CN109271542A - Cover determines method, apparatus, equipment and readable storage medium storing program for executing - Google Patents
Cover determines method, apparatus, equipment and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN109271542A CN109271542A CN201811140229.XA CN201811140229A CN109271542A CN 109271542 A CN109271542 A CN 109271542A CN 201811140229 A CN201811140229 A CN 201811140229A CN 109271542 A CN109271542 A CN 109271542A
- Authority
- CN
- China
- Prior art keywords
- candidate image
- information
- file
- entity
- degree
- 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
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a kind of covers to determine method, apparatus, equipment and readable storage medium storing program for executing.Wherein, method includes: the information for obtaining the candidate image set of file, and obtaining from each candidate image the entity of display；Obtain the content information of the file；Calculate the degree of correlation of the information for the entity that each candidate image is shown and the content information of file；From candidate image set, determine that the degree of correlation meets cover of the candidate image of the first preset requirement as the file, so that the main contents of the cover fitting file determined, realize that picture and text are consistent；Moreover, determining the high cover of the degree of correlation by intelligence, the human cost for manually selecting cover can be saved, the high-quality cover of selection is able to ascend user's clicking rate.
Description
Technical field
The present embodiments relate to computer vision techniques more particularly to a kind of cover to determine method, apparatus, equipment and can
Read storage medium.
Background technique
Cover original refers to one layer outside books and periodicals, refers in particular to be printed on the first of title, author or editor, publisher's title etc. sometimes
Face.In recent years, the electronic documents such as webpage, video, electronics comic books, electronics atlas emerged in multitude, generally using image as this
The cover of a little files.
The prior art arbitrarily chooses an image as cover, for example, arbitrarily choosing one from video generally from file
Cover of a video frame as the video, in another example, cover of the image as the webpage is arbitrarily chosen from webpage.
Since the image in file often has many and picture material numerous and jumbled, the cover quality for causing existing method to be chosen
It is low, it is low with the relevance of file.
Summary of the invention
The embodiment of the present invention provides a kind of cover and determines method, apparatus, equipment and readable storage medium storing program for executing, so that determine
Cover is bonded the main contents of file, realizes that picture and text are consistent.
In a first aspect, the embodiment of the invention provides a kind of covers to determine method, comprising:
The candidate image set of file is obtained, and obtains from each candidate image the information of the entity of display；
Obtain the content information of the file；
Calculate the degree of correlation of the information for the entity that each candidate image is shown and the content information of file；
From candidate image set, determine that the degree of correlation meets envelope of the candidate image of the first preset requirement as the file
Face.
Second aspect, the embodiment of the invention also provides a kind of cover determining device, which includes: the first acquisition mould
Block for obtaining the candidate image set of file, and obtains from each candidate image the information of the entity of display；
Second obtains module, for obtaining the content information of the file；
Computing module, for calculating the degree of correlation of the information for the entity that each candidate image is shown and the content information of file；
Determining module, for from candidate image set, determining that the degree of correlation meets the candidate image work of the first preset requirement
For the cover of the file.
The third aspect, the embodiment of the invention also provides a kind of electronic equipment, the electronic equipment includes:
One or more processors；
Memory, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processing
Device realizes that cover described in any embodiment determines method.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer
Program realizes that cover described in any embodiment determines method when the program is executed by processor.
In the embodiment of the present invention, by obtaining the candidate image set of file, and display is obtained from each candidate image
The information of entity obtains the content information of the file, calculates the information for the entity that each candidate image is shown and the content of file
The degree of correlation of information, to obtain the information of the entity of candidate image and the correlation degree of content information；By from candidate image
In set, determine that the degree of correlation meets cover of the candidate image of the first preset requirement as the file, to scheme according to candidate
The entity information of picture and the correlation degree of content information determine cover, so that the main contents of the cover fitting file determined, real
Existing picture and text are consistent；Moreover, determining the high cover of the degree of correlation by intelligence, the human cost for manually selecting cover can be saved, is chosen
High-quality cover be able to ascend user's clicking rate.
Detailed description of the invention
Fig. 1 is the flow chart that a kind of cover that the embodiment of the present invention one provides determines method；
Fig. 2 is the flow chart that a kind of cover provided by Embodiment 2 of the present invention determines method；
Fig. 3 is the flow chart that a kind of cover that the embodiment of the present invention three provides determines method；
Fig. 4 is a kind of structure chart for cover determining device that the embodiment of the present invention four provides；
Fig. 5 is the structural schematic diagram for a kind of electronic equipment that the embodiment of the present invention five provides.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention rather than limiting the invention.It also should be noted that in order to just
Only the parts related to the present invention are shown in description, attached drawing rather than entire infrastructure.
Embodiment one
Fig. 1 is the flow chart that a kind of cover that the embodiment of the present invention one provides determines method, and the present embodiment is applicable to really
The case where determining the cover of file.It is alternatively possible to which the modification in response to the content information to file operates or to candidate image
The modification of set operates, and executes cover provided in this embodiment and determines method；File can also pulled or distributed
When, it executes cover provided in this embodiment and determines method；Cover provided in this embodiment can also periodically be executed and determine method.
This method can be executed by cover determining device, which can collect by hardware and/or software sharing, and generally
At in the electronic device, specifically including following operation:
S110, the candidate image set for obtaining file, and obtain from each candidate image the information of the entity of display.
Optionally, the file in the present embodiment includes but is not limited to video file, atlas, comic books, audio file and net
Page.Correspondingly, the set and/or file association that the candidate image set of file includes at least two images in file are at least
The set of two images, optionally, the image of file association can be the image that user is directed to this document input.
The present embodiment will determine cover image from these images, describe and distinguish for convenience, these images are known as
Candidate image, the collection that candidate image is constituted are collectively referred to as candidate image set.
