CN106203330A - A kind of vehicle classification method based on convolutional neural networks - Google Patents
A kind of vehicle classification method based on convolutional neural networks Download PDFInfo
- Publication number
- CN106203330A CN106203330A CN201610536734.0A CN201610536734A CN106203330A CN 106203330 A CN106203330 A CN 106203330A CN 201610536734 A CN201610536734 A CN 201610536734A CN 106203330 A CN106203330 A CN 106203330A
- Authority
- CN
- China
- Prior art keywords
- neural networks
- convolutional neural
- model
- layer
- vehicle
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of vehicle classification method based on convolutional neural networks, specifically implement according to following steps: step 1: obtain learning sample, and stamp class label to sample;Step 2: using the learning sample that gets as training data, training convolutional neural networks model, obtain outstanding network model's parameter;Step 3: use the convolutional neural networks model trained, extracts the feature of training data, and the support vector cassification model built with liblinear grader carries out ten folding cross-trainings, obtains support vector cassification model;Step 4: the feature of the auto model that use convolutional neural networks model extraction is to be sorted, then uses support vector cassification model, obtains the class of vehicle belonging to vehicle image to be sorted.The present invention utilizes the full articulamentum output of convolutional neural networks as the character representation of vehicle image, then utilizes SVM classifier to classify it, thus obtains preferable vehicle classification accuracy rate.
Description
Technical field
The invention belongs to machine learning and technical field of computer vision, be specifically related to a kind of based on convolutional neural networks
Vehicle classification method.
Background technology
Along with social life level improves constantly, automobile is as a kind of vehicles, and quantity presents the trend of rapid growth.
But also bring such as vehicle accident, a series of traffic problems such as block up, and brings huge challenge to traffic monitoring simultaneously.
Vehicle identification is the important component part of intelligent transportation system.Along with pattern recognition, image procossing and computer
The development of vision technique, vehicle recongnition technique based on image procossing has obtained increasing concern.Especially at highway safety
In management, the identification of type of vehicle is played at aspects such as traffic flow management, expressway tol lcollection and detection lorry illegal road occupations
Important effect.
The speed of Classification and Identification and accuracy are two most important indexs of vehicle identification, for car based on image procossing
Identify system, extract vehicle characteristics be to affect the most important factor of the two index, be the key of whole identification process.Vehicle
The extraction of feature is affected by factors: but vehicle class does not has obvious distinguishing characteristics, the impact of weather, illumination
Impact etc., this brings greatly challenge to the identification of type of vehicle.
Summary of the invention
It is an object of the invention to provide a kind of vehicle classification method based on convolutional neural networks, solve existing vehicle class
The problem that the accuracy rate of type identification is low.
The technical solution adopted in the present invention is, a kind of vehicle classification method based on convolutional neural networks, specifically according to
Following steps are implemented:
Step 1: obtain learning sample, and stamp class label to sample;
Step 2: learning sample step 1 got, as training data, training convolutional neural networks model, obtains excellent
Elegant network model's parameter;
Step 3: use the convolutional neural networks model trained in step 2, extracts the feature of training data, uses
The support vector cassification model that liblinear grader builds carries out ten folding cross-trainings, obtains support vector cassification mould
Type;
Step 4: use the feature of the auto model that convolutional neural networks model extraction is to be sorted in step 2, then use
The support vector cassification model that step 3 obtains, obtains the class of vehicle belonging to vehicle image to be sorted.
The feature of the present invention also resides in:
The acquisition of step 1 Learning Samples includes gathering taxi on road, bus, minibus, car and non-
The picture of motor vehicles.
