基于进化ResNet的交通标志识别

doi:10.19678/j.issn.1000-3428.0062885

摘要/Abstract

摘要： 卷积神经网络具有较优的图像特征提取性能，被广泛应用于交通标志识别领域。然而，现有交通标志识别算法通常基于专家经验设计改进的图像特征提取网络，需经历图像预处理和模型调参过程，导致模型的复杂度增大。提出一种基于进化ResNet的交通标志识别算法。将ResNet的构建参数嵌入到进化算法中，在架构搜索空间中以构建块作为基本单位，并将网络深度、卷积层通道数、池化层类型和模块构建顺序作为搜索空间的可变参数，利用交叉、变异等遗传算子执行自适应优化搜索，以确保进化搜索的有效性，同时设计适用于交通标志识别的轻量化网络。在德国交通标志数据集上的实验结果表明，该算法的识别精度达到99.41%，而参数量仅为2.37×10⁶，相比Multi-column DNN、MFC、MFC+ELM等算法，在保证识别精度的同时减少网络参数量。

关键词: 交通标志识别, 卷积神经块, 残差块, 进化算法, 进化神经网络

Abstract: Convolutional Neural Network(CNN) has better image feature extraction performances and is widely used in traffic sign recognition.However, existing traffic sign recognition algorithms are typically based on expert experience to design an improved image feature extraction network, requiring image preprocessing and model parameter adjustments, which increases complexity of the model.This study proposes a traffic sign recognition algorithm based on evolutionary ResNet.The construction parameters of ResNet are embedded into the Evolutionary Algorithms(EAs).In the architecture search space, the building block is the basic unit, and the network depth, number of convolution layer channels, pooling layer type, and module construction order are variable parameters of the search space.In order to ensure the effectiveness of evolutionary search, genetic operators such as crossover and mutation are used to perform an adaptive optimization search, and a lightweight network suitable for traffic sign recognition is designed.The Experimental results on the German traffic sign dataset show that the recognition accuracy of the proposed algorithm is 99.41% and using only 2.37×10⁶ parameters.Compared with Multi-column DNN, MFC and MFC+ELM algorithms, it can ensure the recognition accuracy and reduce the amount of network parameters.

Key words: traffic sign recognition, Convolutional Neural Network(CNN), residual block, Evolutionary Algorithms(EAs), evolutionary neural network

中图分类号:

TP391

谢艺蓉, 马永杰. 基于进化ResNet的交通标志识别[J]. 计算机工程, 2022, 48(10): 262-269.

XIE Yirong, MA Yongjie. Traffic Sign Recognition Based on Evolutionary ResNet[J]. Computer Engineering, 2022, 48(10): 262-269.

