CN109544252A - Playback method, system and relevant device - Google Patents
Playback method, system and relevant device Download PDFInfo
- Publication number
- CN109544252A CN109544252A CN201811483294.2A CN201811483294A CN109544252A CN 109544252 A CN109544252 A CN 109544252A CN 201811483294 A CN201811483294 A CN 201811483294A CN 109544252 A CN109544252 A CN 109544252A
- Authority
- CN
- China
- Prior art keywords
- user
- article
- image
- involved
- behavioural characteristic
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
Abstract
The invention discloses a kind of playback method, system and relevant devices, are related to multimedia technology field.Playback method includes: the image for sending acquisition;Receive recommendation information, wherein recommendation information is what article involved in the behavioural characteristic of the user behavior characteristics and user in the image according to acquisition determined;Play recommendation information.During the embodiment of the present invention can browse article under user's line, the possible interested information of user is analyzed according to the behavioural characteristic of user and recommends user, so as to improve recommendation efficiency.
Description
Technical field
The present invention relates to multimedia technology field, in particular to a kind of playback method, system and relevant device.
Background technique
As the improvement of people's living standards, customer descends shops increasingly to pay close attention to shopping experience when doing shopping online.Commodity are wide
It accuses as the effective way for attracting customer, launches effect and have a significant impact to the shopping experience of customer.Xian Xia shops at present
Commdity advertisement can also play relevant information by electronic display other than the poster form of papery.Xian Xia shops at present
E-advertising display screen mainly play the picture being stored in advance or video and be used to publicize certain commodity.
Summary of the invention
The interior of preparatory memory is played it was recognized by the inventor that being served only for Commdity advertisement and showing that the display screen of purpose is generally circulated
Hold, there is no launch corresponding Commdity advertisement for specific consumers.To which the recommendation efficiency of the prior art is lower.
One technical problem to be solved by the embodiment of the invention is that: how to improve recommendation efficiency.
First aspect according to some embodiments of the invention provides a kind of playback method, comprising: sends the figure of acquisition
Picture;Receive recommendation information, wherein recommendation information is that the behavior of the user behavior characteristics and user in the image according to acquisition is special
What article involved in levying determined;Play recommendation information.
In some embodiments, user behavior characteristics include user's posture feature;Playback method further include: obtain acquisition
Occurs the image of user in image;Human body key point information is determined in the image for user occur;According to human body key point information
Determine user's posture feature.
In some embodiments, determine that human body key point information includes: will occur user's in the image for user occur
Image is input to key point prediction model, obtains the human body key point information of output;Wherein, key point prediction model includes series connection
Residual error network ResNet model and attention model, ResNet model the image of different resolution is input to attention model
In, attention model is configured with corresponding weight parameter to the input picture of every kind of resolution ratio.
In some embodiments, playback method further include: determine the article region in the image at current family;According to
Hand key point region in article region and human body key point information, determines involved in the behavioural characteristic of user
Article.
In some embodiments, playback method further include: for touch posture or grab appearance in response to user's posture feature
State, the article according to involved in the behavioural characteristic of user determine recommendation information;It is to put down article appearance in response to user's posture feature
State, the relevant article of the article according to involved in the behavioural characteristic of user determine recommendation information.
In some embodiments, user behavior characteristics include user's expressive features;Playback method further include: in response to user
Expressive features are that positive expression or neutral expression, the article according to involved in the behavioural characteristic of user determine recommendation information;Response
It is negative expression in user's expressive features, the relevant article of the article according to involved in the behavioural characteristic of user determines recommendation
Breath.
In some embodiments, user behavior characteristics include user's posture feature and user's expressive features;Playback method is also
Include: in response to user's posture feature to touch posture or crawl posture, is positive expression or neutrality in user's expressive features
In the case where expression, the article according to involved in the behavioural characteristic of user determines recommendation information, is negative in user's expressive features
In the case where expression, the relevant article of the article according to involved in the behavioural characteristic of user determines recommendation information;In response to user
Posture feature is to put down article posture, and the relevant article of the article according to involved in the behavioural characteristic of user determines recommendation information.
In some embodiments, playback method further include: determine in the image of acquisition involved in the behavioural characteristic of user
The corresponding images of items region of article;Feature is extracted from images of items region;The feature of extraction is searched for from article characteristics library,
To determine article involved in the behavioural characteristic of user.
