Advertisement recommendation method, advertisement display method, device and apparatus

Abstract

The application discloses an advertisement recommendation method, a recommended advertisement display method, a device and equipment, and belongs to the technical field of computers. The method comprises the following steps: acquiring a video content clothing label and a candidate advertisement clothing label, wherein the video content clothing label is used for reflecting the type of a first clothing object in a video frame of a video, and the candidate advertisement clothing label is used for reflecting the type of a second clothing object in a candidate advertisement. According to at least one of a first prediction probability, a second prediction probability and a label value, a first similarity between the video content clothing label and the candidate advertisement clothing label is calculated. According to the first similarity, a recommended advertisement corresponding to the video is determined from the candidate advertisement. The first similarity reflects the similarity between the clothing content in the video content and the clothing content in the advertisement content. The recommended advertisement can be matched with the clothing content in the video watched by the user. The accuracy of the recommended advertisement is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computers, and particularly relates to an advertisement recommendation method, a recommended advertisement display method, an apparatus and a device. BACKGROUND

[0002] With the development of Internet technology, more and more merchants place advertisements through the Internet. For example, when a short video client plays a short video, it will also display advertisements recommended to the user.

[0003] In order to improve the effectiveness of advertisement placement, the server will recommend advertisements with a higher matching degree to the user according to the matching degree between the user and the advertisement. At present, the matching degree between the user feature and the advertisement feature is usually calculated to determine the click probability of the advertisement. Then, the expected revenue of the advertisement is determined according to the product of the single click revenue and the click probability of the advertisement. The server recommends the advertisement with the highest expected revenue to the user according to the expected revenue of the advertisement.

[0004] The recommended advertisement determined by the matching degree between the user feature and the advertisement feature and the single click revenue of the advertisement has no correlation with the behavior of the user, and the accuracy of the recommended advertisement is low. SUMMARY

[0005] The present application provides an advertisement recommendation method, a recommended advertisement display method, an apparatus and a device, which improves the matching degree between the clothing content in the advertisement and the clothing content in the video watched by the user when the advertisement is recommended to the user. The technical solution is as follows:

[0006] According to an aspect of the present application, an advertisement recommendation method is provided, which comprises:

[0007] obtaining a video content clothing label and a candidate advertisement clothing label, the video content clothing label being used to reflect the type of a first clothing object in a video frame of a video, and the candidate advertisement clothing label being used to reflect the type of a second clothing object in a candidate advertisement;

[0008] calculating a first similarity between the video content clothing label and the candidate advertisement clothing label according to at least one of a first prediction probability, a second prediction probability and a label value, the first similarity being used to reflect the similarity between the type of the first clothing object and the type of the second clothing object, the first prediction probability being used to reflect the probability that the first clothing object belongs to the video content clothing label, the second prediction probability being used to reflect the probability that the second clothing object belongs to the candidate advertisement clothing label, and the label value being used to reflect the value of the candidate advertisement clothing label;

[0009] determining a recommended advertisement corresponding to the video in the candidate advertisement according to the first similarity.

[0010] According to another aspect of the present application, a method for recommending advertisement display is provided, the method comprising:

[0011] sending an advertisement request to a server, the advertisement request carrying at least one of a video content costume label and a video label, the video content costume label being used to reflect a type of a first costume object in a video frame of a video;

[0012] receiving a recommended advertisement fed back by the server, the recommended advertisement being determined by the server according to a first similarity between the video content costume label and a candidate advertisement costume label, the first similarity being calculated by the server according to at least one of a first prediction probability, a second prediction probability and a label value, the first prediction probability being used to reflect a probability that the first costume object belongs to the video content costume label, the second prediction probability being used to reflect a probability that a second costume object belongs to the candidate advertisement costume label, the label value being used to reflect a value of the candidate advertisement costume label, the candidate advertisement costume label being used to reflect a type of the second costume object in a candidate advertisement, the first similarity being used to reflect a similarity degree between the type of the first costume object and the type of the second costume object;

[0013] displaying a play interface of the video, the play interface displaying the recommended advertisement.

[0014] According to another aspect of the present application, an apparatus for recommending advertisement is provided, the apparatus comprising:

[0015] an obtaining module, configured to obtain a video content costume label and a candidate advertisement costume label, the video content costume label being used to reflect a type of a first costume object in a video frame of a video, the candidate advertisement costume label being used to reflect a type of a second costume object in a candidate advertisement;

[0016] a calculating module, configured to calculate a first similarity between the video content costume label and the candidate advertisement costume label according to at least one of a first prediction probability, a second prediction probability and a label value, the first similarity being used to reflect a similarity degree between the type of the first costume object and the type of the second costume object, the first prediction probability being used to reflect a probability that the first costume object belongs to the video content costume label, the second prediction probability being used to reflect a probability that the second costume object belongs to the candidate advertisement costume label, the label value being used to reflect a value of the candidate advertisement costume label;

[0017] a first determining module, configured to determine a recommended advertisement corresponding to the video from the candidate advertisement according to the first similarity.

