Dual Attention Network for the Classification of Road Surface Conditions Based on EfficientNet
-
摘要: 针对现有EfficientNet模型应用于沥青路面状态分类时，卷积操作易导致高层特征信息丢失问题，在现有EfficientNet模型的深层结构中引入一种双注意力机制，包含通道注意力模块和位置注意力模块，借助Sigmoid线性单元（Sigmoid linear unit，SiLU）激活函数和余弦学习率衰减策略，提出一种融合双注意力机制EfficientNet（Dual attention network based on EfficientNet，DAEfficientNet）的沥青路面状态分类方法。首先，建立不同天气下5种沥青路面共5 938张图像作为数据集，积雪样本来自开源数据集（Canadian adverse driving conditions dataset，CADCD）。然后，对所提出模型进行训练，并得到沥青路面图像分类结果。最后，利用准确率（Accuracy）、精确率（Precision）、召回率（Recall）、F1 score和特异度（Specificity），将所提出模型与其他现有卷积神经网络模型进行分类效果对比分析。试验结果表明：所提出模型优于其他对比模型，能准确、有效地对不同天气下的沥青路面状态进行分类。
针对现有EfficientNet模型应用于沥青路面状态分类时，卷积操作易导致高层特征信息丢失问题，在现有EfficientNet模型的深层结构中引入一种双注意力机制，包含通道注意力模块和位置注意力模块，借助Sigmoid线性单元(Sigmoid linear unit, SiLU)激活函数和余弦学习率衰减策略，提出一种融合双注意力机制EfficientNet(Dual attention network based on EfficientNet, DAEfficientNet)的沥青路面状态分类方法。首先，建立不同天气下5种沥青路面共5 938张图像作为数据集，积雪样本来自开源数据集(Canadian adverse driving conditions dataset, CADCD)。然后，对所提出模型进行训练，并得到沥青路面图像分类结果。最后，利用准确率(Accuracy)、精确率(Precision)、召回率(Recall)、F1 score和特异度(Specificity)，将所提出模型与其他现有卷积神经网络模型进行分类效果对比分析。试验结果表明：所提出模型优于其他对比模型，能准确、有效地对不同天气下的沥青路面状态进行分类。
Abstract: Given the drawback of information loss of high-level features caused by convolution operations when the existing EfficientNet is applied to classify asphaltroad surfaces, a novel dual attention mechanism combined two types of attention modules, named channel attention module and position attention module respectively, is introduced to the existing EfficientNet, and the dual attention network based on EfficientNet (DAEfficientNet) is proposed using sigmoid linear unit (SiLU) activation functions and a cosine learning rate decay technique. First, a dataset including 5, 938 images of five types of asphalt road surfaces under various weather conditions is constructed. The snow image samples of asphalt road surfaces are from the open-source dataset named Canadian adverse driving conditions dataset (CADCD). Second, the proposed model is trained and the image classification results are produced. Finally, the accuracy, precision, recall, F1 score, and specificity of the analyzed models are calculated to compare the classification performance between the proposed model and the others previous convolutional neural network models. The experimental results show that the proposed method outperforms the others completing methods and achieves higher accuracy and stronger robustness in the task of classification of the five types of road surface images. -
表 1 试验数据集分布
数据集 类别 干燥 微湿 潮湿 积水 积雪 总数 训练集/张 584 510 1 080 955 1 150 4 279 验证集/张 146 127 270 238 287 1 068 测试集/张 81 70 149 132 159 591 总数/张 811 707 1 499 1 325 1 596 5 938 
下载: 导出CSV
表 2 各模型对测试集的总体分类性能
模型 指标 准确率 精确率 召回率 F1 score 特异度 VGG-16 0.956 0 0.954 0 0.943 8 0.948 3 0.988 8 GoogLeNet 0.950 9 0.954 2 0.942 2 0.947 4 0.987 3 ResNet-34 0.961 1 0.964 6 0.949 2 0.955 7 0.989 9 ResNet-50 0.961 1 0.960 0 0.951 3 0.954 8 0.990 1 MobileNet-V2 0.940 8 0.936 8 0.930 7 0.933 4 0.984 9 ShuffleNet-V2 0.962 8 0.965 6 0.952 0 0.957 8 0.990 3 EfficientNet 0.969 5 0.969 0 0.960 2 0.964 0 0.992 2 DAEfficientNet 0.983 1 0.982 5 0.977 5 0.979 8 0.995 7
