CN110704650A - OTA picture tag identification method, electronic device and medium - Google Patents

OTA picture tag identification method, electronic device and medium Download PDF

Info

Publication number
CN110704650A
CN110704650A CN201910934233.1A CN201910934233A CN110704650A CN 110704650 A CN110704650 A CN 110704650A CN 201910934233 A CN201910934233 A CN 201910934233A CN 110704650 A CN110704650 A CN 110704650A
Authority
CN
China
Prior art keywords
ota
picture
label
target
tag
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
CN201910934233.1A
Other languages
Chinese (zh)
Other versions
CN110704650B (en
Inventor
成丹妮
罗超
胡泓
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.)
Ctrip Computer Technology Shanghai Co Ltd
Original Assignee
Ctrip Computer Technology Shanghai Co Ltd
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 Ctrip Computer Technology Shanghai Co Ltd filed Critical Ctrip Computer Technology Shanghai Co Ltd
Priority to CN201910934233.1A priority Critical patent/CN110704650B/en
Publication of CN110704650A publication Critical patent/CN110704650A/en
Application granted granted Critical
Publication of CN110704650B publication Critical patent/CN110704650B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • 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
    • G06F16/51—Indexing; Data structures therefor; Storage structures
    • 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
    • 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
    • G06F16/5866—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Library & Information Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an identification method of an OTA picture label, electronic equipment and a medium, wherein the identification method of the OTA picture label comprises the following steps: constructing an OTA image tag data set according to an OTA image library, wherein the OTA image library comprises an OTA image and a tag corresponding to the OTA image; establishing a label relation model according to the OTA image label data set; establishing an image feature extraction model according to the label relation model, and extracting image features; establishing a label identification model according to the image characteristics; and identifying the label of the OTA picture to be detected by adopting a label identification model. The method is used for automatically identifying the label of the picture by utilizing a deep learning method based on massive picture information and the current label in the OTA hotel.