[0018] Optionally, the computing module is configured to:

[0019] calculate a second similarity between the video content clothing label and the candidate advertisement clothing label by using a text embedding technique;

[0020] calculate the first similarity according to a product of the second similarity, a first weight and a second weight, the first weight being determined according to the first predicted probability, and the second weight being determined according to the second predicted probability and the label value.

[0021] Optionally, the computing module is configured to:

[0022] calculate a second similarity between the video content clothing label and the candidate advertisement clothing label by using a text embedding technique;

[0023] calculate the first similarity according to a product of the second similarity, a first weight and a third weight, the first weight being determined according to the first predicted probability, and the third weight being determined according to the second predicted probability.

[0024] Optionally, the computing module is configured to:

[0025] calculate a second similarity between the video content clothing label and the candidate advertisement clothing label by using a text embedding technique;

[0026] calculate the first similarity according to a product of the second similarity and a fourth weight, the fourth weight being determined according to the label value.

[0027] Optionally, the computing module is configured to:

[0028] map the video content clothing label to a first vector to obtain a first target vector, and map the advertisement clothing label to a second vector to obtain a second target vector;

[0029] calculate a cosine similarity between the first target vector and the second target vector to obtain the second similarity.

[0030] Optionally, the obtaining module is configured to:

[0031] obtain p video frames of the video, p being a positive integer;

[0032] extract a first clothing region from the p video frames by using an image segmentation model, the first clothing region including the first clothing object;

[0033] predict a type of the first clothing object by using an image classification model to obtain the video content clothing label;

[0034] obtain the video content clothing label.

[0035] Optionally, the obtaining module is configured to:

[0036] predict, by the image classification model, a type of the first clothing object in the first clothing region, to obtain n candidate video content clothing labels, n being a positive integer;

[0037] in response to a repeated video label existing in the n candidate video content clothing labels, merge the repeated video label into a de-duplicated video label, to obtain the video content clothing label.

[0038] Optionally, the apparatus further comprises:

[0039] the second determining module is configured to determine an average of all video clothing prediction probabilities corresponding to the repeated video label as the video clothing prediction probability corresponding to the de-duplicated video label.

[0040] Optionally, the obtaining module is configured to:

[0041] obtain q pictures of the candidate advertisement, q being a positive integer;

[0042] obtain a second clothing region from the q pictures by an image segmentation model, the second clothing region including the second clothing object;

[0043] predict, by the image classification model, a type of the second clothing object, to obtain the candidate advertisement clothing label;

[0044] obtain the candidate advertisement clothing label.

[0045] Optionally, the obtaining module is configured to:

[0046] predict, by the image classification model, a type of the second clothing object in the second clothing region, to obtain m pending candidate advertisement clothing labels, m being a positive integer;

[0047] in response to a repeated advertisement label existing in the m pending candidate advertisement clothing labels, merge the repeated advertisement label into a de-duplicated advertisement label, to obtain the candidate advertisement clothing label.

[0048] Optionally, the second determining module is further configured to:

[0049] determine an average of all advertisement clothing prediction probabilities corresponding to the repeated advertisement label as the advertisement clothing prediction probability corresponding to the de-duplicated advertisement label.

[0050] Optionally, the obtaining module is configured to:

[0051] mapping the video content costume label to a mapping table to obtain an industry feature corresponding to the video content costume label;

[0052] determining the candidate advertisement with the same industry feature as the candidate advertisement set;

[0053] obtaining the candidate advertisement costume label according to the candidate advertisement in the candidate advertisement set.

[0054] Optionally, the first determining module is configured to:

[0055] calculate a recommendation score of the candidate advertisement according to the first similarity and a third similarity, the third similarity reflecting a similarity degree between a first feature of a user account receiving the recommended advertisement and a second feature of the candidate advertisement;

[0056] determine the candidate advertisement with the highest recommendation score as the recommended advertisement.

