CN110287350A - Image search method, device and electronic equipment - Google Patents
Image search method, device and electronic equipment Download PDFInfo
- Publication number
- CN110287350A CN110287350A CN201910580710.9A CN201910580710A CN110287350A CN 110287350 A CN110287350 A CN 110287350A CN 201910580710 A CN201910580710 A CN 201910580710A CN 110287350 A CN110287350 A CN 110287350A
- Authority
- CN
- China
- Prior art keywords
- image
- network model
- target image
- layer group
- retrieved
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/535—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- 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
Abstract
A kind of image search method, device and electronic equipment are provided in the embodiment of the present disclosure, belongs to technical field of data processing, this method comprises: the first layer group using first network model carries out feature extraction to image to be retrieved, obtain fisrt feature image；Feature calculation is carried out to target image by the second network model, obtains eigenmatrix relevant to the target image；Based on the eigenmatrix and the fisrt feature image, search operaqtion is carried out using the second layer group of the first network model, whether the search operaqtion is for determining on the image to be retrieved comprising the target image.By the scheme of the disclosure, mark training data is not needed, the retrieval to target image can be realized, improve the efficiency of image recognition.
Description
Technical field
This disclosure relates to technical field of data processing more particularly to a kind of image search method, device and electronic equipment.
Background technique
With the continuous development of Internet technology, video and image product on network become increasingly abundant, and user watches these
Content is no longer limited to TV, can also be watched by the interested content of internet hunt, and the content of magnanimity is for view
The management of frequency image or picture proposes more challenges.
Digital watermark is the use valence that some identification informations are directly embedded into digital carrier, and do not influence original vector
Value is also not easy to be ascertained and modify again.But it can be identified and be recognized by producer.Letter in the carrier is hidden by these
Breath can achieve confirmation creator of content, buyer, transmission secret information or judge the purpose of whether carrier is tampered.Water
Print be protection information security, realize it is anti-fake trace to the source, the effective way of copyright protection, be the important of Investigation of Information Hiding Technology field
Branch and research direction.
When in the picture of magnanimity including different types of watermark, the picture comprising a certain seed type watermark how is searched,
At a great problem of puzzlement people.
Summary of the invention
In view of this, the embodiment of the present disclosure provides a kind of image search method, device and electronic equipment, at least partly solve
Problems of the prior art.
In a first aspect, the embodiment of the present disclosure provides a kind of image search method, comprising:
Feature extraction is carried out to image to be retrieved using the first layer group of first network model, obtains fisrt feature image；
Feature calculation is carried out to target image by the second network model, obtains feature square relevant to the target image
Battle array；
Based on the eigenmatrix and the fisrt feature image, carried out using the second layer group of the first network model
Search operaqtion, whether the search operaqtion is for determining on the image to be retrieved comprising the target image.
It is described to be treated using the first layer group of first network model according to a kind of specific implementation of the embodiment of the present disclosure
It retrieves image and carries out feature extraction, comprising:
In the first layer group, multiple characteristic patterns are set；
Based on the multiple characteristic pattern, the object on the image to be retrieved is detected.
It is described to be based on the multiple characteristic pattern according to a kind of specific implementation of the embodiment of the present disclosure, to described to be checked
Object on rope image is detected, comprising:
The bounding box of multiple fixed sizes is generated in the first layer group, includes the prediction of object in each bounding box
Value；
Non-maxima suppression is carried out to the bounding box, obtains the final predicted value for object；
Based on the final predicted value for object, the type of the object is determined.
It is described to be treated using the first layer group of first network model according to a kind of specific implementation of the embodiment of the present disclosure
It retrieves image and carries out feature extraction, comprising:
Additional convolutional layer is added in the first layer group, the size of the additional convolutional layer is successively successively decreased.
It is described to be treated using the first layer group of first network model according to a kind of specific implementation of the embodiment of the present disclosure
It retrieves image and carries out feature extraction, comprising:
Multiple convolution filters are arranged to the characteristic layer in the first layer group, the convolution filter is for generating fixation
The predicted value of size,.
According to a kind of specific implementation of the embodiment of the present disclosure, second network model that passes through carries out target image
Feature calculation obtains eigenmatrix relevant to the target image, comprising:
Second network model is obtained for the calculated result of the target image；
Using the calculated result as eigenmatrix relevant to the target image.
It is described to be based on the eigenmatrix and the fisrt feature according to a kind of specific implementation of the embodiment of the present disclosure
Image carries out search operaqtion using the second layer group of the first network model, comprising:
Using the eigenmatrix as the convolution kernel of the last one convolutional layer of the second layer group；
Convolutional calculation is carried out using the characteristic pattern of the convolution kernel and second layer group output, obtains multiple characteristic patterns；
Based on the multiple characteristic pattern, final search result is determined.
It is described to be based on the multiple characteristic pattern according to a kind of specific implementation of the embodiment of the present disclosure, it determines finally
Search result, comprising:
Determine multiple datum windows corresponding with the multiple characteristic pattern；
Classification is carried out to the multiple datum windows and returns calculating；
Return calculating based on the classification as a result, whether determine on the image to be retrieved includes the target image.
