A context-adaptive intelligent recommendation method and system

Abstract

The present invention provides a context-adapted intelligent recommendation method and system, which respond to an advertisement push request from a recommendation end, receive interactive push data sent by the recommendation end, obtain a target item sent by a user end, retrieve the corresponding interactive push data based on the target item and display it to the user end; receive customized posture data uploaded by the user end based on the interactive push data, obtain a posture comparison result based on comparison information between the customized posture data and the interactive push data; when it is determined that the posture comparison result is a similar comparison result, replace the demonstration model in the interactive push data according to a context posture adaptation strategy and the customized posture data to obtain context-adapted data; when it is determined that the posture comparison result is a difference comparison result, retrieve a free posture adaptation strategy to adapt the customized model in the customized posture data to obtain context-adapted data, and send the context-adapted data to the user end.

Description

Technical Field

[0001] The present invention relates to data processing technology, and in particular to a context-adaptive intelligent recommendation method and system. Background Art

[0002] With the rapid development of digital marketing technology, the forms of intelligent recommendations are becoming increasingly diverse, including but not limited to notification bar ads, in-app ads, push notification ads, etc. These advertising push methods can be accurately delivered based on multiple factors such as user interests, behaviors, and geographic locations, thereby increasing advertising exposure and user experience, helping users purchase the items they want.

[0003] However, traditional advertising push systems often focus on content recommendations based on user browsing history, search history and other data, but ignore the importance of user active participation and personalized expression. Users can only passively accept advertising content pushed by the system and are unable to customize or interact according to their actual needs and usage scenarios, which weakens the user's sense of participation and experience quality.

[0004] Therefore, how to customize advertisements according to user needs, thereby improving user interactive experience and making advertising content more in line with user actual needs and preferences has become an urgent problem to be solved. Summary of the invention

[0005] The embodiments of the present invention provide a context-adaptive intelligent recommendation method and system, which can customize advertisements according to user needs, thereby improving the user's interactive experience and making the advertisement content more in line with the user's actual needs and preferences.

[0006] A first aspect of an embodiment of the present invention provides a context-adaptive intelligent recommendation method, comprising:

[0007] Respond to the advertising push request of the recommending end, receive the interactive push data sent by the recommending end, obtain the target item sent by the user end, and retrieve the corresponding interactive push data based on the target item to display it to the user end;

[0008] Receiving customized posture data uploaded by the user terminal based on the interactive push data, and obtaining a posture comparison result according to comparison information between the customized posture data and the interactive push data;

[0009] When it is determined that the posture comparison result is a similar comparison result, the demonstration model in the interactive push data is replaced according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data;

[0010] When it is determined that the posture comparison result is a difference comparison result, the free posture adaptation strategy is called to adapt the customized model in the customized posture data to obtain scenario adaptation data, and the scenario adaptation data is sent to the user end.

[0011] Optionally, in a possible implementation manner of the first aspect, the receiving user terminal obtains a posture comparison result based on the customized posture data uploaded by the interactive push data according to comparison information between the customized posture data and the interactive push data, including:

[0012] Receiving the customized posture data uploaded by the user terminal based on the interactive push data, extracting key parts in the customized posture data as customized key parts based on OpenCV, and extracting contours corresponding to the key parts as customized part contours;

[0013] Determine a center point of the customized gesture data, and perform coordinate processing on the customized gesture data based on the center point;

[0014] Acquire the contour coordinates of the customized part contour, and determine the contour center coordinates of the customized key part based on the contour coordinates;

[0015] Select any two customized key parts to combine, obtain a key part combination as a customized part combination, calculate according to the contour center coordinates corresponding to the customized key parts in the customized part combination, and obtain a customized combination distance corresponding to the customized part combination;

[0016] Extracting key parts in the interactive push data as interactive key parts based on OpenCV, and contours corresponding to the key parts as interactive part contours;

[0017] Determine a center point of the interactive push data, and perform coordinate processing on the interactive push data based on the center point;

[0018] Acquire the contour coordinates of the contour of the interactive part, and determine the contour center coordinates of the interactive key part based on the contour coordinates;

[0019] Select any two interactive key parts to combine, obtain a key part combination as an interactive part combination, and calculate according to the contour center coordinates corresponding to the interactive key parts in the interactive part combination to obtain an interactive combination distance corresponding to the interactive part combination;

[0020] A posture comparison result is obtained according to the customized combined distance and the interactive combined distance.

[0021] Optionally, in a possible implementation manner of the first aspect, obtaining a posture comparison result according to the customized combined distance and the interactive combined distance includes:

[0022] Obtaining a part gap value corresponding to each of the key part combinations according to the absolute value of the difference between the interactive combination distance and the corresponding customized combination distance;

[0023] Counting the difference values ​​of the parts to obtain a posture difference value, and when it is determined that the posture difference value is greater than a preset difference value, generating a difference comparison result;

[0024] When it is determined that the posture difference value is less than or equal to the preset difference value, generating a similarity comparison result;

[0025] A posture comparison result is obtained according to the difference comparison result and the similarity comparison result.

[0026] Optionally, in a possible implementation manner of the first aspect, when determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain the situational adaptation data includes:

[0027] When it is determined that the posture comparison result is a similar comparison result, a demonstration fixed area of ​​the demonstration model in the interactive push data is determined according to the target object, and the remaining areas in the demonstration model are used as demonstration replacement areas;

[0028] Acquiring the placement attribute of the target object, and when determining that the placement attribute is a portable attribute, determining a customized deletion area of ​​the customized model in the customized posture data based on the target object, and using the remaining area in the customized model as a customized reserved area;

[0029] Determine the area center point of the demonstration replacement area as the replacement midpoint, and determine the center point of the customized reservation area as the reservation midpoint;

[0030] Deleting and updating the demonstration replacement area in the demonstration model, and deleting and updating the customized deletion area of ​​the customized model;

[0031] Aligning the retained midpoint with the replacement midpoint to obtain scenario-adapted data;

[0032] When it is determined that the placement attribute is a non-portable attribute, the customized model in the customized posture data is used as a character reserved area;

[0033] Acquire the center point of the demonstration replacement area as the replacement midpoint, and determine the center point of the character retention area as the retention midpoint;

[0034] The demonstration replacement area in the demonstration model is deleted and updated, and the retained midpoint is aligned with the replacement midpoint to obtain scenario adaptation data.

[0035] Optionally, in a possible implementation of the first aspect, the method further includes:

[0036] Acquire an area without pixels in the scene adaptation data as an area to be filled, select pixels adjacent to the area to be filled as adjacent pixels, and determine pixel values ​​of the adjacent pixels as screening pixel values;

[0037] Based on the screening pixel values, the number of the adjacent pixel points corresponding to the screening pixel values ​​is counted as a screening number, and the screening pixel value corresponding to the largest screening number is selected as a filling pixel value;

[0038] The filling pixel value is filled into the area to be filled to obtain the filled scene adaptation data.