In step 2 convolutional neural networks model structure particularly as follows:
A: determine the number of plies of convolutional neural networks: convolutional neural networks totally 8 layers, front 5 layers for be used for carrying out convolution and under adopt
The convolutional layer of sample, latter 3 layers is full articulamentum, and the output of a last full articulamentum is the svm classifier with 5 outputs
Device;
B: determine ground floor neural network structure: in each picture input to neural network structure, be all scaled to 227*
The picture of 227 sizes, and three the color dimension inputs of point red, green, blue;At ground floor, 96 convolution filters, each filter are set
The size of ripple device is 11*11, and convolution step-length is 4, is 55*55 through calculating the characteristic pattern size of the 1st layer of output, characteristic pattern
Number be 96;
In order to improve training speed, at a convolutional layer down-sampling layer disposed behind, during down-sampling, select ReLu
Activation primitive:
F (t)=max (0, t) (1)
Wherein t is the feature of input;
Through the feature extraction of ground floor, the characteristic pattern size drawn is 27*27;
C: the number of second layer convolution filter is set to 256, and characteristic pattern input size is the 27*27 that b obtains;Second
Layer output characteristic figure size after convolution and down-sampling are trained is 13*13;Hereafter, using each layer output as next layer
Input, wherein, the convolution filter number of third layer, the 4th layer and layer 5 is respectively set to 384,384 and 256,
The characteristic pattern size finally drawn after 5 layers of training is 6*6;
D: obtain characteristic vector: according to the result of c output, the node number calculating the output of the first two full articulamentum is 6*6*
256=4096;The size obtained by c is that 6*6 characteristic pattern is arranged as a column vector, obtains final characteristic vector;
E: output result: the characteristic vector obtained by d is sent to the SVM of last layer and classifies.
In step 3 with liblinear grader build support vector cassification model particularly as follows:
1.: determine mode classification: using 1-V-1 mode is exactly i.e. together into many points by many 2 classifiers combination
Class device;
2.: SVM uses kernel function RBF;
3.: parameter c and γ to liblinear-2.1 grader are done preferably
Grid.py file is used automatically to do preferably to parameter penalty factor c and nuclear parameter γ, last of terminal demonstration
Row output (c, γ, cross validation precision), for optimum result.
The invention has the beneficial effects as follows: different from the algorithm relying on priori extraction feature in traditional method, the present invention
Convolutional neural networks can under training data drives adaptively construction feature describe, there is higher motility and pervasive
Property.The convolutional neural networks of the present invention can be by original image directly as input, it is to avoid number extra in tional identification algorithm
Data preprocess process.Extract feature by the convolutional neural networks of the present invention and can solve over-fitting, utilize the GPU speed-up computation can
Improve the training time.And by the feature of the network extraction of deep layer it can be avoided that by illumination, the impact of color etc., and can carry
Take abundanter, more abstract characteristics of image.The present invention uses convolutional neural networks and support vector machine to combine, and both can carry
Take abundant feature, suitable grader can be chosen according to the feature of oneself data set again thus improve the standard of whole disaggregated model
Really property.The present invention utilizes the full articulamentum output of convolutional neural networks as the character representation of vehicle image, then utilizes SVM to divide
It is classified by class device, thus obtains preferable vehicle classification accuracy rate.
Accompanying drawing explanation
Fig. 1 is the structural representation of convolutional neural networks in a kind of vehicle classification method based on convolutional neural networks of the present invention
Figure;
Fig. 2 is the thinking figure of 1-V-1 mode in a kind of vehicle classification method based on convolutional neural networks of the present invention;
Fig. 3 is the flow chart of a kind of vehicle classification method based on convolutional neural networks of the present invention.
Detailed description of the invention
The present invention is described in detail with detailed description of the invention below in conjunction with the accompanying drawings.
A kind of vehicle classification method based on convolutional neural networks of the present invention, under computer Ubuntu system, ratio exists
Under windows system, speed is fast, and therefore the present invention illustrates as a example by under Ubuntu system.
Preparation:
Caffe degree of depth learning framework platform is built under computer Ubuntu system.