In one example, this document is video file, then extracts at least two views according to setpoint frequency from video file
Frequency frame, as candidate image set.Setpoint frequency can be preset, such as one 1 video frame of extraction in 1 second, extraction in 2 seconds view
Frequency frame.Further, at least two video frames are extracted from the preset time period in video file according to setpoint frequency.Based on video
The beginning part of file and ending are usually that advertisement, prelude etc. are associated with little video frame, preset time with video content
Section can be the Central Time section of video file, such as 30%-70% period, so as to be drawn into be associated with video content it is larger
Video frame.
In another example, this document is audio file, then obtains the associated singer's image of audio file, disc image etc.
Set.
After getting the candidate image set of file, display is obtained from each candidate image of candidate image set
The information of entity.For example, the entity that candidate image is shown includes personage, object, the scene etc. that candidate image is shown.Optionally,
The information of entity is information of text type, including characters name, object names, scene information etc..
S120, the content information for obtaining file.
Wherein, the content information of file refers to embody the information of file main contents, optionally, the content information of file
It is the information of text type, such as the title content information of file, the clip Text information of file, the brief introduction content of file letter
Breath, classification information of file etc..
It is worth noting that S110 and S120 can be executed successively, or execute side by side, operation order is without limiting.
The degree of correlation of the content information of S130, the information for calculating the entity that each candidate image is shown and file.
In the present embodiment, the degree of correlation refers to the information of entity and the correlation degree of content information, or is referred to as similar
Degree.
Optionally, the degree of correlation of the information for the entity that each candidate image is shown and the content information of file is calculated separately.Phase
Pass degree calculation method includes but is not limited to cosine similarity, Euclidean distance, Pearson came correlation and Spearman rank correlation system
Number.
S140, from candidate image set, determine the degree of correlation meet the first preset requirement candidate image as file
Cover.
Bigger with the content information degree of correlation of file in the present embodiment, corresponding candidate image can more characterize file
Content also more meets the cover requirement of file.
In order to determine the image for being best suitable for cover from candidate cover set, the first preset requirement is preset.Optionally,
First preset requirement includes that the degree of correlation is more than or equal to preset threshold or degree of correlation maximum.Correspondingly, determine that the degree of correlation is greater than
In cover of the candidate image as file of preset threshold；Alternatively, determining the maximum candidate figure of the degree of correlation in candidate image set
As the cover as file.Have multiple if it is larger than or equal to the candidate image of preset threshold, then chooses any one candidate image
Cover as file.
In the embodiment of the present invention, by obtaining the candidate image set of file, and display is obtained from each candidate image
The information of entity obtains the content information of file, calculates the information for the entity that each candidate image is shown and the content information of file
The degree of correlation, to obtain the information of the entity of candidate image and the correlation degree of content information；By from candidate image set
In, determine that the degree of correlation meets cover of the candidate image of the first preset requirement as file, thus according to the entity of candidate image
The correlation degree of information and content information determines cover, so that the main contents of the cover fitting file determined, realize picture and text phase
Symbol；Moreover, determining the high cover of the degree of correlation by intelligence, the human cost for manually selecting cover, the high-quality envelope of selection can be saved
Face is able to ascend user's clicking rate.
Further, the information of entity can be the information of text type, and the content information of file is also possible to text class
The information of type, text are a kind of semantic spaces of higher-dimension, are capable of the information and content information of accurate characterization entity；Then, pass through
The degree of correlation of the information of two text types is calculated, can relatively accurately obtain the correlation degree of candidate image and file, in turn
Improve the accuracy that cover determines.
Embodiment two
Fig. 2 is the flow chart that a kind of cover provided by Embodiment 2 of the present invention determines method.The embodiment of the present invention is above-mentioned
It is refined on the basis of the technical solution of each embodiment.
Further, it is refined as operation " obtaining the information of the entity of display from each candidate image " " to identify each candidate
The entity that image is shown obtains the description information of entity in each candidate image；By the description information of entity in each candidate image or
The information for the entity that keyword in person's description information is shown as each candidate image ", is embodied as reality for the information of entity
The description information of body or the keyword in description information.
Further, operation " is calculated into the related of the information for the entity that each candidate image is shown and the content information of file
Degree " is refined as " content information of file being generated the first text vector；The information for the entity that each candidate image is shown generates
Second text vector；According to the first text vector and each second text vector, the information for the entity that each candidate image is shown is calculated
With the degree of correlation of the content information of file ", to calculate the degree of correlation by text vector.
A kind of cover as shown in Figure 2 determines method, comprising:
S210, the candidate image set for obtaining file.
The entity that S220, each candidate image of identification are shown, obtains the description information of entity in each candidate image.
Optionally, description information is the information of text type, name description information, the description of the title of article including personage
Information, description information of scene etc..
Optionally, each candidate image is input to entity recognition model, obtains the description information of entity in each candidate image.
Optionally, entity recognition model can be convolutional neural networks model.Entity recognition model includes at least: human face recognition model,
One of article identification model and gesture recognition model.
Optionally, human face recognition model is used to carry out the face in each candidate image public figure's Sex criminals, output
The public figure's name recognized.Article identification model is used to carry out each candidate image the classification of multiclass common user item, and output is known
The Item Title being clipped to, such as mountain, desk, ocean etc..Gesture recognition model is used to carry out gesture recognition to each candidate image, defeated
The posture recognized out, such as play soccer, jump square dance, on a treadmill running etc..
S230, using the keyword in the description information or description information of entity in each candidate image as each candidate image
The information of the entity of display.
Optionally, description is believed by natural language processing (Natural Language Processing, NLP) technology
Breath carries out word cutting, obtains description information keyword, the information as entity.For example, " running on a treadmill " from description information
In obtain keyword " treadmill " and " running ".
S240, the content information for obtaining file.
