Research on Traffic Acoustic Event Detection Algorithm Based on Sparse Autoencoder

Abstract

Road traffic monitoring is very important for intelligent transportation. The detection of traffic state based on acoustic information is a new research direction. A vehicles acoustic event classification algorithm based on sparse autoencoder is proposed to analysis the traffic state. Firstly, the multidimensional Mel-cepstrum features and energy features are extracted to form a feature vector of 125 features; Secondly, based on the computed features, the five-layers autoencoder is trained. Finally, vehicle audio samples are collected and the trained autoencoder is tested. The experimental results show that detection rate of the traffic acoustic event reaches 94.9%, which is 12.3% higher than that of the traditional Convolutional Neural Networks (CNN) algorithm.

Full text

aCorresponding author: zhangdaqing_925@163.com
Research on Traffic Acoustic Event Detection Algorithm Based on Sparse
Autoencoder
Xiaodan Zhang1,a, and Yongsheng Chen1and Guichen Tang2
1Research Institute of Highway, Ministry of Transport, Beijing 100088, China
2School of Communication Engineering, Nanjing Institute of Technology, Jiangsu Nanjing, 211167, China
Abstract. Road traffic monitoring is very important for intelligent transportation. The detection of traffic state based
on acoustic information is a new research direction. A vehicles acoustic event classification algorithm based on sparse
autoencoder is proposed to analysis the traffic state. Firstly, the multidimensional Mel-cepstrum features and energy
features are extracted to form a feature vector of 125 features; Secondly, based on the computed features, the five-
layers autoencoder is trained. Finally, vehicle audio samples are collected and the trained autoencoder is tested. The
experimental results show that detection rate of the traffic acoustic event reaches 94.9%, which is 12.3% higher than
that of the traditional Convolutional Neural Networks (CNN) algorithm.
1 Introduction
In order to provide auxiliary research for intelligent
transportation and traffic safety development, an
autoencoder based the acoustic feature for traffic event
detection is proposed. In order to integrate the dynamic
features of the sound, the algorithm extracts the
multidimensional Mel-cepstrum features and energy
features, and forms a 110 dimension feature vector. Then
the 5-layers autoencoder based on the computed features
is trained to improve the robustness. The experimental
results show that the detection rate of traffic acoustic
events reaches 94.9%, and the recognition rate of
collision sounds reaches 97. 9%. The proposed audio
surveillance system may be used to monitor traffic
accidents and save valuable time in rescue mission. In
addition, the system can be embedded in the automatic
driving system, which is conducive to the timely response
of the self-driving car to the traffic state, greatly
improving safety.
The detection of traffic state based on acoustic
information has been an important research direction for
intelligent transportation. Compared to existing
monitoring techniques, acoustic signal processing and
classification techniques have the advantage of being low
cost and unaffected by lighting conditions. Especially in
the case of insufficient light or intermediate obstructions,
acoustic signals have higher information coverage.
Therefore it is an important supplement to existing
monitoring methods. However, compared with the
laboratory environment, the real traffic environment is
complex. For example, the tunnel is a special traffic
environment, that is very different from the open road
environment. How to effectively process the traffic
acoustic data remains a challenge. In 1998, Henryk
Maciejewski et al.[1] studied and designed a
classification system based on wavelet and neural
network. The specific recognition model based the sound
signal was constructed for four different vehicles, and the
recognition accuracy was 73.68%. Audi Ovox et al.
applied sound recognition technology to the field of
intelligent transportation[2] and used voice recognition
technology in the car phones. Xianglong Luo et al.[3]
used empirical mode decomposition (EMD) and support
vector machine (SVM) to identify the vehicle state. In
recent years, some scholars have tried to apply
convolutional neural networks (CNN) to recognize sound
event[4]. Compared with traditional classifiers,
convolutional neural networks have greatly improved
recognition rate and recognition speed. The ConvNet
model[5] has improved the accuracy of nearly 20% on
Esc-50 database. The LSTM+CNN model proposed by
Bae et al. achieved an 84.1% accuracy rate in the
DCASE2016 competition[6]. However, there are still
some problems when the traditional CNN model is
applied to sound event recognition. The CNN adopts a
serial stack structure. The convolutional layer in the
network transforms the low-level feature map into a high-
level feature map layer by layer. Such a network structure
will result in the low-level feature information loss in the
final extracted features.
For the above mentioned problems, a sparse
autoencoder based on acoustic features is proposed to
classify the vehicle state. The original acoustic features
are multidimensional Mel-cepstrum features and energy
features. The autoencoder generates encoded features to
classify the vehicle state from the original acoustic
features, which improve the robustness of the algorithm.
To verify the performance of the proposed algorithm, we
collected a total of 829 samples for experiment, including
three types of data: engine running (normal driving),
brake and crash. Compared with four algorithms, the
recognition rate based on the proposed method can reach
94.9%, which is 11.45% higher than that of the traditional
MATEC Web of Conferences 308, 05002 (2020) https://doi.org/10.1051/matecconf/202030805002
ICTTE 2019
© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (http://creativecommons.org/licenses/by/4.0/).