[0039] Optionally, in a possible implementation of the first aspect, when it is determined that the posture comparison result is a difference comparison result, a free posture adaptation strategy is called to perform adaptation processing on the customized model in the customized posture data to obtain scenario adaptation data, including:

[0040] When it is determined that the posture comparison result is a difference comparison result, identifying the facial orientation information of the customized model in the customized posture data, and determining the orientation information of the customized model based on the facial orientation information, wherein the orientation information includes a front orientation, a side orientation, and a back orientation;

[0041] Retrieving a preset object model for placing an object in an orientation corresponding to the target object based on the orientation information;

[0042] A key part corresponding to the target object is determined as a placement part, and the preset object model is set at the placement part to obtain scenario adaptation data.

[0043] Optionally, in a possible implementation of the first aspect, the method further includes:

[0044] Acquire pixel points of the customized model in the scenario adaptation data as model pixel points, determine the center point of the scenario adaptation data, and construct a coordinate system based on the center point to coordinate the scenario adaptation data;

[0045] Taking the coordinates corresponding to the model pixel points as the model pixel point coordinates, obtaining the minimum vertical coordinate and the maximum vertical coordinate of the model pixel point coordinates;

[0046] Determine the middle part of the customized model, obtain the coordinates of the pixel points corresponding to the middle part as the middle coordinates, and determine the ordinate of the middle coordinates as the middle ordinate;

[0047] An upper scale distance is obtained based on a difference between the maximum ordinate and the middle ordinate, and a lower scale distance is obtained based on a difference between the middle ordinate and the minimum ordinate;

[0048] Obtaining a minimum horizontal coordinate and a maximum horizontal coordinate of the intermediate coordinate, and obtaining an intermediate proportional distance based on a difference between the maximum horizontal coordinate and the minimum horizontal coordinate;

[0049] When it is determined that the placement attribute of the target object is a carry-on attribute, the target object is regarded as a carry-on item;

[0050] Retrieving a placement comparison table, wherein the placement comparison table has preset sizes corresponding to a preset lower ratio interval and a preset middle ratio interval, and preset sizes corresponding to a preset upper ratio interval and a preset middle ratio interval;

[0051] When it is determined that the placement location of the carry-on item is the lower half, the lower proportional distance and the middle proportional distance are compared with the placement comparison table to determine the corresponding preset size as the selected size, and the preset item model is adjusted according to the selected size to obtain the adjusted item model;

[0052] When it is determined that the placement location of the carry-on item is the upper half, the upper proportional distance and the middle proportional distance are compared with the placement comparison table, the corresponding preset size is determined as the selected size, and the preset item model is adjusted according to the selected size to obtain the adjusted item model.

[0053] Optionally, in a possible implementation of the first aspect, the method further includes:

[0054] When it is determined that the placement attribute of the target item is a non-portable item, the target item is treated as a non-portable item;

[0055] Determine the actual full-scale distance according to the difference between the maximum ordinate and the minimum ordinate, and retrieve the preset full-scale distance corresponding to the preset object model;

[0056] Based on the ratio of the actual full-scale distance to the preset full-scale distance, a scale adjustment factor is obtained, and the preset object model is scaled according to the scale adjustment factor to obtain an adjusted object model.

[0057] Optionally, in a possible implementation of the first aspect, the method further includes:

[0058] The active adjustment multiple sent by the user terminal based on the preset object model is received, and the preset object model is scaled according to the active adjustment multiple to obtain an adjusted object model.

[0059] A second aspect of an embodiment of the present invention provides a context-adaptive intelligent recommendation system, including:

[0060] A response module is used to respond to the advertising push request of the recommendation end, receive the interactive push data sent by the recommendation end, obtain the target item sent by the user end, and retrieve the corresponding interactive push data based on the target item to display it to the user end;

[0061] A comparison module, used for receiving the customized posture data uploaded by the user terminal based on the interactive push data, and obtaining a posture comparison result according to the comparison information between the customized posture data and the interactive push data;

[0062] A replacement module, for, when determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data;

[0063] The adaptation module is used to determine that the posture comparison result is a difference comparison result, call the free posture adaptation strategy to adapt the customized model in the customized posture data, obtain the scenario adaptation data, and send the scenario adaptation data to the user end.

[0064] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: a memory, a processor and a computer program, wherein the computer program is stored in the memory, and the processor runs the computer program to execute the first aspect of the present invention and various methods that may be involved in the first aspect.

[0065] According to a fourth aspect of an embodiment of the present invention, a storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the first aspect of the present invention and various methods that may be involved in the first aspect.

[0066] The beneficial effects of the present invention are as follows:

[0067] 1. The present invention can provide users with highly personalized advertising display by receiving customized posture data uploaded by the user end and comparing it with the interactive push data in the advertising push. Users can customize postures according to their preferences and habits, so that the advertising content is closer to the actual needs and interests of users, enhancing the interactivity of advertisements and user participation. In addition, OpenCV is used to automatically extract key parts and contours, coordinate them, and calculate the combined distance, thereby realizing intelligent processing of posture comparison. At the same time, by formulating detailed adaptation strategies, such as situational posture adaptation and free posture adaptation, as well as accurate replacement and adjustment of customized postures and demonstration models, the present invention can efficiently perform advertising push tasks, improve work efficiency, and thus enhance user experience.

[0068] 2. The present invention can support advertisement adaptation according to the user's customized posture, and also provides a variety of adaptation strategies and adjustment methods, such as adaptation processing according to the placement attributes of the target object, scaling processing according to the user's active adjustment multiple, etc. These flexibility and scalability enable the present invention to adapt to different scenarios and needs, and provide more possibilities and choices for advertisement push. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A flow chart of a context-adaptive intelligent recommendation method provided by the present invention;

[0070] Figure 2 A schematic diagram of the structure of a context-adaptive intelligent recommendation system provided by the present invention;

[0071] Figure 3 A schematic diagram of the hardware structure of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] The terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of the present invention and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in sequences other than those illustrated or described herein.

[0074] It should be understood that in various embodiments of the present invention, the size of the sequence number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0075] It should be understood that in the present invention, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0076] It should be understood that in the present invention, "plurality" refers to two or more than two. "And / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "Contains A, B and C", "Contains A, B, C" means that A, B, and C are all included, "Contains A, B or C" means that one of A, B, and C is included, and "Contains A, B and / or C" means that any one, any two, or any three of A, B, and C are included.

[0077] It should be understood that in the present invention, "B corresponding to A", "B corresponding to A", "A corresponds to B" or "B corresponds to A" means that B is associated with A and B can be determined based on A. Determining B based on A does not mean determining B based only on A, but B can also be determined based on A and / or other information. A and B match when the similarity between A and B is greater than or equal to a preset threshold.

