CN106156732A - Object identifying method and object recognition equipment - Google Patents
Object identifying method and object recognition equipment Download PDFInfo
- Publication number
- CN106156732A CN106156732A CN201610516069.9A CN201610516069A CN106156732A CN 106156732 A CN106156732 A CN 106156732A CN 201610516069 A CN201610516069 A CN 201610516069A CN 106156732 A CN106156732 A CN 106156732A
- Authority
- CN
- China
- Prior art keywords
- sub
- frame images
- field picture
- eigenvalue
- practical
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/49—Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of object identifying method and object recognition equipment, this object identifying method includes: according to each eigenvalue of a two field picture of target video, determines the classification of object included by described two field picture；According to the classification of the object included by described two field picture, described two field picture being divided into multiple sub-frame images, wherein, each sub-frame images in the plurality of sub-frame images is corresponding to a kind of the object included by described two field picture；And according at least one recognizer, the practical object included by each sub-frame images described is carried out parallelism recognition.The purchase of user can be got excited and is rapidly converted to purchasing behavior by the present invention, is effectively improved user and clicks on the conversion ratio buying product when watching video.
Description
Technical field
The present invention relates to Internet technical field, particularly relate to a kind of object identifying method and object recognition equipment.
Background technology
Along with the fast development of Internet technology, user can watch video by terminal unit, and especially, user is not
Only can support brand business and make or the artificial particular video program thrown in is done shopping, but also can allow at video platform
Any video scene in, the clothes with star, ornaments, digital product etc. or the cloth occurred in real-time search video scene
Sofa curtain desk lamp in scape etc. " same to money " and the explanation product that people introduced and promoted, and shopping should " same to money " maybe this product
Product, i.e. " see while buy ", thus relative to former " spectators chase after play-login shopping website-search relative article-screening rate of exchange-
Determine purchase " tediously long chain, user without jump out the video watched just can buy this " same to money " or explain orally people be situated between
The product continued and promote.
Above-mentioned " seeing while buy " process is " watch-search for-browse-than p-purchase ".Specifically, process " is seen while buy "
Including: first, user uses terminal unit to watch video；Secondly, user searches for above-mentioned " same to money " on this video pictures or solves
Say the product that people introduces and promotes；Then, after searching this " same to money " or explaining orally the product that people introduced and promoted, use
It is (such as, shown various bottom the half screen of video pictures that family browses on this video pictures shown various shopping informations
Shopping information)；Then, user is by shown various shopping informations and above-mentioned " same to money " or the explanation product that people introduced and promoted
Product are compared, and find the shopping information corresponding with above-mentioned " same to money " or the explanation product that people introduced and promoted；Finally,
User clicks on the shopping information that found to enter shopping entrance to buy above-mentioned " same to money " or to explain orally that people introduced and promoted
Product.
But, above-mentioned " seeing while buy " shopping process exist complex operation, user learning cost is higher, user may lose
Go the purchase intention to this product, the user shortcoming that the conversion ratio of click purchase product is relatively low when watching video.
Summary of the invention
Technical problem
In view of this, the technical problem to be solved in the present invention is to provide a kind of object identifying method and object recognition equipment,
To simplify the shopping process of above-mentioned " seeing while buy ", thus improve user and click on the conversion ratio buying product when watching video.
Solution
In order to solve above-mentioned technical problem, in first aspect, the invention provides a kind of object identifying method, including:
Each eigenvalue of the two field picture according to target video, determines the classification of object included by described two field picture；
According to the classification of the object included by described two field picture, described two field picture is divided into multiple sub-frame images, wherein,
Each sub-frame images in the plurality of sub-frame images is corresponding to a kind of the object included by described two field picture；And
According at least one recognizer, the practical object included by each sub-frame images described is known parallel
Not.
In conjunction with first aspect, in the implementation that the first is possible, described according at least one recognizer, to described
Practical object included by each sub-frame images carries out parallelism recognition, including:
According at least one recognizer, extract the eigenvalue of each sub-frame images described；And
The eigenvalue of each sub-frame images described is mated, according to matching result with the eigenvalue of each practical object
Determine described practical object included by each sub-frame images.
In conjunction with the first possible implementation of first aspect or first aspect, in the implementation that the second is possible
In,
Each eigenvalue at the two field picture according to target video cannot determine the object included by described two field picture
In the case of classification, using with the kind of the object being shaped like of the described object that cannot determine classification as described cannot be true
Make the classification of the object of classification.
In conjunction with the first possible implementation of first aspect or first aspect, at the embodiment that the third is possible
In, also include:
Practical object included by each sub-frame images described adds one respectively for showing the relevant letter of object
The floating layer of breath.
In conjunction with the implementation that the second of first aspect is possible, in the 4th kind of possible implementation, also include:
Practical object included by each sub-frame images described adds one respectively for showing the relevant letter of object
The floating layer of breath.