MATEC Web of Conferences
classification algorithm or 12.3% higher than the
traditional CNN algorithm.
2 Acoustic Features
For traffic acoustic event detection, three types of
vehicles states are important, which are engine running
(normal driving) state, crake state and crash state. The
waveform and spectrogram of three kinds of signal are
shown in Figure 1. From the graph, three states have
some obvious difference. For example, the waveform of
the running signal is flat and its frequency range is below
800Hz. In addition, the waveform and the spectrogram of
the crash signal have the obvious change. After the
analysis, the valid features should be selected to reflect
these differences.
Figure 1.The representation of three types of signals
Among many acoustic features, the Mel-Cepstral
feature is widely used in sound event classification and
speaker recognition. The Mel-Cepstral is a spectral
feature which is calculated base on the non-linear
relationship between the human ear's auditory
characteristics and signal frequency. However, the
standard Mel-Cepstral only reflects the static
characteristics of speech parameters. The dynamic
characteristics of speech can be described by the
differential spectrum of these features. The extracted
features are differentiated to further strip the features, and
obtain the information such as type and speed changes of
the acoustic event. Combining the Mel-Cepstral feature
with its difference can improve the recognition
performance of the system.
In addition, different types of sounds have different
energy values and change trends, so short-time energy
and their statistical features are also computed. Table 1
describes the adopted 110 features.
Table 1. List of Acoustic Features.
No. features
1-95
Mel-Frequency Cepstral Coefficients (MFCC-0 to
MFCC-12), averages of their 1st and 2nd order
differences, their maximums, minimums, ranges,
and standard deviations.
96-110
Short-term energy, averages of its 1st and 2nd order
difference, its maximum, minimum, range, and
standard deviation.
3 Autoencoder for traffic acoustic event
detection
The autoencoder is the neural network composed of
several hidden layers and is an unsupervised learning
algorithm. The general data representation from the input
will be obtained by setting the target value as the input[7].
As shown in Fig.1, the autoencoder includes two parts:
encoder and decoder. A number of hidden layers on the
left of the graph form an encoder. The right hidden layers
form a decoder, which has the same output as the encoder.
In the middle of Figure 2, the output of the middle layer
is the coded feature. In the process of the input
reconfiguration, the data distribution is learned by the
encoding and decoding process. At last, the data is
compressed and coded, so the more compact features[8]
are obtained.
Figure 2. The architecture of autoencoder
For an autoencoder,
i
x
is the input,
i
y
is the output and
θ
represents the parameters. The input
i
x
represents the
110-dimensional audio feature we extracted, and the
output
i
y
is the optimized feature after the sparse
autoencoder, which is also 110-dimensional. During the
training, the goal of parameter optimization is:
2
1
min
N
i i
i
x y



θ(1)
By limiting the expected activation of the hidden unit
to sparsity, a regularization term[9] is added to punish the
deviations between the expected activation degree of the
hidden unit and the target sparsity

. So equation (1) is:
2
1 1
ˆ
min + ( || )
N m
i i j
i j
x y sp
  
 

 
θ(2)
2
MATEC Web of Conferences 308, 05002 (2020) https://doi.org/10.1051/matecconf/202030805002
ICTTE 2019

ICTTE 2019
where
ˆ
( || )
j
sp
 
is the sparsity penalty term, which is
computed as:
1
ˆ
( || ) log (1 )log
ˆ ˆ
1
j
j j
sp
 
   
 

