A Smoke Detection Model Based on Improved YOLOv5

被引:58
作者
Wang, Zhong [1 ]
Wu, Lei [1 ]
Li, Tong [1 ]
Shi, Peibei [1 ]
机构
[1] Hefei Normal Univ, Sch Comp Sci & Technol, Hefei 230601, Peoples R China
基金
中国国家自然科学基金;
关键词
smoke detection; YOLOv5; dynamic anchor; attention mechanism; loss function; CONVOLUTIONAL NEURAL-NETWORK;
D O I
10.3390/math10071190
中图分类号
O1 [数学];
学科分类号
0701 ; 070101 ;
摘要
Fast and accurate smoke detection is very important for reducing fire damage. Due to the complexity and changeable nature of smoke scenes, existing smoke detection technology has the problems of a low detection rate and a high false negative rate, and the robustness and generalization ability of the algorithms are not high. Therefore, this paper proposes a smoke detection model based on the improved YOLOv5. First, a large number of real smoke and synthetic smoke images were collected to form a dataset. Different loss functions (GIoU, DIoU, CIoU) were used on three different models of YOLOv5 (YOLOv5s, YOLOv5m, YOLOv5l), and YOLOv5m was used as the baseline model. Then, because of the problem of small numbers of smoke training samples, the mosaic enhancement method was used to randomly crop, scale and arrange nine images to form new images. To solve the problem of inaccurate anchor box prior information in YOLOv5, a dynamic anchor box mechanism is proposed. An anchor box was generated for the training dataset through the k-means++ clustering algorithm. The dynamic anchor box module was added to the model, and the size and position of the anchor box were dynamically updated in the network training process. Aiming at the problem of unbalanced feature maps in different scales of YOLOv5, an attention mechanism is proposed to improve the network detection performance by adding channel attention and spatial attention to the original network structure. Compared with the traditional deep learning algorithm, the detection performance of the improved algorithm in this paper was is 4.4% higher than the mAP of the baseline model, and the detection speed reached 85 FPS, which is obviously better and can meet engineering application requirements.
引用
收藏
页数:13
相关论文
共 30 条
[1]   Image-based smoke detection using feature mapping and discrimination [J].
Asiri, Norah ;
Bchir, Ouiem ;
Ben Ismail, Mohamed Maher ;
Zakariah, Mohammed ;
Alotaibi, Yousef A. .
SOFT COMPUTING, 2021, 25 (05) :3665-3674
[2]  
Chen J., 2013, VISUAL COMMUNICATION, P1, DOI DOI 10.1109/VCIP.2013.6706406
[3]   Global2Salient: Self-adaptive feature aggregation for remote sensing smoke detection [J].
Chen, Shikun ;
Cao, Yichao ;
Feng, Xiaoqiang ;
Lu, Xiaobo .
NEUROCOMPUTING, 2021, 466 :202-220
[4]   Higher Order Linear Dynamical Systems for Smoke Detection in Video Surveillance Applications [J].
Dimitropoulos, Kosmas ;
Barmpoutis, Panagiotis ;
Grammalidis, Nikos .
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2017, 27 (05) :1143-1154
[5]   Smoke Detection on Video Sequences Using Convolutional and Recurrent Neural Networks [J].
Filonenko, Alexander ;
Kurnianggoro, Laksono ;
Jo, Kang-Hyun .
COMPUTATIONAL COLLECTIVE INTELLIGENCE, ICCCI 2017, PT II, 2017, 10449 :558-566
[6]  
Filonenko A, 2017, C HUM SYST INTERACT, P64, DOI 10.1109/HSI.2017.8004998
[7]  
Frizzi S, 2016, IEEE IND ELEC, P877, DOI 10.1109/IECON.2016.7793196
[8]  
Hohberg S.P., 2015, THESIS FREIE U BERLI
[9]  
Joo Cheoi Kyung, 2015, [JOURNAL OF KOREA MULTIMEDIA SOCIETY, 멀티미디어학회논문지], V18, P359, DOI 10.9717/kmms.2015.18.3.359
[10]  
김영진, 2016, [Journal of the Korea Institute Of Information and Communication Engineering, 한국정보통신학회논문지], V20, P1649, DOI 10.6109/jkiice.2016.20.9.1649