In second aspect, the invention provides a kind of object recognition equipment, including:
Determine unit, for each eigenvalue of the two field picture according to target video, determine included by described two field picture
The classification of object；
With described, cutting unit, determines that unit is connected, for the classification according to the object included by described two field picture, by institute
Stating two field picture and be divided into multiple sub-frame images, wherein, each sub-frame images in the plurality of sub-frame images is corresponding to described
One kind of the object included by two field picture；And
Parallelism recognition unit, is connected with described cutting unit, for according at least one recognizer, to described each
Practical object included by sub-frame images carries out parallelism recognition.
In conjunction with second aspect, in the implementation that the first is possible, described parallelism recognition unit includes:
Extraction module, for according at least one recognizer, extracting the eigenvalue of each sub-frame images described；And
Determine module, be connected with described extraction module, for by the eigenvalue of each sub-frame images described and each reality
The eigenvalue of object mates, and determines the practical object included by each sub-frame images described according to matching result.
In conjunction with the first possible implementation of second aspect or second aspect, in the implementation that the second is possible
In,
Determine that unit cannot determine described two field picture institute according to each eigenvalue of a two field picture of target video described
Including object classification in the case of, described determine unit by with being shaped like of the described object that cannot determine classification
The kind of object is as the classification of the described object that cannot determine classification.
In conjunction with the first possible implementation of second aspect or second aspect, in the implementation that the third is possible
In, also include:
Adding device, is connected with described parallelism recognition unit, in the reality included by each sub-frame images described
One is added respectively for the floating layer showing object-related information on object.
In conjunction with the implementation that the second of second aspect is possible, in the 4th kind of possible implementation, also include:
Adding device, is connected with described parallelism recognition unit, in the reality included by each sub-frame images described
One is added respectively for the floating layer showing object-related information on object.
Beneficial effect
The object identifying method of the embodiment of the present invention and object recognition equipment, included by the two field picture of target video
Two field picture is divided into multiple sub-frame images and the multiple subframes obtained segmentation according to recognizer by the classification of object
The practical object included by each sub-frame images in image carries out parallelism recognition, thus, carrys out parallelism recognition by multithreading
Practical object included by two field picture, user has only to by simple behaviour such as such as point touching practical objects when watching video
Make just can directly buy this practical object, and without passing through " watch-search for-browse-than p-purchase " this complicated tediously long behaviour
Make to buy this practical object, in such manner, it is possible to the purchase of user got excited be rapidly converted to purchasing behavior, be effectively increased use
The conversion ratio buying product is clicked at family when watching video.
According to below with reference to the accompanying drawings detailed description of illustrative embodiments, the further feature of the present invention and aspect being become
Clear.
Accompanying drawing explanation
The accompanying drawing of the part comprising in the description and constituting description together illustrates the present invention's with description
Exemplary embodiment, feature and aspect, and for explaining the principle of the present invention.
Fig. 1 illustrates the flow chart of the object identifying method of according to embodiments of the present invention；
Fig. 2 illustrates the flow chart of the object identifying method of according to embodiments of the present invention two；
Fig. 3 illustrates the structured flowchart of the object recognition equipment of according to embodiments of the present invention three；And
Fig. 4 illustrates the structured flowchart of the object recognition equipment of according to embodiments of the present invention four.
Detailed description of the invention
Various exemplary embodiments, feature and the aspect of the present invention is described in detail below with reference to accompanying drawing.In accompanying drawing identical
Reference represent the same or analogous element of function.Although the various aspects of embodiment shown in the drawings, but remove
Non-specifically is pointed out, it is not necessary to accompanying drawing drawn to scale.
The most special word " exemplary " means " as example, embodiment or illustrative ".Here as " exemplary "
Illustrated any embodiment should not necessarily be construed as preferred or advantageous over other embodiments.
It addition, in order to better illustrate the present invention, detailed description of the invention below gives numerous details.
It will be appreciated by those skilled in the art that do not have some detail, the present invention equally implements.In some instances, for
Method well known to those skilled in the art, means, element and circuit are not described in detail, in order to highlight the purport of the present invention.
Embodiment 1
Fig. 1 illustrates the flow chart of the object identifying method of according to embodiments of the present invention.As it is shown in figure 1, this Object identifying
Method mainly may include that
Step S100, each eigenvalue according to a two field picture of target video, determine the class of object included by two field picture
Not.
Step S120, classification according to the object included by two field picture, be divided into multiple sub-frame images by two field picture, its
In, each sub-frame images in multiple sub-frame images is corresponding to a kind of the object included by two field picture.
Wherein, user can use terminal unit to play video.Terminal unit includes but not limited to that user can pass through
The modes such as keyboard, mouse, remote controller, touch pad, handwriting equipment carry out the electronic product of man-machine interaction, such as, computer, hands
Machine, personal digital assistant (English: Personal Digital Assistant, PDA), notebook, desktop computer, intelligence be called for short:
Energy TV etc..Video such as refers to the dynamic image propagated with video file formats such as WMV, RM, RMVB, FLV and MOV, example
As all kinds of movie and video programs, news, advertisement, FLASH animation, auto heterodyne DV, chat video, game video, monitor video can be included
Deng.