[0057] According to another aspect of the present application, a recommended advertisement display device is provided, and the device comprises:

[0058] a sending module configured to send an advertisement request to a server, the advertisement request carrying at least one of a video content costume label and a video identifier, the video content costume label reflecting a type of a first costume object in a video frame of a video;

[0059] a receiving module configured to receive a recommended advertisement fed back by the server, the recommended advertisement being determined by the server according to a first similarity between the video content costume label and a candidate advertisement costume label, the first similarity being calculated by the server according to at least one of a first prediction probability, a second prediction probability and a label value, the first prediction probability reflecting a probability that the first costume object belongs to the video content costume label, the second prediction probability reflecting a probability that a second costume object belongs to the candidate advertisement costume label, the label value reflecting a value of the candidate advertisement costume label, the candidate advertisement costume label reflecting a type of the second costume object in a candidate advertisement, and the first similarity reflecting a similarity degree between the type of the first costume object and the type of the second costume object;

[0060] a display module configured to display a playing interface of the video, the playing interface displaying the recommended advertisement.

[0061] According to another aspect of the present application, a computer device is provided, which comprises a processor and a memory, the memory storing at least one instruction, at least one program, a code set or an instruction set, which are loaded and executed by the processor to implement the advertisement recommendation method or the advertisement display method according to the above aspects.

[0062] According to another aspect of the present application, a computer storage medium is provided, which stores at least one instruction, at least one program, a code set or an instruction set, which, when loaded and executed by a processor of a computer device, implement the advertisement recommendation method or the advertisement display method according to the above aspects.

[0063] According to another aspect of the present application, a computer program product is provided, which comprises computer instructions stored in a computer readable storage medium. The processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device perform the advertisement recommendation method or the advertisement display method provided in various optional implementation manners of the above aspects.

[0064] The technical solutions provided by the present application have at least the following beneficial effects:

[0065] The first similarity between the video content clothing label and the candidate advertisement clothing label is calculated, and the recommended advertisement corresponding to the video is determined according to the first similarity. The first similarity reflects the similarity between the types of the first clothing object and the second clothing object, that is, the similarity between the clothing content in the video content and the clothing content in the advertisement content. The recommended advertisement can be matched with the clothing content in the video watched by the user, and the possibility that the user is interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved. BRIEF DESCRIPTION OF DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0067] Figure 1 is a flowchart of recommending an advertisement to a user according to an embodiment of the present application;

[0068] Figure 2 is a structural schematic diagram of an advertisement recommendation system according to an embodiment of the present application;

[0069] Figure 3 is a flowchart of a method for recommending an advertisement according to an embodiment of the present application;

[0070] Figure 4 is a flowchart of a method for displaying a recommended advertisement according to an embodiment of the present application;

[0071] Figure 5 is a flowchart of another method for recommending an advertisement according to an embodiment of the present application;

[0072] Figure 6 is a flowchart of a method for detecting a video content costume tag of a video according to an embodiment of the present application;

[0073] Figure 7 is a flowchart of a method for predicting a type of a first costume object according to an embodiment of the present application;

[0074] Figure 8 is a flowchart of a method for determining a candidate advertisement costume tag of a candidate advertisement according to an embodiment of the present application;

[0075] Figure 9 is a flowchart of a method for predicting a type of a second costume object according to an embodiment of the present application;

[0076] Figure 10 is a flowchart of a method for obtaining a candidate advertisement costume tag according to an embodiment of the present application;

[0077] Figure 11 is a flowchart of a method for calculating a first similarity according to an embodiment of the present application;

[0078] Figure 12 is a flowchart of another method for calculating a first similarity according to an embodiment of the present application;

[0079] Figure 13 is a flowchart of yet another method for calculating a first similarity according to an embodiment of the present application;

[0080] Figure 14 is a structural diagram of an advertisement recommendation device according to an embodiment of the present application;

[0081] Figure 15 is a structural diagram of another advertisement recommendation device according to an embodiment of the present application;

[0082] Figure 16 is a structural diagram of a device for displaying a recommended advertisement according to an embodiment of the present application;

[0083] Figure 17 is a structural diagram of a server according to an embodiment of the present application;

[0084] Figure 18 is a structural schematic diagram of a terminal provided by an embodiment of the present application.

[0085] The accompanying drawings, which are incorporated into and form a part of the specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the application. DETAILED DESCRIPTION

[0086] For the purpose, technical solutions and advantages of the present application to be clearer, the embodiments of the present application will be described in further detail below with reference to the drawings.

[0087] Figure 1 is a flowchart of a process of recommending an advertisement to a user provided by an embodiment of the present application. As shown in Figure 1 Step S1, a server extracts a video content clothing label of a video. The video content clothing label is used to reflect the type of a first clothing object in a video frame of the video. Optionally, when the server is a short video server, the video refers to any short video in the short video server. The first clothing object includes clothing such as a T-shirt, trousers, a shirt and a coat appearing in the video frame of the video. For example, when the first clothing object includes a T-shirt and trousers, the video content clothing label includes t-shirt and trousers.