In some embodiments, playback method further include: identify the user in the image of acquisition;In the image for obtaining acquisition
User history recommend article, so as to by history recommend article information be determined as this recommendation recommendation information.
The second aspect according to some embodiments of the invention provides a kind of playback terminal, comprising: transmitter is configured
For the image for sending acquisition;Receiver is configured as receiving recommendation information, wherein recommendation information is in the image according to acquisition
User behavior characteristics and user behavioural characteristic involved in article determine;Player is configured as playing recommendation
Breath.
In terms of third according to some embodiments of the invention, a kind of Play Server is provided, comprising: receiver is matched
It is set to the image for receiving acquisition;Processor is configured as the row of the user behavior characteristics and user in the image according to acquisition
Article involved in being characterized determines recommendation information;Transmitter is configured as sending recommendation information, so as to terminal plays recommendation
Breath.
In some embodiments, user behavior characteristics include user's posture feature;The processor of server is further matched
It is set in the image for obtaining acquisition and the image of user occurs;Human body key point information is determined in the image for user occur;According to
Human body key point information determines user's posture feature.
In some embodiments, the processor of server is configured to the image for user occur being input to key
Point prediction model obtains the human body key point information of output;Wherein, key point prediction model includes concatenated residual error network
The image of different resolution is input in attention model by ResNet model and attention model, ResNet model, attention
Model is configured with corresponding weight parameter to the input picture of every kind of resolution ratio.
In some embodiments, the processor of server is configured to determine the article in the image at current family
Region;According to the hand key point region in article region and human body key point information, the row of user is determined
It is characterized related article.
In some embodiments, user behavior characteristics include user's posture feature and user's expressive features;The place of server
Reason device is configured to be positive to touch posture or crawl posture in user's expressive features in response to user's posture feature
In the case where face expression or neutral expression, the article according to involved in the behavioural characteristic of user determines recommendation information, in user's table
In the case that feelings feature is negative expression, the relevant article of the article according to involved in the behavioural characteristic of user determines recommendation
Breath;It is to put down article posture in response to user's posture feature, the relevant article of the article according to involved in the behavioural characteristic of user
Determine recommendation information.
In some embodiments, the processor of server is configured to the behavior of user in the image for determining acquisition
The corresponding images of items region of article involved in feature;Feature is extracted from images of items region;It is searched from article characteristics library
The feature that rope extracts, to determine article involved in the behavioural characteristic of user.
In some embodiments, the processor of server is configured to the user in the image of identification acquisition;It obtains
The history of the user in the image of acquisition is taken to recommend article, to recommend history the information of article to be determined as pushing away for this recommendation
Recommend information.
The 4th aspect according to some embodiments of the invention, provides a kind of play system, comprising: aforementioned playout terminal,
And any one aforementioned Play Server.
The 5th aspect according to some embodiments of the invention, provides a kind of playing device, comprising:
Memory;And it is coupled to the processor of memory, processor is configured as based on finger stored in memory
It enables, executes any one aforementioned playback method.
The 6th aspect according to some embodiments of the invention, provides a kind of computer readable storage medium, stores thereon
There is computer program, wherein the program realizes any one aforementioned playback method when being executed by processor.
Some embodiments in foregoing invention have the following advantages that or the utility model has the advantages that the embodiment of the present invention can be in user
During browsing article under line, the possible interested information of user is analyzed according to the behavioural characteristic of user and recommends user,
So as to improve recommendation efficiency.
By referring to the drawings to the detailed description of exemplary embodiment of the present invention, other feature of the invention and its
Advantage will become apparent.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is the flow diagram according to the playback method of some embodiments of the invention.
Fig. 2 is the flow diagram according to the playback method of yet other embodiments of the invention.
Fig. 3 A is the flow diagram according to the user behavior characteristics detection method of some embodiments of the invention.
Fig. 3 B is the flow diagram that method is determined according to the human body key point information of some embodiments of the invention.
Fig. 4 is the process of the determination method of the article according to involved in the behavioural characteristic of the user of some embodiments of the invention
Schematic diagram.
Fig. 5 is the flow diagram according to the item identification method of some embodiments of the invention.
Fig. 6 is the flow diagram that method is determined according to the recommendation information of some embodiments of the invention.
Fig. 7 is the flow diagram according to the playback method of other embodiments of the invention.
Fig. 8 is the structural schematic diagram according to the playback terminal of some embodiments of the invention.