http://www.ecice06.com/CN/Y2022/V48/I10/262

图/表 17

参考文献

[1] 刘占文, 赵祥模, 李强, 等.基于图模型与卷积神经网络的交通标志识别方法[J].交通运输工程学报, 2016, 16(5):122-131. LIU Z W, ZHAO X M, LI Q, et al.Traffic sign recognition method based on graphical model and convolutional neural network[J].Journal of Traffic and Transportation Engineering, 2016, 16(5):122-131.(in Chinese)
[2] LI L F, WANG Y R.Improved LeNet-5 convolutional neural network traffic sign recognition[J].International Core Journal of Engineering, 2021, 7(4):114-121.
[3] CHENG P, LIU W, ZHANG Y F, et al.LOCO:local context based Faster R-CNN for small traffic sign detection[C]//Proceedings of International Conference on Multimedia Modeling.Berlin, Germany:Springer, 2018:329-341.
[4] 尉天成, 陈小锋, 殷元亮.基于多尺度卷积神经网络的道路交通标志识别方法研究[J].西北工业大学学报, 2021, 39(4):891-900. WEI T C, CHEN X F, YIN Y L.Research on traffic sign recognition method based on multi-scale convolution neural network[J].Journal of Northwestern Polytechnical University, 2021, 39(4):891-900.(in Chinese)
[5] 马永杰, 程时升, 马芸婷, 等.多尺度特征融合与极限学习机结合的交通标志识别[J].液晶与显示, 2020, 35(6):572-582. MA Y J, CHENG S S, MA Y T, et al.Traffic sign recognition based on multi-scale feature fusion and extreme learning machine[J].Chinese Journal of Liquid Crystals and Displays, 2020, 35(6):572-582.(in Chinese)
[6] 薛之昕, 郑英豪, 肖建, 等.基于多尺度卷积神经网络的交通标志识别[J].计算机工程, 2020, 46(3):261-266. XUE Z X, ZHENG Y H, XIAO J, et al.Traffic sign recognition based on multi-scale convolutional neural network[J].Computer Engineering, 2020, 46(3):261-266.(in Chinese)
[7] LIU Y Q, SUN Y N, XUE B, et al.A survey on evolutionary neural architecture search[EB/OL].[2021-10-08].https://arxiv.org/abs/2008.10937v1.
[8] DARWISH A, HASSANIEN A E, DAS S.A survey of swarm and evolutionary computing approaches for deep learning[J].Artificial Intelligence Review, 2020, 53(3):1767-1812.
[9] STANLEY K O, CLUNE J, LEHMAN J, et al.Designing neural networks through neuroevolution[J].Nature Machine Intelligence, 2019, 1(1):24-35.
[10] STANLEY K O, MIIKKULAINEN R.Evolving neural networks through augmenting topologies[J].Evolutionary Computation, 2002, 10(2):99-127.
[11] REAL E, AGGARWAL A, HUANG Y P, et al.Regularized evolution for image classifier architecture search[EB/OL].[2021-10-08].https://arxiv.org/pdf/1802.01548.pdf.
[12] SUN Y N, XUE B, ZHANG M J, et al.Completely automated CNN architecture design based on blocks[J].IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(4):1242-1254.
[13] SUN Y N, XUE B, ZHANG M J, et al.Automatically designing CNN architectures using the genetic algorithm for image classification[J].IEEE Transactions on Cybernetics, 2020, 50(9):3840-3854.
[14] SUN Y N, WANG H D, XUE B, et al.Surrogate-assisted evolutionary deep learning using an end-to-end random forest-based performance predictor[J].IEEE Transactions on Evolutionary Computation, 2020, 24(2):350-364.
[15] SUN Y N, SUN X, FANG Y H, et al.A novel training protocol for performance predictors of evolutionary neural architecture search algorithms[J].IEEE Transactions on Evolutionary Computation, 2021, 25(3):524-536.
[16] ZHANG H Y, JIN Y C, CHENG R, et al.Efficient evolutionary search of attention convolutional networks via sampled training and node inheritance[J].IEEE Transactions on Evolutionary Computation, 2021, 25(2):371-385.
[17] 马永杰, 刘培培.基于DenseNet进化的卷积神经网络图像分类算法[J].激光与光电子学进展, 2020, 57(24):50-57. MA Y J, LIU P P.Convolutional neural network based on DenseNet evolution for image classification algorithm[J].Laser & Optoelectronics Progress, 2020, 57(24):50-57.(in Chinese)
[18] HE K M, ZHANG X Y, REN S Q, et al.Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:770-778.
[19] 马永杰, 李雪燕, 宋晓凤.基于改进深度卷积神经网络的交通标志识别[J].激光与光电子学进展, 2018, 55(12):250-257. MA Y J, LI X Y, SONG X F.Traffic sign recognition based on improved deep convolution neural network[J].Laser & Optoelectronics Progress, 2018, 55(12):250-257.(in Chinese)
[20] STALLKAMP J, SCHLIPSING M, SALMEN J, et al.The German traffic sign recognition benchmark:a multi-class classification competition[C]//Proceedings of International Joint Conference on Neural Networks.Washington D.C., USA:IEEE Press, 2011:1453-1460.
[21] CIREŞAN D, MEIER U, MASCI J, et al.Multi-column deep neural network for traffic sign classification[J].Neural Networks, 2012, 32:333-338.
[22] 王晓斌, 黄金杰, 刘文举.基于优化卷积神经网络结构的交通标志识别[J].计算机应用, 2017, 37(2):530-534. WANG X B, HUANG J J, LIU W J.Traffic sign recognition based on optimized convolutional neural network architecture[J].Journal of Computer Applications, 2017, 37(2):530-534.(in Chinese)
[23] 宋青松, 张超, 田正鑫, 等.基于多尺度卷积神经网络的交通标志识别[J].湖南大学学报(自然科学版), 2018, 45(8):131-137. SONG Q S, ZHANG C, TIAN Z X, et al.Traffic sign recognition based on multi-scale convolutional neural network[J].Journal of Hunan University (Natural Sciences), 2018, 45(8):131-137.(in Chinese)
[24] 孙伟, 杜宏吉, 张小瑞, 等.基于CNN多层特征和ELM的交通标志识别[J].电子科技大学学报, 2018, 47(3):343-349. SUN W, DU H J, ZHANG X R, et al.Traffic sign recognition method based on multi-layer feature CNN and extreme learning machine[J].Journal of University of Electronic Science and Technology of China, 2018, 47(3):343-349.(in Chinese)

选择文件类型/文献管理软件名称

选择包含的内容