Description

OTA picture tag identification method, electronic device and medium
Technical Field
The invention belongs to the technical field of identification of OTA picture tags, and particularly relates to an identification method of an OTA picture tag, electronic equipment and a medium.
Background
Images can provide visual representation and communication of information, and thus are applied in large scale in OTA (online travel) scenarios. Effectively and accurately displaying the images can greatly improve the user experience, thereby improving the conversion rate of the user. The content in the OTA image is usually complex, and a single image often contains multiple semantics and has multiple tag attributes. In the prior art, the image label is usually identified manually, which consumes a long time, and has low efficiency, high cost and low accuracy.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides an identification method of an OTA picture label, electronic equipment and a medium.
The invention solves the technical problems through the following technical scheme:
the invention provides an identification method of an OTA picture label, which comprises the following steps:
constructing an OTA image tag data set according to an OTA image library, wherein the OTA image library comprises an OTA image and a tag corresponding to the OTA image;
establishing a label relation model according to the OTA image label data set;
establishing an image feature extraction model according to the label relation model, and extracting image features;
establishing a label identification model according to the image characteristics;
and identifying the label of the OTA picture to be detected by adopting a label identification model.
Preferably, the step of constructing the OTA image tag data set according to the OTA picture library comprises:
counting the occurrence times of the labels in the OTA picture library;
acquiring a target label, wherein the target label is a label with the occurrence frequency larger than a preset numerical value;
and constructing an OTA image tag data set, wherein the OTA image tag data set comprises a target tag and a target picture, and the target picture is a picture corresponding to the target tag in an OTA picture library.
Preferably, the step of building a tag relationship model from the OTA image tag data set comprises:
by one-hot (a kind of code)Mode) of the target label, and constructing a label relation directed graph according to the occurrence frequency of the target label, wherein in the label relation directed graph, the transformation coefficient z from the ith target label to the jth target labelijThe following formula is satisfied:
zij＝qij/qi
wherein q isiRepresenting the number of target pictures in the OTA image tag dataset that include the ith target tag, qijRepresenting the number of target pictures in the OTA image label data set which simultaneously comprise the ith target label and the jth target label, i ∈ [1, N]，j∈[1,N]And N is the number of categories of the target tag. The target tag comprises content which can be identified in the picture, attributes of the content and shooting environments of the picture, such as mountains, seas, days, nights, local buildings, panoramic buildings and the like.
Preferably, the step of establishing a tag relationship model according to the OTA image tag data set further comprises:
converting the target label into a label relation matrix W through graph convolution transformation, wherein the label relation matrix W is used for representing a label relation model,
Figure BDA0002221184300000021
the dimension of D is consistent with the output characteristic dimension of the characteristic transformation function F (X) of the convolutional neural network.
Figure BDA0002221184300000022
For characterizing the real number domain.
Preferably, the transfer function of the image feature extraction model is:
L＝W·F(X)，
wherein, F (X) is a feature transformation function of the convolutional neural network.
Preferably, the transmission function of the tag identification model is:
Figure BDA0002221184300000023
preferably, the step of establishing a tag identification model according to the image features further comprises:
training a convolutional neural network according to an objective loss function, wherein the objective loss function is as follows:
Figure BDA0002221184300000024
wherein y is the number of the ith target label contained in the target pictureiWhen the ith target tag is not included in the target picture, y is 1i＝0，And representing the probability that the target picture obtained according to the label identification model comprises the ith target label.
Preferably, the step of identifying the tag of the OTA picture to be detected by using the tag identification model includes:
target probability of obtaining OTA picture to be detected by adopting label identification model
Figure BDA0002221184300000032
Probability of object
Figure BDA0002221184300000033
The probability of the OTA picture to be tested including the kth target label is determined if
Figure BDA0002221184300000034
Then a corresponding target label tau is set for the OTA picture to be detectedkFor the target threshold corresponding to the kth target tag, k ∈ [1, N]。
Preferably, after the tag identification model is constructed, the identification method further comprises the following steps:
acquiring the target threshold value taukThe target threshold τkIs F1The maximum value of the score corresponds to a threshold value.
The invention also provides electronic equipment which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the computer program to realize the OTA picture tag identification method.
The invention also provides a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the OTA picture tag identification method of the invention.
The positive progress effects of the invention are as follows: according to the invention, based on massive picture information and current tags in the OTA hotel, the automatic tag identification is carried out on the picture by using a deep learning method, the information of the picture can be rapidly and accurately mined, the hidden tag of the picture can be efficiently extracted, the operation and maintenance cost can be greatly saved, the accuracy of picture display can be ensured, the personalized display and retrieval requirements can be met, and the service experience of a user in the OTA scene can be effectively improved.
Drawings
Fig. 1 is a flowchart of an OTA picture tag identification method according to embodiment 1 of the present invention.
Fig. 2 is a schematic structural diagram of an electronic device according to embodiment 2 of the present invention.
Detailed Description
The invention is further illustrated by the following examples, which are not intended to limit the scope of the invention.
Example 1
The embodiment provides an identification method of an OTA picture tag. Referring to fig. 1, the identification method of the OTA picture tag includes the following steps:
and S101, constructing an OTA image tag data set according to the OTA picture library. The OTA picture library comprises OTA pictures and labels corresponding to the OTA pictures.
And S102, establishing a label relation model according to the OTA image label data set.
And S103, establishing an image feature extraction model according to the label relation model, and extracting image features.
And step S104, establishing a label identification model according to the image characteristics.
And S105, identifying the label of the OTA picture to be detected by adopting a label identification model.
In specific implementation, in step S101, the number of times of occurrence of the tag in the OTA picture library is counted; acquiring a target label, wherein the target label is a label with the occurrence frequency larger than a preset numerical value; and constructing an OTA image tag data set, wherein the OTA image tag data set comprises a target tag and a target picture, and the target picture is a picture corresponding to the target tag in an OTA picture library. In a preferred embodiment, the predetermined value is 500, i.e., the number of target tags appearing in the OTA image tag data set is not less than 500.
The existing tags are various, and the tags include identifiable contents in pictures (especially contents with high occurrence frequency) and attributes thereof and shooting environments of the pictures, such as mountains, seas, days, nights, local buildings, panoramic buildings and the like. The relevance exists between different labels, so that the dependency relationship between the labels is effectively mined, and the accuracy of label identification can be improved. Therefore, in step S102, the target tag is converted into a tag vector in a one-hot form, and a tag relationship directed graph is constructed according to the number of times of occurrence of the target tag, in which a transform coefficient z from an ith target tag to a jth target tag is in the tag relationship directed graphijThe following formula is satisfied:
zij＝qij/qi