Can buy in target video conveniently and efficiently for the ease of user when using terminal unit viewing target video
The product (object) occurred, it is possible, firstly, to extract each eigenvalue of the two field picture of target video in advance, it is then possible to according to carrying
The each eigenvalue taken out determines the kind of the object included by two field picture, finally, and can be according to included by the two field picture determined
The kind of object, two field picture is divided into multiple sub-frame images.Wherein, target video refers to that user uses terminal unit
The video of viewing.
Specifically, such as SKB feature extraction algorithm, quickly approximation SIFT feature extraction algorithm, BRIEF feature can be used
Extraction algorithm scheduling algorithm extracts each eigenvalue of two field picture respectively.Wherein, eigenvalue such as can include a feature (e.g., angle
Point, obvious point, the marginal point of picture), line feature (e.g., the edge line segment of picture), region feature, textural characteristics, color characteristic etc..
After each eigenvalue extracting two field picture, the object included by two field picture can be determined according to each eigenvalue
Classification.For example, it is assumed that extract the eigenvalue A of two field picture, eigenvalue B and eigenvalue C, and eigenvalue A represent one-piece dress,
Eigenvalue B represents that pan, eigenvalue C represent desk, then can determine that this according to eigenvalue A, eigenvalue B and eigenvalue C
Two field picture includes this three class object (or product) of one-piece dress, pan and desk.
In a kind of possible implementation, each eigenvalue at the two field picture according to target video cannot determine frame
In the case of the classification of the object included by image, by the kind with the object being shaped like of the object that cannot determine classification
Classification as the object that cannot determine classification.For example, it is assumed that cannot determine according to the eigenvalue D of the two field picture extracted
The classification of the object included by two field picture, being shaped like represented by the shape represented by eigenvalue D and eigenvalue E, and root
The kind being capable of determining that the object of two field picture according to eigenvalue E is desk, then can determine that two field picture includes according to eigenvalue D
This class object of desk.
After the classification of the object determined included by two field picture, can be according to the classification of the object included by two field picture
Two field picture is divided into multiple sub-frame images.For example, it is assumed that determine this two field picture include one-piece dress, pan and desk this three
Class object, then two field picture can be divided into corresponding to one-piece sub-frame images 1, corresponding to pan sub-frame images 2, with
And the sub-frame images 3 corresponding to desk.
Step S140, according at least one recognizer, the practical object included by each sub-frame images is carried out also
Row identifies.
Specifically, each that segmentation is obtained by existing a kind of recognizer such as face recognition algorithms can be used
Practical object included by sub-frame images carries out parallelism recognition, it would however also be possible to employ existing multiple recognizer such as recognition of face
The practical object included by each sub-frame images that segmentation is obtained by algorithm, object recognition algorithm etc. carries out parallelism recognition.
For example, it is assumed that in the step s 120 two field picture is divided into corresponding to one-piece sub-frame images 1, corresponds to
The sub-frame images 2 of pan and three sub-frame images of the sub-frame images 3 corresponding to desk, then can perform as follows simultaneously
Three Object identifying operations: use face recognition algorithms that the practical object included by sub-frame images 1 is identified, uses object
Practical object included by sub-frame images 2 is identified and uses object recognition algorithm to sub-frame images 3 institute by recognizer
Including practical object be identified, and can be with the practical object included by parallelism recognition goes out sub-frame images 1 for " green even clothing
Skirt ", practical object included by sub-frame images 2 practical object included by " blue pan " and sub-frame images 3 be
" circular table ".
In a kind of possible implementation, this object identifying method can also include: step S160, in each subframe
One is added respectively for the floating layer showing object-related information on practical object included by image.
Specifically, the reality included by each sub-frame images in parallelism recognition goes out multiple sub-frame images that segmentation obtains
After the object of border, can add a floating layer on the practical object identified, this floating layer is used for showing object-related information, should
The link of buying of object-related information e.g. practical object links the name of the practical object being associated with the purchase of practical object
Title, Quick Response Code etc..Object-related information be practical object buy link or object-related information and practical object
In the case of buying the title of the practical object that link is associated, user can be come directly by point touching object-related information
This practical object is carried out shopping place an order.In the case of object-related information is Quick Response Code, user can be by utilizing terminal
Device scan object-related information directly this practical object to be carried out shopping and is placed an order.
The object identifying method of the embodiment of the present invention, according to the classification of the object included by the two field picture of target video by
Two field picture is divided into multiple sub-frame images and according to recognizer to splitting included by each sub-frame images obtained
Practical object carries out parallelism recognition, thus, carrys out the practical object included by parallelism recognition two field picture by multithreading, and user is seeing
Have only to when seeing video by such as point touching practical object or utilize two added on terminal device scans practical object
This practical object just can be directly bought in the dimension simple operations such as code, and without passing through " watch-search for-browse-than p-purchase " this
This practical object is bought in one complicated tediously long operation, in such manner, it is possible to the purchase of user is got excited be rapidly converted to purchasing behavior,
It is effectively increased user and clicks on the conversion ratio buying product when watching video.