[0078] Depending on the context, "if" as used herein may be interpreted as "when" or "when" or "in response to determining" or "in response to detecting."

[0079] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0080] The present invention provides a context-adaptive intelligent recommendation method, such as Figure 1 As shown, steps S1-S4 are included:

[0081] S1, responding to the advertisement push request of the recommending end, receiving the interactive push data sent by the recommending end, and obtaining the target item sent by the user end, and based on the target item, retrieving the corresponding interactive push data to display to the user end.

[0082] The recommending end may be a mobile end that sends advertisements, for example, advertiser A, and the advertisement push request is a request for pushing advertisements.

[0083] The interactive push data is advertising image data that can be pushed by the recommendation end and can interact with the user end, such as a picture of a model with a shoulder bag sent by a shoulder bag advertiser. The interactive push data has corresponding interactive buttons, such as an upload button, and subsequent users can upload corresponding data based on the interactive push data to facilitate user customization.

[0084] Among them, the user end is the mobile end corresponding to the user, which can be a mobile phone, computer, etc., and the target item is the item required by the user, such as a top, a shoulder bag, etc.

[0085] It should be noted that the current method of advertising push mainly relies on users' browsing history, search history and other data to recommend content. However, users cannot interact with it, and the display cannot be customized according to user needs. Generally, the advertisements pushed to users are relatively fixed and cannot be customized according to the user's usage scenario.

[0086] Therefore, the present invention will respond to the advertising push request of the recommendation end, thereby receiving the interactive push data sent by the recommendation end, such as the character product display picture sent by the recommendation end, for example, the model picture with the advertiser's shoulder bag, etc., and obtain the target item sent by the user end, such as the shoulder bag, and then retrieve the corresponding interactive push data based on the target item to display it to the user end.

[0087] For example, when the server receives an advertising push request from the recommendation end, it means that the advertiser needs to push advertising. At this time, the server receives the shoulder bag model picture sent by the recommendation end. When the server receives a signal that a user wants to buy a shoulder bag, all shoulder bag model pictures related to the shoulder bag are retrieved based on the shoulder bag. Subsequently, the user can trigger the interactive button in the shoulder bag model picture and upload his own image, thereby showing the user the scenario of him using the shoulder bag, thereby improving the user experience and interactivity.

[0088] S2, receiving the customized posture data uploaded by the user terminal based on the interactive push data, and obtaining a posture comparison result according to comparison information between the customized posture data and the interactive push data.

[0089] The customized posture data is user image data uploaded by the user, such as a selfie image of the user.

[0090] It is understandable that the server will receive customized posture data uploaded by the user based on the interactive push data. For example, after the interactive button of the interactive push data is triggered, or the hyperlink on the picture is triggered, the posture comparison result will be obtained based on the comparison information of the customized posture data and the interactive push data.

[0091] It should be noted that when users collect customized posture data, they can collect it through the preset character model outline, so that the subsequent character posture is closer to the model posture, and collecting it through the outline facilitates subsequent size adjustment.

[0092] In some embodiments, step S2 (receiving customized gesture data uploaded by the user terminal based on the interactive push data, and obtaining a gesture comparison result according to comparison information between the customized gesture data and the interactive push data) includes S21-S29:

[0093] S21, receiving the customized posture data uploaded by the user terminal based on the interactive push data, extracting key parts in the customized posture data as customized key parts based on OpenCV, and contours corresponding to the key parts as customized part contours.

[0094] Among them, the key parts are the body parts corresponding to the character in the customized posture data, such as hands, legs, etc., and the customized part contours are the contours corresponding to the key parts in the customized posture data.

[0095] It should be noted that OpenCV is a powerful tool library for image processing, which provides a wealth of image processing operations, such as image filtering, geometric transformation, color space conversion, image recognition and contour extraction, etc. These are existing technologies and will not be described in detail here.

[0096] It can be understood that the server will receive the customized posture data uploaded by the user based on the interactive push data. For example, after the user triggers the interactive button of the interactive push data, the selfie image data corresponding to the user is uploaded. Subsequently, OpenCV can be used to extract the key parts in the customized posture data as customized key parts, and the contours corresponding to the key parts as customized part contours, such as the hand contour corresponding to the hand, the leg contour corresponding to the leg, etc.

[0097] S22, determining a center point of the customized gesture data, and performing coordinate processing on the customized gesture data based on the center point.

[0098] It is understandable that the server will determine the center point of the customized posture data and coordinate the customized posture data based on the center point. For example, the center point of the image of the person's selfie image is determined, and a coordinate system is constructed with the center point of the image as the origin of the coordinate system to coordinate the person's selfie image, so that the coordinates corresponding to each pixel in the image can be obtained.

[0099] S23, acquiring the contour coordinates of the customized part contour, and determining the contour center coordinates of the customized key part based on the contour coordinates.

[0100] It is understandable that the server will obtain the contour coordinates of the customized part contour, and determine the contour center coordinates of the customized key part based on the contour coordinates.

[0101] For example, obtain the corresponding hand contour coordinates on the hand contour, and then obtain the contour center horizontal coordinate based on the average of the maximum and minimum values ​​of the horizontal coordinate of the hand contour coordinates, and obtain the contour center vertical coordinate based on the average of the maximum and minimum values ​​of the vertical coordinate, so as to obtain the contour center coordinates of the customized key parts.

[0102] It is easy to understand that the present invention determines the center coordinates of the customized part contour according to the maximum and minimum values ​​of the abscissa and the maximum and minimum values ​​of the ordinate of the contour coordinates of the customized part contour.

[0103] S24, selecting any two customized key parts for combination to obtain a key part combination as a customized part combination, and calculating according to the contour center coordinates corresponding to the customized key parts in the customized part combination to obtain a customized combination distance corresponding to the customized part combination.

[0104] It is understandable that the server will select any two customized key parts to combine and obtain the key part combination as the customized part combination, for example, the head and the hands, the hands and the legs, etc. are combined.

[0105] Subsequently, calculation may be performed based on the contour center coordinates corresponding to the customized key parts in the customized part combination to obtain the customized combination distance corresponding to the customized part combination.

[0106] It is not difficult to understand that the contour center coordinates corresponding to each contour are known. Therefore, by calculating the distance between two points based on the contour center coordinates of the two parts, the customized combination distance corresponding to the customized part combination, that is, the distance between the two points, can be obtained.

[0107] Through the above implementation, the distance between various parts of the character in the customized posture data can be known, and then compared with the distance between various parts of the model. If the difference is small, it means that the posture is relatively similar, such as the hand is very close to the head.

[0108] S25, extracting key parts in the interactive push data as interactive key parts based on OpenCV, and contours corresponding to the key parts as interactive part contours.

[0109] It can be understood that, consistent with the principle in step S21, the server will extract the key parts in the interactive push data as interactive key parts based on OpenCV, and the contours corresponding to the key parts as interactive part contours.