Build convolutional neural networks model, as shown in Figure 1:
A: determine the number of plies of convolutional neural networks: convolutional neural networks totally 8 layers, front 5 layers for be used for carrying out convolution and under adopt
The convolutional layer (convolutional) of sample, latter 3 layers is full articulamentum (full-connected), a last full articulamentum
Output output be have 5 output SVM (Support Vector Machine, support vector machine) graders;
B: determine ground floor neural network structure: in each picture input to neural network structure, be all scaled to 227*
The picture of 227 sizes, and three the color dimension inputs of point red, green, blue;At ground floor, 96 convolution filters, each filter are set
The size of ripple device is 11*11, and convolution step-length is 4, is 55*55 through calculating the characteristic pattern size of the 1st layer of output, characteristic pattern
Number be 96;
In order to improve training speed, at a convolutional layer down-sampling layer disposed behind, during down-sampling, select ReLu
Activation primitive:
F (t)=max (0, t) (1)
Wherein t is the feature of input;
Through the feature extraction of ground floor, the characteristic pattern size drawn is 27*27;
C: the number of second layer convolution filter is set to 256, and characteristic pattern input size is the 27*27 that b obtains;Second
Layer output characteristic figure size after convolution and down-sampling are trained is 13*13;Hereafter, using each layer output as next layer
Input, wherein, the convolution filter number of third layer, the 4th layer and layer 5 is respectively set to 384,384 and 256,
The characteristic pattern size finally drawn after 5 layers of training is 6*6;
D: obtain characteristic vector: according to the result of c output, the node number calculating the output of the first two full articulamentum is 6*6*
256=4096;The size obtained by c is that 6*6 characteristic pattern is arranged as a column vector, obtains final characteristic vector;
E: output result: the characteristic vector obtained by d is sent to the SVM of last layer and classifies.
Build support vector cassification model
Selecting the liblinear-2.1 grader increased income, arranging parameter c is 100 (through test of many times, when c is 100
Accuracy rate the highest)
1.: determine mode classification: as in figure 2 it is shown, using 1-V-1 mode is exactly i.e. to one by many 2 classifiers combination
Rise and become multi-categorizer;
2.: SVM uses kernel function RBF (Radial Basis Function, RBF);
3.: parameter c and γ to liblinear-2.1 grader are done preferably
Grid.py file is used automatically to do preferably to parameter penalty factor c and nuclear parameter γ, last of terminal demonstration
Row output (c, γ, cross validation precision), for optimum result.
A kind of vehicle classification method based on convolutional neural networks of the present invention, flow process is as it is shown on figure 3, specifically according to following step
Rapid enforcement:
Step 1: obtain learning sample, including gathering taxi, bus, minibus, car and non-machine on road
The picture of motor-car totally 5 classes, every class collection 300~400 pictures are as training set, and additionally every class gathers 150 pictures conducts again
Test set, picture size, form are not fixed, and are stamped class label to sample, and then sample is converted into what caffe identified
The formatted data of lmdb.
Step 2: learning sample step 1 got is as training data, training convolutional neural networks CNN
(Convolutional Neural Network) model, adjusts in solver.prototxt and train_val.prototxt
The numerical value in face thus obtain outstanding network model's parameter.
Step 3: use the convolutional neural networks model trained in step 2, extracts the feature of training data, uses
The support vector cassification model that liblinear grader builds carries out ten folding cross-trainings, obtains support vector cassification mould
Type;
Step 4: use the feature of the auto model that convolutional neural networks model extraction is to be sorted in step 2, then use
The support vector cassification model that step 3 obtains, obtains the class of vehicle belonging to vehicle image to be sorted.
The convolutional neural networks of the present invention can be by original image directly as input, it is to avoid extra in tional identification algorithm
Process of data preprocessing.Extract feature by the convolutional neural networks of the present invention and can solve over-fitting, utilize GPU accelerometer
Calculation can improve the training time.And by the feature of the network extraction of deep layer it can be avoided that by illumination, the impact of color etc., and energy
The characteristics of image that enough extractions are abundanter, more abstract.The present invention uses convolutional neural networks and support vector machine to combine, and both may be used
To extract abundant feature, suitable grader can be chosen according to the feature of oneself data set again thus improve whole disaggregated model
Accuracy.The present invention utilizes the full articulamentum output of convolutional neural networks as the character representation of vehicle image, then utilizes
It is classified by SVM classifier, thus obtains preferable vehicle classification accuracy rate.
