|
|
基于通道特征增强的火灾视频识别
|
Abstract:
火灾对全球人民的生命财产安全造成了巨大的威胁。在火灾检测领域中，使用计算机视觉技术检测火灾对保障人民的生命和财产安全具有重要意义。针对经典的火灾识别方法无法高效地利用火焰运动特征的问题，提出基于通道特征增强的Video Swin Transformer (Video Swin Transformer based on Channel Feature Enhancement, VST-CFE)网络。VST-CFE主要包含Video Swin Transformer (VST)块和通道特征增强(Channel Feature Enhancement, CFE)块。为了利用在三维窗口划分时VST块丢失的火焰运动信息，设计了CFE块。通过建立通道信息的语义模型，CFE块增强了描述火焰运动的能力，从而提升了VST-CFE网络识别火焰的准确率。在LVFD数据集上开展大量的实验，实验结果表明VST-CFE优于基准方法VST。在该数据集上，VST-CFE的F1分数是88.16%，比基准方法VST的F1分数提高了1.75%。
Fires pose a huge threat to the safety of people’s lives and property around the world. In the field of fire detection, the usage of computer vision technology to detect fires is of great significance for ensuring the safety of people’s lives and property. Aiming at the problem that classic fire recognition methods cannot efficiently utilize the motion feature of flames, a Video Swin Transformer based on Channel Feature Enhancement (VST-CFE) network is proposed. VST-CFE mainly includes the Video Swin Transformer (VST) block and the Channel Feature Enhancement (CFE) block. To utilize the motion information of flames lost in the VST block during 3D window partitioning, the CFE block is designed. By establishing the semantic model of channel information, the CFE block enhances the ability to describe flame motion, thereby improving the accuracy of the VST-CFE network in recognizing flames. Extensive experiments are conducted on the LVFD dataset, and the experimental results demonstrate that the VST-CFE method outperforms the baseline method VST. On this dataset, the F1 score of VST-CFE is 88.16%, which is 1.75% higher than the F1 score of the baseline method.
| [1] | Vaswani, A., Shazeer, N., Parmar, N., et al. (2017) Attention Is All You Need. Proceedings of the 31st International Conference on Neural Information Processing Systems, December 2017, 6000-6010. https://dl.acm.org/doi/10.5555/3295222.3295349 |
| [2] | Liu, Z., Lin, Y., Cao, Y., et al. (2021) Swin Transformer: Hierarchical Vision Transformer Using Shifted Windows. Proceedings of the IEEE/CVF International Conference on Computer Vision, Montreal, 10-17 October 2021, 10012-10022. https://doi.org/10.1109/ICCV48922.2021.00986 |
| [3] | Liu, Z., Ning, J., Cao, Y., et al. (2022) Video Swin Transformer. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, 18-24 June 2022, 3202-3211. https://doi.org/10.1109/CVPR52688.2022.00320 |
| [4] | Jadon, A., Omama, M., Varshney, A., et al. (2019) Firenet: A Specialized Lightweight Fire & Smoke Detection Model for Real-Time Iot Applications. https://arxiv.org/abs/1905.11922 |
| [5] | Shees, A., Ansari, M.S., Varshney, A., et al. (2023) Firenet-V2: Improved Lightweight Fire Detection Model for Real-Time Iot Applications. Procedia Computer Science, 218, 2233-2242. https://doi.org/10.1016/j.procs.2023.01.199 |
| [6] | Aliser, A. and Duranay, Z.B. (2024) Fire/Flame Detection with Attention-Based Deep Semantic Segmentation. Iranian Journal of Science And Technology, Transactions of Electrical Engineering, 1-13. https://doi.org/10.1007/s40998-024-00697-y |
| [7] | Li, R., Hu, Y., Li, L., et al. (2024) SMWE-Gfpnnet: A High-Precision and Robust Method for Forest Fire Smoke Detection. Knowledge-Based Systems, 289, Article ID: 111528. https://doi.org/10.1016/j.knosys.2024.111528 |
| [8] | Jin, L., Yu, Y., Zhou, J., et al. (2024) SWVR: A Lightweight Deep Learning Algorithm for Forest Fire Detection and Recognition. Forests, 15, 204. https://doi.org/10.3390/f15010204 |
| [9] | Li, B., Xu, F., Li, X., et al. (2024) Early Stage Fire Detection System Based on Shallow Guide Deep Network. Fire Technology, 1-19. https://doi.org/10.1007/s10694-024-01549-1 |
| [10] | Lin, Q., Li, Z., Zeng, K., et al. (2024) Firematch: A Semi-Supervised Video Fire Detection Network Based on Consistency and Distribution Alignment. Expert Systems with Applications, 248, Article ID: 123409. https://doi.org/10.1016/j.eswa.2024.123409 |
| [11] | Zheng, H., Wang, G., Xiao, D., et al. (2024) FTA-DETR: An Efficient and Precise Fire Detection Framework Based on an End-to-End Architecture Applicable to Embedded Platforms. Expert Systems with Applications, 248, Article ID: 123394. https://doi.org/10.1016/j.eswa.2024.123394 |
| [12] | Liu, J., Yin, J. and Yang, Z. (2024) Fire Detection and Flame-Centre Localisation Algorithm Based on Combination of Attention-Enhanced Ghost Mode and Mixed Convolution. Applied Sciences, 14, 989. https://doi.org/10.3390/app14030989 |
| [13] | Kim, H.C., Lam, H.K., Lee, S.H., et al. (2024) Early Fire Detection System by Using Automatic Synthetic Dataset Generation Model Based on Digital Twins. Applied Sciences, 14, 1801. https://doi.org/10.3390/app14051801 |
| [14] | El-Madafri, I., Pe？a, M. and Olmedo-Torre, N. (2024) Dual-Dataset Deep Learning for Improved Forest Fire Detection: A Novel Hierarchical Domain-Adaptive Learning Approach. Mathematics, 12, 534. https://doi.org/10.3390/math12040534 |
| [15] | Xu, Y., Li, J., Zhang, L., et al. (2024) CNTCB-Yolov7: An Effective Forest Fire Detection Model Based on Convnextv2 and CBAM. Fire, 7, 54. https://doi.org/10.3390/fire7020054 |
| [16] | Woo, S., Debnath, S., Hu, R., et al. (2023) Convnext V2: Co-Designing and Scaling Convnets with Masked Autoencoders. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, BC, 17-24 June 2023, 16133-16142. https://doi.org/10.1109/CVPR52729.2023.01548 |
| [17] | Wang, C.Y., Bochkovskiy, A. and Liao, H.Y.M. (2023) Yolov7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, BC, 17-24 June 2023, 7464-7475. https://doi.org/10.1109/CVPR52729.2023.00721 |
| [18] | 陈庆典, 钟晨, 刘慧, 等. 基于Firenet的古建筑火灾检测方法研究及改进[J]. 消防科学与技术, 2024, 43(2): 183-188. |
| [19] | Bertasius, G., Wang, H. and Torresani, L. (2021) Is Space-Time Attention All You Need for Video Understanding? ICML, 2, 4. |