Embodiment 2
Fig. 2 illustrates the flow chart of the object identifying method of according to embodiments of the present invention two.As in figure 2 it is shown, this Object identifying
Method mainly may include that
Step S200, each eigenvalue according to a two field picture of target video, determine the class of object included by two field picture
Not.The associated description illustrated in step S100 that may refer in above-described embodiment 1 of this step.
Step S220, classification according to the object included by two field picture, be divided into multiple sub-frame images by two field picture, its
In, each sub-frame images in multiple sub-frame images is corresponding to a kind of the object included by two field picture.This step
Illustrate the associated description in step S120 that may refer in above-described embodiment 1.
Step S240, according at least one recognizer, extract the eigenvalue of each sub-frame images.
Specifically, after two field picture is divided into multiple sub-frame images, each sub-frame images can be extracted respectively
Eigenvalue, wherein, (e.g., eigenvalue such as can include a feature (e.g., angle point, obvious point, the marginal point of picture), line feature
The edge line segment of picture), region feature, textural characteristics, color characteristic etc..For example, it is assumed that two field picture is split in step S220
For corresponding to one-piece sub-frame images 1, the sub-frame images 2 corresponding to pan and the sub-frame images 3 corresponding to desk
Three sub-frame images, then can extract the eigenvalue of these three sub-frame images simultaneously.
Step S260, the eigenvalue of the eigenvalue of each sub-frame images with each practical object is mated, according to
Join result and determine the practical object included by each sub-frame images.
Specifically, the eigenvalue of the eigenvalue of each sub-frame images with each practical object can be mated, wherein,
The eigenvalue algorithm that used of coupling can include matching algorithm based on edge feature, matching algorithm based on provincial characteristics,
Matching algorithm etc. based on a feature, and determine the practical object included by each sub-frame images according to matching result.
For example, it is assumed that the similarity of the eigenvalue of the eigenvalue A of sub-frame images and practical object 1 more than predetermined threshold and
The eigenvalue B of sub-frame images is more than predetermined threshold with the similarity of the eigenvalue of practical object 2, then may determine that this subframe pattern
The eigenvalue A of picture and the eigenvalue of practical object 1 match and the feature of eigenvalue B and practical object 2 of this sub-frame images
Value matches, thus can determine that this sub-frame images includes practical object 1 and practical object 2.
In a kind of possible implementation, this object identifying method can also include: step S280, in each subframe
One is added respectively for the floating layer showing object-related information on practical object included by image.
Specifically, the reality included by each sub-frame images in parallelism recognition goes out multiple sub-frame images that segmentation obtains
After the object of border, can add a floating layer on the practical object identified, this floating layer is used for showing object-related information, should
The link of buying of object-related information e.g. practical object links the name of the practical object being associated with the purchase of practical object
Title, Quick Response Code etc..Object-related information be practical object buy link or object-related information and practical object
In the case of buying the title of the practical object that link is associated, user can be come directly by point touching object-related information
This practical object is carried out shopping place an order.In the case of object-related information is Quick Response Code, user can be by utilizing terminal
Device scan object-related information directly this practical object to be carried out shopping and is placed an order.
The object identifying method of the embodiment of the present invention, according to the classification of the object included by the two field picture of target video by
Two field picture is divided into multiple sub-frame images and the feature according to each sub-frame images in the multiple sub-frame images extracted
Value determines the practical object included by each sub-frame images with the matching result of the eigenvalue of each practical object, thus, logical
Crossing multithreading and carry out the practical object included by parallelism recognition two field picture, user has only to by such as clicking on tactile when watching video
Touch practical object or utilize the simple operationss such as the Quick Response Code added on terminal device scans practical object just can directly buy
This practical object, and right without buying this reality by " watch-search for-browse-than p-purchase " this complicated tediously long operation
As, in such manner, it is possible to the purchase of user got excited be rapidly converted to purchasing behavior, it is effectively increased user at viewing video time point
Hit the conversion ratio buying product.
Embodiment 3
Fig. 3 illustrates the structured flowchart of the object recognition equipment of according to embodiments of the present invention three.The object that the present embodiment provides
Identify that device 300 is for realizing the object identifying method shown in Fig. 1.As it is shown on figure 3, this object recognition equipment 300 is the most permissible
Including:
Determining unit 310, for each eigenvalue according to a two field picture of target video, determine included by two field picture is right
The classification of elephant.
Cutting unit 330, and determines that unit 310 is connected, for the classification according to the object included by two field picture, by frame figure
As being divided into multiple sub-frame images, wherein, each sub-frame images in multiple sub-frame images is corresponding to included by two field picture
One kind of object.
