基于自动编码机-分类器的视频交通状态自动识别

doi:10.13229/j.cnki.jdxbgxb20200039

Abstract

Abstract:

In order to recognize road traffic state timely and effectively， a method combining auto-encoder with classifiers was proposed for traffic state recognition from videos. Firstly， traffic video and image data sets were built， based on which auto-encoder structure parameter tests and optimizations were conducted over hidden layer and reduced data dimensionality. Then， a quantitative evaluation method for auto-encoders was put forward， thus the best auto-encoder model was presented as A^*. At last， four traffic state recognition models were constructed by combining A^* with Linear Classifier （LC）， Support Vector Machine （SVM）， Deep Neural Network （DNN） and DNN-LC respectively. The before mentioned models and CNN models including AlexNet， LeNet， GoogLeNet and VGG16 were trained and tested. Results show that， precision and recall of the proposed models are 94.5%~97.1%， and their F1 values are 94.4%~97.1%. Furthermore， AlexNet performs best among the four CNN models， with precision， recall and F1 value equal to 94%. Therefore， combiningA^* and commonly used classifiers may reach or surpass the traffic state recognition effects of complicated CNN models. The proposed methods are convenient for training and testing with low computation cost， which are suitable for traffic state recognition from videos or images.

Key words: engineering of communications and transportation system, traffic status recognition, auto-encoder, traffic video, deep learning

CLC Number:

U491.1

Bo PENG,Yuan-yuan ZHANG,Yu-ting WANG,Ju TANG,Ji-ming XIE. Automatic traffic state recognition from videos based on auto⁃encoder and classifiers[J].Journal of Jilin University(Engineering and Technology Edition), 2021, 51(3): 886-892.

Figures/Tables 14

Fig.1

Fig.2

Fig.3

Table 1

Fig.4

Fig.5

Table 2

Table 3

Lˉ0.005 and slope kD of alternative models"

模型	Ⅰ	Ⅱ	Ⅲ	Ⅳ	Ⅴ	Ⅵ	Ⅶ	Ⅷ
$L ˉ 0.005$ /10^-3	3.00	2.17	2.11	2.17	2.81	2.78	2.46	2.80
$k D$ /10^-7	-1.95	-1.45	-1.38	-1.42	-1.80	-1.78	-1.57	-1.77

Table 3

Table 4

Average evaluation indexes of A*-classifier for each traffic states in data set A2"

模型	$P ˉ$	$R ˉ$	$F ˉ 1$
A^*-SVM	0.962	0.958	0.960
A^*-DNN	0.962	0.961	0.962
A^*-Linear	0.945	0.945	0.944
A^*-DNN_ Linear	0.971	0.971	0.971

Table 4

Fig.6

Fig.7

Fig.8

Fig.9

Table 5

Test effect of CNN models"

模型名称	模型结构	精确率	召回率	$F 1$	训练测试时长/h
AlexNet	5个卷积层、3个池化层、3个全连接层，1个分类层	0.94	0.94	0.94	8.5
LeNet	3个卷积层、2个池化层、1个全连接层，1个分类层	0.82	0.62	0.71	8.3
GoogLeNet	卷积层、池化层等22个网络层，1个分类层	0.37	0.35	0.36	16.8
VGG16	13个卷积层、3个全连接层和5个池化层，1个分类层	0.11	0.33	0.17	16.2
A^*-分类器	5个编码隐藏层+分类器(线性分类、SVM、DNN等)	0.95~0.97	0.95~0.97	0.94~0.97	2.3~2.7

Table 5

References 12

1	Li L, Hu J, Huang Q, et al. A fuzzy hidden Markov model for traffic status classification based on video features[C]∥IEEE Conference on Computational Engineering in Systems Applications, Beijing, China, 2006: 2050-2055.
2	Quiros A R F, Bedruz R A, Uy A C, et al. Machine vision of traffic state estimation using fuzzy logic[C]∥IEEE Region 10 Conference, Marina Bay Sands, Singapore, 2016: 2104-2109.
3	崔华, 袁超, 魏泽发, 等. 利用FCM对静态图像进行交通状态识别[J]. 西安电子科技大学学报:自然科学版, 2017, 44(6): 85-90.
	Cui Hua, Yuan Chao, Wei Ze-fa, et al. Traffic state recognition using state images and FCM[J]. Journal of Xidian University(Natural Science Edition), 2017, 44(6): 85-90.
4	Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks[J]. Advances in Neural Information Processing Systems, 2012, 25(2): 1097-1105.
5	Ostrom Q T, Bauchet L, Davis F G, et al. The epidemiology of gliomain in adults: a "state of the science" review[J]. Neuro-Oncology, 2014, 16(7): 896-913.
6	Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions[C]∥IEEE Conference on Computer Vision and Pattern Recognition(CVPR), Boston, USA, 2015: 1-9.
7	李映, 龚红丽, 梁佳熙, 等. 基于KSVD和PCA的SAR图像目标特征提取[J]. 吉林大学学报:工学版, 2010, 40(5): 1336-1339.
	Li Ying, Gong Hong-li, Liang Jia-xi, et al. SAR image target feature extraction based on KSVD and PCA[J]. Journal of Jilin University(Engineering and Technology Edition), 2010, 40(5): 1336-1339.
8	侯阿临, 廖庆, 靳志娟, 等. 计算全息图的人工神经网络压缩算法[J]. 吉林大学学报:工学版, 2013, 43(): 21-24.
	Hou A-lin, Liao Qing, Jin Zhi-juan, et al. Compression algorithm of computer-generated hologram based on artificial neural network[J]. Journal of Jilin University(Engineering and Technology Edition), 2013, 43(Sup.1): 21-24.
9	Gao S, Zhang Y, Jia K, et al. Single sample face recognition via learning deep supervised autoencoders[J]. IEEE Transactions on Information Forensics and Security, 2015, 10(10): 2108-2118.
10	Zhao C, Wan X, Zhao G, et al. Spectral-spatial classification of hyperspectral imagery based on stacked sparse autoencoder and random forest[J]. European Journal of Remote Sensing, 2017, 50(1): 47-63.
11	Ge P, Ren C X, Dai D Q, et al. Dual adversarial autoencoders for clustering[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(4): 1417-1424.
12	Cheng H, Koc L, Harmsen J, et al. Wide & deep learning for recommender systems[C]∥Conference on Recommender Systems, New York, USA, 2016: 7-10.