Claims (4)
1. a vehicle classification method based on convolutional neural networks, it is characterised in that specifically implement according to following steps:
Step 1: obtain learning sample, and stamp class label to sample;
Step 2: learning sample step 1 got, as training data, training convolutional neural networks model, obtains outstanding
Network model's parameter;
Step 3: use the convolutional neural networks model trained in step 2, extracts the feature of training data, uses
The support vector cassification model that liblinear grader builds carries out ten folding cross-trainings, obtains support vector cassification mould
Type;
Step 4: use the feature of the auto model that convolutional neural networks model extraction is to be sorted in step 2, then use step 3
The support vector cassification model obtained, obtains the class of vehicle belonging to vehicle image to be sorted.
A kind of vehicle classification method based on convolutional neural networks the most according to claim 1, it is characterised in that described step
The acquisition of rapid 1 Learning Samples includes gathering taxi, bus, minibus, car and the figure of bicycle on road
Sheet.
A kind of vehicle classification method based on convolutional neural networks the most according to claim 1, it is characterised in that described step
In rapid 2 convolutional neural networks model structure particularly as follows:
A: determine the number of plies of convolutional neural networks: convolutional neural networks totally 8 layers, first 5 layers is to be used for carrying out convolution and down-sampling
Convolutional layer, latter 3 layers is full articulamentum, and the output of a last full articulamentum is the SVM classifier with 5 outputs;
B: determine ground floor neural network structure: in each picture input to neural network structure, be all scaled to 227*227 big
Little picture, and three the color dimension inputs of point red, green, blue;At ground floor, 96 convolution filters are set, each wave filter
Size is 11*11, and convolution step-length is 4, is 55*55 through calculating the characteristic pattern size of the 1st layer of output, the number of characteristic pattern
It it is 96;
In order to improve training speed, at a convolutional layer down-sampling layer disposed behind, ReLu is selected to activate during down-sampling
Function:
F (t)=max (0, t) (1)
Wherein t is the feature of input;
Through the feature extraction of ground floor, the characteristic pattern size drawn is 27*27;
C: the number of second layer convolution filter is set to 256, and characteristic pattern input size is the 27*27 that b obtains;Second layer warp
Cross convolution and down-sampling training after output characteristic figure size be 13*13;Hereafter, using defeated as next layer of output of each layer
Entering, wherein, the convolution filter number of third layer, the 4th layer and layer 5 is respectively set to 384,384 and 256, warp
The characteristic pattern size finally drawn after crossing 5 layers of training is 6*6;
D: obtain characteristic vector: according to the result of c output, the node number calculating the output of the first two full articulamentum is 6*6*256
=4096;The size obtained by c is that 6*6 characteristic pattern is arranged as a column vector, obtains final characteristic vector;
E: output result: the characteristic vector obtained by d is sent to the SVM of last layer and classifies.
A kind of vehicle classification method based on convolutional neural networks the most according to claim 1, it is characterised in that described step
In rapid 3 with liblinear grader build support vector cassification model particularly as follows:
1.: determine mode classification: using 1-V-1 mode is exactly i.e. together into multi-categorizer by many 2 classifiers combination;
2.: SVM uses kernel function RBF;
3.: parameter c and γ to liblinear-2.1 grader are done preferably