Second aspect, the embodiment of the present disclosure provide a kind of image retrieving apparatus, comprising:
Extraction module carries out feature extraction to image to be retrieved for the first layer group using first network model, obtains
Fisrt feature image；
Computing module obtains and the target figure for carrying out feature calculation to target image by the second network model
As relevant eigenmatrix；
Retrieval module utilizes the first network model for being based on the eigenmatrix and the fisrt feature image
Second layer group carry out search operaqtion, whether the search operaqtion for determining on the image to be retrieved comprising the target figure
Picture.
The third aspect, the embodiment of the present disclosure additionally provide a kind of electronic equipment, which includes:
At least one processor；And
The memory being connect at least one processor communication；Wherein,
The memory is stored with the instruction that can be executed by least one processor, and the instruction is by least one processor
It executes, so that at least one processor is able to carry out the figure in any implementation of aforementioned first aspect or first aspect
As search method.
Fourth aspect, the embodiment of the present disclosure additionally provide a kind of non-transient computer readable storage medium, the non-transient meter
Calculation machine readable storage medium storing program for executing stores computer instruction, and the computer instruction is for making the computer execute aforementioned first aspect or the
Image search method in any implementation of one side.
5th aspect, the embodiment of the present disclosure additionally provide a kind of computer program product, which includes
The calculation procedure being stored in non-transient computer readable storage medium, the computer program include program instruction, when the program
When instruction is computer-executed, the computer is made to execute the image in aforementioned first aspect or any implementation of first aspect
Search method.
Image retrieval scheme in the embodiment of the present disclosure, including the use of the first layer group of first network model to figure to be retrieved
As carrying out feature extraction, fisrt feature image is obtained；By the second network model to target image carry out feature calculation, obtain with
The relevant eigenmatrix of the target image；Based on the eigenmatrix and the fisrt feature image, first net is utilized
The second layer group of network model carries out search operaqtion, and whether the search operaqtion is for determining on the image to be retrieved comprising described
Target image.By the scheme of the disclosure, mark training data is not needed, the retrieval to target image can be realized, improve
The efficiency of image recognition.
Detailed description of the invention
It, below will be to needed in the embodiment attached in order to illustrate more clearly of the technical solution of the embodiment of the present disclosure
Figure is briefly described, it should be apparent that, the accompanying drawings in the following description is only some embodiments of the present disclosure, for this field
For those of ordinary skill, without creative efforts, it can also be obtained according to these attached drawings other attached drawings.
Fig. 1 is a kind of image retrieval flow diagram that the embodiment of the present disclosure provides；
Fig. 2 is a kind of schematic diagram for network model for image retrieval that the embodiment of the present disclosure provides；
Fig. 3 is another image retrieval flow diagram that the embodiment of the present disclosure provides；
Fig. 4 is another image retrieval flow diagram that the embodiment of the present disclosure provides；
Fig. 5 is the image retrieving apparatus structural schematic diagram that the embodiment of the present disclosure provides；
Fig. 6 is the electronic equipment schematic diagram that the embodiment of the present disclosure provides.
Specific embodiment
The embodiment of the present disclosure is described in detail with reference to the accompanying drawing.
Illustrate embodiment of the present disclosure below by way of specific specific example, those skilled in the art can be by this specification
Disclosed content understands other advantages and effect of the disclosure easily.Obviously, described embodiment is only the disclosure
A part of the embodiment, instead of all the embodiments.The disclosure can also be subject to reality by way of a different and different embodiment
It applies or applies, the various details in this specification can also be based on different viewpoints and application, in the spirit without departing from the disclosure
Lower carry out various modifications or alterations.It should be noted that in the absence of conflict, the feature in following embodiment and embodiment can
To be combined with each other.Based on the embodiment in the disclosure, those of ordinary skill in the art are without creative efforts
Every other embodiment obtained belongs to the range of disclosure protection.
It should be noted that the various aspects of embodiment within the scope of the appended claims are described below.Ying Xian
And be clear to, aspect described herein can be embodied in extensive diversified forms, and any specific structure described herein
And/or function is only illustrative.Based on the disclosure, it will be understood by one of ordinary skill in the art that one described herein
Aspect can be independently implemented with any other aspect, and can combine the two or both in these aspects or more in various ways.
For example, carry out facilities and equipments in terms of any number set forth herein can be used and/or practice method.In addition, can make
With other than one or more of aspect set forth herein other structures and/or it is functional implement this equipment and/or
Practice the method.
It should also be noted that, diagram provided in following embodiment only illustrates the basic structure of the disclosure in a schematic way
Think, component count, shape and the size when only display is with component related in the disclosure rather than according to actual implementation in schema are drawn
System, when actual implementation kenel, quantity and the ratio of each component can arbitrarily change for one kind, and its assembly layout kenel can also
It can be increasingly complex.
In addition, in the following description, specific details are provided for a thorough understanding of the examples.However, fields
The skilled person will understand that the aspect can be practiced without these specific details.