[0110] S26, determining a center point of the interactive push data, and performing coordinate processing on the interactive push data based on the center point.

[0111] It is understandable that, consistent with the principle in step S22, the interactive push data will be processed in coordinates.

[0112] S27, obtaining the contour coordinates of the contour of the interactive part, and determining the center coordinates of the contour of the interactive key part based on the contour coordinates.

[0113] It can be understood that, consistent with the principle in step S23, the server will obtain the contour coordinates of the contour of the interactive part, and determine the contour center coordinates of the interactive key part based on the contour coordinates.

[0114] S28, selecting any two interactive key parts for combination to obtain a key part combination as an interactive part combination, and calculating according to the contour center coordinates corresponding to the interactive key parts in the interactive part combination to obtain an interactive combination distance corresponding to the interactive part combination.

[0115] It is not difficult to understand that, consistent with the principle in step S24, the server will select any two interactive key parts for combination, obtain the key part combination as the interactive part combination, and calculate according to the contour center coordinates corresponding to the interactive key parts in the interactive part combination to obtain the interactive combination distance corresponding to the interactive part combination.

[0116] S29, obtaining a posture comparison result according to the customized combination distance and the interactive combination distance.

[0117] In some embodiments, step S29 (obtaining a gesture comparison result according to the customized combined distance and the interactive combined distance) includes S291-S294:

[0118] S291, obtaining a part gap value corresponding to each of the key part combinations according to the absolute value of the difference between the interactive combination distance and the corresponding customized combination distance.

[0119] It is understandable that the server will select the absolute value of the difference between the interactive combination distance corresponding to the same key part combination and the corresponding customized combination distance, so as to obtain the part difference value corresponding to each key part combination, for example, the difference value between the hand and the head.

[0120] S292, counting the part difference values ​​to obtain a posture difference value, and when it is determined that the posture difference value is greater than a preset difference value, generating a difference comparison result.

[0121] It is understandable that the present invention will count the part difference values ​​corresponding to all key part combinations, so as to obtain the posture difference values ​​corresponding to the model and the user character.

[0122] It is not difficult to understand that the present invention will count the distance differences between the model and the body parts corresponding to the character postures in the user's picture, and summarize them. If the total difference is small, it is considered that the postures of the two are relatively close. If the subsequent difference is large, it is considered that the difference between the postures of the two is large.

[0123] Therefore, when it is determined that the posture difference value is greater than the preset difference value, a difference comparison result is generated, that is, the difference is large.

[0124] S293, when it is determined that the posture difference value is less than or equal to the preset difference value, generating a similarity comparison result.

[0125] It is understandable that the server will generate a similar comparison result when it determines that the posture difference value is less than or equal to the preset difference value, that is, the difference is small.

[0126] S294, obtaining a posture comparison result according to the difference comparison result and the similarity comparison result.

[0127] It is understandable that the server will obtain the posture comparison result according to the difference comparison result and the similarity comparison result. That is, the posture comparison result includes the difference comparison result and the similarity comparison result.

[0128] At the same time, the present invention can also identify the degree of difference between the customized posture data and the posture corresponding to the interactive push data through the existing technology, for example, the pre-trained person detection model (such as YOLO, Faster R-CNN, etc.) detects the person in the photo. Ensure that the outline and position of the person in the photo are accurately identified, and then extract the key points of the person by using a posture estimation model (such as OpenPose, HRNet, etc.). These key points usually include the main joints of the body: head, shoulders, elbows, wrists, hips, knees and ankles, etc., and then a threshold can be set to determine whether the postures of the people in the two photos are consistent according to the similarity score. For example, if the Euclidean distance is less than a certain threshold, or the cosine similarity is greater than a certain threshold, the posture is considered to be consistent, wherein the common method for calculating the similarity can be the Euclidean distance, which is the existing technology and will not be described here.

[0129] S3, when it is determined that the posture comparison result is a similar comparison result, the demonstration model in the interactive push data is replaced according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data.

[0130] In some embodiments, step S3 (when determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data) includes S31-S38:

[0131] S31, when it is determined that the posture comparison result is a similar comparison result, a demonstration fixed area of ​​the demonstration model in the interactive push data is determined according to the target object, and the remaining areas in the demonstration model are used as demonstration replacement areas.

[0132] The demonstration model is a model of a mannequin other than the background in the interactive push data, and the demonstration model includes the items worn by the model. The demonstration fixed area is the area where the corresponding outline of the target item in the interactive push data is located. The demonstration replacement area is other areas other than the corresponding outline of the target item in the demonstration model.

[0133] It is understandable that when the posture comparison result is determined to be a similar comparison result, it means that the two are relatively consistent. Therefore, the present invention will subsequently replace the model in the interactive push data with the user's character.

[0134] For example, when the target item is pants, the pants area of ​​the model will be used as a demonstration fixed area, and the rest of the model will be used as a demonstration replacement area, so as to facilitate the subsequent replacement of the model with the user character.

[0135] To determine the area corresponding to the target object, OpenCV can be used to extract the contour of the corresponding object, and the contour area can be used as a demonstration fixed area.

[0136] S32 obtains the placement attribute of the target object, and when it is determined that the placement attribute is a portable attribute, determines a customized deletion area of ​​the customized model in the customized gesture data based on the target object, and uses the remaining area in the customized model as a customized reserved area.

[0137] The placement attribute is the attribute of the target object for placement, such as a personal item, such as a coat, pants, etc. The customized model is a user character model other than the background in the customized posture data.

[0138] The custom deleted area is the area corresponding to the target object in the custom posture data. The custom reserved area is the remaining area in the custom model except the custom deleted area.

[0139] It is understandable that the server will obtain the placement attribute of the target item, and when it is determined that the placement attribute is a portable attribute, i.e., pants, etc., the server will determine the customized deletion area of ​​the customized model in the customized posture data based on the target item, and use the remaining area in the customized model as the customized reserved area. Because the user character will be updated later and the character will be shown wearing the advertiser's pants, only the upper body of the user needs to be updated, so the lower body of the pants worn by the user is used as the customized deletion area, and the upper body part that needs to be updated is reserved as the customized reserved area.

[0140] S33 determines the area center point of the demonstration replacement area as the replacement midpoint, and determines the center point of the customized reservation area as the reservation midpoint.

[0141] It can be understood that the server determines the center point of the demonstration replacement area as the replacement midpoint, and determines the center point of the customized reserved area as the reserved midpoint.

[0142] It is not difficult to understand that the center point of the upper body in the model image and the center point of the user's upper body clothing are determined in turn to facilitate subsequent alignment updates.

[0143] S34 deletes and updates the demonstration replacement area in the demonstration model, and deletes and updates the customized deletion area of ​​the customized model.

[0144] It is understandable that the present invention deletes and updates the demonstration replacement area in the demonstration model, for example, the area where the model's upper body is located, and deletes and updates the customized deletion area of ​​the customized model, for example, the area where the user's lower body is located.