Optionally, the content information of file includes file content itself or content keyword.
Optionally, by natural language processing (Natural Language Processing, NLP) technology in file
Hold, such as title, abstract, brief introduction or classification, carries out word cutting, obtain content keyword, the content information as file.Example
Such as, keyword " TV play A " and " end " are obtained from title " TV play A end ".
S250, the content information of file is generated into the first text vector.
Optionally, the content information of file is mapped as by N-dimensional vector using vector space model.Further, by file
Content itself or content keyword are mapped as N-dimensional text vector, referred to as the first text vector.When content keyword have it is multiple
When, the first text vector includes multiple vectors of corresponding each keyword.
Information the second text vector of generation of S260, the entity for showing each candidate image.
Optionally, use vector space model by the information MAP of entity for N-dimensional vector.Further, retouching entity
It states the keyword in information or description information and is mapped as N-dimensional text vector, referred to as the second text vector.When in description information
When keyword has multiple, the second text vector includes multiple vectors of corresponding each keyword.
S270, according to the first text vector and each second text vector, calculate the information for the entity that each candidate image is shown
With the degree of correlation of the content information of file.
Optionally, for each second text vector, each second text of each vector sum in the first text vector is calculated
The degree of correlation of each vector in vector, chooses the maximum degree of correlation, as the final degree of correlation.
Optionally, cosine similarity is used to calculate the degree of correlation between two text vectors: with each in the first text vector
The angle of each vector is to consider angle in each second text vector of vector sum, respectively with inner product (each corresponding element of two vectors
Be multiplied summation) compare the product of two vector field homoemorphisms for calculated result, the cosine similarity between two text vectors is obtained, by each second text
The content information of the information and file for the entity that maximum cosine similarity is shown respectively as each candidate image in this vector
The degree of correlation.
Optionally, the inverse of the distance of the first text vector and each second text vector is calculated, it is aobvious as each candidate image
The degree of correlation of the content information of the information and file of the entity shown.
Specifically, it is calculated using NLP technology every in each second text vector of each vector sum in the first text vector
Distance of a vector, such as literal distance, Levenstein distance, Jaro distance etc., by minimum range in each second text vector
The entity shown respectively as each candidate image reciprocal information and file content information the degree of correlation.
As it can be seen that the first text vector and the second text vector are closer, the degree of correlation is higher, and corresponding candidate image is determined
It is bigger for the probability of cover.
S280, from candidate image set, determine the degree of correlation meet the first preset requirement candidate image as file
Cover.
In the present embodiment, by using the keyword in the description information or description information of entity in each candidate image as
The content information of file is generated the first text vector, each candidate image is shown by the information for the entity that each candidate image is shown
The information of the entity shown generates the second text vector, according to the first text vector and each second text vector, calculates each candidate figure
The degree of correlation of the content information of the information and file of entity as shown in improves to calculate the degree of correlation according to text vector
The accuracy of relatedness computation, and then improve the accuracy that cover determines.
Embodiment three
Fig. 3 is the flow chart that a kind of cover that the embodiment of the present invention three provides determines method.The embodiment of the present invention is above-mentioned
It is optimized on the basis of the technical solution of each embodiment.
Further, additional operation " the image quality information for obtaining each candidate image ", and will " from candidate image set,
Determine that the degree of correlation meets cover of the candidate image of the first preset requirement as file " it is refined as " being shown according to each candidate image
Entity information and file content information the degree of correlation and corresponding image quality information, calculate the score of each candidate image；
From candidate image set, determine that score meets cover of the candidate image of the second preset requirement as file ", to pass through image quality
Information and the comprehensive determining cover of the degree of correlation.
A kind of cover as shown in Figure 3 determines method, comprising:
S310, the candidate image set for obtaining file, and obtain from each candidate image the information of the entity of display.
S320, the content information for obtaining file.
S330, the image quality information for obtaining each candidate image.
Optionally, image quality information includes but is not limited to clarity, aesthetics and whether there is or not texts.
Optionally, each candidate image is input to image quality identification model, obtains the image quality information of candidate image.Image quality identification
Model can be convolutional neural networks model.Image quality identification model includes at least: aesthetics identification model, clarity identification model
One of with Text region model.
Optionally, aesthetics identification model is used to carry out aesthetics identification to each candidate image, exports aesthetics.Aesthetics
The training process of identification model includes: to obtain multiple image patterns, and mark the aesthetics 1-10 of each image pattern, and numerical value is higher
Illustrate more beautiful.Aesthetics identification model is trained according to multiple image patterns and aesthetics, makes aesthetics identification model
Output approach the aesthetics of mark.
Clarity identification model is used to carry out clarity identification to each candidate image, exports clarity.Clarity identifies mould
The training process of type includes: to obtain multiple image patterns, and mark the clarity 1-10 of each image pattern, and the higher explanation of numerical value is more
Clearly.Clarity identification model is trained according to multiple image patterns and clarity, makes the output of clarity identification model
Approach the clarity of mark.
Text region model is used to carry out each candidate image the identification of text existence, and whether there is or not texts, such as 1 expression for output
Text is indicated without text, 0.
As it can be seen that the image quality value of information is bigger, illustrate that image quality is better, the probability that corresponding candidate image is confirmed as cover is got over
Greatly.
It is worth noting that S330 can be executed after the candidate image set for obtaining file, before S350, this implementation
Example, which provides only, a kind of optional executes sequence.
The degree of correlation of the content information of S340, the information for calculating the entity that each candidate image is shown and file.
The degree of correlation of the content information of S350, the information of the entity shown according to each candidate image and file, and it is corresponding
Image quality information, calculate the score of each candidate image.
Optionally, the degree of correlation of the content information of the information and file of the entity each candidate image shown, and it is corresponding
Image quality information be weighted summation, obtain the score of each candidate image.