The embodiment of the present disclosure provides a kind of image search method.Image search method provided in this embodiment can be by a meter
Device is calculated to execute, which can be implemented as software, or be embodied as the combination of software and hardware, which can
To be integrally disposed in server, terminal device etc..
Referring to Fig. 1, a kind of image search method that the embodiment of the present disclosure provides includes the following steps:
S101 carries out feature extraction to image to be retrieved using the first layer group of first network model, obtains fisrt feature
Image.
The training of deep learning model generally requires a large amount of training data, and the acquisition of mass data is logical in practical application
It is often more difficult.How data volume it is less even without data in the case where, realize for specify watermark detection have weight
The realistic meaning wanted.
For this purpose, the scheme of the disclosure utilizes first network model and second in the case where not a large amount of training sample
The mutually matched mode of network model realizes the quick-searching to image to be retrieved, to improve the efficiency of data retrieval.
First network model can realize by building neural network model, as an example, first network model
It may include convolutional layer, pond layer, sample level.
Convolutional layer major parameter includes the size of convolution kernel and the quantity of input feature vector figure, if each convolutional layer may include
The characteristic pattern of dry same size, for same layer characteristic value by the way of shared weight, the convolution kernel in every layer is in the same size.Volume
Lamination carries out convolutional calculation to input picture, and extracts the spatial layout feature of input picture.
It can be connect with sample level behind the feature extraction layer of convolutional layer, sample level is used to ask the office of input picture vector
Portion's average value simultaneously carries out Further Feature Extraction, by the way that sample level is connect with convolutional layer, can guarantee neural network model for
Input audio vector has preferable robustness.
In order to accelerate the training speed of neural network model, pond layer is additionally provided with behind convolutional layer, pond layer uses
The mode in maximum pond handles the output result of convolutional layer, can preferably extract the Invariance feature of input picture.
According to the actual needs, multiple convolutional layers, pond layer, sample level can be selected to form in first network model
First layer group extracts the characteristics of image on image to be retrieved by first layer group.
S102 carries out feature calculation to target image by the second network model, obtains relevant to the target image
Eigenmatrix.
Target image is the subject matter of image retrieval, and as an example, target image can be watermark recovery, and at this time
Image to be retrieved is the image with watermark, and the process of image retrieval is from retrieving in different types of watermarking images
Specific watermarking images.
A component part of the target image as image to be retrieved, usually occupies lesser face in the target image
Product is usually more difficult accurately to retrieve target image on image to be retrieved in the case where smaller training sample.Is constructed thus
Two network models, directly extract the feature of target image using the second network model, and by the calculated target figure of the second model
Input of the feature of picture as first network model, to quickly retrieve target image.
Second network model can realize by building neural network model, as an example, second network model
It may include convolutional layer, pond layer, sample level.
Convolutional layer major parameter includes the size of convolution kernel and the quantity of input feature vector figure, if each convolutional layer may include
The characteristic pattern of dry same size, for same layer characteristic value by the way of shared weight, the convolution kernel in every layer is in the same size.Volume
Lamination carries out convolutional calculation to input picture, and extracts the spatial layout feature of input picture.
It can be connect with sample level behind the feature extraction layer of convolutional layer, sample level is used to ask the office of input picture vector
Portion's average value simultaneously carries out Further Feature Extraction, by the way that sample level is connect with convolutional layer, can guarantee neural network model for
Input audio vector has preferable robustness.
In order to accelerate the training speed of neural network model, pond layer is additionally provided with behind convolutional layer, pond layer uses
The mode in maximum pond handles the output result of convolutional layer, can preferably extract the Invariance feature of input picture.
By the second network model, target image can be calculated, to obtain feature relevant to target object
Matrix can further retrieve target image by this feature matrix.
S103 is based on the eigenmatrix and the fisrt feature image, utilizes the second layer of the first network model
Group carries out search operaqtion, and whether the search operaqtion is for determining on the image to be retrieved comprising the target image.
After obtaining eigenmatrix, using this feature matrix as the input of first network model, pass through this feature matrix and the
The second layer group of one network model is classified and is returned to fisrt feature image, to obtain the retrieval about target image
As a result.Second layer group includes at least a convolutional layer in first network model.
Specifically, by eigenmatrix (information comprising target image) conduct of the target image of the second network model output
The convolution kernel of the last layer convolution in first network model carries out convolution with the characteristic pattern of the layer second from the bottom of first network model
It calculates, generates the last layer characteristic pattern, datum windows (anchor) is generated using the characteristic pattern of the last layer, to datum windows
(anchor) classified and returned, realize the detection to target image (for example, watermark), to realize in no target image mark
In the case where infusing data, target image is detected.
By scheme above, referring to fig. 2, by the output of the second network model as first network model second layer group
Input, using the output of first layer group, common searched targets image.Training is completed in first network model and the second network model
Afterwards, for new target image, mark training data is not needed, it only need to be using new target image as the defeated of the second network model
Enter the detection that can be realized to such target image.