[0145] S35 performs alignment processing on the retained midpoint and the replacement midpoint to obtain scenario adaptation data.

[0146] It can be understood that the retained midpoint is aligned with the replacement midpoint to obtain scene adaptation data, that is, the center points of the image corresponding to the user's retained upper body and the model's upper body are aligned to obtain scene adaptation data.

[0147] S36: When it is determined that the placement attribute is a non-portable attribute, the customized model in the customized posture data is used as a character reserved area.

[0148] It can be understood that when the placement attribute is determined to be a non-portable attribute, the customized model in the customized posture data is used as a character reserved area.

[0149] It is not difficult to understand that if the attribute of the item is determined to be a non-portable attribute, such as a shoulder bag, etc., and therefore the entire character needs to be directly updated, the customized model in the customized posture data is used as the character retention area.

[0150] S37 obtains the area center point of the demonstration replacement area as the replacement midpoint, and determines the center point of the character retention area as the retention midpoint.

[0151] It is understandable that the server will obtain the center point of the model replacement area as the replacement midpoint. If it is a shoulder bag, then it is the area corresponding to the entire model except the shoulder bag, and obtain the center point corresponding to the area as the replacement midpoint. And determine the center point of the character retention area as the retention midpoint.

[0152] S38 deletes and updates the demonstration replacement area in the demonstration model, aligns the retained midpoint with the replacement midpoint, and obtains scenario adaptation data.

[0153] It is understandable that the server will delete and update the demonstration replacement area in the demonstration model, that is, delete the area outside the shoulder bag in the model, and then update the entire user character, align the retained midpoint with the replacement midpoint to obtain scenario-adapted data.

[0154] Through the above-mentioned implementation, the present invention can perform different processing according to different items required by users, so that users can interact with advertisements, thereby fitting user needs and usage scenarios, so that characters can be updated to the scenes of corresponding commodities, which is convenient for display to users and improves user experience.

[0155] Based on the above embodiment, it also includes: A1-A3:

[0156] A1, obtaining an area without pixels in the scene adaptation data as an area to be filled, selecting pixels adjacent to the area to be filled as adjacent pixels, and determining pixel values ​​of the adjacent pixels as screening pixel values.

[0157] It should be noted that in the process of replacing the user character with the model character, due to individual differences, it is impossible to fill in perfectly, so there are some areas without pixels, that is, there is no corresponding filling image.

[0158] Therefore, the present invention will fill the area according to the pixel values ​​of the surrounding pixels.

[0159] The area to be filled is an area without pixels in the scene adaptation data.

[0160] It is understandable that the server will obtain an area without pixels in the scene adaptation data as an area to be filled, that is, a blank area, and then obtain pixels adjacent to the area to be filled as adjacent pixels, for example, pixels around the blank area, and use the pixel values ​​of the adjacent pixels as screening pixel values.

[0161] A2, based on the screening pixel value, counting the number of the corresponding adjacent pixel points as the screening number, and selecting the screening pixel value corresponding to the largest screening number as the filling pixel value.

[0162] It is understandable that the server will count the number of corresponding adjacent pixels based on the filtering pixel value as the filtering number, select the filtering pixel value corresponding to the largest filtering number as the filling pixel value, and then fill the blank area with the filling pixel value.

[0163] It is not difficult to understand that the present invention will select the pixel value with the largest proportion of pixel values ​​around the blank area as the filling pixel value, so that the final image is more consistent with the surrounding color, thereby making the image of the replaced user character more fitting.

[0164] A3, filling the to-be-filled area with the filling pixel value to obtain filled scene adaptation data.

[0165] S4, when it is determined that the posture comparison result is a difference comparison result, the free posture adaptation strategy is called to adapt the customized model in the customized posture data to obtain scenario adaptation data, and the scenario adaptation data is sent to the user end.

[0166] In some embodiments, step S4 (when judging that the posture comparison result is a difference comparison result, calling the free posture adaptation strategy to adapt the customized model in the customized posture data to obtain the scenario adaptation data) includes S41-S43:

[0167] S41, when it is determined that the posture comparison result is a difference comparison result, the facial orientation information of the customized model in the customized posture data is identified, and the orientation information of the customized model is determined based on the facial orientation information, wherein the orientation information includes front orientation, side orientation and back orientation.

[0168] It can be understood that when the posture comparison result is judged to be a difference comparison result, that is, the user's posture is quite different from the model's posture. At this time, it is impossible to directly update the character to the model picture. Therefore, the present invention will directly take the user character as the basis, and then match the corresponding orientation of the product according to the orientation of the photographed character. For example, if the character is facing forward, the front of the top will be matched to update the upper body of its image. If the character is facing backward, the back of the top will be matched to update the upper body of its image.

[0169] Therefore, the present invention identifies the facial orientation information of the customized model in the customized posture data, and determines the orientation information of the customized model based on the facial orientation information, wherein the orientation information includes front orientation, side orientation, and back orientation.

[0170] The facial orientation information is the orientation information of the user's face, such as the front face, the side face, and the back face.

[0171] The present invention determines the orientation information of the customized model according to different facial orientation information. For example, a frontal face indicates that the person is facing forward at this time.

[0172] S42, based on the orientation information, retrieve a preset object model with an object orientation corresponding to the target object.

[0173] The orientation of the placed object is the orientation corresponding to the object, such as the front and back of a coat. The preset object model is the model corresponding to the target object required by the user, such as a coat.

[0174] It is understandable that the server will retrieve the preset object model for placing the object in the orientation corresponding to the target object based on the orientation information, wherein different orientation information has corresponding preset object models for placing the object in different orientations.

[0175] It is not difficult to understand that the object model corresponding to the orientation of the character is determined according to the orientation of the character, and then the corresponding object, for example, a top, can be directly updated to the character.

[0176] S43, determining a key part corresponding to the target object as a placement part, setting the preset object model at the placement part, and obtaining scenario adaptation data.

[0177] It is understandable that the server will determine the key parts corresponding to the target object as the placement parts, for example, determine the body parts corresponding to the top, and use the upper body as the placement part, and then set the preset object model at the placement part to obtain scenario adaptation data.

[0178] Based on the above embodiment, it also includes B1-B9:

[0179] B1, obtaining pixel points of the customized model in the scenario adaptation data as model pixel points, determining the center point of the scenario adaptation data, and constructing a coordinate system with the center point to coordinate the scenario adaptation data.

[0180] The model pixels are pixels of the customized model.

[0181] It is understandable that the server obtains the pixel points of the customized model in the scenario adaptation data as model pixel points, and determines the center point of the scenario adaptation data, that is, the center point of the image displayed to the user, and constructs a coordinate system with the center point to coordinate the scenario adaptation data.

