–1586. https://doi.org/10.1109/CVPR42600.2020.00165.
27
Wang A, Chen H, Lin Z J, et al. RepViT: revisiting mobile CNN from ViT perspective[Z]. arxiv: 2307.09283, 2023. https://doi.org/10.48550/arXiv.2307.09283.
2026年第53卷第1期
PDF下载
110
42
引用本文
BibTeX
文章信息
doi: 10.12086/oee.2026.250152
  • 接收时间：2025-05-26
  • 首发时间：2026-10-22
  • 出版时间：2026-01-25
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期：2025-05-26
  • 修回日期：2025-09-04
  • 录用日期：2025-09-08
基金
作者信息
    1安徽建筑大学电子与信息工程学院，安徽 合肥 230601
    2安徽建筑大学安徽省古建筑智能感知与高维建模国际联合研究中心，安徽 合肥 230601

通讯作者:

参考文献
分享链接
https://castjournals.cast.org.cn/joweb/oee/CN/10.12086/oee.2026.250152
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数
科
Family		 属数
Number of
genus		 种数
Number of
species		 占总种数比例
Percentage of
total species (%)		 属
Genus		 种数
Number of
species		 占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
相关链接
论文
最新文章
热读文章
热点专题
内参
蓝皮书
产业发展报告
传播力报告
期刊分级目录
资讯
学术新闻
行业新闻
学术会议
行业会议
关于
关于我们
联系我们
北京市海淀区学院南路86号
100081
010-62662699（运营及技术支持）
qkjq@cast.org.cn

中国科学技术协会版权所有 京ICP 备10216604号-4 海淀分局备案 1101084647
科技导报社主办 中国科协信息中心技术支持