Fig. 9 is the structural schematic diagram according to the Play Server of some embodiments of the invention.
Figure 10 is the structural schematic diagram according to the play system of some embodiments of the invention.
Figure 11 is the structural schematic diagram according to the playing device of other embodiments of the invention.
Figure 12 is the structural schematic diagram according to the playing device of yet other embodiments of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Below
Description only actually at least one exemplary embodiment be it is illustrative, never as to the present invention and its application or make
Any restrictions.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
Unless specifically stated otherwise, positioned opposite, the digital table of the component and step that otherwise illustrate in these embodiments
It is not limited the scope of the invention up to formula and numerical value.
Simultaneously, it should be appreciated that for ease of description, the size of various pieces shown in attached drawing is not according to reality
Proportionate relationship draw.
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable
In the case of, the technology, method and apparatus should be considered as authorizing part of specification.
It is shown here and discuss all examples in, any occurrence should be construed as merely illustratively, without
It is as limitation.Therefore, the other examples of exemplary embodiment can have different values.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, then in subsequent attached drawing does not need that it is further discussed.
Fig. 1 is the flow diagram according to the playback method of some embodiments of the invention.As shown in Figure 1, the embodiment
Playback method includes step S102~S106.
In step s 102, the image of acquisition is sent.Image can be acquired by the lower camera arranged online, camera
It can be for one or multiple.The places such as shop, exhibition under line can be set in camera.
In step S104, recommendation information is received, wherein recommendation information is that the user behavior in the image according to acquisition is special
What article involved in sign and the behavioural characteristic of user determined.
User behavior characteristics may include one or more, such as can be user's posture feature, user's expressive features etc.
Deng.User's posture feature for example may include the posture of user, movement etc., such as user touches article, grabs and take, put
Inferior operation.The article that article involved in the behavioural characteristic of user for example can be touched, be grabbed, put down for user, can also be with
Article etc. before the position stopped for a long time for user.
Recommendation information can be involved by the information of article involved in the behavioural characteristic of user or the behavioural characteristic of user
And the relevant information of article, these information can for article brief introduction, price, evaluation, picture, placement position, advertisement, promotion letter
Breath etc..
In step s 106, recommendation information is played.
Method through the foregoing embodiment, it is special according to the behavior of user during article can be browsed under user's line
The possible interested information of sign analysis user simultaneously recommends user, so as to improve recommendation efficiency.
In some embodiments, server side can analyze received image, so that it is determined that recommendation information.Below
The embodiment of playback method of the present invention is described with reference to Fig. 2.
Fig. 2 is the flow diagram according to the playback method of yet other embodiments of the invention.As shown in Fig. 2, the embodiment
Playback method include step S202~S206.
In step S202, the image of acquisition is received.
In step S204, according to involved in the behavioural characteristic of user behavior characteristics and user in the image of acquisition
Article determines recommendation information.
In step S206, recommendation information is sent, so as to terminal plays recommendation information.
In some embodiments, user behavior characteristics include user's posture feature.User of the present invention is described below with reference to Fig. 3
The embodiment of behavioural characteristic detection method.
Fig. 3 A is the flow diagram according to the user behavior characteristics detection method of some embodiments of the invention.Such as Fig. 3 A institute
Show, the user behavior characteristics detection method of the embodiment includes step S302~S306.
In step s 302, it obtains and occurs the image of user in the image of acquisition.
Due to there is no user's appearance in parts of images, can detect first in the image of acquisition whether have user.?
In some embodiments, detection image can be come using a kind of YOLO (You Only Look Once, object detection algorithm) algorithm
In whether someone occur.Furthermore SSD (Single Shot MultiBox Detector, single-lens more box detections also can be used
Device), Faster RCNN (Faster Regions with Convolutional Neural Network features, base
The methods of fast convolution neural network in region).
In step s 304, human body key point information is determined in the image for user occur.Human body key point information is for example
It can be trunk, the key point of hand etc..The posture of user can be sketched the contours of by human body key point information.
In step S306, user's posture feature is determined according to human body key point information.
After user's posture feature has been determined, it may further determine that the article of user's interaction, such as user are touched, grabbed
Article under taking, putting,
Method through the foregoing embodiment can determine user's posture feature, and then can determine the friendship of user and commodity
Mutual process.So as to reflecting interest of the user to the commodity interacted, improving the accuracy rate of recommendation and recommending efficiency.