[0088] In step S2, the server extracts a candidate advertisement clothing label of a candidate advertisement. The candidate advertisement clothing label is used to reflect the type of a second clothing object in the candidate advertisement. Optionally, the candidate advertisement refers to any picture advertisement in the server, and the candidate advertisement includes at least one picture. When the server's corresponding client displays the candidate advertisement in a user interface, the pictures of the candidate advertisement are displayed.

[0089] In step S3, the server retrieves candidate advertisements of the same type based on the video content clothing label. Optionally, the server maps the video content clothing label to a mapping table to obtain an industry feature corresponding to the video content clothing label. Then, the candidate advertisements with the same industry feature are determined as a candidate advertisement set. For example, when the video content clothing label includes t-shirt, the industry feature corresponding to the video content clothing label is upper garment. At this time, the candidate advertisements in the candidate advertisement set determined are advertisements of upper garments.

[0090] In step S4, the server obtains the video content clothing label and the candidate advertisement clothing label. Optionally, the server obtains a video identifier of a video being played in the client, and obtains the video content clothing label of the video according to the video identifier. The server obtains the candidate advertisement clothing label of the candidate advertisement from the candidate advertisement set corresponding to the video content clothing label.

[0091] In step S5, the server calculates a first similarity between the clothing tag in the video content and the clothing tags in the candidate advertisements. This first similarity reflects the degree of similarity between the types of the first clothing object and the types of the second clothing object. Optionally, the server calculates a second similarity between the clothing tags in the video content and the clothing tags in the candidate advertisements using text embedding technology. The first similarity is then calculated based on the product of the second similarity, a first weight, and a third weight. The first weight is determined based on the probability that the first clothing object belongs to the clothing tag in the video content, and the third weight is determined based on the probability that the second clothing object belongs to the clothing tag in the candidate advertisements.

[0092] In step S6, the server determines the recommended advertisement based on a first similarity and a third similarity. The third similarity reflects the degree of similarity between a first feature of the user account receiving the recommended advertisement and a second feature of the candidate advertisement. The first feature includes the user's age, education level, and interests, while the second feature includes the industry of the candidate advertisement, the category of the product in the candidate advertisement, and the copywriting in the candidate advertisement. Optionally, the recommendation score of the candidate advertisement is obtained by summing the product of the first similarity and its weight, and the third similarity. The candidate advertisement with the highest recommendation score is determined as the recommended advertisement. Optionally, the weight of the first similarity is determined by the server based on the importance of the relevance between the video content and the candidate advertisement content.

[0093] By calculating the first similarity between clothing tags in video content and clothing tags in candidate ads, and then determining recommended ads based on the first and second similarity scores, the system aims to match recommended ads with the clothing content in the videos users watch. The first similarity reflects the degree of similarity between the types of clothing items in the first category and the types of clothing items in the second category, i.e., the similarity between the clothing content in the video content and the clothing content in the ad content. This approach ensures that recommended ads match the clothing content in the videos users watch, increasing the likelihood of users being interested in the recommended ads and improving the accuracy of ad recommendations.

[0094] Figure 2 This is a schematic diagram of the structure of an advertising recommendation system provided in an illustrative embodiment of this application, as shown below. Figure 2 As shown, the system includes: server 210 and terminal 220.

[0095] Optionally, server 210 may be a single server, a server cluster consisting of several servers, or a virtual server in a cloud computing service center, etc., without limitation. Terminal 220 is a terminal device with a display screen, such as a smartphone, tablet, desktop computer, or laptop. Server 210 and terminal 220 establish a connection via a wired or wireless network. Figure 2 The number of terminals that establish connections with server 210 is for illustrative purposes only and is not intended to limit the advertising recommendation system provided in this application embodiment.

[0096] It should be noted that the terminal 220 is installed with a client, and the terminal 220 connects with the server 210 through the client, and the server 210 is a server corresponding to the client. Alternatively, when the server 210 is a short video server, the client installed on the terminal 220 is a short video client.

[0097] Figure 3 is a flow diagram of an advertisement recommendation method provided by an embodiment of the present application. The method can be used for a server in the system as shown in Figure 2 , and the method comprises the following steps. Figure 3

[0098] Step 301, obtaining a video content costume label and a candidate advertisement costume label.

[0099] The video content costume label is used to reflect the type of a first costume object in a video frame of a video. Alternatively, the video refers to any video in the server. For example, any short video in a short video server. The first costume object includes costumes such as T-shirts, trousers, shirts, and coats appearing in the video frame of the video. The video content costume label can finely classify the first costume object, such as striped T-shirts, camouflage trousers, and hooded sportswear, etc. The video content costume label corresponds to a video identifier of the video, and the video identifier is used to indicate the video corresponding to the video content costume label. The video identifier includes the name of the video, the number of the video in the server, and other information capable of uniquely identifying the video.