wherein q isiRepresenting the number of target pictures in the OTA image tag dataset that include the ith target tag, qijRepresenting the number of target pictures in the OTA image label data set which simultaneously comprise the ith target label and the jth target label, i ∈ [1, N]，j∈[1,N]And N is the number of categories of the target tag. For example, a label representing a picture containing "mountain" is in a different category than a label representing a picture containing "day". And converting the target label into a label relation matrix W through graph convolution transformation, wherein the label relation matrix W is used for representing a label relation model.
Figure BDA0002221184300000051
The dimension of D is consistent with the output characteristic dimension of the characteristic transformation function F (X) of the convolutional neural network.
Figure BDA0002221184300000052
For characterizing the real number domain.
The pictures (including images) are dense large-size data, and the effective extraction of high-level semantic features in the pictures is greatly helpful for accurately identifying the labels. Therefore, in step S103, a deep neural network including 55 convolutional layers is constructed as BaseNet (basic network) for image feature extraction. Specifically, model preliminary training is carried out on a large-scale scene classification data set based on public, a training set and a verification set are divided, the model iterates on the training set until convergence is achieved, the weight of a convolutional layer is fixed after the effect on the divided verification set reaches the optimal value, and features before input into a full connection layer are extracted and used for subsequent multi-label recognition image feature extraction. Namely, inputting a picture X, outputting a feature vector through the feature transformation F of the convolution layer, and then performing dot-product on the label relation matrix W to obtain an output feature L after feature change, namely, the transfer function of the image feature extraction model is as follows:
L＝W·F(X)，
wherein, F (X) is a feature transformation function of the convolutional neural network.
In step S104, in order to adapt to the multi-label learning task, the image feature vector is input to the fully connected layer, including N nodes. And dividing a training set to train the network weight of the full connection layer based on the OTA image label data set constructed in the step S101. Finally, outputting an N-dimensional feature vector, and accessing sigmoid (an activation function of a neural network) as an activation function output by the last layer, wherein the probability of the output model predicting the tag, that is, the transfer function of the tag identification model is as follows:
Figure BDA0002221184300000053
outputting an N-dimensional feature vector by the label identification model
Figure BDA0002221184300000054
Each dimension represents the probability that the target picture contains the corresponding target label.
In the training process, an objective loss function is defined, a convolutional neural network is trained according to the objective loss function, and the objective loss function is as follows:
Figure BDA0002221184300000061
wherein y is the number of the ith target label contained in the target pictureiWhen the ith target tag is not included in the target picture, y is 1i0. That is, the vector Y is { Y ═ Y1,y2,y3,…yNAnd the method is used for representing whether the target picture contains a corresponding target label or not.
Figure BDA0002221184300000062
And the probability that the target picture obtained by the identification according to the label identification model comprises the ith target label is represented.
After the tag identification model is constructed, the identification method of the embodiment further includes the following steps:
acquiring the target threshold value taukThe target threshold τkIs F1The maximum value of the score corresponds to a threshold value.
Wherein, F1The expression of the score is:
Figure BDA0002221184300000063
where precision represents accuracy and recall represents recall.
Different alternative threshold values tau 'are adopted for the 1 st target label'1Acquiring tag identification model at different τ'1Precision and recall of an OTA test picture library so as to obtain different tau'1Respectively corresponding to F1And (4) scoring. Wherein, F1Alternative threshold τ 'corresponding to maximum value of score'1Target threshold value tau corresponding to the 1 st target label1. According to a similar mode, a target threshold value tau corresponding to each target label can be obtainedk，k∈[1,N]。
In step S105, a target probability of the OTA picture to be detected is obtained by adopting a label identification model
Figure BDA0002221184300000064
Probability of object
Figure BDA0002221184300000065
The probability of the OTA picture to be tested including the kth target label is determined if
Figure BDA0002221184300000066
And setting a corresponding target label for the OTA picture to be detected.
The identification method of the OTA picture tag in the embodiment is based on massive picture information under the OTA and the current existing tag, and the picture is automatically tagged by using a deep learning method. Practice shows that the identification method of the OTA picture labels can achieve 95% of average accuracy, and the identification of a single picture with multiple labels takes 120 milliseconds. The identification method of the OTA picture label can quickly and accurately mine the information of the picture, efficiently extracts the hidden label of the picture, greatly saves the operation and maintenance cost, ensures the accuracy of picture display, can meet the requirements of personalized display and retrieval, and effectively improves the service experience of users in the OTA scene.
Example 2
Fig. 2 is a schematic structural diagram of an electronic device provided in this embodiment. The electronic device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to realize the identification method of the OTA picture tag of the embodiment 1. The electronic device 30 shown in fig. 2 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiment of the present invention.
As shown in FIG. 2, electronic device 30 may take the form of a general purpose computing device, which may be a server device, for example. The components of the electronic device 30 may include, but are not limited to: the at least one processor 31, the at least one memory 32, and a bus 33 connecting the various system components (including the memory 32 and the processor 31).
The bus 33 includes a data bus, an address bus, and a control bus.
The memory 32 may include volatile memory, such as Random Access Memory (RAM)321 and/or cache memory 322, and may further include Read Only Memory (ROM) 323.
Memory 32 may also include a program/utility 325 having a set (at least one) of program modules 324, such program modules 324 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
The processor 31 executes various functional applications and data processing, such as the identification method of the OTA picture tag according to embodiment 1 of the present invention, by executing the computer program stored in the memory 32.
The electronic device 30 may also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). Such communication may be through input/output (I/O) interfaces 35. Also, model-generating device 30 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via network adapter 36. As shown, network adapter 36 communicates with the other modules of model-generating device 30 via bus 33. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the model-generating device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.
It should be noted that although in the above detailed description several units/modules or sub-units/modules of the electronic device are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.
Example 3
The present embodiment provides a computer-readable storage medium on which a computer program is stored, which when executed by a processor implements the steps of the OTA picture tag identification method of embodiment 1.
More specific examples, among others, that the readable storage medium may employ may include, but are not limited to: a portable disk, a hard disk, random access memory, read only memory, erasable programmable read only memory, optical storage device, magnetic storage device, or any suitable combination of the foregoing.
In a possible implementation, the invention can also be implemented in the form of a program product comprising program code for causing a terminal device to perform the steps of implementing the identification method of OTA picture tags of embodiment 1 when said program product is run on said terminal device.
Where program code for carrying out the invention is written in any combination of one or more programming languages, the program code may be executed entirely on the user device, partly on the user device, as a stand-alone software package, partly on the user device and partly on a remote device or entirely on the remote device.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.