In some embodiments, step S304 can be implemented as step S3042, as shown in Figure 3B.In step S3042
In, the image for user occur is input to key point prediction model, obtains the human body key point information of output.Crucial point prediction mould
Type includes concatenated ResNet (residual error network) model and attention model.ResNet model inputs the image of different resolution
It into attention model, such as can be ResNet101;Attention model is to the input picture of every kind of resolution ratio configured with corresponding
Weight parameter.
The characteristics of ResNet model is that the output of first different layers can be used as the input of posterior same layer, different layers
Output there are a variety of resolution ratio, so as to realize feature pyramid structure, allow network preferably combine it is high-rise and
The information of low layer is learnt.And the characteristics of image that the image attention of different resolution is different, for example, the image of low resolution closes
Infuse global information, high-resolution image attention local detail.Pass through the corresponding weight ginseng of image configurations for different resolution
Number, can make network carry out the study stressed and training to different types of information.
The present invention can also in advance instruct key point prediction model using the training data for being labelled with human body key point
Practice.It, can be according to the human body key point of ResNet model prediction and the mark of training data in the training process of some embodiments
Note value calculates the loss function of first stage, the mark value of human body key point and training data further according to attention model prediction
The loss function for calculating second stage, is trained model according to the loss function in two stages.In some embodiments,
The loss function of first stage can be enabled to calculate the loss of whole human body key points, enable the loss function of second stage calculate difficult to estimate
The loss of key point is counted, so as to the prediction accuracy and computational efficiency of further lift scheme.Meter key point difficult to estimate can be with
It is determined according to the size of the corresponding penalty values of the key point.
In some embodiments, it can further determine that user's by the position of the posture of user and human body key point
Article involved in behavioural characteristic.Below with reference to the determination method for the article that Fig. 4 behavioural characteristic for describing user of the present invention is related to.
Fig. 4 is the process of the determination method of the article according to involved in the behavioural characteristic of the user of some embodiments of the invention
Schematic diagram.As shown in figure 4, the article of the embodiment determines that method includes step S402~404.
In step S402, the article region in the image at current family is determined.In some embodiments, it can adopt
The article in image is determined with the algorithm of target detection for detecting commodity.
In step s 404, according to the hand key point region in article region and human body key point information,
Determine article involved in the behavioural characteristic of user.
In some embodiments, the intersection that can calculate the region and article region where hand key point, passes through
Intersection area information is compared with preset threshold value.If intersection area information is greater than threshold value, illustrate that the article is to use
Article involved in the behavioural characteristic at family.Intersection area information for example can be the size of intersection area, or intersection area
The ratio equal proportion information of domain size and hand key point region or article region.
Method through the foregoing embodiment can determine that user is interacted by the hand key point position of user
Article improve recommendation efficiency to provide foundation for subsequent information recommendation process.
In some embodiments, commodity detection algorithm can detecte which position in image there are articles.If necessary
Determine that these articles are, it is also necessary to further be identified.Item identification method of the present invention is described below with reference to Fig. 5
Embodiment.
Fig. 5 is the flow diagram according to the item identification method of some embodiments of the invention.As shown in figure 5, the implementation
The item identification method of example includes step S502~S506.
In step S502, the corresponding images of items of article involved in the behavioural characteristic of user in the image of acquisition is determined
Region.For example, can determine images of items region using algorithm of target detection.
In step S504, feature is extracted from images of items region.Such as object can be extracted by deep neural network
The feature of product.The feature of extraction may include color, texture, edge, angle point etc., and which is not described herein again.
In step S506, the feature of extraction is searched for from article characteristics library, to determine involved by the behavioural characteristic of user
Article.Feature in article characteristics library including article and the corresponding relationship between the mark of article.By carrying out feature
Match, can determining the behavioural characteristic of user, what is involved is which or which articles.
In some embodiments, can classify according to user's posture feature, to determine the current specific appearance of user
State.For example, user's posture feature can be input to in advance trained posture disaggregated model, it may thereby determine that user is current
Posture is to touch article, pick up article, put down article or other postures.
After the specific posture of user has been determined, different types of letter can be recommended for user according to the difference of posture
Breath.It in some embodiments, is to touch posture or crawl posture in response to user's posture feature, according to the behavioural characteristic of user
Related article determines recommendation information;It is to put down article posture in response to user's posture feature, according to the behavioural characteristic of user
The relevant article of related article determines recommendation information.To be worked as with recommended user when user generates interest to article
The information of the article of preceding interaction;When user is unsatisfied with current item, the information of other relative articles can be recommended.By this
The method of embodiment can determine the content recommended according to the posture of user in real time, improve the accuracy of recommendation.