Wherein, user can use terminal unit to play video.Terminal unit includes but not limited to that user can pass through
The modes such as keyboard, mouse, remote controller, touch pad, handwriting equipment carry out the electronic product of man-machine interaction, such as, computer, hands
Machine, personal digital assistant (English: Personal Digital Assistant, PDA), notebook, desktop computer, intelligence be called for short:
Energy TV etc..Video such as refers to the dynamic image propagated with video file formats such as WMV, RM, RMVB, FLV and MOV, example
As all kinds of movie and video programs, news, advertisement, FLASH animation, auto heterodyne DV, chat video, game video, monitor video can be included
Deng.
Can buy in target video conveniently and efficiently for the ease of user when using terminal unit viewing target video
The product (object) occurred, first, object recognition equipment 300 can extract each eigenvalue of the two field picture of target video in advance,
Then, it is determined that unit 310 can determine the kind of the object included by two field picture according to each eigenvalue extracted, finally, divide
Cut unit 330 and according to the kind of the object included by the two field picture determined, two field picture can be divided into multiple sub-frame images.
Wherein, target video refers to the video that user uses terminal unit watching.
Specifically, object recognition equipment 300 can use such as SKB feature extraction algorithm, quickly approximation SIFT feature to carry
Take algorithm, BRIEF feature extraction algorithm scheduling algorithm extracts each eigenvalue of two field picture respectively.Wherein, eigenvalue is the most permissible
Including a feature (e.g., angle point, obvious point, the marginal point of picture), line feature (e.g., the edge line segment of picture), region feature, texture
Feature, color characteristic etc..
After each eigenvalue extracting two field picture, determine that unit 310 can determine two field picture according to each eigenvalue
The classification of included object.For example, it is assumed that extract the eigenvalue A of two field picture, eigenvalue B and eigenvalue C, and eigenvalue
A represents that one-piece dress, eigenvalue B represent that pan, eigenvalue C represent desk, it is determined that unit 310 is according to eigenvalue A, eigenvalue
B and eigenvalue C can determine that this two field picture includes this three class object (or product) of one-piece dress, pan and desk.
In a kind of possible implementation, determining the unit 310 each eigenvalue according to a two field picture of target video
In the case of the classification of object included by two field picture cannot be determined, determine unit 310 by with cannot determine the right of classification
The kind of the object being shaped like of elephant is as the classification of the object that cannot determine classification.For example, it is assumed that determine unit 310
Eigenvalue D according to the two field picture extracted cannot determine the classification of the object included by two field picture, the shape represented by eigenvalue D
Being shaped like represented by shape and eigenvalue E, and determine that unit 310 is capable of determining that the object of two field picture according to eigenvalue E
Kind be desk, it is determined that according to eigenvalue D, unit 310 can determine that two field picture includes this class object of desk.
After determining the classification that unit 310 determines the object included by two field picture, cutting unit 330 can be according to frame
Two field picture is divided into multiple sub-frame images by the classification of the object included by image.For example, it is assumed that determine that unit 310 determines this
Two field picture includes this three class object of one-piece dress, pan and desk, then cutting unit 330 two field picture can be divided into corresponding to
One-piece sub-frame images 1, the sub-frame images 2 corresponding to pan and the sub-frame images 3 corresponding to desk.
Parallelism recognition unit 350, is connected with described cutting unit 330, for according at least one recognizer, to each
Practical object included by individual sub-frame images carries out parallelism recognition.
Specifically, that existing a kind of recognizer such as face recognition algorithms can be used is right for parallelism recognition unit 350
The practical object included by each sub-frame images that segmentation obtains carries out parallelism recognition, and parallelism recognition unit 350 can also be adopted
Each subframe pattern segmentation obtained with existing multiple recognizer such as face recognition algorithms, object recognition algorithm etc.
As included practical object carries out parallelism recognition.
For example, it is assumed that two field picture is divided into corresponding to one-piece sub-frame images 1, corresponds to by cutting unit 330
The sub-frame images 2 of pan and three sub-frame images of the sub-frame images 3 corresponding to desk, then parallelism recognition unit 350 can
To perform the operation of following three Object identifying simultaneously: use face recognition algorithms that the practical object included by sub-frame images 1 is entered
Row identifies, uses object recognition algorithm be identified the practical object included by sub-frame images 2 and use object identification to calculate
Practical object included by sub-frame images 3 is identified by method, and parallelism recognition unit 350 can go out subframe pattern with parallelism recognition
Be " green one-piece dress " as the practical object included by 1, practical object included by sub-frame images 2 be " blueness pan ", with
And the practical object included by sub-frame images 3 is " circular table ".
In a kind of possible implementation, this object recognition equipment 300 can also include: adding device 370, with described
Parallelism recognition unit 350 connects, for adding a use on the practical object included by each sub-frame images described respectively
Floating layer in display object-related information.
