CN105975941A - Multidirectional vehicle model detection recognition system based on deep learning - Google Patents

Multidirectional vehicle model detection recognition system based on deep learning Download PDF

Info

Publication number
CN105975941A
CN105975941A CN201610316159.3A CN201610316159A CN105975941A CN 105975941 A CN105975941 A CN 105975941A CN 201610316159 A CN201610316159 A CN 201610316159A CN 105975941 A CN105975941 A CN 105975941A
Authority
CN
China
Prior art keywords
vehicle
network
loss
picture
training
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.)
Granted
Application number
CN201610316159.3A
Other languages
Chinese (zh)
Other versions
CN105975941B (en
Inventor
李鸿升
范峻铭
周辉
胡欢
曹滨
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.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
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 University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201610316159.3A priority Critical patent/CN105975941B/en
Publication of CN105975941A publication Critical patent/CN105975941A/en
Application granted granted Critical
Publication of CN105975941B publication Critical patent/CN105975941B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • 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

Abstract

The invention discloses a multidirectional vehicle model detection recognition system based on deep learning. For an automobile image taken at a non-fixed view angle, the system firstly detects the automobile position in the image through employing a vehicle detection positioning algorithm based on a convolution neural network, and then transmits the image of a part where an automobile is located to a vehicle model recognition network based on the convolution neural network for the recognition of the brand and model of the automobile. The system is robust for the photographing direction, illumination condition and scale deformation of the automobile image, and is greatly high in accuracy of the detection and recognition of the automobile.