In some embodiments, user behavior characteristics can also include user's expressive features.Pass through the table current to user
Feelings are identified, can be recommended according to the recognition result of user's expression.
It in some embodiments, is positive expression or neutral expression in response to user's expressive features, according to the behavior of user
Article involved in feature determines recommendation information;It is negative expression in response to user's expressive features, according to the behavioural characteristic of user
The relevant article of related article determines recommendation information.Positive expression for example can indicate full for smile, laugh, happy etc.
The expression of meaning, neutral expression can be for example poker-faced, hesitation etc., and negative expression for example can be to frown, curl one's lip, sighing
Etc. the expression of meaning with thumb down.
It is thus possible to determine the content recommended in real time according to the expression of user, the accuracy of recommendation is improved.
The embodiment of the present invention can be combined with the posture of user and expression is recommended.This hair is described below with reference to Fig. 6
Bright recommendation information determines the embodiment of method.
Fig. 6 is the flow diagram that method is determined according to the recommendation information of some embodiments of the invention.As shown in fig. 6, should
The recommendation information of embodiment determines that method includes step S602~S608.
In step S602, user's posture feature in image is determined.If in user's posture feature be touch posture or
Person grabs posture, executes step S604;If user's posture feature is to put down article posture, step S608 is executed.
In step s 604, user's expressive features in image are determined.If user's expressive features be positive expression or in
Property expression, execute step S606;If user's expressive features are negative expression, step S608 is executed.
In step S606, the article according to involved in the behavioural characteristic of user determines recommendation information.
In step S608, the relevant article of the article according to involved in the behavioural characteristic of user determines recommendation information.
Method through the foregoing embodiment can synthetically determine recommendation information in conjunction with the posture and expression of user, thus
The content of recommendation can be quickly determined according to the variation of user's current signature.
In some embodiments, for that there may be the user of purchase intention again, history can also be carried out and recommend article
It repeats to recommend.The embodiment of playback method of the present invention is described below with reference to Fig. 7.
Fig. 7 is the flow diagram according to the playback method of other embodiments of the invention.As shown in fig. 7, the embodiment
Playback method include step S702~S704.
In step S702, the user in the image of acquisition is identified.In some embodiments, it can be calculated by recognition of face
Method is identified.Network side can store the information of the user once identified, so that the user identified in image is
Old user or new user.
In step S704, the history for obtaining the user in the image of acquisition recommends article, so that history is recommended article
Information be determined as this recommendation recommendation information.
In addition, the more matchmakers used when broadcasting can also be switched according to the gender information of the face identified, age information etc.
Voxel material, such as the style of display interface, the style of background music etc..
Method through the foregoing embodiment can recommend history to recommend article for user, improve the success rate of recommendation and push away
Recommend efficiency.
The embodiment of playback terminal of the present invention is described below with reference to Fig. 8.
Fig. 8 is the structural schematic diagram according to the playback terminal of some embodiments of the invention.As shown in figure 8, the embodiment
Playback terminal 800 includes: transmitter 8100, is configured as sending the image of acquisition;Receiver 8200 is configured as receiving and recommend
Information, wherein recommendation information is involved in the behavioural characteristic of the user behavior characteristics and user in the image according to acquisition
What article determined;Player 8300 is configured as playing recommendation information.
The embodiment of Play Server of the present invention is described below with reference to Fig. 9.
Fig. 9 is the structural schematic diagram according to the Play Server of some embodiments of the invention.As shown in figure 9, the embodiment
Play Server 900 include: receiver 9100, be configured as receive acquisition image;Processor 9200, is configured as basis
User behavior characteristics in the image of acquisition and article involved in the behavioural characteristic of user determine recommendation information;Transmitter
9300, it is configured as sending recommendation information, so as to terminal plays recommendation information.
In some embodiments, user behavior characteristics include user's posture feature;The processor 9200 of server is further
It is configured as the image of user occur in the image for obtaining acquisition;Human body key point information is determined in the image for user occur;
User's posture feature is determined according to human body key point information.