Claims (11)

1. A method for identifying an OTA picture tag is characterized by comprising the following steps:
constructing an OTA image tag data set according to an OTA image library, wherein the OTA image library comprises an OTA image and a tag corresponding to the OTA image;
establishing a label relation model according to the OTA image label data set;
establishing an image feature extraction model according to the label relation model, and extracting image features;
establishing a label identification model according to the image characteristics;
and identifying the label of the OTA picture to be detected by adopting the label identification model.
2. The OTA picture tag identification method of claim 1, wherein the step of constructing the OTA picture tag data set according to the OTA picture library comprises:
counting the occurrence times of the labels in the OTA picture library;
acquiring a target label, wherein the target label is the label of which the occurrence frequency is greater than a preset value;
and constructing the OTA image tag data set, wherein the OTA image tag data set comprises the target tag and a target picture, and the target picture is the picture corresponding to the target tag in the OTA picture library.
3. The OTA picture tag identification method of claim 2, wherein the step of building a tag relationship model from the OTA picture tag data set comprises:
converting the target tags into tag vectors in a one-hot form, and constructing a tag relation directed graph according to the occurrence times of the target tags, wherein in the tag relation directed graph, the conversion coefficient z from the ith target tag to the jth target tag isijThe following formula is satisfied:
zij＝qij/qi
wherein q isiRepresenting the number of said target pictures in said OTA image tag dataset comprising the ith said target tag, qijRepresenting the number of the target pictures in the OTA image label dataset comprising the ith and jth target labels simultaneously, i ∈ [1, N]，j∈[1,N]And N is the number of the categories of the target label.
4. The OTA picture tag identification method of claim 3, wherein the step of building a tag relationship model from the OTA picture tag data set further comprises:
converting the target label into a label relation matrix W through graph convolution transformation, wherein the label relation matrix W is used for representing the label relation model,
Figure FDA0002221184290000021
the dimension of D is consistent with the output feature dimension of the feature transformation function f (x) of the convolutional neural network,
Figure FDA0002221184290000022
for characterizing the real number domain.
5. The OTA picture tag identification method of claim 4, wherein the transfer function of the image feature extraction model is:
L＝W·F(X)。
6. the OTA picture tag identification method of claim 5, wherein the transfer function of the tag identification model is:
Figure FDA0002221184290000023
7. the OTA picture tag identification method of claim 6, wherein the step of establishing a tag identification model according to the image features further comprises:
training the convolutional neural network according to an objective loss function, wherein the objective loss function is as follows:
Figure FDA0002221184290000024
wherein y is the number of the ith target label contained in the target pictureiWhen the ith target tag is not included in the target picture, y is 1i＝0，yiE (0,1) represents whether the ith target label is contained in the diagram,
Figure FDA0002221184290000025
and representing the probability that the target picture obtained according to the label identification model comprises the ith target label.
8. The OTA picture tag identification method of claim 7, wherein the step of identifying the tag of the OTA picture to be detected using the tag identification model comprises:
obtaining the target probability of the OTA picture to be detected by adopting the label identification modelThe target probability
Figure FDA0002221184290000027
The probability that the OTA picture to be detected comprises the kth target label is determined, if
Figure FDA0002221184290000028
Setting the corresponding target label, tau, for the OTA picture to be detectedkFor the target threshold corresponding to the kth target label, k ∈ [1, N]。
9. The identification method of OTA picture tags according to claim 8, wherein after building said tag identification model, said identification method further comprises the steps of:
acquiring the target threshold value taukThe target threshold τkIs F1The maximum value of the score corresponds to a threshold value.
10. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for identifying an OTA picture tag of any of claims 1 to 9 when executing the computer program.
11. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for identification of OTA picture tags according to any of claims 1 to 9.
CN201910934233.1A 2019-09-29 2019-09-29 OTA picture tag identification method, electronic equipment and medium Active CN110704650B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910934233.1A CN110704650B (en) 2019-09-29 2019-09-29 OTA picture tag identification method, electronic equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910934233.1A CN110704650B (en) 2019-09-29 2019-09-29 OTA picture tag identification method, electronic equipment and medium