Description

A kind of multi-direction vehicle detection identification system based on degree of depth study
Technical field
The present invention relates to computer vision and degree of depth learning art field, particularly relate to a kind of based on the degree of depth learn in many ways Identification system is detected to vehicle.
Background technology
Intelligent traffic monitoring system is a prior development direction of current traffic monitoring industry, mainly relies on computer Picture under monitoring camera is shot by the technology such as vision and degree of depth study is automatically analyzed, thus judges hypervelocity, rushes The acts of violating regulations such as red light, and can automatically identify the brand and model of vehicles peccancy, thus greatly facilitate traffic monitoring.
The degree of depth learns formally to propose in 2006, is a popular domain inside machine learning in recent years, originates from many Layer artificial neural network, has been successfully applied to the fields such as computer vision, natural language processing and intelligent search the most.Wherein roll up Long-pending neutral net has been achieved for the achievement attracted people's attention in field of image recognition, all gains ground, compare in a lot of application Traditional method is greatly improved.And due to the classification problem requirement machine learning model tool of vehicle particular type number under fine granularity Having extremely strong feature extraction ability and generalization ability, conventional machines learning model is difficult to accomplish preferable effect, the most naturally Take into account use degree of depth learning model to solve this problem.
In vehicle cab recognition field, traditional method usually requires that automobile image shooting angle is fixed, and generally uses feature more Significantly headstock front viewing angles, and require that the resolution of image is higher, although also can obtain good recognition accuracy, But the vehicle cab recognition problem under monitoring environment is just seemed unable, because the image under monitoring environment is complex, car The problem such as shooting angle unfixing, ambient lighting, image resolution ratio be difficult to solve.
Through retrieval to prior art, it has been found that some begin to use convolutional neural networks to do the side of vehicle cab recognition Method, but the vehicle location problem before identification, the most still use traditional algorithm of target detection, detect vehicle location it After, then the image feeding neutral net of vehicle location part is carried out vehicle cab recognition.The accuracy rate of its tradition location algorithm is compared In convolutional neural networks often to differ from upper much, and the vehicle cab recognition carried out on this basis, its accuracy rate is the most correspondingly subject to To impact.
Summary of the invention
The purpose of the present invention is that provides a kind of multi-direction vehicle based on degree of depth study to solve the problems referred to above Detection identification system, emphasis solves the unfixed problem in vehicle image taking visual angle, and the vehicle cab recognition under all directions is all There is high accuracy rate.
For achieving the above object, the present invention is by the following technical solutions:
A kind of multi-direction vehicle detection recognition methods based on degree of depth study, comprises the steps:
1. collect vehicle picture, mark out by vehicle location by hand, build the data set with vehicle brand mark；
2. using marked go out automobile position picture as training set, training vehicle detection location network；
3., using the vehicle brand mark of automobile position part picture in picture and its correspondence as training set, train vehicle cab recognition net Network；
4., after training is complete, two cascades is got up, input an auto graph, carry out a propagated forward, Obtain automobile brand model.
In technique scheme, the data set mode of described structure is, collects the vehicle picture from all directions shooting 300,000, more than 1,000 kind of vehicle, and obtain the vehicle brand mark of its correspondence；Mark out by vehicle location by hand.
In technique scheme, the manual method marked by vehicle location is the minimum rectangle frame taking and comprising whole vehicle body, Mark using the rectangle frame upper left corner and lower right corner point coordinates as position.
In technique scheme, described vehicle detection location network training process is as follows:
(4.1) network is initialized, treat training parameter with in the Gauss distribution random initializtion network that standard deviation is 0.01；
(4.2) picture sample being inputted network, the loss that vehicle location Labeling Coordinate is given region recommendation network simultaneously calculates Part and last classification, the loss calculating section of recurrence device；Carry out a propagated forward, export target frame coordinate according to network The loss of region recommendation network and the summation of the loss of last classification recurrence is obtained with the mathematic interpolation of actual callout box coordinate loss；
(4.3) back-propagation algorithm, the whole network of directional trim declined according to gradient descent algorithm are used along summation loss Parameter value；
(4.4) change image pattern and be iteratively repeated above-mentioned steps, until after summation loss drops to appropriate level, deconditioning.
In technique scheme, described vehicle cab recognition network comprises the following steps:
(5.1) network is initialized, with the xavier algorithm that standard deviation is 0.1 according to neuronal quantity random initializtion net Network is treated training parameter；
(5.2) it is 224 × 224 sizes using picture as samples normalization, and inputs network, give three by vehicle mark simultaneously Softmax grader；Carry out a propagated forward, differentiate the difference meter of probability and actual vehicle according to softmax output vehicle Calculation obtains loss；
(5.3) back-propagation algorithm is used, according to the ginseng of the whole network of directional trim that gradient descent algorithm declines along loss Numerical value；
(5.4) changing picture, picture is iteratively repeated above-mentioned steps as sample, until after loss drops to appropriate level, stopping Training.
In technique scheme, be respectively trained two networks complete after, two cascades are got up, wherein first After network output, need to be the size of 224 × 224 by image normalization, then input second network, all directions are shot Include the image input system of vehicle, final i.e. its concrete brand vehicle exportable.
In sum, owing to have employed technique scheme, the invention has the beneficial effects as follows:
Multi-direction vehicle detection identification system based on degree of depth study of the present invention, vehicle detection position portion uses convolution god Carry out feature extraction through network, automobile position coordinate is returned, simultaneously to whether being that automobile is classified.Through training Afterwards, finally multidirectional automobile image all can be carried out vehicle location effectively, reach good effect；Vehicle cab recognition portion Divide the degree of depth convolutional neural networks by multilamellar, feature is carried out abstract expression, use softmax grader to classify, Eventually under the complicated environmental conditions such as different directions, illumination condition, deformation, vehicle cab recognition rate of accuracy reached to more than 97%.
Accompanying drawing explanation
Fig. 1 is that vehicle of the present invention detection identifies system block diagram；
Fig. 2 is that vehicle detection of the present invention positions network structure；
Fig. 3 is vehicle cab recognition network structure of the present invention.
Detailed description of the invention
All features disclosed in this specification, or disclosed all methods or during step, except mutually exclusive Feature and/or step beyond, all can combine by any way.
Below in conjunction with Fig. 1, Fig. 2, the present invention is elaborated.
As it is shown in figure 1, the present invention proposes a set of multi-direction model recognition system based on degree of depth study, whole system framework Comprise the steps of:
1, model data collection is built
The mode building data set is, collects the vehicle figure agreement that contracts a film or TV play to an actor or actress 300,000 from all directions shooting from network, more than 1, 000 kind of vehicle, and obtain the vehicle brand mark of its correspondence；Mark out by vehicle location, in order to train vehicle detection by hand Location network.Concrete notation methods is to take the minimum rectangle frame comprising whole vehicle body, sits with the rectangle frame upper left corner and bottom right angle point It is denoted as marking for position.Data set is divided into training set and test set two parts according to the ratio of 6:1 by the present embodiment.
2, vehicle detection location network
The step for relate to vehicle detection location network structure design and training method.The vehicle inspection that the present embodiment is used Measure position network structure such as Fig. 2, mainly by convolution stage, RPN region recommendation network, area-of-interest pond layer, full articulamentum Forming with a softmax grader, the convolution stage is made up of 5 convolutional layers, and its parameter is:
This network mainly roughly selects out target candidate frame by RPN region recommendation network, then by last corresponding for target frame layer Convolutional layer characteristic sends into follow-up pond, fully-connected network, then determines whether vehicle mesh through softmax grader Mark, simultaneously by the position of a recurrence device fine setting target frame.
As follows to the training process of this network:
(1) network is initialized, treat training parameter with in the Gauss distribution random initializtion network that standard deviation is 0.01；
(2) image pattern is inputted network, vehicle location Labeling Coordinate is given the loss calculating part of region recommendation network simultaneously Divide and last classification, the loss calculating section of recurrence device；Carry out a propagated forward, according to network output target frame coordinate with The mathematic interpolation of actual callout box coordinate obtains the loss and the loss of last classification recurrence of region recommendation network；
(3) back-propagation algorithm is used, according to the parameter of the whole network of directional trim that gradient descent algorithm declines along loss Value；
(4) change image pattern and be iteratively repeated above-mentioned steps, until after loss drops to appropriate level, deconditioning.
3, vehicle cab recognition network
The step for relate to vehicle cab recognition network structure design and training method.The vehicle cab recognition net that the present embodiment is used Network structure such as Fig. 3, its primary structure includes: two convolution stages, nine inception modules, 3 softmax classification subnets Network, its parameter is:
This network is in addition to last softmax classified part, and the part at relatively shallow-layer also adds two softmax classification Part, for solving the training problem that the network number of plies is the deepest, to ensure that shallower network portion also can sufficiently be trained.? Later softmax layer is as final output, and the first two only uses when training.
As follows to the training process of this network:
(1) network is initialized, with the xavier algorithm that standard deviation is 0.1 according to neuronal quantity random initializtion network In treat training parameter；
(2) image pattern is normalized to 224 × 224 sizes, and inputs network, give three by vehicle mark simultaneously Softmax grader；Carry out a propagated forward, differentiate the difference meter of probability and actual vehicle according to softmax output vehicle Calculation obtains loss；
(3) back-propagation algorithm is used, according to the parameter of the whole network of directional trim that gradient descent algorithm declines along loss Value；
(4) change image pattern and be iteratively repeated above-mentioned steps, until after loss drops to appropriate level, deconditioning.
4, train complete, cascade constitute vehicle detection identification system
Be respectively trained two networks complete after, two cascades are got up, wherein after the output of first network, need to will scheme As being normalized to the size of 224 × 224, then input second network.The image including vehicle all directions shot is defeated Enter system, final i.e. its concrete brand vehicle exportable.Test sample collection contains the vehicle figure of various ambient lighting and deformation Picture, after testing about 10,000 test samples, obtains vehicle cab recognition rate of accuracy reached more than 97%.
Technical scheme is not limited to the restriction of above-mentioned specific embodiment, every does according to technical scheme The technology deformation gone out, within each falling within protection scope of the present invention.

Claims (6)

1. a multi-direction vehicle detection recognition methods based on degree of depth study, it is characterised in that: comprise the steps:
1. collect vehicle picture, mark out by vehicle location by hand, build the data set with vehicle brand mark；
2. using marked go out automobile position picture as training set, training vehicle detection location network；
3., using the vehicle brand mark of automobile position part picture in picture and its correspondence as training set, train vehicle cab recognition net Network；
4., after training is complete, two cascades is got up, input an auto graph, carry out a propagated forward, Obtain automobile brand model.
Multi-direction vehicle detection identification system based on degree of depth study the most according to claim 1, it is characterised in that: described The data set mode built is, collects 300,000, the vehicle picture from all directions shooting, more than 1,000 kind of vehicle, and obtains Take the vehicle brand mark of its correspondence；Mark out by vehicle location by hand.
Multi-direction vehicle detection identification system based on degree of depth study the most according to claim 2, it is characterised in that: manual The method marked by vehicle location is the minimum rectangle frame taking and comprising whole vehicle body, with the rectangle frame upper left corner and lower right corner point coordinates Mark as position.
Multi-direction vehicle detection identification system based on degree of depth study the most according to claim 1, it is characterised in that: described Vehicle detection location network training process as follows:
(4.1) network is initialized, treat training parameter with in the Gauss distribution random initializtion network that standard deviation is 0.01；
(4.2) picture sample being inputted network, the loss that vehicle location Labeling Coordinate is given region recommendation network simultaneously calculates Part and last classification, the loss calculating section of recurrence device；Carry out a propagated forward, export target frame coordinate according to network The loss of region recommendation network and the summation of the loss of last classification recurrence is obtained with the mathematic interpolation of actual callout box coordinate loss；
(4.3) back-propagation algorithm, the whole network of directional trim declined according to gradient descent algorithm are used along summation loss Parameter value；
(4.4) change image pattern and be iteratively repeated above-mentioned steps, until after summation loss drops to appropriate level, deconditioning.
Multi-direction vehicle detection identification system based on degree of depth study the most according to claim 1, it is characterised in that: described Vehicle cab recognition network comprise the following steps:
(5.1) network is initialized, with the xavier algorithm that standard deviation is 0.1 according to neuronal quantity random initializtion net Network is treated training parameter；
(5.2) it is 224 × 224 sizes using picture as samples normalization, and inputs network, give three by vehicle mark simultaneously Softmax grader；Carry out a propagated forward, differentiate the difference meter of probability and actual vehicle according to softmax output vehicle Calculation obtains loss；
(5.3) back-propagation algorithm is used, according to the ginseng of the whole network of directional trim that gradient descent algorithm declines along loss Numerical value；
(5.4) changing picture, picture is iteratively repeated above-mentioned steps as sample, until after loss drops to appropriate level, stopping Training.
Multi-direction vehicle detection identification system based on degree of depth study the most according to claim 1, it is characterised in that: respectively After training two networks complete, two cascades are got up, wherein after first network output, need to be by image normalization It is the size of 224 × 224, then inputs second network, the image input system including vehicle that all directions are shot, End its concrete brand vehicle the most exportable.
CN201610316159.3A 2016-05-31 2016-05-31 A kind of multi-direction vehicle detection identifying system based on deep learning Expired - Fee Related CN105975941B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610316159.3A CN105975941B (en) 2016-05-31 2016-05-31 A kind of multi-direction vehicle detection identifying system based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610316159.3A CN105975941B (en) 2016-05-31 2016-05-31 A kind of multi-direction vehicle detection identifying system based on deep learning

Publications (2)

Publication Number Publication Date
CN105975941A true CN105975941A (en) 2016-09-28
CN105975941B CN105975941B (en) 2019-04-12

Family

ID=56992471

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610316159.3A Expired - Fee Related CN105975941B (en) 2016-05-31 2016-05-31 A kind of multi-direction vehicle detection identifying system based on deep learning

Country Status (1)

Country Link
CN (1) CN105975941B (en)

Cited By (41)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106529511A (en) * 2016-12-13 2017-03-22 北京旷视科技有限公司 Image structuring method and device
CN106528662A (en) * 2016-10-20 2017-03-22 中山大学 Quick retrieval method and system of vehicle image on the basis of feature geometric constraint
CN106570477A (en) * 2016-10-28 2017-04-19 中国科学院自动化研究所 Vehicle model recognition model construction method based on depth learning and vehicle model recognition method based on depth learning
CN106600969A (en) * 2016-12-19 2017-04-26 安徽百诚慧通科技有限公司 Intelligent road comprehensive prevention and control method and system
CN106874418A (en) * 2017-01-24 2017-06-20 成都容豪电子信息科技有限公司 A kind of automobile model data set method for building up for serving deep learning
CN106909901A (en) * 2017-02-28 2017-06-30 北京京东尚科信息技术有限公司 The method and device of detection object from image
CN107038448A (en) * 2017-03-01 2017-08-11 中国科学院自动化研究所 Target detection model building method
CN107316007A (en) * 2017-06-07 2017-11-03 浙江捷尚视觉科技股份有限公司 A kind of monitoring image multiclass object detection and recognition methods based on deep learning
CN107545263A (en) * 2017-08-02 2018-01-05 清华大学 A kind of object detecting method and device
CN108171246A (en) * 2017-12-21 2018-06-15 北京科技大学 A kind of clothes salient region detecting method
CN108427920A (en) * 2018-02-26 2018-08-21 杭州电子科技大学 A kind of land and sea border defense object detection method based on deep learning
CN108615401A (en) * 2018-06-11 2018-10-02 西安电子科技大学 The non-homogeneous light parking situation recognition methods in interior based on deep learning
CN108875911A (en) * 2018-05-25 2018-11-23 同济大学 One kind is parked position detecting method
CN109086708A (en) * 2018-07-25 2018-12-25 深圳大学 A kind of parking space detection method and system based on deep learning
CN109101934A (en) * 2018-08-20 2018-12-28 广东数相智能科技有限公司 Model recognizing method, device and computer readable storage medium
CN109284749A (en) * 2017-07-19 2019-01-29 微软技术许可有限责任公司 Refine image recognition
CN109447171A (en) * 2018-11-05 2019-03-08 电子科技大学 A kind of vehicle attitude classification method based on deep learning
CN109583392A (en) * 2018-12-05 2019-04-05 北京纵目安驰智能科技有限公司 A kind of method for detecting parking stalls, device and storage medium
CN109624850A (en) * 2017-10-05 2019-04-16 斯特拉德视觉公司 Monitor the method for the blind spot of vehicle and the blind spot monitoring device using this method
CN109670501A (en) * 2018-12-10 2019-04-23 中国科学院自动化研究所 Object identification and crawl position detection method based on depth convolutional neural networks
CN109726717A (en) * 2019-01-02 2019-05-07 西南石油大学 A kind of vehicle comprehensive information detection system
CN109753999A (en) * 2018-12-21 2019-05-14 西北工业大学 For the fine granularity model recognizing method of visual angle auto graph
CN109829398A (en) * 2019-01-16 2019-05-31 北京航空航天大学 A kind of object detection method in video based on Three dimensional convolution network
CN109829469A (en) * 2018-11-08 2019-05-31 电子科技大学 A kind of vehicle checking method based on deep learning
CN109993031A (en) * 2017-12-29 2019-07-09 杭州海康威视数字技术股份有限公司 A kind of animal-drawn vehicle target is driven against traffic regulations behavioral value method, apparatus and camera
CN110084285A (en) * 2019-04-08 2019-08-02 安徽艾睿思智能科技有限公司 Fish fine grit classification method based on deep learning
CN110119726A (en) * 2019-05-20 2019-08-13 四川九洲视讯科技有限责任公司 A kind of vehicle brand multi-angle recognition methods based on YOLOv3 model
CN110197589A (en) * 2019-05-29 2019-09-03 杭州诚道科技股份有限公司 A kind of illegal detection method of making a dash across the red light based on deep learning
CN110400464A (en) * 2019-07-15 2019-11-01 杭州亿圣信息技术有限公司 A kind of cell disobeys the monitoring method of parking
CN107437062B (en) * 2017-06-27 2019-11-12 浙江工业大学 A kind of multi-direction vehicle rough localization method of still image
CN110556000A (en) * 2018-06-04 2019-12-10 义硕智能股份有限公司 Artificial intelligence traffic detection system
CN110598749A (en) * 2019-08-13 2019-12-20 广东数鼎科技有限公司 Image vehicle recognition implementation system based on convolutional neural network AI algorithm
CN110826411A (en) * 2019-10-10 2020-02-21 电子科技大学 Vehicle target rapid identification method based on unmanned aerial vehicle image
CN110969655A (en) * 2019-10-24 2020-04-07 百度在线网络技术（北京）有限公司 Method, device, equipment, storage medium and vehicle for detecting parking space
CN111083362A (en) * 2019-12-13 2020-04-28 山东浪潮通软信息科技有限公司 Method for realizing automatic focusing of vehicle entering and exiting warehouse
CN111079640A (en) * 2019-12-09 2020-04-28 合肥工业大学 Vehicle type identification method and system based on automatic amplification sample
CN111093123A (en) * 2019-12-09 2020-05-01 华中科技大学 Flexible optical network time domain equalization method and system based on composite neural network
CN111401128A (en) * 2020-01-16 2020-07-10 杭州电子科技大学 Method for improving vehicle recognition rate
CN112258472A (en) * 2020-10-20 2021-01-22 大连理工大学 Automatic scoring method for automobile exterior shape
US11386680B2 (en) 2020-03-28 2022-07-12 Wipro Limited System and method of identifying vehicle brand and model
US11961335B1 (en) 2020-06-26 2024-04-16 Harris County Toll Road Authority Dual mode electronic toll road system

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140085475A1 (en) * 2011-05-19 2014-03-27 The Regents Of The University Of California Dynamic bayesian networks for vehicle classification in video
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
CN104809443A (en) * 2015-05-05 2015-07-29 上海交通大学 Convolutional neural network-based license plate detection method and system
CN105354568A (en) * 2015-08-24 2016-02-24 西安电子科技大学 Convolutional neural network based vehicle logo identification method
CN105488517A (en) * 2015-11-30 2016-04-13 杭州全实鹰科技有限公司 Vehicle brand model identification method based on deep learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140085475A1 (en) * 2011-05-19 2014-03-27 The Regents Of The University Of California Dynamic bayesian networks for vehicle classification in video
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
CN104809443A (en) * 2015-05-05 2015-07-29 上海交通大学 Convolutional neural network-based license plate detection method and system
CN105354568A (en) * 2015-08-24 2016-02-24 西安电子科技大学 Convolutional neural network based vehicle logo identification method
CN105488517A (en) * 2015-11-30 2016-04-13 杭州全实鹰科技有限公司 Vehicle brand model identification method based on deep learning

Cited By (57)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106528662A (en) * 2016-10-20 2017-03-22 中山大学 Quick retrieval method and system of vehicle image on the basis of feature geometric constraint
CN106570477A (en) * 2016-10-28 2017-04-19 中国科学院自动化研究所 Vehicle model recognition model construction method based on depth learning and vehicle model recognition method based on depth learning
CN106570477B (en) * 2016-10-28 2019-07-16 中科视语(北京)科技有限公司 Vehicle cab recognition model building method and model recognizing method based on deep learning
CN106529511A (en) * 2016-12-13 2017-03-22 北京旷视科技有限公司 Image structuring method and device
CN106600969A (en) * 2016-12-19 2017-04-26 安徽百诚慧通科技有限公司 Intelligent road comprehensive prevention and control method and system
CN106874418A (en) * 2017-01-24 2017-06-20 成都容豪电子信息科技有限公司 A kind of automobile model data set method for building up for serving deep learning
CN106909901B (en) * 2017-02-28 2020-06-05 北京京东尚科信息技术有限公司 Method and device for detecting object from image
CN106909901A (en) * 2017-02-28 2017-06-30 北京京东尚科信息技术有限公司 The method and device of detection object from image
CN107038448B (en) * 2017-03-01 2020-02-28 中科视语(北京)科技有限公司 Target detection model construction method
CN107038448A (en) * 2017-03-01 2017-08-11 中国科学院自动化研究所 Target detection model building method
CN107316007A (en) * 2017-06-07 2017-11-03 浙江捷尚视觉科技股份有限公司 A kind of monitoring image multiclass object detection and recognition methods based on deep learning
CN107316007B (en) * 2017-06-07 2020-04-03 浙江捷尚视觉科技股份有限公司 Monitoring image multi-class object detection and identification method based on deep learning
CN107437062B (en) * 2017-06-27 2019-11-12 浙江工业大学 A kind of multi-direction vehicle rough localization method of still image
US11670071B2 (en) 2017-07-19 2023-06-06 Microsoft Technology Licensing, Llc Fine-grained image recognition
CN109284749A (en) * 2017-07-19 2019-01-29 微软技术许可有限责任公司 Refine image recognition
CN107545263A (en) * 2017-08-02 2018-01-05 清华大学 A kind of object detecting method and device
CN109624850A (en) * 2017-10-05 2019-04-16 斯特拉德视觉公司 Monitor the method for the blind spot of vehicle and the blind spot monitoring device using this method
CN108171246A (en) * 2017-12-21 2018-06-15 北京科技大学 A kind of clothes salient region detecting method
CN109993031B (en) * 2017-12-29 2021-07-20 杭州海康威视数字技术股份有限公司 Method and device for detecting target illegal driving behaviors of livestock-powered vehicle and camera
CN109993031A (en) * 2017-12-29 2019-07-09 杭州海康威视数字技术股份有限公司 A kind of animal-drawn vehicle target is driven against traffic regulations behavioral value method, apparatus and camera
CN108427920A (en) * 2018-02-26 2018-08-21 杭州电子科技大学 A kind of land and sea border defense object detection method based on deep learning
CN108875911B (en) * 2018-05-25 2021-06-18 同济大学 Parking space detection method
CN108875911A (en) * 2018-05-25 2018-11-23 同济大学 One kind is parked position detecting method
CN110556000A (en) * 2018-06-04 2019-12-10 义硕智能股份有限公司 Artificial intelligence traffic detection system