[0182] B2. Taking the coordinates corresponding to the model pixel point as the model pixel point coordinates, and obtaining the minimum vertical coordinate and the maximum vertical coordinate of the model pixel point coordinates.

[0183] It is understandable that the server will use the coordinates corresponding to the model pixel points as the model pixel point coordinates, and obtain the minimum vertical coordinate and the maximum vertical coordinate of the model pixel point coordinates, such as the coordinates of the top of the head and the coordinates of the feet of the character.

[0184] B3, determine the middle part of the customized model, obtain the coordinates of the pixel points corresponding to the middle part as the middle coordinates, and determine the ordinate of the middle coordinates as the middle ordinate.

[0185] The middle part is the middle position of the user-customized model, such as the waist.

[0186] It is understandable that the server will determine the middle part of the customized model, obtain the coordinates of the pixel points corresponding to the middle part as the middle coordinates, and determine the ordinate of the middle coordinates as the middle ordinate.

[0187] B4, obtaining an upper proportional distance based on a difference between the maximum ordinate and the middle ordinate, and obtaining a lower proportional distance based on a difference between the middle ordinate and the minimum ordinate.

[0188] It can be understood that the upper proportional distance, i.e., the distance of the upper body, is obtained based on the difference between the maximum ordinate and the middle ordinate, and the lower proportional distance, i.e., the distance of the lower body, is obtained based on the difference between the middle ordinate and the minimum ordinate.

[0189] B5, obtaining the minimum horizontal coordinate and the maximum horizontal coordinate of the intermediate coordinate, and obtaining the intermediate proportional distance based on the difference between the maximum horizontal coordinate and the minimum horizontal coordinate.

[0190] It can be understood that the minimum horizontal coordinate and the maximum horizontal coordinate of the intermediate coordinate are obtained, and the intermediate proportional distance, that is, the waist circumference distance, is obtained based on the difference between the maximum horizontal coordinate and the minimum horizontal coordinate.

[0191] B6, when it is determined that the placement attribute of the target object is a carry-on attribute, the corresponding target object is regarded as a carry-on item.

[0192] B7, calling a placement comparison table, wherein the placement comparison table has preset sizes corresponding to a preset lower ratio interval and a preset middle ratio interval, and preset sizes corresponding to a preset upper ratio interval and a preset middle ratio interval.

[0193] Among them, the placement comparison table is a size comparison table for placing corresponding target objects, wherein the placement comparison table has preset sizes corresponding to a preset lower proportion interval and a preset middle proportion interval, and preset sizes corresponding to a preset upper proportion interval and a preset middle proportion interval.

[0194] B8, when it is determined that the placement location of the carry-on item is the lower half, based on the lower proportional distance and the middle proportional distance, the corresponding preset size is compared with the placement comparison table to determine the selected size, and the preset item model is adjusted according to the selected size to obtain the adjusted item model.

[0195] It is not difficult to understand that when the placement location of the carry-on item is determined to be the lower half, for example, pants, the lower proportional distance and the middle proportional distance are compared with the placement comparison table to determine the corresponding preset size as the selected size. For example, the selected size may be XL or L. The preset item model is adjusted according to the selected size to obtain an adjusted item model. The server will subsequently adjust the size of the retrieved preset item model according to the corresponding size to obtain the adjusted item model.

[0196] B9, when it is determined that the placement location of the carry-on item is the upper half, the upper proportional distance and the middle proportional distance are compared with the placement comparison table, the corresponding preset size is determined as the selected size, and the preset item model is adjusted according to the selected size to obtain the adjusted item model.

[0197] It is not difficult to understand that when it is determined that the placement location of the carry-on item is the upper part, for example, a top, the upper proportional distance and the middle proportional distance are compared with the placement comparison table to determine the corresponding preset size as the selected size. For example, the selected size can be XL or L. The preset item model is adjusted according to the selected size to obtain the adjusted item model.

[0198] The present invention can adaptively adjust the size of clothes according to the length of the upper and lower clothes and the middle waist circumference so that the clothes are more fitted.

[0199] Based on the above embodiment, C1-C3 are also included:

[0200] C1. When it is determined that the placement attribute of the target object is a non-portable attribute, the corresponding target object is regarded as a non-portable object.

[0201] It is understandable that when the server determines that the placement attribute of the target item is a non-personal item, the server will treat the corresponding target item as a non-personal item, such as a shoulder bag.

[0202] C2, determining the actual full-scale distance according to the difference between the maximum ordinate and the minimum ordinate, and retrieving the preset full-scale distance corresponding to the preset object model.

[0203] The preset full-scale distance is a height distance preset manually, and the preset object model has a corresponding preset full-scale distance.

[0204] It can be understood that the difference between the maximum vertical coordinate and the minimum vertical coordinate determines the actual full-scale distance, that is, the height distance. Subsequently, the preset full-scale distance corresponding to the preset object model will be called up, for example, how big a shoulder bag corresponds to a preset height distance.

[0205] C3, obtaining a ratio adjustment factor based on the ratio of the actual full-scale distance to the preset full-scale distance, and scaling the preset object model according to the ratio adjustment factor to obtain an adjusted object model.

[0206] It can be understood that the proportional adjustment multiple is obtained based on the ratio of the actual full-scale distance and the preset full-scale distance. It is not difficult to understand that the taller the height, the larger the corresponding proportional adjustment multiple, and the shorter the height, the smaller the corresponding proportional adjustment multiple. Subsequently, the preset object model is scaled according to the proportional adjustment multiple to obtain the adjusted object model.

[0207] Based on the above embodiment, it also includes:

[0208] The active adjustment multiple sent by the user terminal based on the preset object model is received, and the preset object model is scaled according to the active adjustment multiple to obtain an adjusted object model.

[0209] Among them, the active adjustment multiple is the adjustment multiple actively set by the user end.

[0210] It is not difficult to understand that, based on all the above schemes, the server can receive the user's active adjustment, and therefore, can receive the active adjustment multiple sent by the user end based on the preset object model, and then scale the preset object model according to the active adjustment multiple to obtain the adjusted object model.

[0211] See also Figure 2 , is a schematic diagram of a structure of a context-adaptive intelligent recommendation system provided by an embodiment of the present invention, the context-adaptive intelligent recommendation system comprising:

[0212] A response module is used to respond to the advertising push request of the recommendation end, receive the interactive push data sent by the recommendation end, obtain the target item sent by the user end, and retrieve the corresponding interactive push data based on the target item to display it to the user end;

[0213] A comparison module, used for receiving the customized posture data uploaded by the user terminal based on the interactive push data, and obtaining a posture comparison result according to the comparison information between the customized posture data and the interactive push data;

[0214] A replacement module, for, when determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data;

[0215] The adaptation module is used to determine that the posture comparison result is a difference comparison result, call the free posture adaptation strategy to adapt the customized model in the customized posture data, obtain the scenario adaptation data, and send the scenario adaptation data to the user end.

