Fire Video Intelligent Monitoring Method Based on Moving Target Enhancement and PRV-YOLO Network.
Saved in:
| Title: | Fire Video Intelligent Monitoring Method Based on Moving Target Enhancement and PRV-YOLO Network. |
|---|---|
| Authors: | Wang, Hongyi1 (AUTHOR) wanghongyi@tiangong.edu.cn, Li, Anjing2 (AUTHOR), Yang, Yang2 (AUTHOR), Zhu, Xinjun1 (AUTHOR), Song, Limei2 (AUTHOR) |
| Source: | Fire Technology. Jul2025, Vol. 61 Issue 4, p1463-1489. 27p. |
| Subjects: | Video monitors, Image intensifiers, Artificial intelligence, Image processing, Smoke, Tunnel ventilation |
| Abstract: | Different from objects with clear boundaries in target detection, the fire and smoke generated by fire are variable in shape and hard to be detected by traditional methods. To detect the fire and smoke accurately and timely, a fire identification method based on moving target enhancement and the PRV-YOLO network was proposed in this work. By considering the motion information of smoke and fire in the video data, a PCLAHE-KNN moving target enhancement algorithm is designed to roughly locate the target in the pre-processing stage. In the recognition stage, the PRV-YOLO network is developed for smoke and fire detection. For PRV-YOLO network, CSPResNeXt module is introduced in the backbone position and the VoVGSCSP module is used in the head position, which improves the detection speed and reduces the computation load of the model. Meanwhile, the priority boundary frame loss function PIoU is proposed to improve the regression speed and the accuracy of the detection model. The experimental results have shown that the proposed method has advantages in fire video monitoring, especially in terms of sensitivity to smoke in the early stages of a fire. [ABSTRACT FROM AUTHOR] |
| Copyright of Fire Technology is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Different from objects with clear boundaries in target detection, the fire and smoke generated by fire are variable in shape and hard to be detected by traditional methods. To detect the fire and smoke accurately and timely, a fire identification method based on moving target enhancement and the PRV-YOLO network was proposed in this work. By considering the motion information of smoke and fire in the video data, a PCLAHE-KNN moving target enhancement algorithm is designed to roughly locate the target in the pre-processing stage. In the recognition stage, the PRV-YOLO network is developed for smoke and fire detection. For PRV-YOLO network, CSPResNeXt module is introduced in the backbone position and the VoVGSCSP module is used in the head position, which improves the detection speed and reduces the computation load of the model. Meanwhile, the priority boundary frame loss function PIoU is proposed to improve the regression speed and the accuracy of the detection model. The experimental results have shown that the proposed method has advantages in fire video monitoring, especially in terms of sensitivity to smoke in the early stages of a fire. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 00152684 |
| DOI: | 10.1007/s10694-024-01650-5 |