Wherein, when image quality information includes multiple information, the weight of image quality information includes the corresponding power of each information
Weight.Optionally, the weight of the weight of image quality information and the degree of correlation can require to be arranged according to practical cover.For example, cover requires
Closer to file content, then the weight of the degree of correlation is arranged larger.In another example cover requires image quality more preferable, then image quality is believed
The weight of breath is arranged larger.
Optionally, the weight of the weight of image quality information and the degree of correlation can also be obtained by study.Specifically include following three
Step operation:
Step 1: obtaining multiple image patterns and corresponding target score.
Optionally, multiple image patterns are obtained from sample file, such as multiple video frames are extracted from Sample video.People
Work gives a mark to every image pattern according to image quality information and the degree of correlation, as target score.Such as 0-4 points, the bigger expression of score value
More meet cover requirement.
Step 2: the degree of correlation of the content information of the information and file for the entity that every image pattern of acquisition is shown and every
The image quality information of image pattern.
It is obtained in the information and file for the entity that every image pattern is shown using the method that the various embodiments described above provide
Hold the degree of correlation of information and the image quality information of every image pattern, details are not described herein again.
Step 3: the degree of correlation of the content information of the information and file of the entity shown by every image pattern, and
The image quality information fit object score of every image pattern, obtains the weight of the information of entity and the weight of image quality information.
Optionally, using logistical regression fit object score.Logistic regression is linear separable problem
A kind of disaggregated model for being easily achieved and haveing excellent performance is one of the most widely used disaggregated model.Building prediction first
FunctionZ is the weighted sum of the degree of correlation and image quality information, and weight is variable.φ (Z) is predicted value.Then, will
Target score maps between 0-1, by gradient descent method, so that predicted value approaches the mapping value of target score, to obtain
Suitable weight.
S360, from candidate image set, determine that score meets envelope of the candidate image as file of the second preset requirement
Face.
Optionally it is determined that score is more than or equal to cover of the candidate image of preset fraction as file；Alternatively, determining candidate
Cover of the maximum candidate image of score as file in image collection.If score is more than or equal to the candidate image of preset fraction
Have multiple, then therefrom chooses cover of any candidate image as file.
In the present embodiment, the information of the entity by being shown according to each candidate image is related to the content information of file
Degree and corresponding image quality information, calculate the score of each candidate image, so that synthesis pertinence and image quality information, determine file
Cover so that determine cover be bonded file content while, image quality is also preferable.
In the above embodiments, after the information of entity for obtaining display in each candidate image, further includes: according to each
The information for the entity that candidate image is shown judges the legitimacy of corresponding candidate image.Correspondingly, from candidate image set,
Determine the degree of correlation meet the first preset requirement candidate image as the cover of file include: from candidate image set, determination
The degree of correlation meets the first preset requirement and cover of the legal candidate image as file.
Optionally, information unauthorized, such as Drug Reference, gambling information, violence information, pornography etc. are collected in advance.
Information such as sporocarp is consistent with information unauthorized, then determines that corresponding candidate image is illegal；As sporocarp information with not
Legal information is inconsistent, then determines that corresponding candidate image is legal.In turn, determine that the degree of correlation meets the first preset requirement and legal
Cover of the candidate image as file, can guarantee the legitimacy for the cover chosen in this way.
Example IV
Fig. 4 is a kind of structure chart for cover determining device that the embodiment of the present invention four provides.The embodiment of the present invention is suitable for
The case where determining the cover of file.As shown in figure 4, the device specifically includes: first, which obtains module 410, second, obtains module
420, computing module 430 and determining module 440.
First obtains module 410, for obtaining the candidate image set of file, and display is obtained from each candidate image
The information of entity；
Second obtains module 420, for obtaining the content information of file；
Computing module 430, for calculating the related of the information for the entity that each candidate image is shown and the content information of file
Degree；
Determining module 440, for from candidate image set, determining that the degree of correlation meets the candidate image of the first preset requirement
Cover as file.
In the embodiment of the present invention, by obtaining the candidate image set of file, and display is obtained from each candidate image
The information of entity obtains the content information of file, calculates the information for the entity that each candidate image is shown and the content information of file
The degree of correlation, to obtain the information of the entity of candidate image and the correlation degree of content information；By from candidate image set
In, determine that the degree of correlation meets cover of the candidate image of the first preset requirement as file, thus according to the entity of candidate image
The correlation degree of information and content information determines cover, so that the main contents of the cover fitting file determined, realize picture and text phase
Symbol；Moreover, determining the high cover of the degree of correlation by intelligence, the human cost for manually selecting cover, the high-quality envelope of selection can be saved
Face is able to ascend user's clicking rate.
Further, the information of entity is the information of text type, and the content information of file is also the information of text type,
Text is a kind of semantic space of higher-dimension, is capable of the information and content information of accurate characterization entity；Then, by calculating two texts
The degree of correlation of the information of this type can relatively accurately obtain the correlation degree of candidate image and file, and then it is true to improve cover
Fixed accuracy.
Optionally, first module 410 is obtained when obtaining the information of entity of display from each candidate image, it is specific to use
In: it identifies the entity that each candidate image shows, obtains the description information of entity in each candidate image；By entity in each candidate image
Description information or description information in the information of entity that is shown as each candidate image of keyword.Further, first
Module 410 is obtained in the entity for identifying that each candidate image is shown, when obtaining the description information of entity in each candidate image, is specifically used
In: each candidate image is input to entity recognition model, obtains the description information of entity in each candidate image, entity recognition model
It includes at least: one of human face recognition model, article identification model and gesture recognition model.
