CN108875812B - Driver behavior classification method based on branch convolutional neural network - Google Patents

Driver behavior classification method based on branch convolutional neural network Download PDF

Info

Publication number
CN108875812B
CN108875812B CN201810557923.5A CN201810557923A CN108875812B CN 108875812 B CN108875812 B CN 108875812B CN 201810557923 A CN201810557923 A CN 201810557923A CN 108875812 B CN108875812 B CN 108875812B
Authority
CN
China
Prior art keywords
neural network
layer
driver
branch
convolutional neural
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810557923.5A
Other languages
Chinese (zh)
Other versions
CN108875812A (en
Inventor
陈征
赵付舟
杨超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo University of Technology
Original Assignee
Ningbo University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ningbo University of Technology filed Critical Ningbo University of Technology
Priority to CN201810557923.5A priority Critical patent/CN108875812B/en
Publication of CN108875812A publication Critical patent/CN108875812A/en
Application granted granted Critical
Publication of CN108875812B publication Critical patent/CN108875812B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Traffic Control Systems (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention relates to a driver behavior classification method based on a branch convolutional neural network, which comprises the following steps: 1) processing the test sample based on the constructed trunk convolutional neural network to obtain a primary classification result of the driver behavior; 2) and processing the test sample again based on the constructed branch convolutional neural network according to the subdivision requirement of the driver behavior, so as to realize the subdivision of the driver behavior. The deep learning-based classification method for the driver behaviors does not need to manually extract the characteristic parameters related to the types of the drivers, can reduce the calculated amount according to the actual classification requirement, can extract the effective characteristic parameters related to the types of the drivers, and accurately finishes the classification of the driver behaviors.