Using grid.py file to be automatically parameter penalty factor c and nuclear parameter γ preferably, last column of terminal demonstration is defeated
Go out (c, γ, cross validation precision), for optimum result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610536734.0A CN106203330A (en) | 2016-07-08 | 2016-07-08 | A kind of vehicle classification method based on convolutional neural networks |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610536734.0A CN106203330A (en) | 2016-07-08 | 2016-07-08 | A kind of vehicle classification method based on convolutional neural networks |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106203330A true CN106203330A (en) | 2016-12-07 |
Family
ID=57473766
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610536734.0A Pending CN106203330A (en) | 2016-07-08 | 2016-07-08 | A kind of vehicle classification method based on convolutional neural networks |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106203330A (en) |
Cited By (25)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106599848A (en) * | 2016-12-16 | 2017-04-26 | 南京理工大学 | Depth visual feature and support vector machine-based terrain texture recognition algorithm |
CN106650801A (en) * | 2016-12-09 | 2017-05-10 | 西南交通大学 | GPS data-based method for classifying multiple types of vehicles |
CN106682649A (en) * | 2017-01-24 | 2017-05-17 | 成都容豪电子信息科技有限公司 | Vehicle type recognition method based on deep learning |
CN106919949A (en) * | 2017-01-23 | 2017-07-04 | 中国石油大学(华东) | A kind of real-time vehicle matching process based on convolutional neural networks |
CN107067011A (en) * | 2017-03-20 | 2017-08-18 | 北京邮电大学 | A kind of vehicle color identification method and device based on deep learning |
CN107229942A (en) * | 2017-04-16 | 2017-10-03 | 北京工业大学 | A kind of convolutional neural networks rapid classification method based on multiple graders |
CN107610091A (en) * | 2017-07-31 | 2018-01-19 | 阿里巴巴集团控股有限公司 | Vehicle insurance image processing method, device, server and system |
CN108108768A (en) * | 2017-12-29 | 2018-06-01 | 清华大学 | Photovoltaic glass defect classification method and device based on convolutional neural networks |
CN108764289A (en) * | 2018-04-26 | 2018-11-06 | 福建天晴数码有限公司 | A kind of UI exceptions picture classification method and system based on convolutional neural networks |
CN108804988A (en) * | 2017-05-04 | 2018-11-13 | 上海荆虹电子科技有限公司 | A kind of remote sensing image scene classification method and device |
CN109117826A (en) * | 2018-09-05 | 2019-01-01 | 湖南科技大学 | A kind of vehicle identification method of multiple features fusion |
WO2019034129A1 (en) * | 2017-08-18 | 2019-02-21 | 北京市商汤科技开发有限公司 | Neural network structure generation method and device, electronic equipment and storage medium |
CN109753983A (en) * | 2017-11-07 | 2019-05-14 | 北京京东尚科信息技术有限公司 | Image classification method, device and computer readable storage medium |
CN110147731A (en) * | 2019-04-16 | 2019-08-20 | 深圳云天励飞技术有限公司 | Vehicle type recognition method and Related product |
CN110651276A (en) * | 2017-03-17 | 2020-01-03 | 纽拉拉股份有限公司 | Tagging and online incremental real-time learning of data streams for deep neural networks and neural network applications |
CN110764064A (en) * | 2019-11-08 | 2020-02-07 | 哈尔滨工业大学 | Radar interference signal identification method based on deep convolutional neural network integration |
CN110807493A (en) * | 2019-11-06 | 2020-02-18 | 上海眼控科技股份有限公司 | Optimization method and equipment of vehicle classification model |
WO2020083073A1 (en) * | 2018-10-23 | 2020-04-30 | 苏州科达科技股份有限公司 | Non-motorized vehicle image multi-label classification method, system, device and storage medium |
CN111161814A (en) * | 2019-12-18 | 2020-05-15 | 浙江大学 | DRGs automatic grouping method based on convolutional neural network |
CN111259821A (en) * | 2020-01-19 | 2020-06-09 | 山西国芯科技有限公司 | Cloud computing model method based on deep learning |