In some embodiments, the processor 9200 of server is configured to for the image for user occur being input to
Key point prediction model obtains the human body key point information of output;Wherein, key point prediction model includes concatenated residual error network
The image of different resolution is input in attention model by ResNet model and attention model, ResNet model, attention
Model is configured with corresponding weight parameter to the input picture of every kind of resolution ratio.
In some embodiments, the processor 9200 of server is configured in the image for determining current family
Article region;According to the hand key point region in article region and human body key point information, user is determined
Behavioural characteristic involved in article.
In some embodiments, user behavior characteristics include user's posture feature and user's expressive features;The place of server
Reason device 9200 is configured in response to user's posture feature to touch posture or crawl posture, in user's expressive features
In the case where for positive expression or neutral expression, the article according to involved in the behavioural characteristic of user determines recommendation information, with
In the case that family expressive features are negative expression, the relevant article of the article according to involved in the behavioural characteristic of user, which determines, to be recommended
Information;It is to put down article posture in response to user's posture feature, the relevant object of the article according to involved in the behavioural characteristic of user
Product determine recommendation information.
In some embodiments, the processor 9200 of server is configured to user in the image for determining acquisition
The corresponding images of items region of article involved in behavioural characteristic;Feature is extracted from images of items region;From article characteristics library
The feature that middle search is extracted, to determine article involved in the behavioural characteristic of user.
In some embodiments, the processor 9200 of server is configured to the use in the image of identification acquisition
Family;The history for obtaining the user in the image of acquisition recommends article, pushes away to recommend the information of article to be determined as this history
The recommendation information recommended.
The embodiment of play system of the present invention is described below with reference to Figure 10.
Figure 10 is the structural schematic diagram according to the play system of some embodiments of the invention.As shown in Figure 10, the embodiment
Play system 100 include playback terminal 1010 and Play Server 1020.Playback terminal 1010 and Play Server 1020
Specific embodiment can refer to previous embodiment, and which is not described herein again.
Figure 11 is according to the structural schematic diagram of the playing device of other embodiments of the invention, and playing device can be broadcasting
Terminal or Play Server.As shown in figure 11, the playing device 110 of the embodiment includes: memory 1110 and is coupled to this
The processor 1120 of memory 1110, processor 1120 are configured as based on the instruction being stored in memory 1110, before execution
State the playback method in any one embodiment.
Wherein, memory 1110 is such as may include system storage, fixed non-volatile memory medium.System storage
Device is for example stored with operating system, application program, Boot loader (BootLoader) and other programs etc..
Figure 12 is according to the structural schematic diagram of the playing device of yet other embodiments of the invention, and playing device can be broadcasting
Terminal or Play Server.As shown in figure 12, the playing device 120 of the embodiment includes: memory 1210 and processor
1220, it can also include input/output interface 1230, network interface 1240, memory interface 1250 etc..These interfaces 1230,
It can for example be connected by bus 1260 between 1240,1250 and memory 1210 and processor 1220.Wherein, input and output
The input-output equipment such as interface 1230 is display, mouse, keyboard, touch screen provide connecting interface.Network interface 1240 is each
Kind networked devices provide connecting interface.The external storages such as memory interface 1250 is SD card, USB flash disk provide connecting interface.
The embodiment of the present invention also provides a kind of computer readable storage medium, is stored thereon with computer program, special
Sign is that the program realizes any one aforementioned playback method when being executed by processor.
Those skilled in the art should be understood that the embodiment of the present invention can provide as method, system or computer journey
Sequence product.Therefore, complete hardware embodiment, complete software embodiment or combining software and hardware aspects can be used in the present invention
The form of embodiment.Moreover, it wherein includes the calculating of computer usable program code that the present invention, which can be used in one or more,
Machine can use the meter implemented in non-transient storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of calculation machine program product.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It is interpreted as to be realized by computer program instructions each in flowchart and/or the block diagram
The combination of process and/or box in process and/or box and flowchart and/or the block diagram.It can provide these computer journeys
Sequence instruct to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices processor with
A machine is generated, so that the instruction generation executed by computer or the processor of other programmable data processing devices is used for
Realize the dress for the function of specifying in one or more flows of the flowchart and/or one or more blocks of the block diagram
It sets.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (20)
1. a kind of playback method, comprising:
Send the image of acquisition;
Receive recommendation information, wherein the recommendation information is user behavior characteristics and user in the image according to acquisition
What article involved in behavioural characteristic determined;
Play the recommendation information.