Description

Driver behavior classification method based on branch convolutional neural network
Technical Field
The invention relates to a driver behavior classification method, in particular to a driver behavior classification method based on a branch convolutional neural network, and belongs to the technical field of road safety research in an intelligent traffic system.
Background
The driving behavior of the driver greatly affects the safety of the road, and most traffic accidents are caused by the wrong behavior of the driver. Different drivers have different driving habits, and therefore identifying the type of driver is an important research direction.
In the traditional driver behavior classification method, a multi-purpose clustering method and a neural network method are adopted, and the clustering method generally needs to manually extract characteristic parameters for representing the types of drivers, so that a large amount of professional preparation work is needed in the early stage to obtain a proper parameter set. The common neural network method has a large calculation amount or can not extract enough effective characteristic parameters, so that the classification effect is influenced.
Disclosure of Invention
In view of the above problems, the present invention aims to provide a driver behavior classification method based on a branch convolutional neural network, which can effectively extract the characteristic parameters of the driver type and then accurately complete the driver behavior classification.
In order to achieve the purpose, the invention adopts the following technical scheme: a driver behavior classification method based on a branch convolutional neural network comprises the following steps:
1) processing the test sample based on the constructed trunk convolutional neural network to obtain a primary classification result of the driver behavior;
2) and processing the test sample again based on the constructed branch convolutional neural network according to the subdivision requirement of the driver behavior, so as to realize the subdivision of the driver behavior.
Further, the specific process of the step 1) is as follows:
1.1) preliminary determination of the specific type of driver, including A1Type A2Type …, AkMolding;
1.2) selecting training samples: (x)1,y1),(x2,y2),…,(xn,yn) Wherein x isiCharacteristic parameter, y, representing driver behavioriA type representing driver behavior;
1.3) constructing a backbone convolutional neural network:
1.4) processing the test sample by adopting a trunk convolution neural network to establish an objective function
Figure BDA0001681881900000011
In the formula, f (x)i,θ1) Representing a neural network of the backboneModel, theta1For the parameters of the main neural network, the method of stochastic gradient descent is adopted to calculate theta in the objective function1And obtaining a classification machine for the primary classification of the driver behaviors.
Further, the specific process of step 1.3) is as follows:
1.3.1) construction of the input layer: the device is used for determining the number of the neurons of the input layer, and the number of the neurons of the input layer is determined according to the characteristic parameters of a driver;
1.3.2) construction of the convolutional layer: the step of constructing the convolution layer is to determine the size and the step length of the convolution kernel, and the size of the convolution kernel is determined according to the size of the input data scale and the type of the data;
1.3.3) construction of a pooling layer: constructing the pooling layer is to complete the determination of the pooling size and step size and the pooling type;
1.3.4) constructing a full connection layer;
1.3.5) setting a single convolution layer, a single pooling layer and a single full-connection layer to test the training sample;
1.3.6) adding a convolution layer and a pooling layer, testing the training sample again, if the precision increment of the model is not more than 0.01 after the convolution layer and the pooling layer are added, determining that the current neural network is approximately optimal, otherwise, continuously adding the convolution layer and the pooling layer to test the training sample until the optimal neural network is found as a main convolutional neural network.
Further, the specific process of step 2) is as follows:
2.1) constructing a branch convolution neural network:
2.2) retraining the test sample by adopting the branch convolution neural network to establish an objective function
Figure BDA0001681881900000021
Wherein, g (x'j,θ2) Being a model of a branched convolutional neural network, theta2For the parameters of the branch convolution neural network, the random gradient descent method is adopted to calculate the neural network parameter theta in the target function2Sorter for obtaining driver behavior subdivisionA device.
Further, the specific process of step 2.1) is as follows: adding several layers to several layers of the main convolutional neural network to form branch convolutional neural network, and determining A1,…,AkExtracting class AqClass subdivision into B1，B2，…，BmAnd (3) determining the output from the u layer to the v layer on the main neural network and the newly selected driver characteristics as the input of the branch neural network according to the subdivided types, wherein the new training sample is (x)1′,y1′),(x′2,y′2),…,(x′m,y′m) Wherein, x'jRepresenting a new behavioral characteristic parameter, y 'of the driver'jAnd representing the behavior types subdivided by the driver, setting a single pooling layer, a single convolutional layer and a single full-connection layer for the branch neural network, testing the training sample through the branch neural network, adding one convolutional layer and one pooling layer each time, and determining the optimal neural network as the branch convolutional neural network until the accuracy increase of the model is less than or equal to 0.01.
Due to the adoption of the technical scheme, the invention has the following advantages: 1. the deep learning-based classification method for the driver behaviors does not need to manually extract the characteristic parameters related to the types of the drivers. 2. The invention can reduce the calculation amount according to the actual classification requirement, and can extract the sufficient and effective characteristic parameters related to the driver type to accurately finish the classification of the driver behaviors. 3. The invention can flexibly add the type of the driving behavior on the premise of not changing the structure of the whole neural network. The invention can be widely applied to intelligent traffic systems.
Drawings
Fig. 1 is a flow chart illustrating a driver behavior classification method according to the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. It is to be understood, however, that the drawings are provided solely for the purposes of promoting an understanding of the invention and that they are not to be construed as limiting the invention.
Since the construction of the convolutional neural network depends on specific problems, the method provided in the embodiment of the present invention can modify the parameters of the convolutional neural network according to specific situations in practical applications. The invention provides a driver behavior classification method based on a branch convolutional neural network, which comprises the following steps:
1. and processing the test sample based on the constructed trunk convolutional neural network to obtain a primary classification result of the driver behavior.
1) Preliminary determination of the specific driver type, including A1Type A2Type …, AkAnd (4) molding.
2) Selecting a training sample: (x)1,y1),(x2,y2),…,(xn,yn) Wherein x isiCharacteristic parameter, y, representing driver behavioriIndicating the type of driver behavior.
3) Constructing a backbone convolutional neural network:
3.1) constructing an input layer: the method is used for determining the number of the neurons of the input layer, wherein the number of the neurons of the input layer is mainly determined according to the characteristic parameters of a driver.
3.2) constructing a convolution layer: the convolutional layer is constructed by mainly determining the size and the step size of a convolutional kernel, and determining the size of the convolutional kernel according to the size of the input data scale and the type of data, wherein the size of the convolutional kernel is 2 by 2, and the step size is 1.