CN111860219A (en) * | 2020-06-30 | 2020-10-30 | 杭州科度科技有限公司 | High-speed road occupation judging method and device and electronic equipment |
WO2020232840A1 (en) * | 2019-05-23 | 2020-11-26 | 厦门市美亚柏科信息股份有限公司 | Vehicle multi-attribute identification method and device employing neural network structure search, and medium |
CN112115902A (en) * | 2020-09-25 | 2020-12-22 | 广州市派客朴食信息科技有限责任公司 | Dish identification method based on single-stage target detection algorithm |
CN112215291A (en) * | 2020-10-19 | 2021-01-12 | 中国计量大学 | Method for extracting and classifying medical image features under cascade neural network |
CN113642655A (en) * | 2021-08-18 | 2021-11-12 | 杭州电子科技大学 | Small sample image classification method based on support vector machine and convolutional neural network |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101794378A (en) * | 2010-01-26 | 2010-08-04 | 浙江大学 | Rubbish image filtering method based on image encoding |
CN104463241A (en) * | 2014-10-31 | 2015-03-25 | 北京理工大学 | Vehicle type recognition method in intelligent transportation monitoring system |
CN104537387A (en) * | 2014-12-16 | 2015-04-22 | 广州中国科学院先进技术研究所 | Method and system for classifying automobile types based on neural network |
CN104657748A (en) * | 2015-02-06 | 2015-05-27 | 中国石油大学(华东) | Vehicle type recognition method based on convolutional neural network |
CN105117739A (en) * | 2015-07-29 | 2015-12-02 | 南京信息工程大学 | Clothes classifying method based on convolutional neural network |
CN105184271A (en) * | 2015-09-18 | 2015-12-23 | 苏州派瑞雷尔智能科技有限公司 | Automatic vehicle detection method based on deep learning |
CN105224951A (en) * | 2015-09-30 | 2016-01-06 | 深圳市华尊科技股份有限公司 | A kind of vehicle type classification method and sorter |
CN105335702A (en) * | 2015-10-15 | 2016-02-17 | 浙江捷尚视觉科技股份有限公司 | Bayonet vehicle type recognition method based on statistical learning |
CN105512676A (en) * | 2015-11-30 | 2016-04-20 | 华南理工大学 | Food recognition method at intelligent terminal |
CN105574543A (en) * | 2015-12-16 | 2016-05-11 | 武汉烽火众智数字技术有限责任公司 | Vehicle brand and model identifying method and system based on deep learning |
-
2016
- 2016-07-08 CN CN201610536734.0A patent/CN106203330A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101794378A (en) * | 2010-01-26 | 2010-08-04 | 浙江大学 | Rubbish image filtering method based on image encoding |
CN104463241A (en) * | 2014-10-31 | 2015-03-25 | 北京理工大学 | Vehicle type recognition method in intelligent transportation monitoring system |
CN104537387A (en) * | 2014-12-16 | 2015-04-22 | 广州中国科学院先进技术研究所 | Method and system for classifying automobile types based on neural network |
CN104657748A (en) * | 2015-02-06 | 2015-05-27 | 中国石油大学(华东) | Vehicle type recognition method based on convolutional neural network |
CN105117739A (en) * | 2015-07-29 | 2015-12-02 | 南京信息工程大学 | Clothes classifying method based on convolutional neural network |
CN105184271A (en) * | 2015-09-18 | 2015-12-23 | 苏州派瑞雷尔智能科技有限公司 | Automatic vehicle detection method based on deep learning |
CN105224951A (en) * | 2015-09-30 | 2016-01-06 | 深圳市华尊科技股份有限公司 | A kind of vehicle type classification method and sorter |
CN105335702A (en) * | 2015-10-15 | 2016-02-17 | 浙江捷尚视觉科技股份有限公司 | Bayonet vehicle type recognition method based on statistical learning |
CN105512676A (en) * | 2015-11-30 | 2016-04-20 | 华南理工大学 | Food recognition method at intelligent terminal |
CN105574543A (en) * | 2015-12-16 | 2016-05-11 | 武汉烽火众智数字技术有限责任公司 | Vehicle brand and model identifying method and system based on deep learning |
Non-Patent Citations (3)
Title |
---|
张海彬: "基于卡口图像的车型识别方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
王茜等: "基于深度神经网络的汽车车型识别", 《现代计算机》 * |
邓柳: "基于深度卷积神经网络的车型识别", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (37)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106650801A (en) * | 2016-12-09 | 2017-05-10 | 西南交通大学 | GPS data-based method for classifying multiple types of vehicles |
CN106599848A (en) * | 2016-12-16 | 2017-04-26 | 南京理工大学 | Depth visual feature and support vector machine-based terrain texture recognition algorithm |
