LFNet: An end-to-end lightweight model for real-time fire detection in embedded firefighting systems
Source abstract
Abstract
This study introduces LFNet, a novel lightweight end-to-end model designed for real-time fire detection in embedded firefighting systems. LFNet addresses the challenges of detecting flames in complex scenarios, including blurred object boundaries, smoke occlusion, small targets, and cluttered backgrounds. By leveraging a star-shaped operation and Multi-Level Contextual Attention in the backbone, Partial Convolution (PConv) for feature fusion, and a weighted Bidirectional Feature Pyramid Network for detection head optimization, LFNet achieves high detection accuracy while maintaining low computational demands. The model was trained and evaluated on the LFD dataset, which combines the public DFS dataset with 6000 self-collected real fire images, enhancing robustness in diverse fire scenarios. Experimental results demonstrate that on the LFD dataset, compared with the baseline model, LFNet improves precision by 11.8%, recall by 7.6%, reduces the number of parameters by 30%, decreases the floating-point operations (GFLOPs) by 30%, and increases the frame rate by 60%. LFNet’s efficiency and accuracy make it a promising solution for early fire detection in applications such as forest fire monitoring and industrial fire safety.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Fire Detection and Safety Systems · Image Enhancement Techniques · Fire effects on ecosystems
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Hunan Automotive Engineering Vocational University 1 , Zhuzhou, Hunan 412001,
College of Forestry, Central South University of Forestry and Technology 2 , Changsha 410004,
School of Transportation and Electrical Engineering, Hunan University of Technology 3 , Zhuzhou, Hunan 412007,
Funding
Natural Science Foundation of Hunan Province · Science and Technology Project of Hunan Provincial Sports Bureau
References (69)
Show source-provided references
1. Pedestrian detection using hybrid DWT and HOG in multiple scenarios
2. RCNN architecture for forest fire detection
3. Classification of different vehicles in traffic using RGB and depth images: A fast RCNN approach
4. Op mask R-CNN: An advanced mask R-CNN network for cattle individual recognition on large farms
5. Small object detection algorithm based on improved YOLOv8 for remote sensing10.1109/JSTARS.2023.3339235 ↗
6. YOLOv8-QSD: An improved small object detection algorithm for autonomous vehicles based on YOLOv810.1109/TIM.2024.3379090 ↗
7. A dataset for fire and smoke object detection10.1007/s11042-022-13580-x ↗
8. Rewrite the stars
9. Mixed local channel attention for object detection10.1016/j.engappai.2023.106442 ↗
10. Run, don’t walk: Chasing higher flops for faster neural networks
11. EfficientViT: Memory efficient vision transformer with cascaded group attention
12. Robust little flame detection on real-time video surveillance system
13. Flame detection algorithm based on a saliency detection technique with prior information in the YCbCr color space
14. Fire flame detection based on color model and motion estimation
15. Computer vision and smoke sensor based fire detection system
16. Research on flame detection algorithm based on multi - feature fusion
17. An electric fire flame detection algorithm based on temporal convolution
18. Intelligent multi-sensor detection system for monitoring indoor building fires10.1109/JSEN.2021.3124266 ↗
19. Optimization of a fire detection system based on radial sector scanning using a UV sensor10.1109/LSENS.2022.3225436 ↗
20. A UAV forest fire detection method based on dual-light vision images
21. Smoke and flame identification method for the entire process of grassland fire based onYOLOv5m-D and static and dynamic characteristics10.1109/ACCESS.2023.3346504 ↗
22. Self-extinction characteristics of fire extinguishing induced by nitrogen injection rescue in an enclosed urban utility tunnel10.1016/j.csite.2024.104478 ↗
23. Sniper: Efficient multi-scale training
24. Video flame detection method based on twostream convolutional neural network
25. Faster R-CNN based indoor flame detection for firefighting robot
26. Dynamic convolution: Attention over convolution kernels
27. A hierarchical deep convolutional neural network and gated recurrent unit framework for structural damage detection10.1016/j.ins.2020.05.090 ↗
28. Dynamic head: Unifying object detection heads with attentions
29. Diverse branch block: Building a convolution as an inception-like unit
30. ATTt squeeze U-Net: A lightweight network for forest fire detection and recognition10.1109/ACCESS.2021.3050628 ↗
31. A data-driven structural damage detection framework based on parallel convolutional neural network and bidirectional gated recurrent unit10.1016/j.ins.2021.02.064 ↗
32. H. Wei , “DWRSeg: Rethinking efficient acquisition of multi-scale contextual information for real-time semantic segmentation,” arXiv:2212.01173 (2022).