2. playback method according to claim 1, wherein the user behavior characteristics include user's posture feature;
The playback method further include:
It obtains in the image of acquisition and the image of user occurs;
Human body key point information is determined in the image for user occur;
User's posture feature is determined according to human body key point information.
3. playback method according to claim 2, wherein described to determine human body key point letter in the image for user occur
Breath includes:
The image for user occur is input to key point prediction model, obtains the human body key point information of output;
Wherein, the key point prediction model includes concatenated residual error network ResNet model and attention model, ResNet mould
The image of different resolution is input in attention model by type, and attention model is configured with the input picture of every kind of resolution ratio
Corresponding weight parameter.
4. playback method according to claim 2, further includes:
Determine the article region in the image for user occur;
According to the hand key point region in article region and human body key point information, the behavioural characteristic of user is determined
Related article.
5. playback method according to claim 2, further includes:
It is to touch posture or grab posture in response to user's posture feature, the article according to involved in the behavioural characteristic of user is true
Determine recommendation information;
It is to put down article posture in response to user's posture feature, the relevant article of the article according to involved in the behavioural characteristic of user
Determine recommendation information.
6. playback method according to claim 1, wherein the user behavior characteristics include user's expressive features;
The playback method further include:
It is positive expression or neutral expression in response to user's expressive features, the article according to involved in the behavioural characteristic of user determines
Recommendation information;
It is negative expression in response to user's expressive features, the relevant article of the article according to involved in the behavioural characteristic of user determines
Recommendation information.
7. playback method according to claim 1, wherein the user behavior characteristics include user's posture feature and user
Expressive features;
The playback method further include:
It is to touch posture or crawl posture in response to user's posture feature, is positive expression or neutral table in user's expressive features
In the case where feelings, the article according to involved in the behavioural characteristic of user determines recommendation information, is negative table in user's expressive features
In the case where feelings, the relevant article of the article according to involved in the behavioural characteristic of user determines recommendation information;
It is to put down article posture in response to user's posture feature, the relevant article of the article according to involved in the behavioural characteristic of user
Determine recommendation information.
8. playback method according to claim 1, further includes:
Determine the corresponding images of items region of article involved in the behavioural characteristic of user in the image of acquisition;
Feature is extracted from the images of items region;
The feature of extraction is searched for, from article characteristics library to determine article involved in the behavioural characteristic of user.
9. playback method according to claim 1, further includes:
Identify the user in the image of acquisition;
The history for obtaining the user in the image of acquisition recommends article, to recommend the history information of article to be determined as this
The recommendation information of secondary recommendation.
10. a kind of playback terminal, comprising:
Transmitter is configured as sending the image of acquisition;
Receiver is configured as receiving recommendation information, wherein the recommendation information is the user behavior in the image according to acquisition
What article involved in feature and the behavioural characteristic of user determined;
Player is configured as playing the recommendation information.
11. a kind of Play Server, comprising:
Receiver is configured as receiving the image of acquisition;
Processor is configured as object involved in the behavioural characteristic of the user behavior characteristics and user in the image according to acquisition
Product determine recommendation information;
Transmitter is configured as sending recommendation information, so as to terminal plays recommendation information.
12. Play Server according to claim 11, wherein the user behavior characteristics include user's posture feature;
The processor of the server is configured to the image of user occur in the image for obtaining acquisition;There is user
Image in determine human body key point information;User's posture feature is determined according to human body key point information.
13. Play Server according to claim 12, wherein the processor of the server be configured to by
The image for user occur is input to key point prediction model, obtains the human body key point information of output;Wherein, the key point is pre-
Surveying model includes concatenated residual error network ResNet model and attention model, and ResNet model is defeated by the image of different resolution
Enter into attention model, attention model is configured with corresponding weight parameter to the input picture of every kind of resolution ratio.
14. Play Server according to claim 12, wherein the processor of the server is configured to really
Article region in the fixed image for user occur;According to the hand in article region and human body key point information
Key point region determines article involved in the behavioural characteristic of user.
15. Play Server according to claim 11, wherein the user behavior characteristics include user's posture feature and
User's expressive features;
The processor of the server is configured in response to user's posture feature to touch posture or crawl posture,
In the case where user's expressive features are positive expression or neutral expression, the article according to involved in the behavioural characteristic of user is determined
Recommendation information, in the case where user's expressive features are negative expression, the article according to involved in the behavioural characteristic of user is related
Article determine recommendation information;It is to put down article posture in response to user's posture feature, according to involved by the behavioural characteristic of user
The relevant article of article determine recommendation information.
