A lightweight global awareness deep network model for flame and smoke detection

Abstract

Aiming at the trouble of low detection accuracy and the problem of large model size, this paper proposes a lightweight flame-and-smoke detection model depending on global awareness of images. The proposed method replaces the Conv+BatchNorm+SiLU (CBS) module of original you only look once version 5 (YOLOv5) in the backbone with DSConv+BatchNorm+SiLU (DBS), and the C3 module with GC3, and thus constructs a lightweight backbone network. Besides, involution (InvC3) module is proposed to enhance the global modeling ability and compress the model size, and a module using adaptive receptive fields, named FConv, is proposed to enhance the model’s perception capacity for foreground complex flame-and-smoke information in feature maps. Experimental results show that the proposed model increases the mean average precision of all categories at 0.5 IOU (mAP@0.5) to 70.8%, the mAP@0.5: 0.95 to 39.7%, reduces the number of parameters to 3.57M and the amount of calculation to 7.4 giga floating-point operations per second (GFLOPs) under the premise of ensuring the detection speed. It has been verified that the model can achieve high-precision real-time detection of flame and smoke.

References

[1] GaurA, SinghA, KumarA, et al.. Video flame and smoke based fire detection algorithms: a literature review[J]. Fire technology, 2020, 56: 1943-1980

[2] FoggiaP, SaggeseA, VentoM. Real-time fire detection for video-surveillance applications using a combination of experts based on color, shape, and motion[J]. IEEE transactions on circuits and systems for video technology, 2015, 25(9):1545-1556

[3] WangT, BuL, YangZ, et al.. A new fire detection method using a multi-expert system based on color dispersion, similarity and centroid motion in indoor environment[J]. IEEE/CAA journal of automatica sinica, 2019, 7(1):263-275

[4] XiongW. Research on fire detection and image information processing system based on image processing[C], 2020, New York, IEEE: 106-109

[5] LiC, CaoY, PengY. Research on automatic driving target detection based on YOLOv5s[C], 2022, Bristol, IOP Publishing: 0120472171(1)

[6] LI S, LI L. DRT-Unet: a segmentation network for aiding brain tumor diagnosis[J]. Security & communication networks, 2022.

[7] REN S, HE K, GIRSHICK R, et al. Faster R-CNN towards real-time object detection with region proposal networks[J]. Advances in neural information processing systems, 2015, 28.

[8] LiuW, AnguelovD, ErhanD, et al.. SSD: single shot multibox detector[C], 2016, Cham, Springer International Publishing: 21-37

[9] RedmonJ, DivvalaS, GirshickR, et al.. You only look once: unified, real-time object detection[C], 2016, New York, IEEE: 779-788

[10] LinT Y, GoyalP, GirshickR, et al.. Focal loss for dense object detection[C], 2017, New York, IEEE: 2980-2988

[11] LintY, DollárP, GirshickR, et al.. Feature pyramid networks for object detection[C], 2017, New York, IEEE: 2117-2125

[12] LiuS, QiL, QinH, et al.. Path aggregation network for instance segmentation[C], 2018, New York, IEEE: 8759-8768

[13] RedmonJ, FarhadiA. YOLO9000: better, faster, stronger[C], 2017, New York, IEEE: 7263-7271

[14] REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. (2018-04-08) [2023-01-14]. http://arxiv.org/abs/1804.02767.

[15] BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. (2020-05-28) [2023-01-14]. http://arxiv.org/abs/2004.10934.

[16] ChaoxiaC, ShangW, ZhangF. Information-guided flame detection based on faster R-CNN[J]. IEEE access, 2020, 8: 58923-58932

[17] ShiL, ZhangH F, YangJ F. Video-based fire and smoke detection based on improved SSD[J]. Computer applications and software, 2021, 38(12):161-167(in Chinese)

[18] HuY, ZhanJ, ZhouG, et al.. Fast forest fire smoke detection using MVMNet[J]. Knowledge-based systems, 2022, 241: 108219

[19] CaiW, WangC, HuangH, et al.. A real-time smoke detection model based on YOLO-SMOKE algorithm[C], 2020, New York, IEEE: 1-3

[20] ZHANG S G, ZHANG F, DING Y, et al. Swin-YOLOv5: research and application of fire and smoke detection algorithm based on YOLOv5[J]. Computational intelligence and neuroscience, 2022.

[21] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30.

[22] HanK, WangY, TianQ, et al.. Ghostnet: more features from cheap operations[C], 2020, New York, IEEE: 1580-1589Virtual

[23] HOWARD A G, ZHU M, CHEN B, et al. Mobilenets: efficient convolutional neural networks for mobile vision applications[EB/OL]. (2017-06-18) [2023-01-14]. http://arxiv.org/abs/1704.04861.

[24] LiD, HuJ, WangC, et al.. Involution: inverting the inherence of convolution for visual recognition[C], 2021, New York, IEEE: 12321-12330Virtual

[25] AvenashR, ViswanathP. Semantic segmentation of satellite images using a modified CNN with hard-swish activation function[C], 2019, Seattle, Semantic Scholar: 413-420

[26] MaN, ZhangX, SunJ. Funnel activation for visual recognition[C], 2020, Cham, Springer International Publishing: 351-368

[27] GlorotX, BordesA, BengioY. Deep sparse rectifier neural networks[C], 2011, Cambridge, JMLR: 315-323

[28] MaN, ZhangX, ZhengH T, et al.. Shufflenet v2: practical guidelines for efficient CNN architecture design[C], 2018, Cham, Springer International Publishing: 116-131

[29] HowardA, SandlerM, ChuG, et al.. Searching for mobilenetv3[C], 2019, New York, IEEE/CVF: 1314-1324

[30] Fire-flame-dataset: version 1.0[EB/OL]. (2019-06-28) [2023-01-14]. https://github.com/DeepQuestAI/Fire-Smoke-Dataset.

[31] Fire-smoke-detect-YOLOv4-v5 and fire-smoke-detect-dataset: version 1.0[EB/OL]. (2022-12-03) [2023-01-14]. https://github.com/gengyanlei/fire-smoke-detect-yolov4.