Optionally, content information of the computing module 430 in the information and file that calculate the entity that each candidate image is shown
When the degree of correlation, it is specifically used for: the content information of file is generated into the first text vector；The entity that each candidate image is shown
Information generates the second text vector；According to the first text vector and each second text vector, the reality that each candidate image is shown is calculated
The degree of correlation of the content information of the information and file of body.Further, computing module 430 is according to the first text vector and Ge
Two text vectors are specifically used for when calculating the degree of correlation of the information for the entity that each candidate image is shown and the content information of file:
The inverse for calculating the distance of the first text vector and each second text vector, the information of the entity shown as each candidate image with
The degree of correlation of the content information of file.
Optionally it is determined that module 440 is in the candidate for from candidate image set, determining the degree of correlation the first preset requirement of satisfaction
When cover of the image as file, it is specifically used for: determines that the degree of correlation is more than or equal to the candidate image of preset threshold as file
Cover；Alternatively, determining cover of the maximum candidate image of the degree of correlation as file in candidate image set.
Optionally, which further includes that third obtains module, for obtaining the image quality information of each candidate image.Correspondingly,
Determining module 44 from candidate image set, is determining envelope of the candidate image of the degree of correlation the first preset requirement of satisfaction as file
When face, it is specifically used for: the degree of correlation of the content information of the information and file of the entity shown according to each candidate image, and it is corresponding
Image quality information, calculate the score of each candidate image；From candidate image set, determine that score meets the time of the second preset requirement
Select cover of the image as file.
Optionally, third obtains module when obtaining the image quality information of each candidate image, is specifically used for: by each candidate image
Be input to image quality identification model, obtain the image quality information of candidate image, image quality identification model includes at least: aesthetics identifies mould
One of type, clarity identification model and Text region model.
Optionally it is determined that module 440 is in the information of the entity shown according to each candidate image and the content information of file
The degree of correlation and corresponding image quality information are specifically used for when calculating the score of each candidate image: showing to each candidate image
The degree of correlation of the content information of the information and file of entity and corresponding image quality information are weighted summation, obtain each candidate
The score of image.
Optionally, which further includes training module, for the information and file in the entity shown to each candidate image
Content information the degree of correlation and corresponding image quality information be weighted before summation, obtain multiple image patterns and correspondence
Target score；Obtain the degree of correlation and every image of the information for the entity that every image pattern is shown and the content information of file
The image quality information of sample；The degree of correlation of the content information of the information and file of the entity shown by every cover sample and every
The image quality information fit object score of cover sample, obtains the weight of the information of entity and the weight of image quality information.
Optionally, which further includes judgment module, for obtained from each candidate image display entity information it
Afterwards, the information of the entity shown according to each candidate image judges the legitimacy of corresponding candidate image.Correspondingly, determining module
440 when from candidate image set, determining that the degree of correlation meets cover of the candidate image of the first preset requirement as file, tool
Body is used for: from candidate image set, determining that the degree of correlation meets the first preset requirement and legal candidate image is as file
Cover.
Optionally, the content information of file is the title content information of file.First, which obtains module 410, is obtaining file
When candidate image set, it is specifically used for: extracts at least two video frames according to setpoint frequency from video file, schemes as candidate
Image set closes.
It is true that cover provided by any embodiment of the invention can be performed in cover determining device provided by the embodiment of the present invention
Determine method, has the corresponding functional module of execution method and beneficial effect.
Embodiment five
Fig. 5 is the structural schematic diagram for a kind of electronic equipment that the embodiment of the present invention five provides.Fig. 5, which is shown, to be suitable for being used in fact
The block diagram of the example electronic device 12 of existing embodiment of the present invention.The electronic equipment 12 that Fig. 5 is shown is only an example, no
The function and use scope for coping with the embodiment of the present invention bring any restrictions.
As shown in figure 5, electronic equipment 12 is showed in the form of universal computing device.The component of electronic equipment 12 may include
But be not limited to: one or more processor or processing unit 16, system storage 28, connect different system components (including
System storage 28 and processing unit 16) bus 18.
Bus 18 indicates one of a few class bus structures or a variety of, including memory bus or Memory Controller,
Peripheral bus, graphics acceleration port, processor or the local bus using any bus structures in a variety of bus structures.It lifts
For example, these architectures include but is not limited to industry standard architecture (ISA) bus, microchannel architecture (MAC)
Bus, enhanced isa bus, Video Electronics Standards Association (VESA) local bus and peripheral component interconnection (PCI) bus.
Electronic equipment 12 typically comprises a variety of computer system readable media.These media can be it is any can be electric
The usable medium that sub- equipment 12 accesses, including volatile and non-volatile media, moveable and immovable medium.
System storage 28 may include the computer system readable media of form of volatile memory, such as arbitrary access
Memory (RAM) 30 and/or cache memory 32.Electronic equipment 12 may further include other removable/not removable
Dynamic, volatile/non-volatile computer system storage medium.Only as an example, storage system 34 can be used for read and write can not
Mobile, non-volatile magnetic media (Fig. 5 do not show, commonly referred to as " hard disk drive ").Although being not shown in Fig. 5, Ke Yiti
For the disc driver for being read and write to removable non-volatile magnetic disk (such as " floppy disk "), and to moving non-volatile light
The CD drive of disk (such as CD-ROM, DVD-ROM or other optical mediums) read-write.In these cases, each driver
It can be connected by one or more data media interfaces with bus 18.Memory 28 may include that at least one program produces
Product, the program product have one group of (for example, at least one) program module, these program modules are configured to perform of the invention each
The function of embodiment.
Program/utility 40 with one group of (at least one) program module 42 can store in such as memory 28
In, such program module 42 include but is not limited to operating system, one or more application program, other program modules and
It may include the realization of network environment in program data, each of these examples or certain combination.Program module 42 is usual
Execute the function and/or method in embodiment described in the invention.
Electronic equipment 12 can also be with one or more external equipments 14 (such as keyboard, sensing equipment, display 24 etc.)