16. Play Server according to claim 11, wherein the processor of the server is configured to really
Surely the corresponding images of items region of article involved in the behavioural characteristic of user in the image acquired;From the images of items region
Middle extraction feature;The feature of extraction is searched for, from article characteristics library to determine article involved in the behavioural characteristic of user.
17. Play Server according to claim 11, wherein the processor of the server is configured to know
The user in image not acquired;The history for obtaining the user in the image of acquisition recommends article, to recommend the history
The information of article is determined as the recommendation information of this recommendation.
18. a kind of play system, comprising:
Playback terminal described in any one of claim 10, and
Play Server described in any one of claim 11~17.
19. a kind of playing device, comprising:
Memory;And
It is coupled to the processor of the memory, the processor is configured to the instruction based on storage in the memory,
Execute such as playback method according to any one of claims 1 to 9.
20. a kind of computer readable storage medium, is stored thereon with computer program, power is realized when which is executed by processor
Benefit require any one of 1~9 described in playback method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811483294.2A CN109544252A (en) | 2018-12-06 | 2018-12-06 | Playback method, system and relevant device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811483294.2A CN109544252A (en) | 2018-12-06 | 2018-12-06 | Playback method, system and relevant device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109544252A true CN109544252A (en) | 2019-03-29 |
Family
ID=65852938
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811483294.2A Pending CN109544252A (en) | 2018-12-06 | 2018-12-06 | Playback method, system and relevant device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109544252A (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160364781A1 (en) * | 2015-04-15 | 2016-12-15 | NetDisruptors, LLC | Commerce Recommendation System |
CN108564414A (en) * | 2018-04-23 | 2018-09-21 | 帷幄匠心科技(杭州)有限公司 | Method of Commodity Recommendation based on behavior under line and system |
-
2018
- 2018-12-06 CN CN201811483294.2A patent/CN109544252A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160364781A1 (en) * | 2015-04-15 | 2016-12-15 | NetDisruptors, LLC | Commerce Recommendation System |
CN108564414A (en) * | 2018-04-23 | 2018-09-21 | 帷幄匠心科技(杭州)有限公司 | Method of Commodity Recommendation based on behavior under line and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Shen et al. | Exemplar-based human action pose correction and tagging | |
US8515982B1 (en) | Annotations for three-dimensional (3D) object data models | |
US8606645B1 (en) | Method, medium, and system for an augmented reality retail application | |
CN110310175A (en) | System and method for mobile augmented reality | |
US20140132595A1 (en) | In-scene real-time design of living spaces | |
US20200402307A1 (en) | System and method for camera based cloth fitting and recommendation | |
CN108830251A (en) | Information correlation method, device and system | |
US10671841B2 (en) | Attribute state classification | |
CN106062673A (en) | Controlling a computing-based device using gestures | |
CN106776619A (en) | Method and apparatus for determining the attribute information of destination object | |
CN109635680A (en) | Multitask attribute recognition approach, device, electronic equipment and storage medium | |
WO2020134102A1 (en) | Article recognition method and device, vending system, and storage medium | |
CN108229559A (en) | Dress ornament detection method, device, electronic equipment, program and medium | |
CN106612230A (en) | Media information promotion method, client and server | |
CN107220876A (en) | Commodity information management method, device and mobile terminal | |
CN112307864A (en) | Method and device for determining target object and man-machine interaction system | |
Ikram et al. | Skeleton based dynamic hand gesture recognition using LSTM and CNN | |
CN110209858A (en) | Exhibiting pictures determination, object search, methods of exhibiting, device, equipment and medium | |
Shadrach et al. | Smart virtual trial room for apparel industry | |
US20170013309A1 (en) | System and method for product placement | |
US20220300774A1 (en) | Methods, apparatuses, devices and storage media for detecting correlated objects involved in image | |
CN110246280B (en) | Human-cargo binding method and device, computer equipment and readable medium | |
CN109544252A (en) | Playback method, system and relevant device | |
Chang et al. | A Decision Tree based Real-time Hand Gesture Recognition Method using Kinect | |
Zhang et al. | Optimization of virtual reality in brand Identity Design and Visual Recognition. based on Image Fusion and Text Assistance |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190329 |
|
RJ01 | Rejection of invention patent application after publication |