33. A unified deep learning framework of multi-scale detectors for geo-spatial object detection in high-resolution satellite images10.1007/s13369-021-06288-x ↗
34. MogaNet: Multi-order gated aggregation network
35. AFPN: Asymptotic feature pyramid network for object detection
36. An emerging fire detection system based on convolutional neural network and aerial-based forest fire identification
37. Fire sensor and surveillance camera-based GTCNN for fire detection system10.1109/JSEN.2023.3244833 ↗
38. MMFNet: Forest fire smoke detection using multiscale convergence coordinated pyramid network with mixed attention and fast-robust NMS10.1109/JIOT.2023.3277511 ↗
39. A robust infrared small target detection method jointing multiple information and noise prediction: Algorithm and benchmark10.1109/TGRS.2023.3295932 ↗
40. TL-YOLOv8: A blueberry fruit detection algorithm based on improved YOLOv8 and transfer learning10.1109/access.2024.3416332 ↗
41. D. Qin , “MobileNetV4–universal models for the mobile ecosystem,” arXiv:2404.10518 (2024).10.1007/978-3-031-73661-2_5 ↗
42. DETRs beat YOLOs on real-time object detection
43. VanillaNet: the power of minimalism in deep learning
44. Transmission line insulator condition detection based on improved lightweight YOLOv8n algorithm10.21203/rs.3.rs-4880334/v1 ↗
45. A lightweight convolutional neural network flame detection algorithm
46. C. Li , “YOLOV6: A single-stage object detection framework for industrial applications,” arXiv:2209.02976 (2022).
47. Characteristics based fire detection system under the effect of electric fields with improved YOLO-v4 and ViBe10.1109/ACCESS.2022.3190867 ↗
48. A defect detection method for a boiler inner wall based on an improved YOLO-v5 network and data augmentation technologies10.1109/ACCESS.2022.3204683 ↗
49. Flame detection algorithm based on YOLOv4
50. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
51. YOLO-SF: YOLO for fire segmentation detection10.1109/ACCESS.2023.3322143 ↗
52. Intelligent flame detection of heating furnace based on improved YOLO v5
53. Flame detection based on improved YOLOV5 network model
54. C.-Y. Wang , I.-H.Yeh, and H.-Y. M.Liao, “YOLOv9: Learning what you want to learn using programmable gradient information,” arXiv:2402.13616 (2024).10.1007/978-3-031-72751-1_1 ↗
55. A. Wang , “YOLOv10: Real-time end-to-end object detection,” arXiv:2405.14458 (2024).
56. DCGC-YOLO: The efficient dual-channel bottleneck structure YOLO detection algorithm for fire detection10.1109/ACCESS.2024.3385856 ↗
57. Smoke and fire detection based on YOLOv7 with convolutional structure reparameterization and lightweighting10.1109/LSENS.2024.3420883 ↗
58. YOLO-ULNet: Ultralightweight network for real-time detection of forest fire on embedded sensing devices10.1109/JSEN.2024.3416548 ↗
59. Improved lightweight flame smoke detection algorithm for YOLOv8n
60. Using popular object detection methods for real time forest fire detection
61. Fire-YOLO: A small target object detection method for fire inspection10.3390/su14094930 ↗
62. Flame image recognition detection based on improved YOLOv7
63. An improved real-time detection method for flame and smoke identification based on YOLOv5
64. YOLO-drone: An optimized YOLOv8 network for tiny UAV object detection10.3390/electronics12173664 ↗
65. YOLO-SE: Improved YOLOv8 for remote sensing object detection and recognition10.3390/app132412977 ↗
66. YOLOv8-cab: Improved YOLOv8 for real-time object detection10.33640/2405-609x.3339 ↗
67. YOLO-MIF: Improved YOLOv8 with multi-information fusion for object detection in gray-scale images10.1016/j.aei.2024.102709 ↗
68. YOLOv8-QSD: An improved small object detection algorithm for autonomous vehicles based on YOLOv810.1109/tim.2024.3379090 ↗
69. EfficientDet: Scalable and efficient object detection