[0216] See also Figure 3 , is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention, the electronic device 30 includes: a processor 31, a memory 32 and a computer program; wherein

[0217] The memory 32 is used to store the computer program, which may also be a flash memory. The computer program is, for example, an application program, a functional module, etc. for implementing the above method.

[0218] The processor 31 is used to execute the computer program stored in the memory to implement each step performed by the device in the above method. For details, please refer to the relevant description in the above method embodiment.

[0219] Optionally, the memory 32 may be independent or integrated with the processor 31 .

[0220] When the memory 32 is a device independent of the processor 31, the device may further include:

[0221] The bus 33 is used to connect the memory 32 and the processor 31 .

[0222] The present invention also provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the methods provided by the various embodiments described above.

[0223] Among them, the readable storage medium can be a computer storage medium or a communication medium. The communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The computer storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application-specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). In addition, the ASIC can be located in a user device. Of course, the processor and the readable storage medium can also exist in a communication device as discrete components. The readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0224] The present invention also provides a program product, which includes an execution instruction, which is stored in a readable storage medium. At least one processor of a device can read the execution instruction from the readable storage medium, and at least one processor executes the execution instruction so that the device implements the methods provided in the above various embodiments.

[0225] In the embodiments of the above-mentioned devices, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0226] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A context-adaptive intelligent recommendation method, characterized in that: include: Respond to the advertising push request of the recommending end, receive the interactive push data sent by the recommending end, obtain the target item sent by the user end, and retrieve the corresponding interactive push data based on the target item to display it to the user end; Receiving customized posture data uploaded by the user terminal based on the interactive push data, and obtaining a posture comparison result according to comparison information between the customized posture data and the interactive push data; When it is determined that the posture comparison result is a similar comparison result, the demonstration model in the interactive push data is replaced according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data; When it is determined that the posture comparison result is a difference comparison result, a free posture adaptation strategy is called to adapt the customized model in the customized posture data to obtain scenario adaptation data, and the scenario adaptation data is sent to the user end; The receiving end receives the customized posture data uploaded by the interactive push data, and obtains a posture comparison result according to the comparison information between the customized posture data and the interactive push data, including: Receiving the customized posture data uploaded by the user terminal based on the interactive push data, extracting key parts in the customized posture data as customized key parts based on OpenCV, and extracting contours corresponding to the key parts as customized part contours; Determine a center point of the customized gesture data, and perform coordinate processing on the customized gesture data based on the center point; Acquire the contour coordinates of the customized part contour, and determine the contour center coordinates of the customized key part based on the contour coordinates; Select any two customized key parts to combine, obtain a key part combination as a customized part combination, calculate according to the contour center coordinates corresponding to the customized key parts in the customized part combination, and obtain a customized combination distance corresponding to the customized part combination; Extracting key parts in the interactive push data as interactive key parts based on OpenCV, and contours corresponding to the key parts as interactive part contours; Determine a center point of the interactive push data, and perform coordinate processing on the interactive push data based on the center point; Acquire the contour coordinates of the contour of the interactive part, and determine the contour center coordinates of the interactive key part based on the contour coordinates; Select any two interactive key parts to combine, obtain a key part combination as an interactive part combination, and calculate according to the contour center coordinates corresponding to the interactive key parts in the interactive part combination to obtain an interactive combination distance corresponding to the interactive part combination; Obtaining a posture comparison result according to the customized combined distance and the interactive combined distance; The step of obtaining a posture comparison result according to the customized combined distance and the interactive combined distance includes: Obtaining a part gap value corresponding to each of the key part combinations according to the absolute value of the difference between the interactive combination distance and the corresponding customized combination distance; Counting the difference values ​​of the parts to obtain a posture difference value, and when it is determined that the posture difference value is greater than a preset difference value, generating a difference comparison result; When it is determined that the posture difference value is less than or equal to the preset difference value, generating a similarity comparison result; Obtaining a posture comparison result according to the difference comparison result and the similarity comparison result; When determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data includes: When it is determined that the posture comparison result is a similar comparison result, a demonstration fixed area of ​​the demonstration model in the interactive push data is determined according to the target object, and the remaining areas in the demonstration model are used as demonstration replacement areas; Acquiring the placement attribute of the target object, and when determining that the placement attribute is a portable attribute, determining a customized deletion area of ​​the customized model in the customized posture data based on the target object, and using the remaining area in the customized model as a customized reserved area; Determine the area center point of the demonstration replacement area as the replacement midpoint, and determine the center point of the customized reservation area as the reservation midpoint; Deleting and updating the demonstration replacement area in the demonstration model, and deleting and updating the customized deletion area of ​​the customized model; Aligning the retained midpoint with the replacement midpoint to obtain scenario-adapted data; When it is determined that the placement attribute is a non-portable attribute, the customized model in the customized posture data is used as a character reserved area; Acquire the center point of the demonstration replacement area as the replacement midpoint, and determine the center point of the character retention area as the retention midpoint; Deleting and updating the demonstration replacement area in the demonstration model, aligning the retained midpoint with the replacement midpoint, and obtaining scenario adaptation data; Acquire an area without pixels in the scene adaptation data as an area to be filled, select pixels adjacent to the area to be filled as adjacent pixels, and determine pixel values ​​of the adjacent pixels as screening pixel values; Based on the screening pixel values, the number of the adjacent pixel points corresponding to the screening pixel values ​​is counted as a screening number, and the screening pixel value corresponding to the largest screening number is selected as a filling pixel value; Filling the filling pixel value into the to-be-filled area to obtain the filled scene adaptation data; When it is determined that the posture comparison result is a difference comparison result, the free posture adaptation strategy is called to adapt the customized model in the customized posture data to obtain the scenario adaptation data, including: When it is determined that the posture comparison result is a difference comparison result, identifying the facial orientation information of the customized model in the customized posture data, and determining the orientation information of the customized model based on the facial orientation information, wherein the orientation information includes a front orientation, a side orientation, and a back orientation; Retrieving a preset object model for placing an object in an orientation corresponding to the target object based on the orientation information; A key part corresponding to the target object is determined as a placement part, and the preset object model is set at the placement part to obtain scenario adaptation data.