Communication, can also be enabled a user to one or more equipment interact with the electronic equipment 12 communicate, and/or with make the electricity
Any equipment (such as network interface card, modem etc.) that sub- equipment 12 can be communicated with one or more of the other calculating equipment
Communication.This communication can be carried out by input/output (I/O) interface 22.Also, electronic equipment 12 can also be suitable by network
Orchestration 20 and one or more network (such as local area network (LAN), wide area network (WAN) and/or public network, such as internet)
Communication.As shown, network adapter 20 is communicated by bus 18 with other modules of equipment/terminal/server 12.It should be bright
It is white, although not shown in the drawings, other hardware and/or software module can be used in conjunction with electronic equipment 12, including but not limited to: micro-
Code, device driver, redundant processing unit, external disk drive array, RAID system, tape drive and data backup
Storage system etc..
Processing unit 16 by the program that is stored in system storage 28 of operation, thereby executing various function application and
Data processing, such as realize that cover provided by the embodiment of the present invention determines method.
Embodiment six
The embodiment of the present invention six additionally provides a kind of computer readable storage medium, is stored thereon with computer program, should
Realize that cover provided by any embodiment of the present invention determines method when program is executed by processor.
The computer storage medium of the embodiment of the present invention, can be using any of one or more computer-readable media
Combination.Computer-readable medium can be computer-readable signal media or computer readable storage medium.It is computer-readable
Storage medium for example may be-but not limited to-the system of electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, device or
Device, or any above combination.The more specific example (non exhaustive list) of computer readable storage medium includes: tool
There are electrical connection, the portable computer diskette, hard disk, random access memory (RAM), read-only memory of one or more conducting wires
(ROM), erasable programmable read only memory (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 this document, computer-readable storage
Medium can be any tangible medium for including or store program, which can be commanded execution system, device or device
Using or it is in connection.
Computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for
By the use of instruction execution system, device or device or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including --- but it is unlimited
In wireless, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
The computer for executing operation of the present invention can be write with one or more programming languages or combinations thereof
Program code, programming language include object oriented program language-such as Java, Smalltalk, C++, are also wrapped
Include conventional procedural programming language-such as " C " language or similar programming language.Program code can be complete
Ground executes on the user computer, partly executes on the user computer, executing as an independent software package, partially existing
Part executes on the remote computer or executes on a remote computer or server completely on subscriber computer.It is being related to
In the situation of remote computer, remote computer can pass through the network of any kind --- including local area network (LAN) or wide area
Net (WAN)-be connected to subscriber computer, or, it may be connected to outer computer (such as utilize ISP
To be connected by internet).
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The invention is not limited to the specific embodiments described herein, be able to carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out by above embodiments to the present invention
It is described in further detail, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, also
It may include more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (15)
1. a kind of cover determines method characterized by comprising
The candidate image set of file is obtained, and obtains from each candidate image the information of the entity of display；
Obtain the content information of the file；
Calculate the degree of correlation of the information for the entity that each candidate image is shown and the content information of file；
From candidate image set, determine that the degree of correlation meets cover of the candidate image of the first preset requirement as the file.
2. the method according to claim 1, wherein the letter of the entity for obtaining display from each candidate image
Breath, comprising:
It identifies the entity that each candidate image is shown, obtains the description information of entity in each candidate image；
The reality that keyword in the description information or description information of entity in each candidate image is shown as each candidate image
The information of body.
3. according to the method described in claim 2, it is characterized in that, the entity that each candidate image of identification is shown, obtains each
The description information of entity in candidate image, comprising:
Each candidate image is input to entity recognition model, obtains the description information of entity in each candidate image, the entity is known
Other model includes at least: one of human face recognition model, article identification model and gesture recognition model.
4. the method according to claim 1, wherein the information for calculating the entity that each candidate image show and
The degree of correlation of the content information of file, comprising:
The content information of the file is generated into the first text vector；
The information for the entity that each candidate image is shown generates the second text vector；
According to first text vector and each second text vector, the information and file of the entity that each candidate image is shown are calculated
Content information the degree of correlation.
5. according to the method described in claim 4, it is characterized in that, described according to first text vector and each second text
Vector calculates the degree of correlation of the information for the entity that each candidate image is shown and the content information of file, comprising:
Calculate the inverse of the distance of first text vector and each second text vector, the entity shown as each candidate image
Information and file content information the degree of correlation.
6. determining that the degree of correlation meets the method according to claim 1, wherein described from candidate image set
Cover of the candidate image of first preset requirement as the file, comprising:
Determine that the degree of correlation is more than or equal to cover of the candidate image of preset threshold as the file；Alternatively,
Determine cover of the maximum candidate image of the degree of correlation as the file in candidate image set.
7. method according to claim 1-6, which is characterized in that the method also includes:
Obtain the image quality information of each candidate image；
Correspondingly, determining that the degree of correlation meets the candidate image of the first preset requirement as the file from candidate image set
Cover include:
The degree of correlation and corresponding image quality of the content information of the information and file of the entity shown according to each candidate image are believed
Breath, calculates the score of each candidate image；
From candidate image set, determine that score meets cover of the candidate image of the second preset requirement as the file.
8. the method according to the description of claim 7 is characterized in that the image quality information for obtaining each candidate image, comprising:
Each candidate image is input to image quality identification model, obtains the image quality information of the candidate image, the image quality identifies mould
Type includes at least: one of aesthetics identification model, clarity identification model and Text region model.
9. the method according to the description of claim 7 is characterized in that the information of the entity shown according to each candidate image with
The degree of correlation of the content information of file and corresponding image quality information, calculate the score of each candidate image, comprising:
The degree of correlation and corresponding image quality information of the content information of the information and file for the entity that each candidate image is shown into
Row weighted sum obtains the score of each candidate image.