It is described to utilize the first of first network model according to a kind of specific implementation of the embodiment of the present disclosure referring to Fig. 3
Layer group carries out feature extraction to image to be retrieved, comprising:
Multiple characteristic patterns are arranged in the first layer group in S301.
In order to improve the recall precision of first layer group, multiple characteristic patterns can be set in first layer group, and these are special
Sign figure is used as convolution kernel, to improve the feature extraction efficiency to image to be retrieved.
S302 is based on the multiple characteristic pattern, detects to the object on the image to be retrieved.
Specifically, may include steps of during realizing step S302:
S3021 generates the bounding box of multiple fixed sizes in the first layer group, includes object in each bounding box
Predicted value；
S3022 carries out non-maxima suppression to the bounding box, obtains the final predicted value for object；
S3033 determines the type of the object based on the final predicted value for object.
It, according to the actual needs, can be in the first layer group of first network model during carrying out feature extraction
Different convolutional layers is set, for example, according to a kind of specific implementation of the embodiment of the present disclosure, it is described to utilize first network model
First layer group feature extraction is carried out to image to be retrieved, comprising: add additional convolutional layer in the first layer group, it is described additional
The size of convolutional layer is successively successively decreased.By the way that additional convolutional layer is arranged, more characteristics of image can be extracted according to the actual needs.
In addition to this, according to a kind of specific implementation of the embodiment of the present disclosure, the first layer of first network model is utilized
During group carries out feature extraction to image to be retrieved, multiple convolution can also be set to the characteristic layer in the first layer group
Filter, for generating the predicted value of fixed size.
According to a kind of specific implementation of the embodiment of the present disclosure, second network model that passes through carries out target image
Feature calculation obtains eigenmatrix relevant to the target image, comprising: obtains second network model for the mesh
The calculated result of logo image；Using the calculated result as eigenmatrix relevant to the target image.
Referring to fig. 4, described based on the eigenmatrix and described according to a kind of specific implementation of the embodiment of the present disclosure
Fisrt feature image carries out search operaqtion using the second layer group of the first network model, comprising:
S401, using the eigenmatrix as the convolution kernel of the last one convolutional layer of the second layer group.
It, being capable of quickly and accurately base by using eigenmatrix as the convolution kernel of the last one convolutional layer of the second layer group
Image to be retrieved is retrieved in the feature of target image.
S402 carries out convolutional calculation using the characteristic pattern of the convolution kernel and second layer group output, obtains multiple spies
Sign figure.
S403 is based on the multiple characteristic pattern, determines final search result.
Specifically, may include steps of during realizing step S403:
S4031 determines multiple datum windows corresponding with the multiple characteristic pattern.
Window convolution is carried out using the multiple characteristic patterns obtained in step S403 and assesses calculation, and the window of 3*3 can be used for example
Mouth convolution kernel carries out convolution feature calculation.When convolution kernel slides into some position of characteristic pattern, in current sliding window mouth
A region on fisrt feature image is mapped to centered on the heart, it is one corresponding with the center in this region on fisrt feature image
Scale and length-width ratio have been deformed into multiple datum windows.
S4031 carries out classification to the multiple datum windows and returns calculating.
Classification can be carried out to datum windows using k-means clustering algorithm or similar classification regression algorithm and return calculating,
Finally determine one or several typical reference windows.
S4031 returns calculating as a result, whether determine on the image to be retrieved includes the target based on the classification
Image.
By comparing the similarity of the datum windows and target image characteristics matrix that finally return, it can be determined that image to be retrieved
On whether include target image, for example, when the big Mr. Yu's preset value of similarity, assert on image to be retrieved comprising the target figure
Picture.
Corresponding with above method embodiment, referring to Fig. 5, the disclosure additionally provides a kind of image retrieving apparatus 50, packet
It includes:
Extraction module 501 carries out feature extraction to image to be retrieved for the first layer group using first network model, obtains
To fisrt feature image.
The training of deep learning model generally requires a large amount of training data, and the acquisition of mass data is logical in practical application
It is often more difficult.How data volume it is less even without data in the case where, realize for specify watermark detection have weight
The realistic meaning wanted.
For this purpose, the scheme of the disclosure utilizes first network model and second in the case where not a large amount of training sample
The mutually matched mode of network model realizes the quick-searching to image to be retrieved, to improve the efficiency of data retrieval.
First network model can realize by building neural network model, as an example, first network model
It may include convolutional layer, pond layer, sample level.
Convolutional layer major parameter includes the size of convolution kernel and the quantity of input feature vector figure, if each convolutional layer may include
The characteristic pattern of dry same size, for same layer characteristic value by the way of shared weight, the convolution kernel in every layer is in the same size.Volume
Lamination carries out convolutional calculation to input picture, and extracts the spatial layout feature of input picture.
It can be connect with sample level behind the feature extraction layer of convolutional layer, sample level is used to ask the office of input picture vector
Portion's average value simultaneously carries out Further Feature Extraction, by the way that sample level is connect with convolutional layer, can guarantee neural network model for
Input audio vector has preferable robustness.