3.3) constructing a pooling layer: the pooling layer is constructed by determining the pooling size, the step size and the pooling type, and determining the pooling size according to the size of the previous layer, wherein the pooling size is 2 by 2, the step size is 2 according to the pooling size, and the pooling type is the maximum pooling.
3.4) constructing a full connection layer: the number of the neurons is mainly considered, and each neuron is connected with the neuron of the previous layer, so that a full connection layer is formed. Besides the fully-connected layer, the ReLU function is adopted as the activation function of other layers of the trunk convolutional neural network, and the Softmax function is adopted as the activation function of the fully-connected layer.
3.5) setting a single convolution layer, a single pooling layer and a single full-connection layer to test the training sample;
3.6) adding a convolution layer and a pooling layer, testing the training sample again, if the precision increment of the model is not more than 0.01 after the convolution layer and the pooling layer are added, determining that the current neural network is approximately optimal, otherwise, continuously adding the convolution layer and the pooling layer to test the training sample until the optimal neural network is found, and setting the number of layers of the optimal neural network as k1(see fig. 1), the model is used as a main convolution neural network to judge the type of the driver.
4) Training the test sample by adopting a trunk convolution neural network to establish an objective function
Figure BDA0001681881900000041
In the formula, f (x)i,θ1) Model representing a neural network of the backbone, θ1For the parameters of the main neural network, the method of stochastic gradient descent is adopted to calculate theta in the objective function1And obtaining a classification machine for the primary classification of the driver behaviors.
2. And constructing a branch convolutional neural network according to the subdivision requirement of the driver behavior, and processing the test sample again based on the constructed branch convolutional neural network to realize the subdivision of the driver behavior.
2.1) the construction process of the branch convolution neural network is as follows:
adding several layers to several layers of the main convolutional neural network to form branch convolutional neural network, and determining A from the above1,…,AkExtracting class AqClass subdivision into B1，B2，…，BmAnd determining the output from the u layer to the v layer on the trunk neural network and the newly selected driver characteristics as the input of the branch neural network according to the subdivided types, wherein the new training sample is (x'1,y′1),(x′2,y′2),…,(x′m,y′m) Wherein, x'jRepresenting a new behavioral characteristic parameter, y 'of the driver'jIndicating the type of behavior subdivided by the driver. Setting a single pooling layer, a single convolution layer and a single full-connection layer for the branch neural network, testing the branch neural network, adding one pooling layer and one full-connection layer each time, determining the optimal neural network when the precision increment of the model is less than or equal to 0.01, and setting the number of layers of the optimal neural network as k2As a branched convolutional neural network.
2.2) retraining the test sample by adopting the branch convolution neural network to establish an objective function
Figure BDA0001681881900000042
Wherein, g (x'j,θ2) Being a model of a branched convolutional neural network, theta2And (3) solving the neural network parameters in the target function by adopting a random gradient descent method for the parameters of the branch convolution neural network to obtain the classification machine for subdividing the behavior of the driver.
The feasibility of the driver behavior classification method based on the branch convolutional neural network of the present invention is described in detail by the following specific embodiments, and the specific process is as follows:
1. constructing a backbone convolutional neural network:
selecting a training sample: (x)1,y1),(x2,y2),…,(xn,yn) Inputting xiComprises the following steps: the age, sex, speed and brake state of driver, and output yiComprises the following steps: one of conservative type, normal type, aggressive type and dangerous type.
Construction of the input layer: the input size is 35x 128;
structure of the convolutional layer 1: 32 kernels, each size 35x5, step size 1;
construction of the pooling layer 1: pooling size was 1X2, with maximum pooling;
structure of the convolutional layer 2: 64 cores, each core size 1X3, step size 1;
construction of the pooling layer 2: pooling size was 1X2, with maximum pooling;
construction of the fully-connected layer: 128 neurons, the excitation function is Sigmoid;
construction of the output layer: softmax as an output;
the driver can be divided into two types of (conservative type and common type) and (aggressive type and dangerous type) by adopting the structure of the main convolutional neural network, a single pooling layer, a single convolutional layer and a single full-link layer are preset to carry out preliminary test, so that the pooling layer and the convolutional layer are increased until the precision increment of the model is not more than 0.01, and the number of layers of the optimal neural network is set as k3Constructing a trunk convolution neural network, training the test sample by adopting the trunk convolution neural network, and establishing a target function
Figure BDA0001681881900000051
And solving the characteristic parameters of the driver type in the target function by adopting a random gradient descent method to obtain a preliminary driver behavior classification result.
2. Construction of branched convolutional neural network
2.1) construction of the branched convolutional neural network 1:
a pooling layer is added to the convolutional layer 1 and then a full link layer is added. Further subdividing the driver types into a conservative type and a common type, determining the output of a convolutional layer 1 on a main neural network and the newly selected driver characteristics as the input of a branch neural network, setting a single pooling layer, a single convolutional layer and a single full-connection layer for the branch neural network, testing the branch neural network, sequentially adding one pooling layer and one convolutional layer, determining an optimal neural network when the precision increment of the model is less than or equal to 0.01, wherein the number of the optimal neural network is 8, adopting the branch convolutional neural network to train the test sample again, and determining an objective function
Figure BDA0001681881900000052
And solving the neural network parameters in the target function by adopting a random gradient descent method to obtain the classification machine for subdividing the driver behaviors.
2.2) construction of the branched convolutional neural network 2:
a pooling layer is added to convolutional layer 2, followed by a full link layer. Further subdividing the driver types into an aggressive type and a dangerous type, determining the output of a convolutional layer 2 on a trunk neural network and newly selected driver characteristics as the input of a branch neural network, setting a single pooling layer, a single convolutional layer and a single full-connection layer for the branch neural network, testing the branch neural network, sequentially adding one pooling layer and one convolutional layer, determining an optimal neural network when the precision increment of the model is less than or equal to 0.01, wherein the number of the optimal neural network is 6, adopting the branch convolutional neural network to train the test sample again, and determining an objective function
Figure BDA0001681881900000053
And solving the neural network parameters in the target function by adopting a random gradient descent method to obtain the classification machine for subdividing the driver behaviors.
The above examples are only intended to illustrate the invention, and the steps of the method may be varied, and all equivalent changes and modifications made on the basis of the technical solution of the invention should not be excluded from the scope of the invention.