Specifically, each height in parallelism recognition unit 350 parallelism recognition goes out multiple sub-frame images that segmentation obtains
After practical object included by two field picture, adding device 370 can add a floating layer on the practical object identified, should
Floating layer is used for showing object-related information, this object-related information e.g. practical object buy link and practical object
Buy and link the title of practical object, the Quick Response Code etc. being associated.Object-related information be practical object purchase link or
In the case of person's object-related information is the title that the purchase with practical object links the practical object being associated, user can be led to
Cross point touching object-related information directly this practical object to be carried out shopping and place an order.It is Quick Response Code in object-related information
In the case of, user can be by utilizing terminal device scans object-related information directly to do shopping down this practical object
Single.
The object recognition equipment of the embodiment of the present invention, according to the classification of the object included by the two field picture of target video by
Two field picture is divided into multiple sub-frame images and according to recognizer to splitting included by each sub-frame images obtained
Practical object carries out parallelism recognition, thus, carrys out the practical object included by parallelism recognition two field picture by multithreading, and user is seeing
Have only to when seeing video by such as point touching practical object or utilize two added on terminal device scans practical object
This practical object just can be directly bought in the dimension simple operations such as code, and without passing through " watch-search for-browse-than p-purchase " this
This practical object is bought in one complicated tediously long operation, in such manner, it is possible to the purchase of user is got excited be rapidly converted to purchasing behavior,
It is effectively increased user and clicks on the conversion ratio buying product when watching video.
Embodiment 4
Fig. 4 illustrates the structured flowchart of the object recognition equipment of according to embodiments of the present invention four.The object that the present embodiment provides
Identify that device 400 is for realizing the object identifying method shown in Fig. 2.As shown in Figure 4, this object recognition equipment 400 is the most permissible
Including:
First determines unit 410, for each eigenvalue of the two field picture according to target video, determines included by two field picture
The classification of object.First determine unit 410 illustrate the phase that may refer to cell 310 really in above-described embodiment 3
Close and describe.
With described first, cutting unit 430, determines that unit 410 is connected, for the class according to the object included by two field picture
Not, two field picture being divided into multiple sub-frame images, wherein, each sub-frame images in multiple sub-frame images corresponds to two field picture
One kind of included object.Cutting unit 430 illustrate the cutting unit that may refer in above-described embodiment 3
The associated description of 330.
Extraction unit 450, is connected with described cutting unit 430, for according at least one recognizer, extracts each
The eigenvalue of sub-frame images.
Specifically, after two field picture is divided into multiple sub-frame images by cutting unit 430, extraction unit 450 can divide
Taking the eigenvalue of each sub-frame images indescribably, wherein, eigenvalue such as can include a feature (e.g., angle point, obvious point, picture
The marginal point in face), line feature (e.g., the edge line segment of picture), region feature, textural characteristics, color characteristic etc..For example, it is assumed that point
Cut unit 430 to be divided into by two field picture corresponding to one-piece sub-frame images 1, corresponding to the sub-frame images 2 of pan and right
Should be in three sub-frame images of the sub-frame images 3 of desk, then extraction unit 450 can extract the spy of these three sub-frame images simultaneously
Value indicative.
Second determines unit 470, is connected with described extraction unit 450, for by the eigenvalue of each sub-frame images with
The eigenvalue of each practical object mates, and determines the practical object included by each sub-frame images according to matching result.
Specifically, second determines that unit 470 can be by the feature of the eigenvalue of each sub-frame images Yu each practical object
Value is mated, and wherein, the algorithm that eigenvalue coupling is used can include matching algorithm based on edge feature, based on region
The matching algorithm of feature, matching algorithm etc. based on a feature, and second determine that unit 470 can come really according to matching result
Fixed practical object included by each sub-frame images.
For example, it is assumed that the similarity of the eigenvalue of the eigenvalue A of sub-frame images and practical object 1 more than predetermined threshold and
The eigenvalue B of sub-frame images is more than predetermined threshold with the similarity of the eigenvalue of practical object 2, then second determines that unit 470 can
The eigenvalue of eigenvalue A Yu practical object 1 to judge this sub-frame images match and this sub-frame images eigenvalue B with
The eigenvalue of practical object 2 matches, thus second determines that unit 470 can determine that this sub-frame images includes practical object 1
With practical object 2.
In a kind of possible implementation, this object recognition equipment 400 can also include: adding device 490, with described
Second determines that unit 470 connects, for adding one on the practical object included by each sub-frame images respectively for showing
Show the floating layer of object-related information.
Specifically, under extraction unit 450 and the second cooperation determining unit 470 parallelism recognition go out segmentation obtain multiple
After the practical object included by each sub-frame images in sub-frame images, adding device 490 can be in the reality identified
Adding a floating layer on object, this floating layer is used for showing object-related information, this object-related information e.g. practical object
Buy link and link the title of the practical object being associated, Quick Response Code etc. with the purchase of practical object.In object-related information it is
The purchase link of practical object or object-related information are the names that the purchase with practical object links the practical object being associated
In the case of title, user can directly this practical object to be carried out shopping by point touching object-related information and place an order.?