In order to accelerate the training speed of neural network model, pond layer is additionally provided with behind convolutional layer, pond layer uses
The mode in maximum pond handles the output result of convolutional layer, can preferably extract the Invariance feature of input picture.
According to the actual needs, multiple convolutional layers, pond layer, sample level can be selected to form in first network model
First layer group extracts the characteristics of image on image to be retrieved by first layer group.
Computing module 502 obtains and the target for carrying out feature calculation to target image by the second network model
The relevant eigenmatrix of image.
Target image is the subject matter of image retrieval, and as an example, target image can be watermark recovery, and at this time
Image to be retrieved is the image with watermark, and the process of image retrieval is from retrieving in different types of watermarking images
Specific watermarking images.
A component part of the target image as image to be retrieved, usually occupies lesser face in the target image
Product is usually more difficult accurately to retrieve target image on image to be retrieved in the case where smaller training sample.Is constructed thus
Two network models, directly extract the feature of target image using the second network model, and by the calculated target figure of the second model
Input of the feature of picture as first network model, to quickly retrieve target image.
Second network model can realize by building neural network model, as an example, second network model
It may include convolutional layer, pond layer, sample level.
Convolutional layer major parameter includes the size of convolution kernel and the quantity of input feature vector figure, if each convolutional layer may include
The characteristic pattern of dry same size, for same layer characteristic value by the way of shared weight, the convolution kernel in every layer is in the same size.Volume
Lamination carries out convolutional calculation to input picture, and extracts the spatial layout feature of input picture.
It can be connect with sample level behind the feature extraction layer of convolutional layer, sample level is used to ask the office of input picture vector
Portion's average value simultaneously carries out Further Feature Extraction, by the way that sample level is connect with convolutional layer, can guarantee neural network model for
Input audio vector has preferable robustness.
In order to accelerate the training speed of neural network model, pond layer is additionally provided with behind convolutional layer, pond layer uses
The mode in maximum pond handles the output result of convolutional layer, can preferably extract the Invariance feature of input picture.
By the second network model, target image can be calculated, to obtain feature relevant to target object
Matrix can further retrieve target image by this feature matrix.
Retrieval module 503 utilizes the first network mould for being based on the eigenmatrix and the fisrt feature image
The second layer group of type carries out search operaqtion, and whether the search operaqtion is for determining on the image to be retrieved comprising the target
Image.
After obtaining eigenmatrix, using this feature matrix as the input of first network model, pass through this feature matrix and the
The second layer group of one network model is classified and is returned to fisrt feature image, to obtain the retrieval about target image
As a result.Second layer group includes at least a convolutional layer in first network model.
Specifically, by eigenmatrix (information comprising target image) conduct of the target image of the second network model output
The convolution kernel of the last layer convolution in first network model carries out convolution with the characteristic pattern of the layer second from the bottom of first network model
It calculates, generates the last layer characteristic pattern, datum windows (anchor) is generated using the characteristic pattern of the last layer, to datum windows
(anchor) classified and returned, realize the detection to target image (for example, watermark), to realize in no target image mark
In the case where infusing data, target image is detected.
By scheme above, referring to fig. 2, by the output of the second network model as first network model second layer group
Input, using the output of first layer group, common searched targets image.Training is completed in first network model and the second network model
Afterwards, for new target image, mark training data is not needed, it only need to be using new target image as the defeated of the second network model
Enter the detection that can be realized to such target image.
Fig. 5 shown device can it is corresponding execute above method embodiment in content, what the present embodiment was not described in detail
Part, referring to the content recorded in above method embodiment, details are not described herein.
Referring to Fig. 6, the embodiment of the present disclosure additionally provides a kind of electronic equipment 60, which includes:
At least one processor；And
The memory being connect at least one processor communication；Wherein,
The memory is stored with the instruction that can be executed by least one processor, and the instruction is by least one processor
It executes, so that at least one processor is able to carry out image search method in preceding method embodiment.
The embodiment of the present disclosure additionally provides a kind of non-transient computer readable storage medium, and the non-transient computer is readable to deposit
Storage media stores computer instruction, and the computer instruction is for executing the computer in preceding method embodiment.
The embodiment of the present disclosure additionally provides a kind of computer program product, and the computer program product is non-temporary including being stored in
Calculation procedure on state computer readable storage medium, the computer program include program instruction, when the program instruction is calculated
When machine executes, the computer is made to execute the image search method in preceding method embodiment.
Below with reference to Fig. 6, it illustrates the structural schematic diagrams for the electronic equipment 60 for being suitable for being used to realize the embodiment of the present disclosure.
Electronic equipment in the embodiment of the present disclosure can include but is not limited to such as mobile phone, laptop, Digital Broadcasting Receiver
Device, PDA (personal digital assistant), PAD (tablet computer), PMP (portable media player), car-mounted terminal are (such as vehicle-mounted
Navigation terminal) etc. mobile terminal and such as number TV, desktop computer etc. fixed terminal.Electronics shown in Fig. 6
Equipment is only an example, should not function to the embodiment of the present disclosure and use scope bring any restrictions.
