Complex Scenes Fire Object Detection Based on Feature Fusion and Channel Attention

IF 2.9 4区 综合性期刊 Q1 Multidisciplinary Arabian Journal for Science and Engineering Pub Date : 2024-08-25 DOI:10.1007/s13369-024-09471-y
Xinrong Cao, Jincai Wu, Jian Chen, Zuoyong Li
{"title":"Complex Scenes Fire Object Detection Based on Feature Fusion and Channel Attention","authors":"Xinrong Cao, Jincai Wu, Jian Chen, Zuoyong Li","doi":"10.1007/s13369-024-09471-y","DOIUrl":null,"url":null,"abstract":"<p>For recognizing small targets, fire-like objects in fire images, and detecting fires across various scenes, we propose a fire detection method based on feature fusion and channel attention. Most existing fire detection methods have specific application scenarios with poor speed or accuracy. To address the issues of poor accuracy when directly applying existing object detection models and the reduced detection speed when improving models for fire targets, our approach aims to balance accurate fire localization with real-time processing. In the backbone of the model, deformable convolution is used to capture rich image information, and channel attention is employed to enhance features. The feature fusion in the neck achieves better localization of small fire targets. The visualized heatmap results indicate the effectiveness of our improved measures. By simultaneously employing multiple improvement measures, our method achieved satisfactory fire detection performance. Experimental results on a self-annotated dataset demonstrate that the best AP@50 of the model can reach 63.9%, the fastest detection speed can reach 114 FPS, and the F1-score is stable at around 63%. Our method strikes a good balance between detection speed and accuracy.\n</p>","PeriodicalId":8109,"journal":{"name":"Arabian Journal for Science and Engineering","volume":"76 1","pages":""},"PeriodicalIF":2.9000,"publicationDate":"2024-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Arabian Journal for Science and Engineering","FirstCategoryId":"103","ListUrlMain":"https://doi.org/10.1007/s13369-024-09471-y","RegionNum":4,"RegionCategory":"综合性期刊","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"Multidisciplinary","Score":null,"Total":0}
引用次数: 0

Abstract

For recognizing small targets, fire-like objects in fire images, and detecting fires across various scenes, we propose a fire detection method based on feature fusion and channel attention. Most existing fire detection methods have specific application scenarios with poor speed or accuracy. To address the issues of poor accuracy when directly applying existing object detection models and the reduced detection speed when improving models for fire targets, our approach aims to balance accurate fire localization with real-time processing. In the backbone of the model, deformable convolution is used to capture rich image information, and channel attention is employed to enhance features. The feature fusion in the neck achieves better localization of small fire targets. The visualized heatmap results indicate the effectiveness of our improved measures. By simultaneously employing multiple improvement measures, our method achieved satisfactory fire detection performance. Experimental results on a self-annotated dataset demonstrate that the best AP@50 of the model can reach 63.9%, the fastest detection speed can reach 114 FPS, and the F1-score is stable at around 63%. Our method strikes a good balance between detection speed and accuracy.

Abstract Image

查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
基于特征融合和通道关注的复杂场景火灾目标检测
为了识别火灾图像中的小目标和类火物体,以及检测各种场景中的火灾,我们提出了一种基于特征融合和通道关注的火灾检测方法。现有的火灾检测方法大多有特定的应用场景,速度或准确性较差。为了解决直接应用现有物体检测模型精度不高,以及改进火灾目标模型时检测速度降低的问题,我们的方法旨在兼顾准确的火灾定位和实时处理。在模型的骨干部分,使用可变形卷积来捕捉丰富的图像信息,并使用通道注意来增强特征。颈部的特征融合可以更好地定位小型火灾目标。可视化热图结果显示了我们改进措施的有效性。通过同时采用多种改进措施,我们的方法取得了令人满意的火灾检测性能。在自标注数据集上的实验结果表明,模型的最佳 AP@50 可以达到 63.9%,最快检测速度可以达到 114 FPS,F1 分数稳定在 63% 左右。我们的方法在检测速度和准确性之间取得了良好的平衡。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
Arabian Journal for Science and Engineering
Arabian Journal for Science and Engineering 综合性期刊-综合性期刊
CiteScore
5.20
自引率
3.40%
发文量
0
审稿时长
4.3 months
期刊介绍: King Fahd University of Petroleum & Minerals (KFUPM) partnered with Springer to publish the Arabian Journal for Science and Engineering (AJSE). AJSE, which has been published by KFUPM since 1975, is a recognized national, regional and international journal that provides a great opportunity for the dissemination of research advances from the Kingdom of Saudi Arabia, MENA and the world.
期刊最新文献
Statistical Analysis and Accurate Prediction of Thermophysical Properties of ZnO-MWCNT/EG-Water Hybrid Nanofluid Using Several Artificial Intelligence Methods Proposing a New Egg-Shaped Profile to Further Enhance the Hydrothermal Performance of Extended Dimple Tubes in Turbulent Flows Violence Detection Using Deep Learning Effects of Iron Ion Ratios on the Synthesis and Adsorption Capacity of the Magnetic Graphene Oxide Nanomaterials Enhancing Elderly Care with Wearable Technology: Development of a Dataset for Fall Detection and ADL Classification During Muslim Prayer Activities
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1