CN106919949A (en) * | 2017-01-23 | 2017-07-04 | 中国石油大学(华东) | A kind of real-time vehicle matching process based on convolutional neural networks |
CN106682649A (en) * | 2017-01-24 | 2017-05-17 | 成都容豪电子信息科技有限公司 | Vehicle type recognition method based on deep learning |
CN110651276A (en) * | 2017-03-17 | 2020-01-03 | 纽拉拉股份有限公司 | Tagging and online incremental real-time learning of data streams for deep neural networks and neural network applications |
CN107067011A (en) * | 2017-03-20 | 2017-08-18 | 北京邮电大学 | A kind of vehicle color identification method and device based on deep learning |
CN107067011B (en) * | 2017-03-20 | 2019-05-03 | 北京邮电大学 | A kind of vehicle color identification method and device based on deep learning |
CN107229942A (en) * | 2017-04-16 | 2017-10-03 | 北京工业大学 | A kind of convolutional neural networks rapid classification method based on multiple graders |
CN107229942B (en) * | 2017-04-16 | 2021-03-30 | 北京工业大学 | Convolutional neural network classification method based on multiple classifiers |
CN108804988A (en) * | 2017-05-04 | 2018-11-13 | 上海荆虹电子科技有限公司 | A kind of remote sensing image scene classification method and device |
CN108804988B (en) * | 2017-05-04 | 2020-11-20 | 深圳荆虹科技有限公司 | Remote sensing image scene classification method and device |
US10846556B2 (en) | 2017-07-31 | 2020-11-24 | Advanced New Technologies Co., Ltd. | Vehicle insurance image processing method, apparatus, server, and system |
WO2019024771A1 (en) * | 2017-07-31 | 2019-02-07 | 阿里巴巴集团控股有限公司 | Car insurance image processing method, apparatus, server and system |
TWI709919B (en) * | 2017-07-31 | 2020-11-11 | 開曼群島商創新先進技術有限公司 | Auto insurance image processing method, device, server and system |
CN107610091A (en) * | 2017-07-31 | 2018-01-19 | 阿里巴巴集团控股有限公司 | Vehicle insurance image processing method, device, server and system |
US11270190B2 (en) | 2017-08-18 | 2022-03-08 | Beijing Sensetime Technology Development Co., Ltd. | Method and apparatus for generating target neural network structure, electronic device, and storage medium |
WO2019034129A1 (en) * | 2017-08-18 | 2019-02-21 | 北京市商汤科技开发有限公司 | Neural network structure generation method and device, electronic equipment and storage medium |
CN109753983A (en) * | 2017-11-07 | 2019-05-14 | 北京京东尚科信息技术有限公司 | Image classification method, device and computer readable storage medium |
CN108108768B (en) * | 2017-12-29 | 2020-09-25 | 清华大学 | Convolutional neural network-based photovoltaic glass defect classification method and device |
CN108108768A (en) * | 2017-12-29 | 2018-06-01 | 清华大学 | Photovoltaic glass defect classification method and device based on convolutional neural networks |
CN108764289B (en) * | 2018-04-26 | 2021-04-30 | 福建天晴数码有限公司 | Method and system for classifying UI (user interface) abnormal pictures based on convolutional neural network |
CN108764289A (en) * | 2018-04-26 | 2018-11-06 | 福建天晴数码有限公司 | A kind of UI exceptions picture classification method and system based on convolutional neural networks |
CN109117826B (en) * | 2018-09-05 | 2020-11-24 | 湖南科技大学 | Multi-feature fusion vehicle identification method |
CN109117826A (en) * | 2018-09-05 | 2019-01-01 | 湖南科技大学 | A kind of vehicle identification method of multiple features fusion |
WO2020083073A1 (en) * | 2018-10-23 | 2020-04-30 | 苏州科达科技股份有限公司 | Non-motorized vehicle image multi-label classification method, system, device and storage medium |
CN110147731A (en) * | 2019-04-16 | 2019-08-20 | 深圳云天励飞技术有限公司 | Vehicle type recognition method and Related product |
WO2020232840A1 (en) * | 2019-05-23 | 2020-11-26 | 厦门市美亚柏科信息股份有限公司 | Vehicle multi-attribute identification method and device employing neural network structure search, and medium |
CN110807493A (en) * | 2019-11-06 | 2020-02-18 | 上海眼控科技股份有限公司 | Optimization method and equipment of vehicle classification model |
CN110764064A (en) * | 2019-11-08 | 2020-02-07 | 哈尔滨工业大学 | Radar interference signal identification method based on deep convolutional neural network integration |