下载: 导出CSV
-
[1] 周海超, 梁晨, 杨建, 等. 提升轮胎抗滑水性能的仿生方法[J]. 机械工程学报, 2015, 51(8): 125-130. doi: 10.3901/JME.2015.08.125ZHOU Haichao, LIANG Chen, YANG Jian, et al. Bionic method for improving tire anti-hydroplaning performance[J]. Journal of Mechanical Engineering, 2015, 51(8): 125-130. doi: 10.3901/JME.2015.08.125 [2] 周海超, 王国林, 姜震, 等. 湿滑状态下轮胎路面摩擦特性的数值分析方法[J]. 机械工程学报, 2020, 56(21): 190-198. doi: 10.3901/JME.2020.21.177ZHOU Haichao, WANG Guolin, JIANG Zhen, et al. Numerical analysis method for friction characteristics of tire-pavement[J]. Journal of Mechanical Engineering, 2020, 56(21): 190-198. doi: 10.3901/JME.2020.21.177 [3] MORSY S, SHAKER A, EL-RABBANY A. Multispectral LiDAR data for land cover classification of urban areas[J]. Sensors, 2017, 17(5): 958. doi: 10.3390/s17050958 [4] AKI M, ROJANAARPA T, NAKANO K, et al. Road surface recognition using laser radar for automatic platooning[J]. IEEE Transactions on Intelligent Transportation Systems, 2016, 17(10): 2800-2810. doi: 10.1109/TITS.2016.2528892 [5] SHIN J, PARK H, KIM T. Characteristics of laser backscattering intensity to detect frozen and wet surfaces on roads[J]. Journal of Sensors, 2019, 2019(1): 1-9. [6] BYSTROV A, HOARE E, TRAN T, et al. Sensors for automotive remote road surface classification[C]//2018 IEEE International Conference on Vehicular Electronics and Safety (ICVES), Sep. 12-14, 2018, Madrid, Spain. IEEE, 2018. [7] NAKASHIMA S, ARAMAKI S, KITAZONO Y, et al. Application of ultrasonic sensors in road surface condition distinction methods[J]. Sensors, 2016, 16(10): 1678. doi: 10.3390/s16101678 [8] KALLIRIS M, KANARACHOS S, KOTSAKIS R, et al. Machine learning algorithms for wet road surface detection using acoustic measurements[C]//Proceedings of the 2019 IEEE International Conference on Mechatronics (ICM), Mar. 18-20, 2019, Ilmenau University of Technology, Ilmenau, Germany. IEEE, 2019: 265-270. [9] 卢俊辉, 王建强, 李克强, 等. 基于路面温度和太阳辐射强度的路面状态识别方法[J]. 农业机械学报, 2010, 41(5): 21-23. doi: 10.3969/j.issn.1000-1298.2010.05.005LU Junhui, WANG Jianqiang, LI Keqiang, et al. Road condition detection based on road temperature and solar radiation[J]. Transactions of the Chinese Society of Agricultural Machinery, 2010, 41(5): 21-23. doi: 10.3969/j.issn.1000-1298.2010.05.005 [10] SHINMOTO Y, TAKAGI J, EGAWA K, et al. Road surface recognition sensor using an optical spatial filter[C]//Proceedings of the 1997 IEEE Conference on Intelligent Transportation Systems(ITSC 997), Nov. 09-12, 1997, Boston, Massachusetts, USA. 1997: 1000-1004. [11] YAMADA M, UEDA K, HORIBA I, et al. A study of the road surface condition detection technique for deployment on a vehicle[J]. Transactions of the Institute of Electrical Engineers of Japan, 2004, 124(3): 753-760. [12] JOKELA M, KUTILA M, LE L. Road condition monitoring system based on a stereo camera[C]//Proceedings of the IEEE 5th International Conference on Intelligent Computer Communication and Processing, Aug. 27-29, 2009, Cluj Napoca, Romania. 