基于迁移学习的火焰图像识别技术研究
|
摘要
针对火灾视频中复杂背景环境下火焰识别问题,提出了一种基于迁移学习的火焰图像智能识别方法。采用以深度学习预训练的模型为基础并经过迁移学习作为特征提取器提出特征,而后进行特征融合并结合传统机器学习分类器方法进行识别的流程。在所提出的流程中采用逻辑回归和Xgboost两种机器学习算法作为最终分类器进行了实验,结果表明识别准确率得到较大的提升。最终识别时只需要输入原始图片,就能够自动得出识别结果,在火灾视频中复杂环境下的火焰识别方面取得了很好的效果。
Aiming at the problem of flame identification under the complex background environment of fire video, a method that employs transfer learning on pre-trained deep learning models to identify the flame in images is proposed. Transfer learning models are used as feature extractors to extract the features in image, and the extracted image features are fused as input to traditional machine learning classifier to identify flame. In the proposed procedure, two machine learning classifiers, logistic regression and Xgboost, are used for experiments. The experimental results show that the flame identification accuracy is improved significantly. In the proposed method, the original images are inputed only, and then it can automatically generate identification results. The proposed method can achieve good results in flame identification under the complex background environment of fire video.Key works:
Aiming at the problem of flame identification under the complex background environment of fire video, a method that employs transfer learning on pre-trained deep learning models to identify the flame in images is proposed. Transfer learning models are used as feature extractors to extract the features in image, and the extracted image features are fused as input to traditional machine learning classifier to identify flame. In the proposed procedure, two machine learning classifiers, logistic regression and Xgboost, are used for experiments. The experimental results show that the flame identification accuracy is improved significantly. In the proposed method, the original images are inputed only, and then it can automatically generate identification results. The proposed method can achieve good results in flame identification under the complex background environment of fire video.Key works:
引文
[1] ?etinA E, Dimitropoulos K, Gouverneur B, et al. Video fire detection-review[J].Digital Signal Processing, 2013,23(6):1827-1843
[2] Chetverikov D, Fazekas S, Haindl M.Dynamic texture as foreground and background[J].Machine Vision and Applications, 2011, 22(5):741-750
[3] 张楠. 基于视频图像的火灾检测与识别方法研究[D]. 广州:华南理工大学,2013
[4] Toulouse T, Rossi L, Celik T, et al. Automatic fire pixel detection using image processing:a comparative analysis of rule-based and machine learning-based methods[J]. Signal Image & Video Processing, 2016, 10(4):647-654
[5] Maksymiv O, Rak T, Menshikova O. Deep convolutional network for detecting probable emergency situations[C]. In: Proceedings of the IEEE 1st International Conference on Data Stream Mining & Processing, Lviv, Ukraine, 2016. 23-27
[6] Frizzi1 S, Kaabi R, Bouchouicha M, et al. Convolutional neural network for video fire and smoke detection[C]. In: Proceedings of the 42nd Annual Conference of the IEEE Industrial Electronics Society, Florence, Italy, 2016.877-882
[7] Tao C, Zhang J, Wang P. Smoke detection based on deep convolutional neural networks[C]. In: Proceedings of the 2016 International Conference on Industrial Informatics-Computing Technology, Intelligent Technology, Industrial Information Integration, Wuhan, China, 2017.150-153
[8] Zheng J, Cao X, Zhang B, et al. Bi-heterogeneous convolutional neural network for UAV-based dynamic scene classification[C]. In: Proceedings of the Integrated Communications, Navigation & Surveillance Conference, Herndon, USA, 2017. 1-28
[9] Yin Z, Wan B, Yuan F, et al. A deep normalization and convolutional neural network for image smoke detection[J].IEEE Access, 2017.5(99):18429-18438
[10] Zhang Q, Xu J, Xu L, et al. Deep convolutional neural networks for forest fire detection[C]. In: Proceedings of the International Forum on Management, Education and Information Technology Application, Guangzhou, China, 2016. 568-575
[11] Chollet F. Xception: deep learning with depthwise separable convolutions[C]. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017. 1800-1807
[12] Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the Inception Architecture for Computer Vision[J]. Computer Science, 2015. 2818-2826
[13] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]. In: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016. 770-778
[2] Chetverikov D, Fazekas S, Haindl M.Dynamic texture as foreground and background[J].Machine Vision and Applications, 2011, 22(5):741-750
[3] 张楠. 基于视频图像的火灾检测与识别方法研究[D]. 广州:华南理工大学,2013
[4] Toulouse T, Rossi L, Celik T, et al. Automatic fire pixel detection using image processing:a comparative analysis of rule-based and machine learning-based methods[J]. Signal Image & Video Processing, 2016, 10(4):647-654
[5] Maksymiv O, Rak T, Menshikova O. Deep convolutional network for detecting probable emergency situations[C]. In: Proceedings of the IEEE 1st International Conference on Data Stream Mining & Processing, Lviv, Ukraine, 2016. 23-27
[6] Frizzi1 S, Kaabi R, Bouchouicha M, et al. Convolutional neural network for video fire and smoke detection[C]. In: Proceedings of the 42nd Annual Conference of the IEEE Industrial Electronics Society, Florence, Italy, 2016.877-882
[7] Tao C, Zhang J, Wang P. Smoke detection based on deep convolutional neural networks[C]. In: Proceedings of the 2016 International Conference on Industrial Informatics-Computing Technology, Intelligent Technology, Industrial Information Integration, Wuhan, China, 2017.150-153
[8] Zheng J, Cao X, Zhang B, et al. Bi-heterogeneous convolutional neural network for UAV-based dynamic scene classification[C]. In: Proceedings of the Integrated Communications, Navigation & Surveillance Conference, Herndon, USA, 2017. 1-28
[9] Yin Z, Wan B, Yuan F, et al. A deep normalization and convolutional neural network for image smoke detection[J].IEEE Access, 2017.5(99):18429-18438
[10] Zhang Q, Xu J, Xu L, et al. Deep convolutional neural networks for forest fire detection[C]. In: Proceedings of the International Forum on Management, Education and Information Technology Application, Guangzhou, China, 2016. 568-575
[11] Chollet F. Xception: deep learning with depthwise separable convolutions[C]. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017. 1800-1807
[12] Szegedy C, Vanhoucke V, Ioffe S, et al. Rethinking the Inception Architecture for Computer Vision[J]. Computer Science, 2015. 2818-2826
[13] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]. In: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016. 770-778