   
(3)
Here,
ˆ
j

is the average activation degree for all
neurons in the hidden layer,

is sparsity,

is the
penalty coefficient, and
m
is the number of all neurons.
By adding the sparsity limitation[10], most of the neurons
in the autoencoder are suppressed, only a small number
of neurons are active, which reduces the redundancy[11]
of the network and increases the robustness of the model.
Table2. Parameters Setting.
Component
No. Layer
#Neurons
Activation
Encoder
1 Fully Connected
Layer
110 ReLU
2 Fully Connected
Layer
96 ReLU
3 Fully Connected
Layer
64 ReLU
Decoder
4 Fully Connected
Layer
96 ReLU
5 Fully Connected
Layer
110 -
The 5-layers autoencoder is constructed for learning
the latent feature representation of traffic acoustic event,
and its parameter settings are shown in Table 2. The
workflow of the proposed autoencoder is shown in Fig.3.
The training process is shown in Fig.3(a). At first, the
autoencoder is trained to study the latent representation of
the original acoustic features from the training set. The
learning strategy is to reconstruct input signals to obtain
the latent representation, and use gradient descent method
to minimize the error between the reconstructed signal
and the input signal. When the deviation reaches the set
requirements, a trained autoencoder is constructed. Then,
a softmax classifier is added to compute the deviation
between its output and the real labels. The gradient of the
deviation is calculated and the back propagation
algorithm is used to tune the parameters of each layer.
After the training is finished, its performance of trained
autoencoder can be estimated by the test data. The test
flow is shown in Fig.3(b).
(a) The training process
(b) The test process
Figure 3. The workflow of the proposed model
4 Experimental analysis
4.1. Experimental data and parameter setting
In this experiment, a total of 829 samples were collected
for the three types of sound, such as engine
running(normal driving), braking and crashing. Among
them, there are 442 engine running (normal driving)
sounds, 176 brake sounds and 211 crash sounds.
In order to improve the effectiveness of the algorithm,
the voice activity detection is firstly done on the sample
to remove the silent segment. The sample is then
resampled at a frequency of 16000 Hz. Next, the sample
is framed and the FFT is calculated, the number of FFT
points is 512, and the frame overlap rate is 50%.
The current general deep learning framework Caffe is
used to build and train the network. Because the selection
of hyperparameters in neural networks has a great impact
on the training and the convergence state of the network,
the final network hyperparameters are determined
through multiple experiments and comparisons.
The comparison classifier and its parameters are: 1)
Random forest[12], the maximum depth is 6, and the
number of base estimators is 100; 2) CovNet[13], a two-
layer convolutional layer with convolution kernels of
(110, 6) and (1, 3), with a full layer of 1,000 neurons; 3)
k-nearest neighbors (KNN); 4) the support vector
machine.
4.2 Comparison of state-of-the-art recognition
algorithms
The experimental evaluate the algorithm performance in a
five-fold cross-validation manner. For a more valid
assessment, Random Forest[14], KNN[15], CNN[16] and
ConvNet performed the same experiments and
comparisons on the same dataset.
Table 3 shows the experimental results for the five
classifiers on the data set. Compared with random forest
and KNN, the autoencoder improved the accuracy by
12.2% and 17.4%, respectively. It can be seen that the
performance of autoencoder is better than the traditional
classifier. The reason is that the data samples are very
comprehensive, and there are many recorded data under
various environments, while the generalization ability of
the traditional classifier is bad. In addition, through data
analysis, the traditional classifier has the lowest
recognition rate for the brake event category, the highest
recognition rate for the driving event category, followed
by the collision event category. So, it may also be caused
by the fact that the brake data is too small, and the data
needs to be supplemented later for further verification.
Compared with traditional CNN and ConvNet, the
recognition accuracy is improved by 12.3% and 3.9%,
respectively. The possible reason is that the autoencoder
uses encoded features to classify the state. These encoded
features are more robust than the original features.
Table 3. Recognition rate of five models on the data set.
Classifier
Overall
recognition rate
Collision
recognition rate
Random Forests
82. 7%
84. 8%
DNN
77. 5%
85. 6%
CNN
82. 6%
90. 1%
ConvNet
91. 0%
92. 3%
Proposed method
94. 9%
97. 9%
3
MATEC Web of Conferences 308, 05002 (2020) https://doi.org/10.1051/matecconf/202030805002
ICTTE 2019