CN111161814A (en) * | 2019-12-18 | 2020-05-15 | 浙江大学 | DRGs automatic grouping method based on convolutional neural network |
CN111259821A (en) * | 2020-01-19 | 2020-06-09 | 山西国芯科技有限公司 | Cloud computing model method based on deep learning |
CN111860219A (en) * | 2020-06-30 | 2020-10-30 | 杭州科度科技有限公司 | High-speed road occupation judging method and device and electronic equipment |
CN111860219B (en) * | 2020-06-30 | 2024-01-05 | 杭州科度科技有限公司 | High-speed channel occupation judging method and device and electronic equipment |
CN112115902A (en) * | 2020-09-25 | 2020-12-22 | 广州市派客朴食信息科技有限责任公司 | Dish identification method based on single-stage target detection algorithm |
CN112215291A (en) * | 2020-10-19 | 2021-01-12 | 中国计量大学 | Method for extracting and classifying medical image features under cascade neural network |
CN113642655A (en) * | 2021-08-18 | 2021-11-12 | 杭州电子科技大学 | Small sample image classification method based on support vector machine and convolutional neural network |
CN113642655B (en) * | 2021-08-18 | 2024-02-13 | 杭州电子科技大学 | Small sample image classification method based on support vector machine and convolutional neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106203330A (en) | A kind of vehicle classification method based on convolutional neural networks | |
WO2022083784A1 (en) | Road detection method based on internet of vehicles | |
CN111274970B (en) | Traffic sign detection method based on improved YOLO v3 algorithm | |
CN105938560B (en) | Convolutional-neural-network-based vehicle model refined classification system | |
CN105160309B (en) | Three lanes detection method based on morphological image segmentation and region growing | |
CN111401148A (en) | Road multi-target detection method based on improved multilevel YO L Ov3 | |
CN109447033A (en) | Vehicle front obstacle detection method based on YOLO | |
CN107633220A (en) | A kind of vehicle front target identification method based on convolutional neural networks | |
CN107886073A (en) | A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks | |
CN109948416A (en) | A kind of illegal occupancy bus zone automatic auditing method based on deep learning | |
CN103679205B (en) | Assume based on shade and the Foregut fermenters method of layering HOG symmetrical feature checking | |
CN111460919B (en) | Monocular vision road target detection and distance estimation method based on improved YOLOv3 | |
CN108009518A (en) | A kind of stratification traffic mark recognition methods based on quick two points of convolutional neural networks | |
CN106022300A (en) | Traffic sign identifying method and traffic sign identifying system based on cascading deep learning | |
CN105335710A (en) | Fine vehicle model identification method based on multi-stage classifier | |
CN104463241A (en) | Vehicle type recognition method in intelligent transportation monitoring system | |
CN109367541A (en) | A kind of intelligent vehicle class people's lane change decision-making technique based on driver behavior pattern | |
CN109190444A (en) | A kind of implementation method of the lane in which the drivers should pay fees vehicle feature recognition system based on video | |
CN107092884A (en) | Rapid coarse-fine cascade pedestrian detection method | |
CN107578091A (en) | A kind of Pedestrians and vehicles real-time detection method based on lightweight depth network | |
CN106886757B (en) | A kind of multiclass traffic lights detection method and system based on prior probability image | |
CN108960074B (en) | Small-size pedestrian target detection method based on deep learning | |
CN107274668A (en) | A kind of congestion in road modeling method based on vehicle detection | |
CN114092917B (en) | MR-SSD-based shielded traffic sign detection method and system | |
CN106845387A (en) | Pedestrian detection method based on self study |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161207 |