As shown in fig. 6, electronic equipment 60 may include processing unit (such as central processing unit, graphics processor etc.) 601,
It can be loaded into random access storage according to the program being stored in read-only memory (ROM) 602 or from storage device 608
Program in device (RAM) 603 and execute various movements appropriate and processing.In RAM 603, it is also stored with the behaviour of electronic equipment 60
Various programs and data needed for making.Processing unit 601, ROM 602 and RAM 603 are connected with each other by bus 604.It is defeated
Enter/export (I/O) interface 605 and is also connected to bus 604.
In general, following device can connect to I/O interface 605: including such as touch screen, touch tablet, keyboard, mouse, figure
As the input unit 606 of sensor, microphone, accelerometer, gyroscope etc.；Including such as liquid crystal display (LCD), loudspeaking
The output device 607 of device, vibrator etc.；Storage device 608 including such as tape, hard disk etc.；And communication device 609.It is logical
T unit 609 can permit electronic equipment 60 and wirelessly or non-wirelessly be communicated with other equipment to exchange data.Although showing in figure
The electronic equipment 60 with various devices is gone out, it should be understood that being not required for implementing or having all devices shown.
It can alternatively implement or have more or fewer devices.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed from network by communication device 609, or from storage device 608
It is mounted, or is mounted from ROM 602.When the computer program is executed by processing unit 601, the embodiment of the present disclosure is executed
Method in the above-mentioned function that limits.
It should be noted that the above-mentioned computer-readable medium of the disclosure can be computer-readable signal media or meter
Calculation machine readable storage medium storing program for executing either the two any combination.Computer readable storage medium for example can be --- but not
Be limited to --- electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or any above combination.Meter
The more specific example of calculation machine readable storage medium storing program for executing can include but is not limited to: have the electrical connection, just of one or more conducting wires
Taking formula computer disk, hard disk, random access storage device (RAM), read-only memory (ROM), erasable type may be programmed read-only storage
Device (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device,
Or above-mentioned any appropriate combination.In the disclosure, computer readable storage medium can be it is any include or storage journey
The tangible medium of sequence, the program can be commanded execution system, device or device use or in connection.And at this
In open, computer-readable signal media may include in a base band or as the data-signal that carrier wave a part is propagated,
In carry computer-readable program code.The data-signal of this propagation can take various forms, including but not limited to
Electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be computer-readable and deposit
Any computer-readable medium other than storage media, the computer-readable signal media can send, propagate or transmit and be used for
By the use of instruction execution system, device or device or program in connection.Include on computer-readable medium
Program code can transmit with any suitable medium, including but not limited to: electric wire, optical cable, RF (radio frequency) etc. are above-mentioned
Any appropriate combination.
Above-mentioned computer-readable medium can be included in above-mentioned electronic equipment；It is also possible to individualism, and not
It is fitted into the electronic equipment.
Above-mentioned computer-readable medium carries one or more program, when said one or multiple programs are by the electricity
When sub- equipment executes, so that the electronic equipment: obtaining at least two internet protocol addresses；Send to Node evaluation equipment includes institute
State the Node evaluation request of at least two internet protocol addresses, wherein the Node evaluation equipment is internet from described at least two
In protocol address, chooses internet protocol address and return；Receive the internet protocol address that the Node evaluation equipment returns；Its
In, the fringe node in acquired internet protocol address instruction content distributing network.
Alternatively, above-mentioned computer-readable medium carries one or more program, when said one or multiple programs
When being executed by the electronic equipment, so that the electronic equipment: receiving the Node evaluation including at least two internet protocol addresses and request；
From at least two internet protocol address, internet protocol address is chosen；Return to the internet protocol address selected；Wherein,
The fringe node in internet protocol address instruction content distributing network received.
The calculating of the operation for executing the disclosure can be write with one or more programming languages or combinations thereof
Machine program code, above procedure design language include object oriented program language-such as Java, Smalltalk, C+
+, it further include conventional procedural programming language-such as " C " language or similar programming language.Program code can
Fully to execute, partly execute on the user computer on the user computer, be executed as an independent software package,
Part executes on the remote computer or executes on a remote computer or server completely on the user computer for part.
In situations involving remote computers, remote computer can pass through the network of any kind --- including local area network (LAN)
Or wide area network (WAN)-is connected to subscriber computer, or, it may be connected to outer computer (such as utilize Internet service
Provider is connected by internet).
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the disclosure, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, program segment or code of table, a part of the module, program segment or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are actually
It can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it to infuse
Meaning, the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart can be with holding
The dedicated hardware based system of functions or operations as defined in row is realized, or can use specialized hardware and computer instruction
Combination realize.
Being described in unit involved in the embodiment of the present disclosure can be realized by way of software, can also be by hard
The mode of part is realized.Wherein, the title of unit does not constitute the restriction to the unit itself under certain conditions, for example, the
One acquiring unit is also described as " obtaining the unit of at least two internet protocol addresses ".
It should be appreciated that each section of the disclosure can be realized with hardware, software, firmware or their combination.