Claims (3)

1. A driver behavior classification method based on a branch convolutional neural network is characterized by comprising the following steps:
1) the method comprises the following steps of processing a test sample based on a constructed trunk convolutional neural network to obtain a primary classification result of driver behaviors, wherein the specific process is as follows:
1.1) preliminary determination of the specific type of driver, including A1Type A2Type …, AkMolding;
1.2) selecting training samples: (x)1,y1),(x2,y2),…,(xn,yn) Wherein x isiCharacteristic parameter, y, representing driver behavioriA type representing driver behavior;
1.3) constructing a backbone convolutional neural network:
1.4) processing the test sample by adopting a trunk convolution neural network to establish an objective function
Figure FDA0003143196250000011
In the formula, f (x)i,θ1) Model representing a neural network of the backbone, θ1For the parameters of the main neural network, the method of stochastic gradient descent is adopted to calculate theta in the objective function1Obtaining a classification machine for preliminarily classifying the behaviors of the driver;
2) according to the subdivision requirement of the driver behavior, the test sample is processed again based on the constructed branch convolutional neural network, so that the subdivision of the driver behavior is realized, and the specific process is as follows:
2.1) constructing a branch convolution neural network:
2.2) retraining the test sample by adopting the branch convolution neural network to establish an objective function
Figure FDA0003143196250000012
Wherein, x'jRepresenting a new behavioral characteristic parameter, y 'of the driver'jThe type of behavior, g (x'j,θ2) Being a model of a branched convolutional neural network, theta2For the parameters of the branch convolution neural network, the random gradient descent method is adopted to calculate the neural network parameter theta in the target function2And obtaining the classification machine for the behavior subdivision of the driver.
2. The method for classifying the behavior of the driver based on the branched convolutional neural network as claimed in claim 1, wherein the specific process of the step 1.3) is as follows:
1.3.1) construction of the input layer: the device is used for determining the number of the neurons of the input layer, and the number of the neurons of the input layer is determined according to the characteristic parameters of a driver;
1.3.2) construction of the convolutional layer: the step of constructing the convolution layer is to determine the size and the step length of the convolution kernel, and the size of the convolution kernel is determined according to the size of the input data scale and the type of the data;
1.3.3) construction of a pooling layer: constructing the pooling layer is to complete the determination of the pooling size and step size and the pooling type;
1.3.4) constructing a full connection layer;
1.3.5) setting a single convolution layer, a single pooling layer and a single full-connection layer to test the training sample;
1.3.6) adding a convolution layer and a pooling layer, testing the training sample again, if the precision increment of the model is not more than 0.01 after the convolution layer and the pooling layer are added, determining that the current neural network is approximately optimal, otherwise, continuously adding the convolution layer and the pooling layer to test the training sample until the optimal neural network is found as a main convolutional neural network.
3. The method for classifying the behavior of the driver based on the branched convolutional neural network as claimed in claim 2, wherein the specific process of the step 2.1) is as follows:
adding several layers to several layers of the main convolutional neural network to form branch convolutional neural network, and determining A1,…,AkExtracting class AqClass subdivision into B1，B2，…，BmDetermining the output from the u layer to the v layer on the trunk neural network and the newly selected driver characteristics as the input of the branch neural network according to the subdivided types, wherein the new training sample is (x'1,y′1),(x′2,y′2),…,(x′m,y′m) And setting a single pooling layer, a single convolutional layer and a single full-link layer for the branch neural network, testing the training sample through the branch neural network, and then adding one convolutional layer and one pooling layer each time until the precision increment of the model is less than or equal to 0.01, and determining the optimal neural network as the branch convolutional neural network.
CN201810557923.5A 2018-06-01 2018-06-01 Driver behavior classification method based on branch convolutional neural network Active CN108875812B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810557923.5A CN108875812B (en) 2018-06-01 2018-06-01 Driver behavior classification method based on branch convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810557923.5A CN108875812B (en) 2018-06-01 2018-06-01 Driver behavior classification method based on branch convolutional neural network