10. according to the method described in claim 9, it is characterized in that, the entity that each candidate image is shown information and text
The degree of correlation of the content information of part and corresponding image quality information are weighted before summation, further includes:
Obtain multiple image patterns and corresponding target score；
Obtain the degree of correlation and every image pattern of the information for the entity that every image pattern is shown and the content information of file
Image quality information；
The degree of correlation and every cover sample of the content information of the information and file of the entity shown by every cover sample
Image quality information fit object score, obtain the weight of the information of entity and the weight of image quality information.
11. method according to claim 1-6, which is characterized in that obtain the reality of display from each candidate image
After the information of body, further includes:
According to the information for the entity that each candidate image is shown, the legitimacy of corresponding candidate image is judged；
Correspondingly, determining that the degree of correlation meets the candidate image of the first preset requirement as the file from candidate image set
Cover include:
From candidate image set, determine that the degree of correlation meets the first preset requirement and legal candidate image is as the file
Cover.
12. method according to claim 1-6, which is characterized in that the content information of the file is the text
The title content information of part, also, the candidate image set for obtaining file includes:
At least two video frames are extracted according to setpoint frequency from video file, as candidate image set.
13. a kind of cover determining device characterized by comprising
First obtains module, for obtaining the candidate image set of file, and obtain from each candidate image the entity of display
Information；
Second obtains module, for obtaining the content information of the file；
Computing module, for calculating the degree of correlation of the information for the entity that each candidate image is shown and the content information of file；
Determining module, for from candidate image set, determining that the degree of correlation meets the candidate image of the first preset requirement as institute
State the cover of file.
14. a kind of electronic equipment, which is characterized in that the electronic equipment includes:
One or more processors；
Memory, for storing one or more programs,
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
Now the cover as described in any in claim 1-12 determines method.
15. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
Realize that the cover as described in any in claim 1-12 determines method when execution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811140229.XA CN109271542A (en) | 2018-09-28 | 2018-09-28 | Cover determines method, apparatus, equipment and readable storage medium storing program for executing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811140229.XA CN109271542A (en) | 2018-09-28 | 2018-09-28 | Cover determines method, apparatus, equipment and readable storage medium storing program for executing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109271542A true CN109271542A (en) | 2019-01-25 |
Family
ID=65199086
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811140229.XA Pending CN109271542A (en) | 2018-09-28 | 2018-09-28 | Cover determines method, apparatus, equipment and readable storage medium storing program for executing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109271542A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN109996091A (en) * | 2019-03-28 | 2019-07-09 | 苏州八叉树智能科技有限公司 | Generate method, apparatus, electronic equipment and the computer readable storage medium of video cover |
| CN110457523A (en) * | 2019-08-12 | 2019-11-15 | 腾讯科技（深圳）有限公司 | The choosing method of cover picture, the training method of model, device and medium |
| CN111143613A (en) * | 2019-12-30 | 2020-05-12 | 携程计算机技术（上海）有限公司 | Method, system, electronic device and storage medium for selecting video cover |
| CN111918130A (en) * | 2020-08-11 | 2020-11-10 | 北京达佳互联信息技术有限公司 | Video cover determining method and device, electronic equipment and storage medium |
| CN112231504A (en) * | 2020-09-30 | 2021-01-15 | 北京三快在线科技有限公司 | Method and device for determining cover picture, electronic equipment and storage medium |
| CN112561334A (en) * | 2020-12-16 | 2021-03-26 | 咪咕文化科技有限公司 | Grading method and device for reading object, electronic equipment and storage medium |
| CN112613293A (en) * | 2020-12-29 | 2021-04-06 | 北京中科闻歌科技股份有限公司 | Abstract generation method and device, electronic equipment and storage medium |
| CN112866801A (en) * | 2021-03-11 | 2021-05-28 | 北京小米移动软件有限公司 | Video cover determining method and device, electronic equipment and storage medium |
| CN113157963A (en) * | 2021-04-28 | 2021-07-23 | 维沃移动通信有限公司 | Image screening method, device electronic equipment and readable storage medium |
| CN113254696A (en) * | 2021-06-30 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Cover image acquisition method and device |
| CN113343832A (en) * | 2021-06-01 | 2021-09-03 | 北京奇艺世纪科技有限公司 | Video cover judging method, device, equipment and computer readable medium |
| CN113591513A (en) * | 2020-04-30 | 2021-11-02 | 北京字节跳动网络技术有限公司 | Method and apparatus for processing image |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060140469A1 (en) * | 2003-05-15 | 2006-06-29 | Warren Joel E | Method for communicating and matching electronic files for financial transactions |
| CN105389593A (en) * | 2015-11-16 | 2016-03-09 | 上海交通大学 | Image object recognition method based on SURF |
| CN106503693A (en) * | 2016-11-28 | 2017-03-15 | 北京字节跳动科技有限公司 | The offer method and device of video front cover |
| CN106973324A (en) * | 2017-03-28 | 2017-07-21 | 深圳市茁壮网络股份有限公司 | A kind of poster generation method and device |
| CN107832725A (en) * | 2017-11-17 | 2018-03-23 | 北京奇虎科技有限公司 | Video front cover extracting method and device based on evaluation index |
| CN108062533A (en) * | 2017-12-28 | 2018-05-22 | 北京达佳互联信息技术有限公司 | Analytic method, system and the mobile terminal of user's limb action |