In the case of object-related information is Quick Response Code, user can be the most right by utilizing terminal device scans object-related information
This practical object carries out shopping and places an order.
The object recognition equipment of the embodiment of the present invention, according to the classification of the object included by the two field picture of target video by
Two field picture is divided into multiple sub-frame images and the feature according to each sub-frame images in the multiple sub-frame images extracted
Value determines the practical object included by each sub-frame images with the matching result of the eigenvalue of each practical object, thus, logical
Crossing multithreading and carry out the practical object included by parallelism recognition two field picture, user has only to by such as clicking on tactile when watching video
Touch practical object or utilize the simple operationss such as the Quick Response Code added on terminal device scans practical object just can directly buy
This practical object, and right without buying this reality by " watch-search for-browse-than p-purchase " this complicated tediously long operation
As, in such manner, it is possible to the purchase of user got excited be rapidly converted to purchasing behavior, it is effectively increased user at viewing video time point
Hit the conversion ratio buying product.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any
Those familiar with the art, in the technical scope that the invention discloses, can readily occur in change or replace, should contain
Cover within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with described scope of the claims.
Claims (10)
1. an object identifying method, it is characterised in that including:
Each eigenvalue of the two field picture according to target video, determines the classification of object included by described two field picture；
According to the classification of the object included by described two field picture, described two field picture is divided into multiple sub-frame images, wherein, described
Each sub-frame images in multiple sub-frame images is corresponding to a kind of the object included by described two field picture；And
According at least one recognizer, the practical object included by each sub-frame images described is carried out parallelism recognition.
Object identifying method the most according to claim 1, it is characterised in that described according at least one recognizer, right
Practical object included by each sub-frame images described carries out parallelism recognition, including:
According at least one recognizer, extract the eigenvalue of each sub-frame images described；And
The eigenvalue of each sub-frame images described is mated with the eigenvalue of each practical object, determines according to matching result
Practical object included by each sub-frame images described.
Object identifying method the most according to claim 1 and 2, it is characterised in that
Each eigenvalue at the two field picture according to target video cannot determine the classification of the object included by described two field picture
In the case of, cannot will determine as described with the kind of the object being shaped like of the described object that cannot determine classification
The classification of the object of classification.
Object identifying method the most according to claim 1 and 2, it is characterised in that also include:
Practical object included by each sub-frame images described adds one respectively for showing object-related information
Floating layer.
Object identifying method the most according to claim 3, it is characterised in that also include:
Practical object included by each sub-frame images described adds one respectively for showing object-related information
Floating layer.
6. an object recognition equipment, it is characterised in that including:
Determine unit, for each eigenvalue of the two field picture according to target video, determine the object included by described two field picture
Classification；
With described, cutting unit, determines that unit is connected, for the classification according to the object included by described two field picture, by described frame
Image is divided into multiple sub-frame images, and wherein, each sub-frame images in the plurality of sub-frame images corresponds to described frame figure
A kind as included object；And
Parallelism recognition unit, is connected with described cutting unit, for according at least one recognizer, to each subframe described
Practical object included by image carries out parallelism recognition.
Object recognition equipment the most according to claim 6, it is characterised in that described parallelism recognition unit includes:
Extraction module, for according at least one recognizer, extracting the eigenvalue of each sub-frame images described；And
Determine module, be connected with described extraction module, for by the eigenvalue of each sub-frame images described and each practical object
Eigenvalue mate, determine the practical object included by each sub-frame images described according to matching result.
8. according to the object recognition equipment described in claim 6 or 7, it is characterised in that
Determine that unit cannot be determined included by described two field picture according to each eigenvalue of a two field picture of target video described
Object classification in the case of, described determine that unit is by the object being shaped like with the described object that cannot determine classification
Kind as the classification of the described object that cannot determine classification.
9. according to the object recognition equipment described in claim 6 or 7, it is characterised in that also include:
Adding device, is connected with described parallelism recognition unit, at the practical object included by each sub-frame images described
Upper interpolation one respectively is for showing the floating layer of object-related information.