2009: 423-428. [13] 顾昊, 李勃, 张潇, 等. 基于偏振测量的路面积水积冰检测方法[J]. 电子测量技术, 2011, 34(7): 99-102. https://www.cnki.com.cn/Article/CJFDTOTAL-DZCL201107026.htmGU Hao, LI Bo, ZHANG Xiao, et al. Detection of road surface water and ice based on polarization measurement[J]. Electronic Measurement Technology, 2011, 34(7): 99-102. https://www.cnki.com.cn/Article/CJFDTOTAL-DZCL201107026.htm [14] COLACE L, SANTONI F, ASSANTO G. A near-infrared optoelectronic approach to detection of road conditions[J]. Optics and Lasers in Engineering, 2013, 51(5): 633-636. doi: 10.1016/j.optlaseng.2013.01.003 [15] COLACE L, SANTONI F, ASSANTO G. Optical road-ice detector operating in the near infrared[J]. Electronics Letters, 2013, 49(5): 338-339. doi: 10.1049/el.2012.3704 [16] JONSSON P, CASSELGREN J, THORNBERG B. Road surface status classification using spectral analysis of NIR camera images[J]. IEEE Sensors Journal, 2015, 15(3): 1641-1656. doi: 10.1109/JSEN.2014.2364854 [17] AMTHOR M, HARTMANN B, DENZLER J. Road condition estimation based on spatio-temporal reflection models[C]//Proceedings of the 37th German Conference on Pattern Recognition (GCPR), Oct. 07-10, 2015, Aachen, Germany. 2015: 3-15. [18] 苑会珍, 葛俊锋, 叶林, 等. 基于线偏振度的非接触式路面状态探测方法[J]. 仪表技术与传感器, 2017(8): 89-91. https://www.cnki.com.cn/Article/CJFDTOTAL-YBJS201708023.htmYUAN Huizhen, GE Junfeng, YE Lin, et al. Noncontact road condition detection method based on degree of linear polarization[J]. Instrument Technique and Sensor, 2017 (8): 89-91. https://www.cnki.com.cn/Article/CJFDTOTAL-YBJS201708023.htm [19] LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553): 436-444. doi: 10.1038/nature14539 [20] LIU Weibo, WANG Zidong, LIU Xiaohui, et al. A survey of deep neural network architectures and their applications[J]. Neurocomputing, 2017, 234: 11-26. [21] YANG H, JANG H, JEONG D. Detection algorithm for road surface condition using wavelet packet transform and SVM[C]//Proceedings of the 19th Korea-Japan Joint Workshop on Frontiers of Computer Vision (FCV 2013), Jan. 30-Feb. 01, 2013, Inha Univ, Incheon, South Korea. 2013: 323-326. [22] 万剑, 赵恺, 王维锋. 基于高维特征和RBP神经网络的湿滑道路图像判别方法[J]. 交通信息与安全, 2013, 31(2): 32-35. https://www.cnki.com.cn/Article/CJFDTOTAL-JTJS201302010.htmWAN Jian, ZHAO Kai, WANG Weifeng. Classification of slippery road images based on high-dimensional features and RBP neural network[J]. Journal of Transport Information and Safety, 2013, 31(2): 32-35. https://www.cnki.com.cn/Article/CJFDTOTAL-JTJS201302010.htm [23] ZHAO J, WU H, CHEN L. Road surface state recognition based on SVM optimization and image segmentation processing[J]. Journal of Advanced Transportation, 2017, 2017: 1-21. [24] HAN X, NGUYEN C, YOU S, et al. Single image water hazard detection using FCN with reflection attention units[C]//15th European Conference on Computer Vision (ECCV). Lecture Notes in Computer Science (LNCS), Sept. 08-14, 2018, Munich, Germany. ECCV, 2018: 105-21. [25] NOLTE M, KISTER N, MAURER M. Assessment of deep