Chinese Journal of Liquid Crystals and Displays, Volume. 38, Issue 1, 118(2023)
Lightweight smoke and fire detection algorithm based on efficient global context network
Fig. 1. Missing and false detection results using YOLOX-based smoke and fire detection algorithm.（a）Missing detection result；（b），（c），（d）False detection results.
Fig. 4. Grad-CAM results for image classification.（a）Original image；（b）Result of ResNet18；（c）Result of EGC-ResNet18.
Fig. 5. Grad-CAM results for object detection.（a）Input image；（b）Result of YOLOX；（c）Result of EGC-YOLOX.
Fig. 6. Effect comparison for smoke and fire detection.（a），（c）Results of YOLOX；（b），（d）Corresponding results of EGC-YOLOX.
Fig. 7. Results of the proposed algorithm to detect smoke and fire in real-world images
Table 1. Performance comparison for lightweight algorithms of object detection
View tableView in ArticleTable 1. Performance comparison for lightweight algorithms of object detection
对比算法 ILR/% ILFAR/% 大小/MB 耗时/ms YOLOv3-Tiny 84.36 12.63 16.5 13 YOLOv4-Tiny 84.48 14.22 11.7 10 YOLOv5s 87.58 8.56 13.7 16 YOLOX 90.76 9.14 34.3 26 NL-YOLOX 94.19 7.98 36.3 31 GC-YOLOX 94.09 5.24 34.5 29 EGC-YOLOX 95.56 4.75 34.3 26
Get Citation
Copy Citation Text
Lun-sheng WEI, Wang-ming XU, Jing-yuan ZHANG, Bin CHEN. Lightweight smoke and fire detection algorithm based on efficient global context network[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(1): 118
Category: Research Articles
Received: Jun. 1, 2022
Accepted: --
Published Online: Feb. 20, 2023
The Author Email: Wang-ming XU (xuwangming@wust.edu.cn)