Object recognition equipment the most according to claim 8, it is characterised in that also include:
Adding device, is connected with described parallelism recognition unit, at the practical object included by each sub-frame images described
Upper interpolation one respectively is for showing the floating layer of object-related information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610516069.9A CN106156732A (en) | 2016-07-01 | 2016-07-01 | Object identifying method and object recognition equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610516069.9A CN106156732A (en) | 2016-07-01 | 2016-07-01 | Object identifying method and object recognition equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106156732A true CN106156732A (en) | 2016-11-23 |
Family
ID=58061042
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610516069.9A Pending CN106156732A (en) | 2016-07-01 | 2016-07-01 | Object identifying method and object recognition equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106156732A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108491258A (en) * | 2018-03-29 | 2018-09-04 | 百度在线网络技术（北京）有限公司 | Method and apparatus for handling image |
| CN109858463A (en) * | 2019-02-22 | 2019-06-07 | 成都云鼎丝路信息技术有限公司 | A kind of user identification method with double engines, system and terminal |
| CN109996107A (en) * | 2017-12-29 | 2019-07-09 | 百度在线网络技术（北京）有限公司 | Video generation method, device and system |
| CN113222995A (en) * | 2020-02-06 | 2021-08-06 | 阿里巴巴集团控股有限公司 | Video processing method, video processing device and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104967885A (en) * | 2015-03-27 | 2015-10-07 | 哈尔滨工业大学深圳研究生院 | Advertisement recommending method and system based on video content |
| CN105373938A (en) * | 2014-08-27 | 2016-03-02 | 阿里巴巴集团控股有限公司 | Method for identifying commodity in video image and displaying information, device and system |
-
2016
- 2016-07-01 CN CN201610516069.9A patent/CN106156732A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105373938A (en) * | 2014-08-27 | 2016-03-02 | 阿里巴巴集团控股有限公司 | Method for identifying commodity in video image and displaying information, device and system |
| CN104967885A (en) * | 2015-03-27 | 2015-10-07 | 哈尔滨工业大学深圳研究生院 | Advertisement recommending method and system based on video content |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109996107A (en) * | 2017-12-29 | 2019-07-09 | 百度在线网络技术（北京）有限公司 | Video generation method, device and system |
| CN108491258A (en) * | 2018-03-29 | 2018-09-04 | 百度在线网络技术（北京）有限公司 | Method and apparatus for handling image |
| CN109858463A (en) * | 2019-02-22 | 2019-06-07 | 成都云鼎丝路信息技术有限公司 | A kind of user identification method with double engines, system and terminal |
| CN109858463B (en) * | 2019-02-22 | 2021-03-26 | 成都云鼎丝路信息技术有限公司 | Dual-engine user identification method, system and terminal |
| CN113222995A (en) * | 2020-02-06 | 2021-08-06 | 阿里巴巴集团控股有限公司 | Video processing method, video processing device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9607010B1 (en) | Techniques for shape-based search of content | |
| US20200311126A1 (en) | Methods to present search keywords for image-based queries | |
| US10083521B1 (en) | Content recommendation based on color match | |
| CN106202317A (en) | Method of Commodity Recommendation based on video and device | |
| US20160225053A1 (en) | Mobile visual commerce system | |
| CN106598998B (en) | Information acquisition method and information acquisition device | |
| Yang et al. | How related exemplars help complex event detection in web videos? | |
| JP2019531547A (en) | Object detection with visual search queries | |
| CN102165404B (en) | Object detection and user settings | |
| US9881084B1 (en) | Image match based video search | |
| CN106202316A (en) | Merchandise news acquisition methods based on video and device | |
| CN106156732A (en) | Object identifying method and object recognition equipment | |
| KR102102571B1 (en) | System and method for providing online shopping platform | |
| CN113766296B (en) | Live broadcast picture display method and device | |
| US10379721B1 (en) | Interactive interfaces for generating annotation information | |
| US20190325497A1 (en) | Server apparatus, terminal apparatus, and information processing method | |
| KR102517961B1 (en) | Method for providing clothing recommendation information based on user-selected clothing, and server and program using the same | |
| TW202141475A (en) | Method of determining item name of object, device, computer equipment and storage medium | |
| CN102982472A (en) | Method and device for automatically recommending similar commodities by terminal | |
| Zhang et al. | A comprehensive survey on computational aesthetic evaluation of visual art images: Metrics and challenges | |
| CN114005077B (en) | Audience evaluation data driven silent product video creation auxiliary method and device | |
| CN105159923A (en) | Video image based article extraction, query and purchasing method | |
| CN113330455A (en) | Finding complementary digital images using conditional generative countermeasure networks | |
| CN105069005A (en) | Data searching method and data searching device | |
| Mameli et al. | Deep learning approaches for fashion knowledge extraction from social media: a review |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 100080 Beijing Haidian District city Haidian street A Sinosteel International Plaza No. 8 block 5 layer A, C Applicant after: Youku network technology (Beijing) Co.,Ltd. Address before: 100080 Beijing Haidian District city Haidian street A Sinosteel International Plaza No. 8 block 5 layer A, C Applicant before: 1VERGE INTERNET TECHNOLOGY (BEIJING) Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200623 Address after: 310052 room 508, floor 5, building 4, No. 699, Wangshang Road, Changhe street, Binjiang District, Hangzhou City, Zhejiang Province Applicant after: Alibaba (China) Co.,Ltd. Address before: 100080 Beijing Haidian District city Haidian street A Sinosteel International Plaza No. 8 block 5 layer A, C Applicant before: Youku network technology (Beijing) Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161123 |
|
| RJ01 | Rejection of invention patent application after publication |