Publications (2)

Publication Number Publication Date
CN108875812A CN108875812A (en) 2018-11-23
CN108875812B true CN108875812B (en) 2021-10-22

Family

ID=64335363

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810557923.5A Active CN108875812B (en) 2018-06-01 2018-06-01 Driver behavior classification method based on branch convolutional neural network

Country Status (1)

Country Link
CN (1) CN108875812B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109543627B (en) * 2018-11-27 2023-08-01 西安电子科技大学 Method and device for judging driving behavior category and computer equipment
CN109544550B (en) * 2018-12-05 2021-10-22 易必祥 CT image-based intelligent detection and identification method and system
CN110322509B (en) * 2019-06-26 2021-11-12 重庆邮电大学 Target positioning method, system and computer equipment based on hierarchical class activation graph
CN110321970A (en) * 2019-07-11 2019-10-11 山东领能电子科技有限公司 A kind of fine-grained objective classification method of multiple features based on branch neural network
CN116092059B (en) * 2022-11-30 2023-10-20 南京通力峰达软件科技有限公司 Neural network-based vehicle networking user driving behavior recognition method and system

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106056071A (en) * 2016-05-30 2016-10-26 北京智芯原动科技有限公司 Method and device for detection of driver' behavior of making call
CN106682602A (en) * 2016-12-16 2017-05-17 深圳市华尊科技股份有限公司 Driver behavior identification method and terminal

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9633282B2 (en) * 2015-07-30 2017-04-25 Xerox Corporation Cross-trained convolutional neural networks using multimodal images

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106056071A (en) * 2016-05-30 2016-10-26 北京智芯原动科技有限公司 Method and device for detection of driver' behavior of making call
CN106682602A (en) * 2016-12-16 2017-05-17 深圳市华尊科技股份有限公司 Driver behavior identification method and terminal

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
《Driver Behavior Recognition Based on Deep Convolutional Neural Networks》;Shiyang Yan；et al;《2016 12th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery》;20161231;第636-641页 *
《Unconstrained Facial Landmark Localization with Backbone-Branches Fully-Convolutional Networks》;Zhujian Liang，et al;《arXiv：1507.03409v1》;20150713;第1-9页 *
《融合面部表情和语音的驾驶员路怒症识别方法研究》;刘鹏;《中国优秀硕士学位论文全文数据库 信息科技辑》;20171115(第11期);第I136-167页 *

Also Published As

Publication number Publication date
CN108875812A (en) 2018-11-23

Similar Documents

Publication Publication Date Title
CN108875812B (en) Driver behavior classification method based on branch convolutional neural network
CN109800736A (en) A kind of method for extracting roads based on remote sensing image and deep learning
CN108280484B (en) Driver acceleration characteristic online classification and identification method
CN109367541B (en) Intelligent vehicle-like person lane change decision-making method based on driver behavior characteristics
CN109543502A (en) A kind of semantic segmentation method based on the multiple dimensioned neural network of depth
CN106446942A (en) Crop disease identification method based on incremental learning
CN107832835A (en) The light weight method and device of a kind of convolutional neural networks
CN105946860B (en) A kind of bend speed prediction method for considering driving style
CN112149796B (en) Driving style identification method based on improved genetic algorithm optimization BP neural network
CN108694408B (en) Driving behavior recognition method based on deep sparse filtering convolutional neural network
CN106570516A (en) Obstacle recognition method using convolution neural network
CN108629369A (en) A kind of Visible Urine Sediment Components automatic identifying method based on Trimmed SSD
CN108447048B (en) Convolutional neural network image feature processing method based on attention layer
CN112465001B (en) Classification method and device based on logistic regression
CN108665093A (en) Traffic accidents Severity forecasting method based on deep learning
CN107945534A (en) A kind of special bus method for predicting based on GMDH neutral nets
CN111563555A (en) Driver driving behavior analysis method and system
CN113378479A (en) Intelligent standard method and system based on automatic driving test intelligent platform vehicle
CN111907523B (en) Vehicle following optimizing control method based on fuzzy reasoning
CN103049679A (en) Method for predicting potential sensitization in protein
CN116738332A (en) Aircraft multi-scale signal classification recognition and fault detection method combining attention mechanism
CN112330114B (en) Aircraft hazard identification method based on mixed deep neural network
CN113858566A (en) Injection molding machine energy consumption prediction method and system based on machine learning
CN107609700A (en) A kind of customer value model optimization method based on machine learning
CN116863418A (en) Road traffic limiting vehicle species division method based on CA_MixNet

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant