CN107886073A - A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks - Google Patents

A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks Download PDF

Info

Publication number
CN107886073A
CN107886073A CN201711107713.8A CN201711107713A CN107886073A CN 107886073 A CN107886073 A CN 107886073A CN 201711107713 A CN201711107713 A CN 201711107713A CN 107886073 A CN107886073 A CN 107886073A
Authority
CN
China
Prior art keywords
layer
mrow
attribute
vehicle
msubsup
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
CN201711107713.8A
Other languages
Chinese (zh)
Other versions
CN107886073B (en
Inventor
唐伦
王耀玮
杨恒
刘云龙
陈前斌
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.)
Jilin Province contemporary Zhide Electronic Information Technology Co.,Ltd.
Shenzhen Wanzhida Technology Transfer Center Co ltd
Original Assignee
Chongqing University of Post and Telecommunications
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 Chongqing University of Post and Telecommunications filed Critical Chongqing University of Post and Telecommunications
Priority to CN201711107713.8A priority Critical patent/CN107886073B/en
Publication of CN107886073A publication Critical patent/CN107886073A/en
Application granted granted Critical
Publication of CN107886073B publication Critical patent/CN107886073B/en
Active 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The present invention relates to a kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks, belong to Computer Vision Recognition technical field.This method comprises the following steps：Neural network structure, including convolutional layer, pond layer and full articulamentum are designed, wherein convolutional layer is responsible for feature extraction with pond layer, in the full articulamentum of last layer by calculating target loss function output category result；The training of neutral net is carried out using fine granularity vehicle data collection and label data collection, training method is supervised learning, and the adjustment of weight matrix and offset is carried out using stochastic gradient descent algorithm；The neural network model trained, for carrying out vehicle attribute identification.Present embodiments can apply to more Attribute Recognitions of vehicle, using fine granularity vehicle data collection and more attribute tags data sets the more abstract high-rise expression of vehicle is obtained by convolutional neural networks, from a large amount of training sample learnings to the stealth characteristics for reflecting vehicle essence to be identified, scalability is stronger, and accuracy of identification is also higher.

Description

A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks
Technical field
The invention belongs to Computer Vision Recognition technical field, is related to a kind of fine granularity vehicle based on convolutional neural networks More attribute recognition approaches.
Background technology
Economic development therewith, automobile turn into the most important vehicles of people, and incident urban traffic blocking is asked Topic is even more getting worse, is occurred in increasing " stifled city ".Intelligent transportation system is considered as that current alleviation traffic pressure is optimal Scheme, and a part of the intelligent transportation system as smart city, digital city, it is mainly used in magnitude of traffic flow monitoring, car Monitoring, freeway toll station management, cell intelligent management, parking lot management, traffic police, police criminal detection investigation etc., its Middle vehicle detection identification is the most key part of intelligent transportation system, is to realize the intelligentized important step of traffic administration, is The important subject of computer vision and pattern-recognition in intelligent transportation system.
Vehicle recongnition technique measures in the full-automatic non-parking charge of vehicle, magnitude of traffic flow Con trolling index, automatic vehicle identification, Accident automatic telemetry on highway, onstream inspection, vehicle location, automobile burglar, inspection and tracking rule-breaking vehicle, illegal row To safeguard traffic safety and urban public security, preventing traffic jam, improve the service rate of charge road and bridge, alleviate traffic anxiety shape Condition etc. all has positive role.
In recent years, because powerful feature representation ability, deep learning have caused extensive concern, solve traditional The bottleneck of linear classifier feature extraction.Convolutional neural networks are applied to vehicle classification identification mission.However, this It is all very preliminary for original convolution neutral net framework to work a bit, and many important practical problems are not begged for By more particularly, to the more Attribute Recognition problems of vehicle.
The content of the invention
In view of this, it is an object of the invention to provide a kind of more attributes of fine granularity vehicle based on convolutional neural networks to know Other method.This method is based on convolutional neural networks structure, and using multitask, study mechanism, introducing spatial measure learn, carried simultaneously Vehicle image higher level of abstraction is taken to express.By obtaining image tripletloss and softmaxloss, and then two kinds of combination learning Loss carrys out training network.
To reach above-mentioned purpose, the present invention provides following technical scheme：
A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks, this method comprise the following steps：
S1：Build the neutral net hierarchical structure of multi-task learning；
S2：Using the vehicle data collection of the fine granularity image classification comprising different vehicle attribute, neutral net, study are trained Triplet Loss and Softmax Loss associated losses are so as to carrying out the adjustment of weight parameter matrix and bias；
S3：By the weight parameter matrix and bias in the obtained each layer trained, neutral net is accordingly assigned to In each layer, for vehicle characteristics extraction and the more Attribute Recognitions of vehicle.
Further, the neutral net hierarchical structure includes input layer, convolutional layer, activation primitive layer, normalization layer, Chi Hua Layer, full articulamentum, slicing layer and classification output layer, wherein, input layer includes original image input layer and label data input layer； Convolutional layer, activation primitive layer, normalization layer, pond layer and full articulamentum are responsible for vehicle characteristics extraction；Slicing layer is responsible for number of tags According to section；Classification layer is responsible for output category result.
Further, the step 2 is specially：Cutting is carried out to label data using slicing layer, label data is special with image Levy data and carry out classification learning, different attribute tasks is completed by the associated losses for learning different attribute；Each learning tasks Shared hidden layer, that is, excavate the inner link of different attribute study, distinguishes the difference between learning tasks again；For the difference of vehicle Attribute, different learning tasks are determined, each learning tasks learns an attribute of vehicle, and vehicle attribute is known Not；Row label mark is entered to each attribute of vehicle, obtains more attribute tags data；
The multi-task learning multiple tasks collateral learning, as a result influences each other once to learn multiple tasks；
The vehicle attribute includes vehicle brand, vehicle, the vehicle adeditive attribute of color；
The training method of the training neutral net is the supervised learning of the data of tape label；
The output that the Triplet Loss are triple network structure Triplet Network is lost, the Softmax Loss is the output loss of Softmax regression functions；Wherein triple network structure belongs to metric learning, is by training and learning Practise, reduce the distance between similar sample, while increase the distance between different classes of sample；
The fine granularity image classification be for coarseness is classified, for identify under similar categorization different subclasses it Between nuance.
Further, the neutral net hierarchical structure includes 2 input layers, 5 convolutional layers, 5 activation primitive layers, 2 Normalize layer, 3 pond layers, 3 full articulamentums, 1 slicing layer and some classification output layers；
Wherein, first convolutional layer is 11x11x3 using 96 sizes, and step-length is the convolution kernel of 4 pixels, is to size 224x224x3 input picture carries out convolutional filtering；
Second layer convolutional layer is in response to normalization and Chi Huahou knot using the output of first convolutional layer as input Fruit, and be filtered using the convolution kernel that 256 sizes are 5x5x48；
Three, the 4th and the 5th convolutional layers are connected with each other, not between any pond layer with normalizing among layer；
The convolution kernel that 3rd convolutional layer has 384 sizes to be 3x3x256 is connected to the output of second convolutional layer, is to return One is changing, Chi Huahou result；
4th convolutional layer possesses the core that 384 sizes are 3x3x192；
5th convolutional layer possesses the core that 256 sizes are 3x3x192；
Pond layer uses size as 3, and step-length is 2 pond window, using overlapping pool strategy；
Activation primitive layer is applied to the output of each convolutional layer and full articulamentum, using unsaturated nonlinear function f (x)= Max (0, x) solves nonlinear problem；The output of each hidden neuron is arranged to zero by dropout layers with 0.5 probability, is kept away Exempt from the generation of over-fitting；
Full articulamentum has 4096 neurons, for the classification in whole convolutional neural networks.
Further, in the neutral net hierarchical structure, the input of input layer includes raw image data and more attributes Label data, wherein raw image data input raw image data, more more attribute tags data of attribute tags data input；
This layer of input data is split into N number of branch in the full articulamentum of last layer, wherein N is the vehicle category for needing to learn Property number；
There is data slicer layer after more attribute tags data input layers, label data cutting is cut into slices for N number of label, each Label section possesses an attribute mark, and wherein N is the vehicle attribute number for needing to learn；
By the raw image data after division and label section combination learning, the classification results of each attribute of final output.
Further, the learning method of the associated losses is：
It is N to define for the data image number of training, and classification number is C, and note image ri label classification is li, it is designated asWherein i=1 ... N, f (ri, c) and it is classification c in the output of last layer of full articulamentum, c=1 ..., C, then all instructions Practice image { ri}iNegative log-likelihood sum is defined as in the Softmax loss outputs of last layer of full articulamentum：
The metric learning is：
A triple is formed using three images, is designated as, wherein beingReference sample,Be withTogether The image of classification, referred to as positive sample,Be withDifferent classes of image, referred to as negative sample；Note image feature representation is f (x), F (x) needs to standardize by L2, that is, meets | | f (x) | |2=1, then the feature expression of sample is respectively in tripleWithα is predetermined threshold value, and triple loss Triplet Loss are expressed as：
The weighing apparatus of final classification result is carried out using the associated losses of two kinds of losses of Softmax Loss and Triplet Loss Amount, associated losses are expressed as：
L=ω Ltriplet+(1-ω)Lsoftmax
Wherein, parameter ω represents Triplet Loss proportions.
The beneficial effects of the present invention are：
(1) present invention learns more with the vehicle attributes of convolutional neural networks to combine, be pointedly designed, Training, improve, while learn and export more attribute of vehicle；
(2) present invention is represented using deep learning extraction vehicle characteristics, reduces human intervention.
Brief description of the drawings
In order that the purpose of the present invention, technical scheme and beneficial effect are clearer, the present invention provides drawings described below and carried out Explanation：
Fig. 1 is implementation method schematic diagram of the present invention；
Fig. 2 is based on LsoftmaxAnd LtripletCombination learning network structure model；
Fig. 3 is the neural network structure model suitable for multi-task learning；
Fig. 4 is the structure chart of the convolutional neural networks of the present invention.
Embodiment
Below in conjunction with accompanying drawing, the preferred embodiments of the present invention are described in detail.
As shown in figure 1, Fig. 1 is implementation method schematic diagram of the present invention.Including two modules：Training module and test module. Training module is the training process of neutral net, and training dataset and label data obtain neural mould by multitask neutral net Type, test module are neutral net test process, and test data set passes through characteristic extraction procedure, passes through the neural model trained Obtain final output.
As shown in Fig. 2 Fig. 2 is LsoftmaxAnd LtripletCombination learning network structure model, step is as follows：
Step 201：Fine granularity vehicle data collection is obtained, vehicle image is divided into trigram modelsWherein it isReference sample,Be withGeneric image, referred to as positive sample,Be withDifferent classes of image, referred to as negative sample, I is image label,
Step 202：Triple picture is sent into neutral net and carries out feature learning, learns to share same hidden layer；
Step 203：Output is normalized in last layer of full articulamentum；
Step 204：Structuring study is carried out with label data；
Step 205：L is obtained using the output of step 203softmax；
Step 206：L is calculated using the output of step 204triplet；
Step 207：To LtripletAnd LsoftmaxCombination learning is carried out, obtains associated losses.
As shown in figure 3, Fig. 3 is multi-task learning network structure model, step is as follows：
Step 301：Raw image data is inputted to neutral net, image reaches after convolution, pond etc. are handled to be connected entirely Connect a layer layer 7；
Step 302：Input label data, come in and gone out after slicing treatment and arrive full articulamentum layer 7, wherein slicing treatment Carry out, cut into slices according to attribute classification, and carry out attribute mark in slicing layer；
Step 303：Division processing is done in characteristics of image input to last layer of full articulamentum, wherein divided number is category Property number, then image feature data respectively learnt with label data；
Step 304：Output obtains the learning outcome of different attribute.
As shown in figure 4, Fig. 4 is the specific neural network structure schematic diagram of the present invention, step is as follows：
Step 401：An input of the original image as neutral net；
Step 402：Another input of label data as neutral net；
Step 403：The convolutional layer of neutral net, wherein 403_1 represent first convolutional layer, and 403_2 represents second volume Lamination, by that analogy, 5 convolutional layers, convolutional layer do process of convolution to this layer of input altogether；
Step 404：The activation primitive layer of neutral net, 404_1 represent first activation primitive, and 404_2 represents second Activation primitive layer, by that analogy, 7 activation primitive layers altogether, respectively after convolutional layer and full articulamentum, wherein activation primitive For f (x)=max (0, x)；
Step 405：The normalization layer of neutral net, 405_1 represent first normalization layer, and 405_2 represents second and returned One changes layer, and by that analogy, 2 normalization layers, the input to this layer altogether is normalized；
Step 406：The pond layer of neutral net, 406_1 represent first pond layer, and 406_2 represents second pond layer, By that analogy, three pond layers altogether；
Step 407：The full articulamentum of neutral net, 407_1 represent the full articulamentum of first layer, and 407_2 represents second and connected entirely Layer is connect, by that analogy, three full articulamentums altogether；
Step 407_3：Last layer of the full articulamentum of neutral net, the image feature data of this layer of input is done at division Reason, division number is is learned attribute number；
Step 408：Neutral net dropout layers, 408_1 represent first dropout layer, and 408_2 represents second The output of neuron is arranged to 0 by dropout layers, dropout layers at random according to 0.5 probability；
Step 409：Neutral net slicing layer, slicing treatment is done to label data input, cut according to different attributes Piece；
Step 410：By step 407_3 output and the output combination learning of step 409, associated losses are calculated.
Finally illustrate, preferred embodiment above is merely illustrative of the technical solution of the present invention and unrestricted, although logical Cross above preferred embodiment the present invention is described in detail, it is to be understood by those skilled in the art that can be Various changes are made to it in form and in details, without departing from claims of the present invention limited range.

Claims (6)

  1. A kind of 1. more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks, it is characterised in that：This method include with Lower step：
    S1：Build the neutral net hierarchical structure of multi-task learning；
    S2：Using the vehicle data collection of the fine granularity image classification comprising different vehicle attribute, neutral net, study are trained Triplet Loss and Softmax Loss associated losses are so as to carrying out the adjustment of weight parameter matrix and bias；
    S3：By the weight parameter matrix and bias in the obtained each layer trained, accordingly it is assigned in neutral net Each layer, for vehicle characteristics extraction and the more Attribute Recognitions of vehicle.
  2. 2. a kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks according to claim 1, it is special Sign is：
    The neutral net hierarchical structure includes input layer, convolutional layer, activation primitive layer, normalization layer, pond layer, full connection Layer, slicing layer and classification output layer, wherein, input layer includes original image input layer and label data input layer；Convolutional layer, swash Function layer, normalization layer, pond layer and full articulamentum living are responsible for vehicle characteristics extraction；Slicing layer is responsible for label data section；Point Class layer is responsible for output category result.
  3. 3. a kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks according to claim 1, it is special Sign is：The step 2 is specially：Cutting is carried out to label data using slicing layer, label data enters with image feature data Row classification learning, different attribute tasks is completed by the associated losses for learning different attribute；Each learning tasks share hidden layer, The inner link of different attribute study is excavated, distinguishes the difference between learning tasks again；For the different attribute of vehicle, it is determined that Different learning tasks, each learning tasks learn an attribute of vehicle, and vehicle attribute is identified；To vehicle Each attribute enters row label mark, obtains more attribute tags data；
    The multi-task learning multiple tasks collateral learning, as a result influences each other once to learn multiple tasks；
    The vehicle attribute includes vehicle brand, vehicle, the vehicle adeditive attribute of color；
    The training method of the training neutral net is the supervised learning of the data of tape label；
    The output that the Triplet Loss are triple network structure Triplet Network is lost, the Softmax Loss is the output loss of Softmax regression functions；Wherein triple network structure belongs to metric learning, is by training and learning Practise, reduce the distance between similar sample, while increase the distance between different classes of sample；
    The fine granularity image classification is for coarseness is classified, for identifying under similar categorization between different subclasses Nuance.
  4. 4. a kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks according to claim 1 or 2, its It is characterised by：
    The neutral net hierarchical structure includes 2 input layers, 5 convolutional layers, 5 activation primitive layers, 2 normalization layers, 3 Pond layer, 3 full articulamentums, 1 slicing layer and some classification output layers；
    Wherein, first convolutional layer is 11x11x3 using 96 sizes, and step-length is the convolution kernel of 4 pixels, is to size 224x224x3 input picture carries out convolutional filtering；
    Second layer convolutional layer is in response to normalization and Chi Huahou result using the output of first convolutional layer as inputting, and It is filtered using the convolution kernel that 256 sizes are 5x5x48；
    Three, the 4th and the 5th convolutional layers are connected with each other, not between any pond layer with normalizing among layer；
    The convolution kernel that 3rd convolutional layer has 384 sizes to be 3x3x256 is connected to the output of second convolutional layer, is normalization , Chi Huahou result；
    4th convolutional layer possesses the core that 384 sizes are 3x3x192；
    5th convolutional layer possesses the core that 256 sizes are 3x3x192；
    Pond layer uses size as 3, and step-length is 2 pond window, using overlapping pool strategy；
    Activation primitive layer is applied to the output of each convolutional layer and full articulamentum, using unsaturated nonlinear function f (x)=max (0, x) solves nonlinear problem；The output of hidden neuron is arranged to zero by dropout layers with 0.5 probability, wherein Dropout layers are that it is temporarily abandoned neutral net unit to according to certain probability from network, avoid training plan with this Close the mechanism that phenomenon occurs and design；
    Full articulamentum has 4096 neurons, for the classification in whole convolutional neural networks.
  5. 5. a kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks according to claim 4, it is special Sign is：In the neutral net hierarchical structure, the input of input layer includes raw image data and more attribute tags data, Wherein raw image data input raw image data, more more attribute tags data of attribute tags data input；
    This layer of input data is split into N number of branch in the full articulamentum of last layer, wherein N is the vehicle attribute for needing to learn Number；
    There is data slicer layer after more attribute tags data input layers, label data cutting is cut into slices for N number of label, each label Section possesses an attribute mark, and wherein N is the vehicle attribute number for needing to learn；
    By the raw image data after division and label section combination learning, the classification results of each attribute of final output.
  6. 6. a kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks according to claim 3, it is special Sign is：The learning method of the associated losses is：
    It is N to define for the data image number of training, and classification number is C, note image riLabel classification be li, it is designated as Wherein i=1 ... N, f (ri, c) and it is classification c in the output of last layer of full articulamentum, c=1 ..., C, then all training images {ri}iNegative log-likelihood sum is defined as in the Softmax loss outputs of last layer of full articulamentum：
    <mrow> <msub> <mi>L</mi> <mrow> <mi>s</mi> <mi>o</mi> <mi>f</mi> <mi>t</mi> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <mi>N</mi> </mfrac> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <mo>-</mo> <mi>l</mi> <mi>o</mi> <mi>g</mi> <mfrac> <msup> <mi>e</mi> <mrow> <mi>f</mi> <mrow> <mo>(</mo> <msub> <mi>r</mi> <mi>i</mi> </msub> <mo>,</mo> <msub> <mi>l</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> </mrow> </msup> <mrow> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>c</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>C</mi> </msubsup> <msup> <mi>e</mi> <mrow> <mi>f</mi> <mrow> <mo>(</mo> <msub> <mi>r</mi> <mi>i</mi> </msub> <mo>,</mo> <mi>c</mi> <mo>)</mo> </mrow> </mrow> </msup> </mrow> </mfrac> </mrow>
    The metric learning is：
    A triple is formed using three images, is designated asWherein it isReference sample,Be withIt is generic Image, referred to as positive sample,Be withDifferent classes of image, referred to as negative sample；Note image feature representation is f (x), f (x) Need to standardize by L2, that is, meet | | f (x) | |2=1, then the feature expression of sample is respectively in tripleWithα is predetermined threshold value, and triple loss Triplet Loss are expressed as：
    <mrow> <msub> <mi>L</mi> <mrow> <mi>t</mi> <mi>r</mi> <mi>i</mi> <mi>p</mi> <mi>l</mi> <mi>e</mi> <mi>t</mi> </mrow> </msub> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>a</mi> </msubsup> <mo>,</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>p</mi> </msubsup> <mo>,</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>n</mi> </msubsup> <mo>)</mo> </mrow> <mo>=</mo> <msubsup> <mi>&amp;Sigma;</mi> <mi>i</mi> <mi>N</mi> </msubsup> <mi>m</mi> <mi>a</mi> <mi>x</mi> <mo>{</mo> <mn>0</mn> <mo>,</mo> <mo>|</mo> <mo>|</mo> <mi>f</mi> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>a</mi> </msubsup> <mo>)</mo> </mrow> <mo>-</mo> <mi>f</mi> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>p</mi> </msubsup> <mo>)</mo> </mrow> <mo>|</mo> <msubsup> <mo>|</mo> <mn>2</mn> <mn>2</mn> </msubsup> <mo>-</mo> <mo>|</mo> <mo>|</mo> <mi>f</mi> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>a</mi> </msubsup> <mo>)</mo> </mrow> <mo>-</mo> <mi>f</mi> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>n</mi> </msubsup> <mo>)</mo> </mrow> <mo>|</mo> <msubsup> <mo>|</mo> <mn>2</mn> <mn>2</mn> </msubsup> <mo>+</mo> <mi>&amp;alpha;</mi> <mo>}</mo> </mrow>
    The measurement of final classification result, connection are carried out using the associated losses of two kinds of losses of Softmax Loss and Triplet Loss Loss is closed to be expressed as：
    L=ω Ltriplet+(1-ω)Lsoftmax
    Wherein, parameter ω represents Triplet Loss proportions.
CN201711107713.8A 2017-11-10 2017-11-10 Fine-grained vehicle multi-attribute identification method based on convolutional neural network Active CN107886073B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711107713.8A CN107886073B (en) 2017-11-10 2017-11-10 Fine-grained vehicle multi-attribute identification method based on convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711107713.8A CN107886073B (en) 2017-11-10 2017-11-10 Fine-grained vehicle multi-attribute identification method based on convolutional neural network

Publications (2)

Publication Number Publication Date
CN107886073A true CN107886073A (en) 2018-04-06
CN107886073B CN107886073B (en) 2021-07-27

Family

ID=61780178

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711107713.8A Active CN107886073B (en) 2017-11-10 2017-11-10 Fine-grained vehicle multi-attribute identification method based on convolutional neural network

Country Status (1)

Country Link
CN (1) CN107886073B (en)

Cited By (54)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108460428A (en) * 2018-04-11 2018-08-28 波奇（上海）信息科技有限公司 A kind of method and apparatus of pet image recognition
CN108492575A (en) * 2018-04-11 2018-09-04 济南浪潮高新科技投资发展有限公司 A kind of intelligent vehicle type identifier method
CN108830254A (en) * 2018-06-27 2018-11-16 福州大学 A kind of detection of fine granularity vehicle and recognition methods based on data balancing strategy and intensive attention network
CN108898060A (en) * 2018-05-30 2018-11-27 珠海亿智电子科技有限公司 Based on the model recognizing method of convolutional neural networks under vehicle environment
CN109002845A (en) * 2018-06-29 2018-12-14 西安交通大学 Fine granularity image classification method based on depth convolutional neural networks
CN109063719A (en) * 2018-04-23 2018-12-21 湖北工业大学 A kind of image classification method of co-ordinative construction similitude and category information
CN109063319A (en) * 2018-07-27 2018-12-21 天津大学 A kind of analogy method of bioecosystem neural network based
CN109086811A (en) * 2018-07-19 2018-12-25 南京旷云科技有限公司 Multi-tag image classification method, device and electronic equipment
CN109165699A (en) * 2018-10-17 2019-01-08 中国科学技术大学 Fine granularity image classification method
CN109190687A (en) * 2018-08-16 2019-01-11 新智数字科技有限公司 A kind of nerve network system and its method for identifying vehicle attribute
CN109325547A (en) * 2018-10-23 2019-02-12 苏州科达科技股份有限公司 Non-motor vehicle image multi-tag classification method, system, equipment and storage medium
CN109359684A (en) * 2018-10-17 2019-02-19 苏州大学 Fine granularity model recognizing method based on Weakly supervised positioning and subclass similarity measurement
CN109359562A (en) * 2018-09-29 2019-02-19 佳都新太科技股份有限公司 Target identification method, device, target identification equipment and storage medium
CN109447981A (en) * 2018-11-12 2019-03-08 平安科技（深圳）有限公司 Image-recognizing method and Related product
CN109583481A (en) * 2018-11-13 2019-04-05 杭州电子科技大学 A kind of attribute recognition approach of the fine granularity clothes based on convolutional neural networks
CN109657726A (en) * 2018-12-25 2019-04-19 焦点科技股份有限公司 A kind of picture multi-tag classification method
CN109657697A (en) * 2018-11-16 2019-04-19 中山大学 Classified optimization method based on semi-supervised learning and fine granularity feature learning
CN109658445A (en) * 2018-12-14 2019-04-19 北京旷视科技有限公司 Network training method, increment build drawing method, localization method, device and equipment
CN109766808A (en) * 2018-12-29 2019-05-17 南通贝法瑞信息科技有限公司 A kind of AI algorithm carrying out automobile mark and identification based on two-dimensional tag
CN109816001A (en) * 2019-01-10 2019-05-28 高新兴科技集团股份有限公司 A kind of more attribute recognition approaches of vehicle based on deep learning, device and equipment
CN109886147A (en) * 2019-01-29 2019-06-14 电子科技大学 A kind of more attribute detection methods of vehicle based on the study of single network multiple-task
CN109903238A (en) * 2019-01-24 2019-06-18 华南理工大学 A kind of fine granularity image processing method, device and storage medium
CN109919320A (en) * 2019-01-23 2019-06-21 西北工业大学 Triplet online learning methods based on Semantic hierarchy
CN110059751A (en) * 2019-04-19 2019-07-26 南京链和科技有限公司 A kind of tire code and tire condition recognition methods based on machine learning
CN110197202A (en) * 2019-04-30 2019-09-03 杰创智能科技股份有限公司 A kind of local feature fine granularity algorithm of target detection
CN110399828A (en) * 2019-07-23 2019-11-01 吉林大学 A kind of vehicle recognition methods again of the depth convolutional neural networks based on multi-angle
CN110633722A (en) * 2018-06-22 2019-12-31 北京深鉴智能科技有限公司 Artificial neural network adjusting method and device
CN110689043A (en) * 2019-08-22 2020-01-14 长沙千视通智能科技有限公司 Vehicle fine granularity identification method and device based on multiple attention mechanism
CN110874270A (en) * 2019-11-18 2020-03-10 郑州航空工业管理学院 Deep learning offline reasoning load balancing method for Internet of vehicles
CN110991349A (en) * 2019-12-05 2020-04-10 中国科学院重庆绿色智能技术研究院 Lightweight vehicle attribute identification method based on metric learning
CN111126453A (en) * 2019-12-05 2020-05-08 重庆邮电大学 Fine-grained image classification method and system based on attention mechanism and cut filling
CN111260060A (en) * 2020-02-20 2020-06-09 武汉大学 Object detection neural network hybrid training method and system based on dynamic intensity
CN111275061A (en) * 2018-12-04 2020-06-12 北京嘀嘀无限科技发展有限公司 Vehicle attribute identification method and model training method and device thereof, and electronic equipment
CN111652293A (en) * 2020-05-20 2020-09-11 西安交通大学苏州研究院 Vehicle weight recognition method for multi-task joint discrimination learning
CN111738367A (en) * 2020-08-17 2020-10-02 成都中轨轨道设备有限公司 Part classification method based on image recognition
CN111753912A (en) * 2020-06-28 2020-10-09 中国矿业大学 Coal slime flotation clean coal ash content prediction method based on deep learning
CN111865815A (en) * 2020-09-24 2020-10-30 中国人民解放军国防科技大学 Flow classification method and system based on federal learning
CN111914911A (en) * 2020-07-16 2020-11-10 桂林电子科技大学 Vehicle re-identification method based on improved depth relative distance learning model
CN112052877A (en) * 2020-08-06 2020-12-08 杭州电子科技大学 Image fine-grained classification method based on cascade enhanced network
CN112200260A (en) * 2020-10-19 2021-01-08 厦门大学 Figure attribute identification method based on discarding loss function
CN112257601A (en) * 2020-10-22 2021-01-22 福州大学 Fine-grained vehicle identification method based on data enhancement network of weak supervised learning
WO2021038333A1 (en) * 2019-08-27 2021-03-04 International Business Machines Corporation System-aware selective quantization for performance optimized distributed deep learning
CN112560173A (en) * 2020-12-08 2021-03-26 北京京航计算通讯研究所 Vehicle weather resistance temperature prediction method and device based on deep learning
CN112712048A (en) * 2021-01-11 2021-04-27 武汉爱科森网络科技有限公司 Method for monitoring and early warning of entering of electric vehicle into building
CN112733581A (en) * 2019-10-28 2021-04-30 普天信息技术有限公司 Vehicle attribute identification method and system
CN112801176A (en) * 2021-01-25 2021-05-14 西安交通大学 Deep learning identification method for helicopter flight attitude imbalance data
CN112818805A (en) * 2021-01-26 2021-05-18 四川天翼网络服务有限公司 Fine-grained vehicle attribute analysis system and method based on feature fusion
CN113037646A (en) * 2021-03-04 2021-06-25 西南交通大学 Train communication network flow identification method based on deep learning
CN113056743A (en) * 2018-09-20 2021-06-29 辉达公司 Training neural networks for vehicle re-recognition
CN113326768A (en) * 2021-05-28 2021-08-31 浙江商汤科技开发有限公司 Training method, image feature extraction method, image recognition method and device
CN113378962A (en) * 2021-06-24 2021-09-10 中山大学 Clothing attribute identification method and system based on graph attention network
CN113505761A (en) * 2021-09-08 2021-10-15 阿里巴巴达摩院(杭州)科技有限公司 Vehicle attribute extraction method and device, electronic equipment and storage medium
WO2021213157A1 (en) * 2020-11-20 2021-10-28 平安科技（深圳）有限公司 Model training method, recognition method and apparatus, and device and storage medium
CN114155496A (en) * 2021-11-29 2022-03-08 西安烽火软件科技有限公司 Vehicle attribute multi-task collaborative identification method based on self-attention

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105574215A (en) * 2016-03-04 2016-05-11 哈尔滨工业大学深圳研究生院 Instance-level image search method based on multiple layers of feature representations
US20160321522A1 (en) * 2015-04-30 2016-11-03 Canon Kabushiki Kaisha Devices, systems, and methods for pairwise multi-task feature learning
CN106407352A (en) * 2016-09-06 2017-02-15 广东顺德中山大学卡内基梅隆大学国际联合研究院 Traffic image retrieval method based on depth learning
CN106529402A (en) * 2016-09-27 2017-03-22 中国科学院自动化研究所 Multi-task learning convolutional neural network-based face attribute analysis method
CN106599869A (en) * 2016-12-22 2017-04-26 安徽大学 Vehicle attribute identification method based on multi-task convolutional neural network
US20170140248A1 (en) * 2015-11-13 2017-05-18 Adobe Systems Incorporated Learning image representation by distilling from multi-task networks
CN106709528A (en) * 2017-01-10 2017-05-24 深圳大学 Method and device of vehicle reidentification based on multiple objective function deep learning
CN106897390A (en) * 2017-01-24 2017-06-27 北京大学 Target precise search method based on depth measure study
CN106934392A (en) * 2017-02-28 2017-07-07 西交利物浦大学 Vehicle-logo recognition and attribute forecast method based on multi-task learning convolutional neural networks

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160321522A1 (en) * 2015-04-30 2016-11-03 Canon Kabushiki Kaisha Devices, systems, and methods for pairwise multi-task feature learning
US20170140248A1 (en) * 2015-11-13 2017-05-18 Adobe Systems Incorporated Learning image representation by distilling from multi-task networks
CN105574215A (en) * 2016-03-04 2016-05-11 哈尔滨工业大学深圳研究生院 Instance-level image search method based on multiple layers of feature representations
CN106407352A (en) * 2016-09-06 2017-02-15 广东顺德中山大学卡内基梅隆大学国际联合研究院 Traffic image retrieval method based on depth learning
CN106529402A (en) * 2016-09-27 2017-03-22 中国科学院自动化研究所 Multi-task learning convolutional neural network-based face attribute analysis method
CN106599869A (en) * 2016-12-22 2017-04-26 安徽大学 Vehicle attribute identification method based on multi-task convolutional neural network
CN106709528A (en) * 2017-01-10 2017-05-24 深圳大学 Method and device of vehicle reidentification based on multiple objective function deep learning
CN106897390A (en) * 2017-01-24 2017-06-27 北京大学 Target precise search method based on depth measure study
CN106934392A (en) * 2017-02-28 2017-07-07 西交利物浦大学 Vehicle-logo recognition and attribute forecast method based on multi-task learning convolutional neural networks

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
YIDA WANG 等: ""Self-restraint object recognition by model based CNN learning"", 《2016 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP)》 *
曾成 等: ""基于多任务CNN的人脸活体多属性检测"", 《科学技术与工程》 *

Cited By (81)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108492575A (en) * 2018-04-11 2018-09-04 济南浪潮高新科技投资发展有限公司 A kind of intelligent vehicle type identifier method
CN108460428A (en) * 2018-04-11 2018-08-28 波奇（上海）信息科技有限公司 A kind of method and apparatus of pet image recognition
CN109063719B (en) * 2018-04-23 2022-02-01 湖北工业大学 Image classification method combining structure similarity and class information
CN109063719A (en) * 2018-04-23 2018-12-21 湖北工业大学 A kind of image classification method of co-ordinative construction similitude and category information
CN108898060A (en) * 2018-05-30 2018-11-27 珠海亿智电子科技有限公司 Based on the model recognizing method of convolutional neural networks under vehicle environment
CN110633722A (en) * 2018-06-22 2019-12-31 北京深鉴智能科技有限公司 Artificial neural network adjusting method and device
CN108830254A (en) * 2018-06-27 2018-11-16 福州大学 A kind of detection of fine granularity vehicle and recognition methods based on data balancing strategy and intensive attention network
CN109002845A (en) * 2018-06-29 2018-12-14 西安交通大学 Fine granularity image classification method based on depth convolutional neural networks
CN109002845B (en) * 2018-06-29 2021-04-20 西安交通大学 Fine-grained image classification method based on deep convolutional neural network
CN109086811A (en) * 2018-07-19 2018-12-25 南京旷云科技有限公司 Multi-tag image classification method, device and electronic equipment
CN109086811B (en) * 2018-07-19 2021-06-22 南京旷云科技有限公司 Multi-label image classification method and device and electronic equipment
CN109063319A (en) * 2018-07-27 2018-12-21 天津大学 A kind of analogy method of bioecosystem neural network based
CN109063319B (en) * 2018-07-27 2023-04-07 天津大学 Simulation method of biological ecosystem based on neural network
CN109190687A (en) * 2018-08-16 2019-01-11 新智数字科技有限公司 A kind of nerve network system and its method for identifying vehicle attribute
CN113056743A (en) * 2018-09-20 2021-06-29 辉达公司 Training neural networks for vehicle re-recognition
CN109359562A (en) * 2018-09-29 2019-02-19 佳都新太科技股份有限公司 Target identification method, device, target identification equipment and storage medium
CN109165699B (en) * 2018-10-17 2020-06-26 中国科学技术大学 Fine-grained image classification method
CN109359684A (en) * 2018-10-17 2019-02-19 苏州大学 Fine granularity model recognizing method based on Weakly supervised positioning and subclass similarity measurement
CN109359684B (en) * 2018-10-17 2021-10-29 苏州大学 Fine-grained vehicle type identification method based on weak supervision positioning and subcategory similarity measurement
CN109165699A (en) * 2018-10-17 2019-01-08 中国科学技术大学 Fine granularity image classification method
CN109325547A (en) * 2018-10-23 2019-02-12 苏州科达科技股份有限公司 Non-motor vehicle image multi-tag classification method, system, equipment and storage medium
CN109447981A (en) * 2018-11-12 2019-03-08 平安科技（深圳）有限公司 Image-recognizing method and Related product
CN109583481A (en) * 2018-11-13 2019-04-05 杭州电子科技大学 A kind of attribute recognition approach of the fine granularity clothes based on convolutional neural networks
CN109583481B (en) * 2018-11-13 2021-08-10 杭州电子科技大学 Fine-grained clothing attribute identification method based on convolutional neural network
CN109657697A (en) * 2018-11-16 2019-04-19 中山大学 Classified optimization method based on semi-supervised learning and fine granularity feature learning
CN109657697B (en) * 2018-11-16 2023-01-06 中山大学 Classification optimization method based on semi-supervised learning and fine-grained feature learning
CN111275061A (en) * 2018-12-04 2020-06-12 北京嘀嘀无限科技发展有限公司 Vehicle attribute identification method and model training method and device thereof, and electronic equipment
CN109658445A (en) * 2018-12-14 2019-04-19 北京旷视科技有限公司 Network training method, increment build drawing method, localization method, device and equipment
CN109657726A (en) * 2018-12-25 2019-04-19 焦点科技股份有限公司 A kind of picture multi-tag classification method
CN109766808A (en) * 2018-12-29 2019-05-17 南通贝法瑞信息科技有限公司 A kind of AI algorithm carrying out automobile mark and identification based on two-dimensional tag
CN109816001A (en) * 2019-01-10 2019-05-28 高新兴科技集团股份有限公司 A kind of more attribute recognition approaches of vehicle based on deep learning, device and equipment
CN109919320B (en) * 2019-01-23 2022-04-01 西北工业大学 Triplet network learning method based on semantic hierarchy
CN109919320A (en) * 2019-01-23 2019-06-21 西北工业大学 Triplet online learning methods based on Semantic hierarchy
CN109903238A (en) * 2019-01-24 2019-06-18 华南理工大学 A kind of fine granularity image processing method, device and storage medium
CN109886147A (en) * 2019-01-29 2019-06-14 电子科技大学 A kind of more attribute detection methods of vehicle based on the study of single network multiple-task
CN110059751A (en) * 2019-04-19 2019-07-26 南京链和科技有限公司 A kind of tire code and tire condition recognition methods based on machine learning
CN110197202A (en) * 2019-04-30 2019-09-03 杰创智能科技股份有限公司 A kind of local feature fine granularity algorithm of target detection
CN110399828B (en) * 2019-07-23 2022-10-28 吉林大学 Vehicle re-identification method based on multi-angle deep convolutional neural network
CN110399828A (en) * 2019-07-23 2019-11-01 吉林大学 A kind of vehicle recognition methods again of the depth convolutional neural networks based on multi-angle
CN110689043A (en) * 2019-08-22 2020-01-14 长沙千视通智能科技有限公司 Vehicle fine granularity identification method and device based on multiple attention mechanism
US11551054B2 (en) 2019-08-27 2023-01-10 International Business Machines Corporation System-aware selective quantization for performance optimized distributed deep learning
GB2600872A (en) * 2019-08-27 2022-05-11 Ibm System-aware selective quantization for performance optimized distributed deep learning
WO2021038333A1 (en) * 2019-08-27 2021-03-04 International Business Machines Corporation System-aware selective quantization for performance optimized distributed deep learning
CN112733581B (en) * 2019-10-28 2024-05-21 普天信息技术有限公司 Vehicle attribute identification method and system
CN112733581A (en) * 2019-10-28 2021-04-30 普天信息技术有限公司 Vehicle attribute identification method and system
CN110874270B (en) * 2019-11-18 2022-03-11 郑州航空工业管理学院 Deep learning offline reasoning load balancing method for Internet of vehicles
CN110874270A (en) * 2019-11-18 2020-03-10 郑州航空工业管理学院 Deep learning offline reasoning load balancing method for Internet of vehicles
CN110991349B (en) * 2019-12-05 2023-02-10 中国科学院重庆绿色智能技术研究院 Lightweight vehicle attribute identification method based on metric learning
CN111126453B (en) * 2019-12-05 2022-05-03 重庆邮电大学 Fine-grained image classification method and system based on attention mechanism and cut filling
CN110991349A (en) * 2019-12-05 2020-04-10 中国科学院重庆绿色智能技术研究院 Lightweight vehicle attribute identification method based on metric learning
CN111126453A (en) * 2019-12-05 2020-05-08 重庆邮电大学 Fine-grained image classification method and system based on attention mechanism and cut filling
CN111260060A (en) * 2020-02-20 2020-06-09 武汉大学 Object detection neural network hybrid training method and system based on dynamic intensity
CN111652293A (en) * 2020-05-20 2020-09-11 西安交通大学苏州研究院 Vehicle weight recognition method for multi-task joint discrimination learning
CN111753912A (en) * 2020-06-28 2020-10-09 中国矿业大学 Coal slime flotation clean coal ash content prediction method based on deep learning
CN111914911B (en) * 2020-07-16 2022-04-08 桂林电子科技大学 Vehicle re-identification method based on improved depth relative distance learning model
CN111914911A (en) * 2020-07-16 2020-11-10 桂林电子科技大学 Vehicle re-identification method based on improved depth relative distance learning model
CN112052877B (en) * 2020-08-06 2024-04-09 杭州电子科技大学 Picture fine granularity classification method based on cascade enhancement network
CN112052877A (en) * 2020-08-06 2020-12-08 杭州电子科技大学 Image fine-grained classification method based on cascade enhanced network
CN111738367B (en) * 2020-08-17 2020-11-13 成都中轨轨道设备有限公司 Part classification method based on image recognition
CN111738367A (en) * 2020-08-17 2020-10-02 成都中轨轨道设备有限公司 Part classification method based on image recognition
CN111865815B (en) * 2020-09-24 2020-11-24 中国人民解放军国防科技大学 Flow classification method and system based on federal learning
CN111865815A (en) * 2020-09-24 2020-10-30 中国人民解放军国防科技大学 Flow classification method and system based on federal learning
CN112200260A (en) * 2020-10-19 2021-01-08 厦门大学 Figure attribute identification method based on discarding loss function
CN112200260B (en) * 2020-10-19 2022-06-14 厦门大学 Figure attribute identification method based on discarding loss function
CN112257601B (en) * 2020-10-22 2023-02-21 福州大学 Fine-grained vehicle identification method based on data enhancement network of weak supervised learning
CN112257601A (en) * 2020-10-22 2021-01-22 福州大学 Fine-grained vehicle identification method based on data enhancement network of weak supervised learning
WO2021213157A1 (en) * 2020-11-20 2021-10-28 平安科技（深圳）有限公司 Model training method, recognition method and apparatus, and device and storage medium
CN112560173A (en) * 2020-12-08 2021-03-26 北京京航计算通讯研究所 Vehicle weather resistance temperature prediction method and device based on deep learning
CN112712048A (en) * 2021-01-11 2021-04-27 武汉爱科森网络科技有限公司 Method for monitoring and early warning of entering of electric vehicle into building
CN112801176B (en) * 2021-01-25 2022-12-06 西安交通大学 Deep learning identification method for helicopter flight attitude imbalance data
CN112801176A (en) * 2021-01-25 2021-05-14 西安交通大学 Deep learning identification method for helicopter flight attitude imbalance data
CN112818805A (en) * 2021-01-26 2021-05-18 四川天翼网络服务有限公司 Fine-grained vehicle attribute analysis system and method based on feature fusion
CN112818805B (en) * 2021-01-26 2023-08-01 四川天翼网络股份有限公司 Fine-grained vehicle attribute analysis system and method based on feature fusion
CN113037646A (en) * 2021-03-04 2021-06-25 西南交通大学 Train communication network flow identification method based on deep learning
CN113326768B (en) * 2021-05-28 2023-12-22 浙江商汤科技开发有限公司 Training method, image feature extraction method, image recognition method and device
CN113326768A (en) * 2021-05-28 2021-08-31 浙江商汤科技开发有限公司 Training method, image feature extraction method, image recognition method and device
CN113378962B (en) * 2021-06-24 2023-06-30 中山大学 Garment attribute identification method and system based on graph attention network
CN113378962A (en) * 2021-06-24 2021-09-10 中山大学 Clothing attribute identification method and system based on graph attention network
CN113505761A (en) * 2021-09-08 2021-10-15 阿里巴巴达摩院(杭州)科技有限公司 Vehicle attribute extraction method and device, electronic equipment and storage medium
CN114155496A (en) * 2021-11-29 2022-03-08 西安烽火软件科技有限公司 Vehicle attribute multi-task collaborative identification method based on self-attention
CN114155496B (en) * 2021-11-29 2024-04-26 西安烽火软件科技有限公司 Vehicle attribute multitasking collaborative recognition method based on self-attention

Also Published As

Publication number Publication date
CN107886073B (en) 2021-07-27

Similar Documents

Publication Publication Date Title
CN107886073A (en) A kind of more attribute recognition approaches of fine granularity vehicle based on convolutional neural networks
CN108875674B (en) Driver behavior identification method based on multi-column fusion convolutional neural network
CN107194346A (en) A kind of fatigue drive of car Forecasting Methodology
Peng et al. Uncertainty evaluation of object detection algorithms for autonomous vehicles
CN106803069A (en) Crowd&#39;s level of happiness recognition methods based on deep learning
CN107633220A (en) A kind of vehicle front target identification method based on convolutional neural networks
CN110188807A (en) Tunnel pedestrian target detection method based on cascade super-resolution network and improvement Faster R-CNN
CN110232316A (en) A kind of vehicle detection and recognition method based on improved DSOD model
CN106203330A (en) A kind of vehicle classification method based on convolutional neural networks
CN105938560A (en) Convolutional-neural-network-based vehicle model refined classification system
CN106485214A (en) A kind of eyes based on convolutional neural networks and mouth state identification method
CN104463241A (en) Vehicle type recognition method in intelligent transportation monitoring system
CN107239730A (en) The quaternary number deep neural network model method of intelligent automobile Traffic Sign Recognition
CN102915453B (en) Real-time feedback and update vehicle detection method
CN112434723B (en) Day/night image classification and object detection method based on attention network
CN111738044A (en) Campus violence assessment method based on deep learning behavior recognition
CN110321897A (en) Divide the method for identification non-motor vehicle abnormal behaviour based on image, semantic
Kumeda et al. Vehicle accident and traffic classification using deep convolutional neural networks
Mase et al. Capturing uncertainty in heavy goods vehicles driving behaviour
CN109993058A (en) The recognition methods of road signs based on multi-tag classification
CN117197896A (en) Medical waste temporary storage station abnormal behavior real-time detection method based on deep learning
Karthigadevi et al. Speed limit signboard recognition and audio alert for driver using CNN
CN114241458A (en) Driver behavior recognition method based on attitude estimation feature fusion
Dorrani Traffic Scene Analysis and Classification using Deep Learning
Chen et al. Multi-modal neural network for traffic event detection

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
TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20240117

Address after: Room 407-408, 4th floor, No. 939, Chuangxin Road, high tech Zone, Changchun City, Jilin Province, 130000

Patentee after: Jilin Province contemporary Zhide Electronic Information Technology Co.,Ltd.

Address before: 1003, Building A, Zhiyun Industrial Park, No. 13 Huaxing Road, Henglang Community, Dalang Street, Longhua District, Shenzhen City, Guangdong Province, 518000

Patentee before: Shenzhen Wanzhida Technology Transfer Center Co.,Ltd.

Effective date of registration: 20240117

Address after: 1003, Building A, Zhiyun Industrial Park, No. 13 Huaxing Road, Henglang Community, Dalang Street, Longhua District, Shenzhen City, Guangdong Province, 518000

Patentee after: Shenzhen Wanzhida Technology Transfer Center Co.,Ltd.

Address before: 400065 Chongqing Nan'an District huangjuezhen pass Chongwen Road No. 2

Patentee before: CHONGQING University OF POSTS AND TELECOMMUNICATIONS