Publications (2)

Publication Number Publication Date
CN110704650A true CN110704650A (en) 2020-01-17
CN110704650B CN110704650B (en) 2023-04-25

Family

ID=69198005

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910934233.1A Active CN110704650B (en) 2019-09-29 2019-09-29 OTA picture tag identification method, electronic equipment and medium

Country Status (1)

Country Link
CN (1) CN110704650B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113011494A (en) * 2021-03-18 2021-06-22 北京百度网讯科技有限公司 Feature processing method, device, equipment and storage medium
CN113177525A (en) * 2021-05-27 2021-07-27 杭州有赞科技有限公司 AI electronic scale system and weighing method

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8873867B1 (en) * 2012-07-10 2014-10-28 Google Inc. Assigning labels to images
CN105354307A (en) * 2015-11-06 2016-02-24 腾讯科技（深圳）有限公司 Image content identification method and apparatus
US20160125274A1 (en) * 2014-10-31 2016-05-05 Bolei Zhou Discovering visual concepts from weakly labeled image collections
CN107861971A (en) * 2017-09-15 2018-03-30 广州唯品会研究院有限公司 A kind of product search method and device
CN108764208A (en) * 2018-06-08 2018-11-06 Oppo广东移动通信有限公司 Image processing method and device, storage medium, electronic equipment
CN109754015A (en) * 2019-01-02 2019-05-14 京东方科技集团股份有限公司 Neural network and correlation technique, medium and equipment for the identification of paintings multi-tag
CN109816009A (en) * 2019-01-18 2019-05-28 南京旷云科技有限公司 Multi-tag image classification method, device and equipment based on picture scroll product
CN109871909A (en) * 2019-04-16 2019-06-11 京东方科技集团股份有限公司 Image-recognizing method and device
CN109918513A (en) * 2019-03-12 2019-06-21 北京百度网讯科技有限公司 Image processing method, device, server and storage medium
CN110084296A (en) * 2019-04-22 2019-08-02 中山大学 A kind of figure expression learning framework and its multi-tag classification method based on certain semantic
CN110110189A (en) * 2018-02-01 2019-08-09 北京京东尚科信息技术有限公司 Method and apparatus for generating information