The above, the only specific embodiment of the disclosure, but the protection scope of the disclosure is not limited thereto, it is any
Those familiar with the art is in the technical scope that the disclosure discloses, and any changes or substitutions that can be easily thought of, all answers
Cover within the protection scope of the disclosure.Therefore, the protection scope of the disclosure should be subject to the protection scope in claims.
Claims (11)
1. a kind of image search method characterized by comprising
Feature extraction is carried out to image to be retrieved using the first layer group of first network model, obtains fisrt feature image；
Feature calculation is carried out to target image by the second network model, obtains eigenmatrix relevant to the target image；
Based on the eigenmatrix and the fisrt feature image, retrieved using the second layer group of the first network model
Operation, whether the search operaqtion is for determining on the image to be retrieved comprising the target image.
2. the method according to claim 1, wherein the first layer group using first network model is to be checked
Rope image carries out feature extraction, comprising:
In the first layer group, multiple characteristic patterns are set；
Based on the multiple characteristic pattern, the object on the image to be retrieved is detected.
3. according to the method described in claim 2, it is characterized in that, described be based on the multiple characteristic pattern, to described to be retrieved
Object on image is detected, comprising:
The bounding box of multiple fixed sizes is generated in the first layer group, includes the predicted value of object in each bounding box；
Non-maxima suppression is carried out to the bounding box, obtains the final predicted value for object；
Based on the final predicted value for object, the type of the object is determined.
4. the method according to claim 1, wherein the first layer group using first network model is to be checked
Rope image carries out feature extraction, comprising:
Additional convolutional layer is added in the first layer group, the size of the additional convolutional layer is successively successively decreased.
5. the method according to claim 1, wherein the first layer group using first network model is to be checked
Rope image carries out feature extraction, comprising:
Multiple convolution filters are arranged to the characteristic layer in the first layer group, the convolution filter is for generating fixed size
Predicted value,.
6. the method according to claim 1, wherein second network model that passes through is special to target image progress
Sign calculates, and obtains eigenmatrix relevant to the target image, comprising:
Second network model is obtained for the calculated result of the target image；
Using the calculated result as eigenmatrix relevant to the target image.
7. the method according to claim 1, wherein described be based on the eigenmatrix and the fisrt feature figure
Picture carries out search operaqtion using the second layer group of the first network model, comprising:
Using the eigenmatrix as the convolution kernel of the last one convolutional layer of the second layer group；
Convolutional calculation is carried out using the characteristic pattern of the convolution kernel and second layer group output, obtains multiple characteristic patterns；
Based on the multiple characteristic pattern, final search result is determined.
8. determining final inspection the method according to the description of claim 7 is characterized in that described be based on the multiple characteristic pattern
Hitch fruit, comprising:
Determine multiple datum windows corresponding with the multiple characteristic pattern；
Classification is carried out to the multiple datum windows and returns calculating；
Return calculating based on the classification as a result, whether determine on the image to be retrieved includes the target image.
9. a kind of image retrieving apparatus characterized by comprising
Extraction module carries out feature extraction to image to be retrieved for the first layer group using first network model, obtains first
Characteristic image；
Computing module obtains and the target image phase for carrying out feature calculation to target image by the second network model
The eigenmatrix of pass；
Retrieval module utilizes the of the first network model for being based on the eigenmatrix and the fisrt feature image
Two layers of group carry out search operaqtion, and whether the search operaqtion is for determining on the image to be retrieved comprising the target image.
10. a kind of electronic equipment, which is characterized in that the electronic equipment includes:
At least one processor；And
The memory being connect at least one described processor communication；Wherein,
The memory is stored with the instruction that can be executed by least one described processor, and described instruction is by described at least one
It manages device to execute, so that at least one described processor is able to carry out image search method described in aforementioned any claim 1-8.