MATEC Web of Conferences
In addition, Table 3 additionally compares the
recognition efficiency of various algorithms for collision
sounds. The experimental results show that the proposed
algorithm's recognition rate of collision sound is 3% more
than the overall recognition rate, reaching 97. 9%. This
shows that the algorithm can be effectively applied to the
traffic incident warning system to realize the timely alarm
function of traffic accidents based on traffic state
detection.
5 Conclusions
In order to provide auxiliary research for intelligent
transportation and traffic safety development, an
autoencoder based the acoustic feature for traffic event
detection is proposed. In order to integrate the dynamic
features of the sound, the algorithm extracts the
multidimensional Mel-cepstrum features and energy
features, and forms a 110 dimension feature vector. Then
the 5-layers autoencoder based on the computed features
is trained to improve the robustness. The experimental
results show that the detection rate of traffic acoustic
events reaches 94.9%, and the recognition rate of
collision sounds reaches 97. 9%. The proposed audio
surveillance system may be used to monitor traffic
accidents and save valuable time in rescue mission. In
addition, the system can be embedded in the automatic
driving system, which is conducive to the timely response
of the self-driving car to the traffic state, greatly
improving safety.
Acknowledgments
The work was supported by Science and Technology
Innovation Project of Research Institute of Highway,
Ministry of Transport (2018-E0021).
References
1. Maciejewski Henryk, Mazurkiewicz Jacek, Skowron
Krzysztof, Walkowiak Tomasz, Neural Networks for
Vehicle Recognition, in Proc. of the 6th International
Conference on Microelectronics for Neural Networks,
Evolutionary and Fuzzy Systems, 1998, pp.292–296.
2. Zhang Dian Ye, Jian Prof Jin, Zhi-Zheng Assoc Prof
Guo. Exploration into Road Traffic Accident
Prevention Research System. China Safety Science
Journal, Vol.17, No.7, 2007, pp.132-138.
3. Luo Xiang Long, Niu. Vehicle recognition by
acoustic signals based on EMD and SVM. Applied
Acoustics, Vol.29, No.3, 2010, pp.178-183.
4. Salamon Justin, Bello Juan. Deep Convolutional
Neural Networks and Data Augmentation for
Environmental Sound Classification. IEEE Signal
Processing Letters, Vol. 99, 2016, pp.1-4.
5. Piczak Karol J. Environmental sound classification
with convolutional neural networks. IEEE
International Workshop on Machine Learning for
Signal Processing, 2015, pp.1-4.
6. H Bae S, I Choi, S Kim N, Acoustic scene
classification using parallel combination of LSTM
and CNN, Proceedings of the Detection and
Classification of Acoustic Scenes and Events, 2016,
pp.11-15.
7. Xu J, Xiang L, Liu Q, et al. Stacked Sparse
Autoencoder (SSAE) for Nuclei Detection on Breast
Cancer Histopathology Images, IEEE Transactions
on Medical Imaging, Vol.35, No.1, 2016, pp.119-130.
8. Chandar A P S, Lauly S, Larochelle H, et al. An
autoencoder approach to learning bilingual word
representations, International Conference on Neural
Information Processing Systems. MIT Press, 2014,
pp.1853-1861.
9. Goodfellow I J, Le Q V, Saxe A M, et al. Measuring
invariances in deep networks, International
Conference on Neural Information Processing
Systems, 2009, pp.646-654.
10. Mairal J, Bach F, Ponce J, et al. Online Learning for
Matrix Factorization and Sparse Coding, Journal of
Machine Learning Research, Vol.11, No.1, 2009,
pp.19-60.
11. Hinton G E, Salakhutdinov R R. Reducing the
Dimensionality of Data with Neural Networks,
Science, 313(5786), 2006, pp.504-507.
12. Phan Huy, Maaß Marco, Mazur Radoslaw, Mertins
Alfred. Random Regression Forests for Acoustic
Event Detection and Classification. IEEE/ACM
Transactions on acoustic Speech & Language
Processing, Vol.23, No.1, 2015, pp.20-31.
13. Garcia-Pedrajas N, Hervas-Martinez C, Munoz-Perez
J. COVNET: a cooperative coevolutionary model for
evolving artificial neural networks. IEEE
Transactions on Neural Networks, Vol.14, No.3,
2003, pp.575-596.
14. Pal, M. Random forest classifier for remote sensing
classification. International Journal of Remote
Sensing, 2005, 26(1):217-222.
15. Zhang M L , Zhou Z H . ML-KNN: A lazy learning
approach to multi-label learning. Pattern Recognition,
2007, 40(7):2038-2048.
16. Wei Y , Zhao Y , Lu C , et al. Cross-Modal Retrieval
With CNN Visual Features: A New Baseline. IEEE
Transactions on Cybernetics, 2017, 47(2):449-460.
4
MATEC Web of Conferences 308, 05002 (2020) https://doi.org/10.1051/matecconf/202030805002
ICTTE 2019