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8873867B1 (en) * 2012-07-10 2014-10-28 Google Inc. Assigning labels to images
US20160125274A1 (en) * 2014-10-31 2016-05-05 Bolei Zhou Discovering visual concepts from weakly labeled image collections
CN105354307A (en) * 2015-11-06 2016-02-24 腾讯科技（深圳）有限公司 Image content identification method and apparatus
CN107861971A (en) * 2017-09-15 2018-03-30 广州唯品会研究院有限公司 A kind of product search method and device
CN110110189A (en) * 2018-02-01 2019-08-09 北京京东尚科信息技术有限公司 Method and apparatus for generating information
CN108764208A (en) * 2018-06-08 2018-11-06 Oppo广东移动通信有限公司 Image processing method and device, storage medium, electronic equipment
CN109754015A (en) * 2019-01-02 2019-05-14 京东方科技集团股份有限公司 Neural network and correlation technique, medium and equipment for the identification of paintings multi-tag
CN109816009A (en) * 2019-01-18 2019-05-28 南京旷云科技有限公司 Multi-tag image classification method, device and equipment based on picture scroll product
CN109918513A (en) * 2019-03-12 2019-06-21 北京百度网讯科技有限公司 Image processing method, device, server and storage medium
CN109871909A (en) * 2019-04-16 2019-06-11 京东方科技集团股份有限公司 Image-recognizing method and device
CN110084296A (en) * 2019-04-22 2019-08-02 中山大学 A kind of figure expression learning framework and its multi-tag classification method based on certain semantic

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
P CUI等: "General Knowledgfe embedded image respresentation learning", 《IEEE TRANSACTIONS ON MULTIMEDIA》 *
XIAO KE等: "End-to-End Automatic Image Annotation Based on Deep CNN and multilabel Data Augmentation", 《IEEE TRANSACTIONS ON MULTIMEDIA》 *
杨阳等: "基于深度学习的图像自动标注算法", 《数据采集与处理》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113011494A (en) * 2021-03-18 2021-06-22 北京百度网讯科技有限公司 Feature processing method, device, equipment and storage medium
CN113011494B (en) * 2021-03-18 2024-02-27 北京百度网讯科技有限公司 Feature processing method, device, equipment and storage medium
CN113177525A (en) * 2021-05-27 2021-07-27 杭州有赞科技有限公司 AI electronic scale system and weighing method

Also Published As

Publication number Publication date
CN110704650B (en) 2023-04-25

Similar Documents

Publication Publication Date Title
CN109104620B (en) Short video recommendation method and device and readable medium
CN110458107B (en) Method and device for image recognition
WO2021139191A1 (en) Method for data labeling and apparatus for data labeling
CN113065013B (en) Image annotation model training and image annotation method, system, equipment and medium
CN110866564B (en) Season classification method, system, electronic device and medium for multiple semi-supervised images
CN111429204A (en) Hotel recommendation method, system, electronic equipment and storage medium
CN113051914A (en) Enterprise hidden label extraction method and device based on multi-feature dynamic portrait
CN112287144B (en) Picture retrieval method, equipment and storage medium
CN110704650B (en) OTA picture tag identification method, electronic equipment and medium
CN116431878A (en) Vector retrieval service method, device, equipment and storage medium thereof
CN112749556A (en) Multi-language model training method and device, storage medium and electronic equipment
CN113190746B (en) Recommendation model evaluation method and device and electronic equipment
CN107729944B (en) Identification method and device of popular pictures, server and storage medium
CN114187486A (en) Model training method and related equipment
CN113963446A (en) Behavior recognition method and system based on human skeleton
CN111127057B (en) Multi-dimensional user portrait recovery method
CN111460822A (en) Method, device, equipment and storage medium for theme expansion
CN116958852A (en) Video and text matching method and device, electronic equipment and storage medium
CN115017385A (en) Article searching method, device, equipment and storage medium
CN114627416A (en) Video processing method and device
CN110516024B (en) Map search result display method, device, equipment and storage medium
CN113822589A (en) Intelligent interviewing method, device, equipment and storage medium
CN113239215A (en) Multimedia resource classification method and device, electronic equipment and storage medium
CN115238805B (en) Training method of abnormal data recognition model and related equipment
CN109408706B (en) Image filtering method

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