11. a kind of non-transient computer readable storage medium, which stores computer instruction,
The computer instruction is for making the computer execute image search method described in aforementioned any claim 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910580710.9A CN110287350A (en) | 2019-06-29 | 2019-06-29 | Image search method, device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910580710.9A CN110287350A (en) | 2019-06-29 | 2019-06-29 | Image search method, device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110287350A true CN110287350A (en) | 2019-09-27 |
Family
ID=68020124
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910580710.9A Pending CN110287350A (en) | 2019-06-29 | 2019-06-29 | Image search method, device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110287350A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111949145A (en) * | 2020-07-21 | 2020-11-17 | 深圳市鹰硕教育服务股份有限公司 | Intelligent pen image processing method and device and electronic equipment |
| CN112232291A (en) * | 2020-11-06 | 2021-01-15 | 珠海大横琴科技发展有限公司 | Ship retrieval method, device, computer equipment and storage medium |
| WO2021168815A1 (en) * | 2020-02-28 | 2021-09-02 | 华为技术有限公司 | Image retrieval method and image retrieval apparatus |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170236053A1 (en) * | 2015-12-29 | 2017-08-17 | Synopsys, Inc. | Configurable and Programmable Multi-Core Architecture with a Specialized Instruction Set for Embedded Application Based on Neural Networks |
| CN108052881A (en) * | 2017-11-30 | 2018-05-18 | 华中科技大学 | The method and apparatus of multiclass entity object in a kind of real-time detection construction site image |
| CN108960192A (en) * | 2018-07-23 | 2018-12-07 | 北京旷视科技有限公司 | Action identification method and its neural network generation method, device and electronic equipment |
| CN109598231A (en) * | 2018-12-03 | 2019-04-09 | 广州市百果园信息技术有限公司 | A kind of recognition methods of video watermark, device, equipment and storage medium |
| CN109784290A (en) * | 2019-01-23 | 2019-05-21 | 科大讯飞股份有限公司 | A kind of object detection method, device, equipment and readable storage medium storing program for executing |
| CN109886130A (en) * | 2019-01-24 | 2019-06-14 | 上海媒智科技有限公司 | Determination method, apparatus, storage medium and the processor of target object |
-
2019
- 2019-06-29 CN CN201910580710.9A patent/CN110287350A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170236053A1 (en) * | 2015-12-29 | 2017-08-17 | Synopsys, Inc. | Configurable and Programmable Multi-Core Architecture with a Specialized Instruction Set for Embedded Application Based on Neural Networks |
| CN108052881A (en) * | 2017-11-30 | 2018-05-18 | 华中科技大学 | The method and apparatus of multiclass entity object in a kind of real-time detection construction site image |
| CN108960192A (en) * | 2018-07-23 | 2018-12-07 | 北京旷视科技有限公司 | Action identification method and its neural network generation method, device and electronic equipment |
| CN109598231A (en) * | 2018-12-03 | 2019-04-09 | 广州市百果园信息技术有限公司 | A kind of recognition methods of video watermark, device, equipment and storage medium |
| CN109784290A (en) * | 2019-01-23 | 2019-05-21 | 科大讯飞股份有限公司 | A kind of object detection method, device, equipment and readable storage medium storing program for executing |
| CN109886130A (en) * | 2019-01-24 | 2019-06-14 | 上海媒智科技有限公司 | Determination method, apparatus, storage medium and the processor of target object |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021168815A1 (en) * | 2020-02-28 | 2021-09-02 | 华为技术有限公司 | Image retrieval method and image retrieval apparatus |
| CN111949145A (en) * | 2020-07-21 | 2020-11-17 | 深圳市鹰硕教育服务股份有限公司 | Intelligent pen image processing method and device and electronic equipment |
| WO2022016650A1 (en) * | 2020-07-21 | 2022-01-27 | 深圳市鹰硕教育服务有限公司 | Smart pen image processing method, apparatus and electronic device |
| US20230135661A1 (en) * | 2020-07-21 | 2023-05-04 | Shenzhen Eaglesoul Education Service Co., Ltd | Image processing method and apparatus for smart pen, and electronic device |
| CN112232291A (en) * | 2020-11-06 | 2021-01-15 | 珠海大横琴科技发展有限公司 | Ship retrieval method, device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110381368A (en) | Video cover generation method, device and electronic equipment | |
| CN110399848A (en) | Video cover generation method, device and electronic equipment | |
| CN109584276A (en) | Critical point detection method, apparatus, equipment and readable medium | |
| CN108491816A (en) | The method and apparatus for carrying out target following in video | |
| CN110321958A (en) | Training method, the video similarity of neural network model determine method | |
| CN110267097A (en) | Video pushing method, device and electronic equipment based on characteristic of division | |
| CN110287350A (en) | Image search method, device and electronic equipment | |
| CN108846440A (en) | Image processing method and device, computer-readable medium and electronic equipment | |
| CN110222726A (en) | Image processing method, device and electronic equipment | |
| CN111078940B (en) | Image processing method, device, computer storage medium and electronic equipment | |
| CN110189246A (en) | Image stylization generation method, device and electronic equipment | |
| CN110378410A (en) | Multi-tag scene classification method, device and electronic equipment | |
| CN110674349B (en) | Video POI (Point of interest) identification method and device and electronic equipment | |
| CN110401844A (en) | Generation method, device, equipment and the readable medium of net cast strategy | |
| CN110399847A (en) | Extraction method of key frame, device and electronic equipment | |
| CN110189394A (en) | Shape of the mouth as one speaks generation method, device and electronic equipment | |
| CN110415276A (en) | Motion information calculation method, device and electronic equipment | |
| CN112232311B (en) | Face tracking method and device and electronic equipment | |
| CN109495552A (en) | Method and apparatus for updating clicking rate prediction model | |
| CN110222775A (en) | Image processing method, device, electronic equipment and computer readable storage medium | |
| WO2024001545A1 (en) | Song list display information generation method and apparatus, electronic device and storage medium | |
| CN110209658A (en) | Data cleaning method and device | |
| CN114721835A (en) | Method, system, device and medium for predicting energy consumption of edge data center server | |
| CN109389660A (en) | Image generating method and device | |
| CN110288037A (en) | Image processing method, device and electronic equipment |
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 |