Related Articles 15

[1]	Zhen SONG,Jun-liang LI,Gui-qiang LIU. Constant flow prediction method of variable speed hydraulic power source based on deep learning and limitation fuzzy [J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(3): 1106-1110.
[2]	Zhe-ming YUAN,Hong-jie YUAN,Yu-xuan YAN,Qian LI,Shuang-qing LIU,Si-qiao TAN. Automatic recognition and classification of field insects based on lightweight deep learning model [J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(3): 1131-1139.
[3]	Jin-qing LI,Jian ZHOU,Xiao-qiang DI. Learning optical image encryption scheme based on CycleGAN [J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(3): 1060-1066.
[4]	Dian-hai WANG,Xin-yi SHEN,Xiao-qin LUO,Sheng JIN. Offset optimization with minimum average vehicle delay [J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(2): 511-523.
[5]	Hua CHEN,Wei GUO,Jing-wen YAN,Wen-hao ZHUO,Liang-bin WU. A new deep learning method for roads recognition from SAR images [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1778-1787.
[6]	Xian-min SONG,Ming-ye ZHANG,Zhen-jian LI,Xin WANG,Ya-nan ZHANG. Setting of dynamic bus lane and its simulation analysis and evaluation [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1677-1686.
[7]	Hong-wei ZHAO,Xiao-han LIU,Yuan ZHANG,Li-li FAN,Man-li LONG,Xue-bai ZANG. Clothing classification algorithm based on landmark attention and channel attention [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1765-1770.
[8]	Hong-fei JIA,Xin-ru DING,Li-li YANG. Bi-level programming model for optimization design of tidal lane [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(2): 535-542.
[9]	Chao-ying YIN,Chun-fu SHAO,Xiao-quan WANG,Zhi-hua XIONG. Influence of built environment on commuting mode choice considering spatial heterogeneity [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(2): 543-548.
[10]	Da-wei ZHANG,Hai-tao ZHU. An optimization⁃based evacuation model considering pedestrian heterogeneity [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(2): 549-556.
[11]	Yuan-li GU, Yuan ZHANG, Xiao-ping RUI, Wen-qi LU, Meng LI, Shuo WANG. Short⁃term traffic flow prediction based on LSSVMoptimized by immune algorithm [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 1852-1857.
[12]	Yi-ming BIE,Kai JIANG,Ru-ru TANG,Lin-hong WANG,Xin-yu XIONG. Time of interval partition for traffic control at isolated intersection considering impacts of plan transition [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 1844-1851.
[13]	Guo-zhu CHENG, Si-he FENG, Tian-jun FENG. Setting condition of on⁃street parking space occupied vehicle lane [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 1858-1864.
[14]	Quan LIANG,Jian-cheng WENG,Wei ZHOU,Jian RONG. Stability identification of public transport commute passengers based on association rules [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(5): 1484-1491.
[15]	Qian XU,Ying LI,Gang WANG. Pedestrian-vehicle detection based on deep learning [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(5): 1661-1667.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

Comments

Recommended 0

No Suggested Reading articles found!

输入数据维度	降维数据维度	隐藏层数量	模型编号
32×32	3~6，步长为1	6~10，步长为2	1~12
64×64	3~6，步长为1	6~10，步长为2	13~24
128×128	3~6，步长为1	6~10，步长为2	25~36

模型编号	输入数据维度	编码器隐藏层数量	降维数据维度	编码器结构
Ⅰ	32×32	5	3	(32×32,256)、(256,128)、(128,64)、(64,12)、(12,3)
Ⅱ	32×32	5	4	(32×32,256)、(256,128)、(128,64)、(64,12)、(12,4)
Ⅲ	32×32	5	5	(32×32,256)、(256,128)、(128,64)、(64,12)、(12,5)
Ⅳ	32×32	5	6	(32×32,256)、(256,128)、(128,64)、(64,12)、(12,6)
Ⅴ	64×64	5	3	(64×64,256)、(256,128)、(128,64)、(64,12)、(12,3)
Ⅵ	64×64	5	4	(64×64,256)、(256,128)、(128,64)、(64,12)、(12,4)
Ⅶ	64×64	5	5	(64×64,256)、(256,128)、(128,64)、(64,12)、(12,5)
Ⅷ	64×64	5	6	(64×64,256)、(256,128)、(128,64)、(64,12)、(12,6)

Automatic traffic state recognition from videos based on auto⁃encoder and classifiers

RICH HTML

PDF (PC)