| CN108415961A (en) * | 2018-02-06 | 2018-08-17 | 厦门集微科技有限公司 | A kind of advertising pictures recommendation method and device |
-
2018
- 2018-09-28 CN CN201811140229.XA patent/CN109271542A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060140469A1 (en) * | 2003-05-15 | 2006-06-29 | Warren Joel E | Method for communicating and matching electronic files for financial transactions |
| CN105389593A (en) * | 2015-11-16 | 2016-03-09 | 上海交通大学 | Image object recognition method based on SURF |
| CN106503693A (en) * | 2016-11-28 | 2017-03-15 | 北京字节跳动科技有限公司 | The offer method and device of video front cover |
| CN106973324A (en) * | 2017-03-28 | 2017-07-21 | 深圳市茁壮网络股份有限公司 | A kind of poster generation method and device |
| CN107832725A (en) * | 2017-11-17 | 2018-03-23 | 北京奇虎科技有限公司 | Video front cover extracting method and device based on evaluation index |
| CN108062533A (en) * | 2017-12-28 | 2018-05-22 | 北京达佳互联信息技术有限公司 | Analytic method, system and the mobile terminal of user's limb action |
| CN108415961A (en) * | 2018-02-06 | 2018-08-17 | 厦门集微科技有限公司 | A kind of advertising pictures recommendation method and device |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| US11475588B2 (en) | 2019-03-12 | 2022-10-18 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Image processing method and device for processing image, server and storage medium |
| CN109996091A (en) * | 2019-03-28 | 2019-07-09 | 苏州八叉树智能科技有限公司 | Generate method, apparatus, electronic equipment and the computer readable storage medium of video cover |
| CN110457523A (en) * | 2019-08-12 | 2019-11-15 | 腾讯科技（深圳）有限公司 | The choosing method of cover picture, the training method of model, device and medium |
| CN110457523B (en) * | 2019-08-12 | 2022-03-08 | 腾讯科技（深圳）有限公司 | Cover picture selection method, model training method, device and medium |
| CN111143613A (en) * | 2019-12-30 | 2020-05-12 | 携程计算机技术（上海）有限公司 | Method, system, electronic device and storage medium for selecting video cover |
| CN111143613B (en) * | 2019-12-30 | 2024-02-06 | 携程计算机技术（上海）有限公司 | Method, system, electronic device and storage medium for selecting video cover |
| CN113591513A (en) * | 2020-04-30 | 2021-11-02 | 北京字节跳动网络技术有限公司 | Method and apparatus for processing image |
| CN113591513B (en) * | 2020-04-30 | 2024-03-29 | 北京字节跳动网络技术有限公司 | Method and apparatus for processing image |
| CN111918130A (en) * | 2020-08-11 | 2020-11-10 | 北京达佳互联信息技术有限公司 | Video cover determining method and device, electronic equipment and storage medium |
| CN112231504A (en) * | 2020-09-30 | 2021-01-15 | 北京三快在线科技有限公司 | Method and device for determining cover picture, electronic equipment and storage medium |
| CN112561334A (en) * | 2020-12-16 | 2021-03-26 | 咪咕文化科技有限公司 | Grading method and device for reading object, electronic equipment and storage medium |
| CN112613293A (en) * | 2020-12-29 | 2021-04-06 | 北京中科闻歌科技股份有限公司 | Abstract generation method and device, electronic equipment and storage medium |
| CN112866801A (en) * | 2021-03-11 | 2021-05-28 | 北京小米移动软件有限公司 | Video cover determining method and device, electronic equipment and storage medium |
| CN112866801B (en) * | 2021-03-11 | 2023-10-03 | 北京小米移动软件有限公司 | Video cover determining method and device, electronic equipment and storage medium |
| CN113157963A (en) * | 2021-04-28 | 2021-07-23 | 维沃移动通信有限公司 | Image screening method, device electronic equipment and readable storage medium |
| CN113343832A (en) * | 2021-06-01 | 2021-09-03 | 北京奇艺世纪科技有限公司 | Video cover judging method, device, equipment and computer readable medium |
| CN113343832B (en) * | 2021-06-01 | 2024-02-02 | 北京奇艺世纪科技有限公司 | Video cover distinguishing method, device, equipment and computer readable medium |
| CN113254696A (en) * | 2021-06-30 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Cover image acquisition method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109271542A (en) | Cover determines method, apparatus, equipment and readable storage medium storing program for executing | |
| CN110287479B (en) | Named entity recognition method, electronic device and storage medium | |
| CN109117777B (en) | Method and device for generating information | |
| CN111062871B (en) | Image processing method and device, computer equipment and readable storage medium | |
| US10642887B2 (en) | Multi-modal image ranking using neural networks | |
| US7853582B2 (en) | Method and system for providing information services related to multimodal inputs | |
| CN106776503B (en) | Text semantic similarity determination method and device | |
| CN111198958A (en) | Method, device and terminal for matching background music | |
| US20190065492A1 (en) | Zero-shot event detection using semantic embedding | |
| CN109657054A (en) | Abstraction generating method, device, server and storage medium | |
| CN103988202A (en) | Image attractiveness based indexing and searching | |
| CN109660865B (en) | Method and device for automatically labeling videos, medium and electronic equipment | |
| US9639633B2 (en) | Providing information services related to multimodal inputs | |
| WO2023197749A9 (en) | Background music insertion time point determining method and apparatus, device, and storage medium | |
| CN112085120B (en) | Multimedia data processing method and device, electronic equipment and storage medium | |
| CN113515669A (en) | Data processing method based on artificial intelligence and related equipment | |
| CN110263218A (en) | Video presentation document creation method, device, equipment and medium | |
| CN112995757B (en) | Video clipping method and device | |
| CN114358203A (en) | Training method and device for image description sentence generation module and electronic equipment | |
| CN111783734A (en) | Original edition video identification method and device | |
| CN104462151B (en) | Assess the method and relevant apparatus of Homepage Publishing time | |
| CN116955707A (en) | Content tag determination method, device, equipment, medium and program product | |
| CN114329049A (en) | Video search method and device, computer equipment and storage medium | |
| CN115439922A (en) | Object behavior identification method, device, equipment and medium | |
| CN113642536A (en) | Data processing method, computer device and readable storage medium |
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 |