CN108615401B (en) * 2018-06-11 2021-01-05 西安电子科技大学 Deep learning-based indoor non-uniform light parking space condition identification method
CN108615401A (en) * 2018-06-11 2018-10-02 西安电子科技大学 The non-homogeneous light parking situation recognition methods in interior based on deep learning
CN109086708A (en) * 2018-07-25 2018-12-25 深圳大学 A kind of parking space detection method and system based on deep learning
CN109101934A (en) * 2018-08-20 2018-12-28 广东数相智能科技有限公司 Model recognizing method, device and computer readable storage medium
CN109447171A (en) * 2018-11-05 2019-03-08 电子科技大学 A kind of vehicle attitude classification method based on deep learning
CN109829469A (en) * 2018-11-08 2019-05-31 电子科技大学 A kind of vehicle checking method based on deep learning
CN109583392A (en) * 2018-12-05 2019-04-05 北京纵目安驰智能科技有限公司 A kind of method for detecting parking stalls, device and storage medium
CN109670501A (en) * 2018-12-10 2019-04-23 中国科学院自动化研究所 Object identification and crawl position detection method based on depth convolutional neural networks
CN109753999B (en) * 2018-12-21 2022-06-07 西北工业大学 Fine-grained vehicle type identification method for automobile pictures with any visual angles
CN109753999A (en) * 2018-12-21 2019-05-14 西北工业大学 For the fine granularity model recognizing method of visual angle auto graph
CN109726717B (en) * 2019-01-02 2022-03-01 西南石油大学 Vehicle comprehensive information detection system
CN109726717A (en) * 2019-01-02 2019-05-07 西南石油大学 A kind of vehicle comprehensive information detection system
CN109829398B (en) * 2019-01-16 2020-03-31 北京航空航天大学 Target detection method in video based on three-dimensional convolution network
CN109829398A (en) * 2019-01-16 2019-05-31 北京航空航天大学 A kind of object detection method in video based on Three dimensional convolution network
CN110084285A (en) * 2019-04-08 2019-08-02 安徽艾睿思智能科技有限公司 Fish fine grit classification method based on deep learning
CN110119726A (en) * 2019-05-20 2019-08-13 四川九洲视讯科技有限责任公司 A kind of vehicle brand multi-angle recognition methods based on YOLOv3 model
CN110197589A (en) * 2019-05-29 2019-09-03 杭州诚道科技股份有限公司 A kind of illegal detection method of making a dash across the red light based on deep learning
CN110400464A (en) * 2019-07-15 2019-11-01 杭州亿圣信息技术有限公司 A kind of cell disobeys the monitoring method of parking
CN110400464B (en) * 2019-07-15 2021-06-01 杭州亿圣信息技术有限公司 Method for monitoring illegal parking vehicles in community
CN110598749A (en) * 2019-08-13 2019-12-20 广东数鼎科技有限公司 Image vehicle recognition implementation system based on convolutional neural network AI algorithm
CN110826411B (en) * 2019-10-10 2022-05-03 电子科技大学 Vehicle target rapid identification method based on unmanned aerial vehicle image
CN110826411A (en) * 2019-10-10 2020-02-21 电子科技大学 Vehicle target rapid identification method based on unmanned aerial vehicle image
CN110969655A (en) * 2019-10-24 2020-04-07 百度在线网络技术（北京）有限公司 Method, device, equipment, storage medium and vehicle for detecting parking space
CN110969655B (en) * 2019-10-24 2023-08-18 百度在线网络技术（北京）有限公司 Method, device, equipment, storage medium and vehicle for detecting parking space
CN111079640B (en) * 2019-12-09 2022-03-29 合肥工业大学 Vehicle type identification method and system based on automatic amplification sample
CN111079640A (en) * 2019-12-09 2020-04-28 合肥工业大学 Vehicle type identification method and system based on automatic amplification sample
CN111093123B (en) * 2019-12-09 2020-12-18 华中科技大学 Flexible optical network time domain equalization method and system based on composite neural network
CN111093123A (en) * 2019-12-09 2020-05-01 华中科技大学 Flexible optical network time domain equalization method and system based on composite neural network
CN111083362A (en) * 2019-12-13 2020-04-28 山东浪潮通软信息科技有限公司 Method for realizing automatic focusing of vehicle entering and exiting warehouse
CN111401128A (en) * 2020-01-16 2020-07-10 杭州电子科技大学 Method for improving vehicle recognition rate
US11386680B2 (en) 2020-03-28 2022-07-12 Wipro Limited System and method of identifying vehicle brand and model
US11961335B1 (en) 2020-06-26 2024-04-16 Harris County Toll Road Authority Dual mode electronic toll road system
CN112258472A (en) * 2020-10-20 2021-01-22 大连理工大学 Automatic scoring method for automobile exterior shape

