ABSTRACT
A fire detection method based on YOLOv5s fused with Swin Transformer is proposed. This is to address the shortcomings of traditional forest fire detection methods such as poor detection accuracy and low reliability. To achieve real-time detection of forest fires, this paper proposes an improved recognition method for YOLOv5s-SwinT. Based on the application of the Transformer model, it solves the shortcomings of the convolutional neural network such as localization of operation and global feature extraction. It achieves favorable results, but at the same time, there are still disadvantages such as poor detection of small targets. In this paper, the Swin Transformer and YOLOv5 convolutional neural network models are fused and applied to the machine vision task of forest fire detection. The α-IoU loss function is introduced to replace the GIOU loss function, and the CA attention mechanism lightweight module is incorporated into the backbone network to improve the overall network's feature extraction capability as well as to obtain high-quality and highly accurate localized image regions for bounding box generation and prediction, improving problems such as missed detection and poor detection accuracy of small targets. The experimental results show that the improved recognition method incorporating YOLOv5s-SwinT can achieve an mAP value of 74.2% in the forest fire detection task, which is 4.5% higher than YOLOv5s, and the GUI interface is designed to be deployed directly to the PC side to realize real-time fire detection requirements, providing an effective detection method for forest fire detection vision tasks.
- Çetin A E, Dimitropoulos K, Gouverneur B, Video fire detection–review[J]. Digital Signal Processing, 2013, 23(6): 1827-1843.Google ScholarDigital Library
- Yfantis E A. A uav with autonomy, pattern recognition for forest fire prevention, and ai for providing advice to firefighters fighting forest fires[C]//2019 IEEE 9th Annual Computing and Communication Workshop and Conference (CCWC). IEEE, 2019: 0409-0413.Google Scholar
- Celik T. Fast and efficient method for fire detection using image processing[J]. ETRI journal, 2010, 32(6): 881-890.Google Scholar
- Rafiee A, Dianat R, Jamshidi M, Fire and smoke detection using wavelet analysis and disorder characteristics[C]//2011 3rd International conference on computer research and development. IEEE, 2011, 3: 262-265.Google Scholar
- Wang S. J., Pei H. L. Early fire recognition algorithm based on infrared dynamic features of flame images[J]. Modern Electronics Technology,2010,33(08):104-105+110.DOI:10.16652/j.issn.1004-373x.2010.08.044.Google ScholarCross Ref
- Erhan D, Szegedy C, Toshev A, Scalable object detection using deep neural networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2014: 2147-2154.Google Scholar
- Yoo D, Park S, Lee J Y, Attentionnet: Aggregating weak directions for accurate object detection[C]//Proceedings of the IEEE International Conference on Computer Vision. 2015: 2659-2667.Google Scholar
- Najibi M, Rastegari M, Davis L S. G-cnn: an iterative grid based object detector[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 2369-2377.Google Scholar
- Redmon J, Divvala S, Girshick R, You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 779-788.Google Scholar
- Liu W, Anguelov D, Erhan D, Ssd: Single shot multibox detector[C]//European conference on computer vision. Springer, Cham, 2016: 21-37.Google Scholar
- Shen Z, Liu Z, Li J, Dsod: Learning deeply supervised object detectors from scratch[C]//Proceedings of the IEEE international conference on computer vision. 2017: 1919-1927.Google Scholar
- Girshick R, Donahue J, Darrell T, Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2014: 580-587.Google Scholar
- He K, Zhang X, Ren S, Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE transactions on pattern analysis and machine intelligence, 2015, 37(9): 1904-1916.Google ScholarDigital Library
- Girshick R. Fast r-cnn[C]//Proceedings of the IEEE international conference on computer vision. 2015: 1440-1448.Google Scholar
- Ren S, He K, Girshick R, Faster r-cnn: Towards real-time object detection with region proposal networks[J]. Advances in neural information processing systems, 2015, 28.Google Scholar
- Dai J, Li Y, He K, R-fcn: Object detection via region-based fully convolutional networks[J]. Advances in neural information processing systems, 2016, 29.Google Scholar
- Lin T Y, Dollár P, Girshick R, Feature pyramid networks for object detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2117-2125.Google Scholar
- He K, Gkioxari G, Dollár P, Mask r-cnn[C]//Proceedings of the IEEE international conference on computer vision. 2017: 2961-2969.Google Scholar
- Zhang Q, Lin G, Zhang Y, Wildland forest fire smoke detection based on faster R-CNN using synthetic smoke images[J]. Procedia engineering, 2018, 211: 441-446.Google Scholar
- Sun X, Sun L, Huang Y. Forest fire smoke recognition based on convolutional neural network[J]. Journal of Forestry Research, 2021, 32(5): 1921-1927.Google ScholarCross Ref
- Liu Z, Lin Y, Cao Y, Swin transformer: Hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2021: 10012-10022.Google Scholar
Index Terms
- Fusion of YOLOv5s and Swin Transformer for forest fire detection
Recommendations
Forest Fire Detection Based on Improved YOLOv5
AIPR '21: Proceedings of the 2021 4th International Conference on Artificial Intelligence and Pattern RecognitionTraditional forest fire detection based on sensing devices has a small range, low sensitivity and high false alarm rate, while existing forest fire detection methods based on deep learning have poor identification effect and inaccurate identification of ...
Development of neural network committee machines for automatic forest fire detection using lidar
Lidar has considerable potential as an early forest fire detection technique, presenting considerable advantages when compared to the passive detection methods based on infrared cameras currently in common use, due to its higher sensitivity, ability to ...
Forest Fire Detection Algorithm Based on DetNet-FPN Feature Fusion Network
AIPR '22: Proceedings of the 2022 5th International Conference on Artificial Intelligence and Pattern RecognitionThe occurrence of forest fire has caused a large area of forest damage, casualties and economic losses, and forest fire detection is the key to the timely warning of fire. The problem of small target loss still exists in the forest fire detection ...
Comments