2. The method according to claim 1, characterized in that Also includes: Acquire pixel points of the customized model in the scenario adaptation data as model pixel points, determine the center point of the scenario adaptation data, and construct a coordinate system based on the center point to coordinate the scenario adaptation data; Taking the coordinates corresponding to the model pixel points as the model pixel point coordinates, obtaining the minimum vertical coordinate and the maximum vertical coordinate of the model pixel point coordinates; Determine the middle part of the customized model, obtain the coordinates of the pixel points corresponding to the middle part as the middle coordinates, and determine the ordinate of the middle coordinates as the middle ordinate; An upper scale distance is obtained based on a difference between the maximum ordinate and the middle ordinate, and a lower scale distance is obtained based on a difference between the middle ordinate and the minimum ordinate; Obtaining a minimum horizontal coordinate and a maximum horizontal coordinate of the intermediate coordinate, and obtaining an intermediate proportional distance based on a difference between the maximum horizontal coordinate and the minimum horizontal coordinate; When it is determined that the placement attribute of the target object is a carry-on attribute, the target object is regarded as a carry-on item; Retrieving a placement comparison table, wherein the placement comparison table has preset sizes corresponding to a preset lower ratio interval and a preset middle ratio interval, and preset sizes corresponding to a preset upper ratio interval and a preset middle ratio interval; When it is determined that the placement location of the carry-on item is the lower half, the lower proportional distance and the middle proportional distance are compared with the placement comparison table to determine the corresponding preset size as the selected size, and the preset item model is adjusted according to the selected size to obtain the adjusted item model; When it is determined that the placement location of the carry-on item is the upper half, the upper proportional distance and the middle proportional distance are compared with the placement comparison table, the corresponding preset size is determined as the selected size, and the preset item model is adjusted according to the selected size to obtain the adjusted item model.

3. The method according to claim 2, characterized in that Also includes: When it is determined that the placement attribute of the target item is a non-portable item, the target item is treated as a non-portable item; Determine the actual full-scale distance according to the difference between the maximum ordinate and the minimum ordinate, and retrieve the preset full-scale distance corresponding to the preset object model; Based on the ratio of the actual full-scale distance to the preset full-scale distance, a scale adjustment factor is obtained, and the preset object model is scaled according to the scale adjustment factor to obtain an adjusted object model.

4. The method according to claim 1, characterized in that: Also includes: The active adjustment multiple sent by the user terminal based on the preset object model is received, and the preset object model is scaled according to the active adjustment multiple to obtain an adjusted object model.

5. A context-adaptive intelligent recommendation system, characterized in that: include: A response module is used to respond to the advertising push request of the recommendation end, receive the interactive push data sent by the recommendation end, obtain the target item sent by the user end, and retrieve the corresponding interactive push data based on the target item to display it to the user end; A comparison module, used for receiving the customized posture data uploaded by the user terminal based on the interactive push data, and obtaining a posture comparison result according to the comparison information between the customized posture data and the interactive push data; A replacement module, for, when determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data; The adaptation module is used to call the free posture adaptation strategy to adapt the customized model in the customized posture data when judging that the posture comparison result is a difference comparison result, obtain the scenario adaptation data, and send the scenario adaptation data to the user end; The receiving end receives the customized posture data uploaded by the interactive push data, and obtains a posture comparison result according to the comparison information between the customized posture data and the interactive push data, including: Receiving the customized posture data uploaded by the user terminal based on the interactive push data, extracting key parts in the customized posture data as customized key parts based on OpenCV, and extracting contours corresponding to the key parts as customized part contours; Determine a center point of the customized gesture data, and perform coordinate processing on the customized gesture data based on the center point; Acquire the contour coordinates of the customized part contour, and determine the contour center coordinates of the customized key part based on the contour coordinates; Select any two customized key parts to combine, obtain a key part combination as a customized part combination, calculate according to the contour center coordinates corresponding to the customized key parts in the customized part combination, and obtain a customized combination distance corresponding to the customized part combination; Extracting key parts in the interactive push data as interactive key parts based on OpenCV, and contours corresponding to the key parts as interactive part contours; Determine a center point of the interactive push data, and perform coordinate processing on the interactive push data based on the center point; Acquire the contour coordinates of the contour of the interactive part, and determine the contour center coordinates of the interactive key part based on the contour coordinates; Select any two interactive key parts to combine, obtain a key part combination as an interactive part combination, and calculate according to the contour center coordinates corresponding to the interactive key parts in the interactive part combination to obtain an interactive combination distance corresponding to the interactive part combination; Obtaining a posture comparison result according to the customized combined distance and the interactive combined distance; The step of obtaining a posture comparison result according to the customized combined distance and the interactive combined distance includes: Obtaining a part gap value corresponding to each of the key part combinations according to the absolute value of the difference between the interactive combination distance and the corresponding customized combination distance; Counting the difference values ​​of the parts to obtain a posture difference value, and when it is determined that the posture difference value is greater than a preset difference value, generating a difference comparison result; When it is determined that the posture difference value is less than or equal to the preset difference value, generating a similarity comparison result; Obtaining a posture comparison result according to the difference comparison result and the similarity comparison result; When determining that the posture comparison result is a similar comparison result, replacing the demonstration model in the interactive push data according to the situational posture adaptation strategy and the customized posture data to obtain situational adaptation data includes: When it is determined that the posture comparison result is a similar comparison result, a demonstration fixed area of ​​the demonstration model in the interactive push data is determined according to the target object, and the remaining areas in the demonstration model are used as demonstration replacement areas; Acquiring the placement attribute of the target object, and when determining that the placement attribute is a portable attribute, determining a customized deletion area of ​​the customized model in the customized posture data based on the target object, and using the remaining area in the customized model as a customized reserved area; Determine the area center point of the demonstration replacement area as the replacement midpoint, and determine the center point of the customized reservation area as the reservation midpoint; Deleting and updating the demonstration replacement area in the demonstration model, and deleting and updating the customized deletion area of ​​the customized model; Aligning the retained midpoint with the replacement midpoint to obtain scenario-adapted data; When it is determined that the placement attribute is a non-portable attribute, the customized model in the customized posture data is used as a character reserved area; Acquire the center point of the demonstration replacement area as the replacement midpoint, and determine the center point of the character retention area as the retention midpoint; Deleting and updating the demonstration replacement area in the demonstration model, aligning the retained midpoint with the replacement midpoint, and obtaining scenario adaptation data; Acquire an area without pixels in the scene adaptation data as an area to be filled, select pixels adjacent to the area to be filled as adjacent pixels, and determine pixel values ​​of the adjacent pixels as screening pixel values; Based on the screening pixel values, the number of the adjacent pixel points corresponding to the screening pixel values ​​is counted as a screening number, and the screening pixel value corresponding to the largest screening number is selected as a filling pixel value; Filling the filling pixel value into the to-be-filled area to obtain the filled scene adaptation data; When it is determined that the posture comparison result is a difference comparison result, the free posture adaptation strategy is called to adapt the customized model in the customized posture data to obtain the scenario adaptation data, including: When it is determined that the posture comparison result is a difference comparison result, identifying the facial orientation information of the customized model in the customized posture data, and determining the orientation information of the customized model based on the facial orientation information, wherein the orientation information includes a front orientation, a side orientation, and a back orientation; Retrieving a preset object model for placing an object in an orientation corresponding to the target object based on the orientation information; A key part corresponding to the target object is determined as a placement part, and the preset object model is set at the placement part to obtain scenario adaptation data.

Images