Also Published As

Publication number Publication date
CN105975941B (en) 2019-04-12

Similar Documents

Publication Publication Date Title
CN105975941A (en) Multidirectional vehicle model detection recognition system based on deep learning
Li et al. Automatic pavement crack detection by multi-scale image fusion
CN110942000B (en) Unmanned vehicle target detection method based on deep learning
CN106845487A (en) A kind of licence plate recognition method end to end
CN108830171B (en) Intelligent logistics warehouse guide line visual detection method based on deep learning
CN106340044B (en) Join automatic calibration method and caliberating device outside video camera
CN108898610A (en) A kind of object contour extraction method based on mask-RCNN
CN106022232A (en) License plate detection method based on deep learning
CN111611874B (en) Face mask wearing detection method based on ResNet and Canny
CN104809443A (en) Convolutional neural network-based license plate detection method and system
Shinzato et al. Fast visual road recognition and horizon detection using multiple artificial neural networks
CN105989334B (en) Road detection method based on monocular vision
CN106934455B (en) Remote sensing image optics adapter structure choosing method and system based on CNN
CN111079518B (en) Ground-falling abnormal behavior identification method based on law enforcement and case handling area scene
CN107025657A (en) A kind of vehicle action trail detection method based on video image
Chen et al. Facial expression recognition based on edge detection
CN113326735B (en) YOLOv 5-based multi-mode small target detection method
CN108875819B (en) Object and component joint detection method based on long-term and short-term memory network
CN108776777A (en) The recognition methods of spatial relationship between a kind of remote sensing image object based on Faster RCNN
KR101908481B1 (en) Device and method for pedestraian detection
CN108520203A (en) Multiple target feature extracting method based on fusion adaptive more external surrounding frames and cross pond feature
CN105740844A (en) Insulator cracking fault detection method based on image identification technology
CN106407951A (en) Monocular vision-based nighttime front vehicle detection method
CN105354816B (en) A kind of electronic units fix method and device
CN111310720A (en) Pedestrian re-identification method and system based on graph metric learning

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20190412

Termination date: 20200531

CF01 Termination of patent right due to non-payment of annual fee