convolutional neural networks for road surface classification[C]//21st IEEE International Conference on Intelligent Transportation Systems (ITSC), Nov. 04-07, 2018, Maui, Hawaii, USA. IEEE, 2018: 381-386. [26] PAN G, FU L, YU R, et al. Winter road surface condition recognition using a pretrained deep convolutional network[C]//Transporation Research Board 97th Annual Metting, Jan. 07-11, 2018, Washington DC, USA. 2018: 18-00838. [27] PAN G, MURESAN M, YU R, et al. Real-time winter road surface condition monitoring using an improved residual CNN[J]. Canadian Journal of Civil Engineering, 2020. [28] DEWANGAN D K, SAHU S. RCNet: Road classification convolutional neural networks for intelligent vehicle system[J]. Intelligent Service Robotics, 2021, 2021: 1-16. [29] WOO S, PARK J, LEE J, et al. CBAM: convolutional block attention module[C]//15th European Conference on Computer Vision (ECCV). Lecture Notes in Computer Science (LNCS), Sept. 08-14, 2018, Munich, Germany. ECCV, 2018: 3-19. [30] TAN M, LE Q. EfficientNet: rethinking model scaling for convolutional neural networks[J]. 2019, arXiv: 1905.11946. [31] FU J, LIU J, TIAN H, et al. Dual attention network for scene segmentation[C]//32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun. 16-20, 2019, long Beach, California, USA. 2019: 3141-3149. [32] RAMACHANDRAN P, ZOPH B, LE Q. Searching for activation functions[J]. 2017, arXiv: 1710.05941. [33] ELFWING S, UCHIBE E, DOYA K. Sigmoid-weighted linear units for neural network function approximation in reinforcement learning[J]. Neural Networks, 2018, 107(SI): 3-11. [34] HE T, ZHANG Z, ZHANG H, et al. Bag of tricks for image classification with convolutional neural networks[C]//32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun. 16-20, 2019, Long Beach, California, USA. 2019: 558-567. [35] LOSHCHILOV I, HUTTER F. SGDR: Stochastic gradient descent with warm restarts[C/CD]//5th International Conference on Learning Representations (ICLR 2017), Apr. 24-26, 2017, Toulon, France. 2017. [36] PITROPOV M, GARCIA D, REBELLO J, et al. Canadian adverse driving conditions dataset[J]. 2020, arXiv: 2001.10117. [37] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. 2014, arXiv: 1409.1556. [38] SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jun. 07-12, 2015, Boston, Massachusetts, USA. 2015: 1-9. [39] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jun 27-30, 2016, Seattle, Washington, USA. 2015: 770-778. [40] SANDLER M, HOWARD A, ZHU M, et al. MobileNetV2: Inverted residuals and linear bottlenecks[C]//31st IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun. 18-23, 2018, Salt Lake City, Utah, USA. 2018: 4510-4520. [41] MA N, ZHANG X, ZHENG H, et al. ShuffleNet V2: practical guidelines for efficient CNN architecture design[C]//15th European Conference on Computer Vision (ECCV). Lecture Notes in Computer Science (LNCS), Sept. 08-14, 2018, Munich, Germany. ECCV, 2018: 122-138. -
下载:
点击查看大图
图(13) / 表(2)
计量
- 文章访问数: 20
- HTML全文浏览量: 15
- PDF下载量: 0
- 被引次数: 0
京公网安备 11010502036328号 京ICP备17033152号