[0100] The candidate advertisement costume label is used to reflect the type of a second costume object in a picture of a candidate advertisement. Alternatively, the candidate advertisement refers to any advertisement in the server, and the candidate advertisement includes at least one picture. The candidate advertisement costume label corresponds to an advertisement identifier of the candidate advertisement, and the advertisement identifier is used to indicate the candidate advertisement corresponding to the candidate advertisement costume label. The advertisement identifier includes the name of the candidate advertisement, the number of the candidate advertisement in the server, and other information capable of uniquely identifying the candidate advertisement.

[0101] Alternatively, when the server receives a video playing request or an advertisement request sent by the client corresponding to the server, the video identifier in the request is obtained, and the video content costume label corresponding to the video is obtained according to the video identifier, and then the candidate advertisement costume label is obtained. Alternatively, the candidate advertisement costume label obtained by the server refers to a candidate advertisement costume label corresponding to the same industry feature as the video content costume label of the video. For example, the industry feature includes upper garment, shoes, and trousers, etc.

[0102] ​Optionally, the server detects a first clothing region including the first clothing object in a video frame of the video, and identifies a type of the first clothing object in the first clothing region, to obtain a video content clothing label of the video. The server detects a second clothing region including a second clothing object in a picture of the detected candidate advertisement, and identifies a type of the second clothing object in the second clothing region, to obtain a candidate advertisement clothing label of the candidate advertisement.

[0103] Step 302, calculating a first similarity between the video content clothing label and the candidate advertisement clothing label according to at least one of the first prediction probability, the second prediction probability, and the label value.

[0104] The first similarity is used to reflect a similarity degree of the type of the first clothing object and the type of the second clothing object. The higher the first similarity, the higher the similarity degree of the type of the first clothing object and the type of the second clothing object, that is, the higher the matching degree of the first clothing object and the second clothing object. The first prediction probability is used to reflect a probability that the first clothing object belongs to the video content clothing label. The second prediction probability is used to reflect a probability that the second clothing object belongs to the candidate advertisement clothing label. The label value is used to reflect a value of the candidate advertisement clothing label.

[0105] Optionally, the server calculates a similarity between texts of the video content clothing label and the candidate advertisement clothing label in a manner of word embedding, to obtain the first similarity. The server calculates the first similarity between the video content clothing label and the candidate advertisement clothing label within a set time after obtaining the video content clothing label and the candidate advertisement clothing label.

[0106] Step 303, determining a recommended advertisement corresponding to the video from the candidate advertisements according to the first similarity.

[0107] Optionally, the server determines the candidate advertisement with the highest first similarity to the video as the recommended advertisement corresponding to the video. That is, the recommended advertisement corresponding to the video is determined according to a similarity degree between clothing content in the video content and clothing content in the advertisement content. Alternatively, the server determines the recommended advertisement according to the first similarity and a third similarity. The third similarity is used to reflect a similarity degree between a first feature of a user account receiving the recommended advertisement and a second feature of the advertisement. The first feature includes an age, an education, and an interest of a user logging in the user account, and the second feature includes an industry to which the candidate advertisement belongs, a category of a commodity in the candidate advertisement, and a script in the candidate advertisement. That is, the recommended advertisement is determined according to the similarity degree between the clothing content in the video content and the clothing content in the advertisement content, and a matching degree between a feature of the video and a feature of the user.

[0108] Optionally, the server sends the recommended advertisement to the client after determining the recommended advertisement. The client is configured to superimpose the recommended advertisement in a user interface in which the video is played.

[0109] To sum up, the advertisement recommendation method provided by the embodiments of the present application calculates the first similarity between the video content costume label and the candidate advertisement costume label, and determines the recommended advertisement corresponding to the video according to the first similarity. The first similarity reflects the similarity between the types of the first costume object and the second costume object, i.e., the similarity between the costume content in the video content and the costume content in the advertisement content. The recommended advertisement can be matched with the costume content in the video watched by the user, and the possibility that the user is interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved.

[0110] Figure 4 is a flowchart of a method for displaying a recommended advertisement provided by the embodiments of the present application. The method can be used in a terminal or a client on a terminal as shown in Figure 2 . As shown in Figure 4 , the method comprises the following steps.

[0111] Step 401: sending an advertisement request to a server, the advertisement request carrying at least one of a video content costume label and a video identifier.

[0112] The video content costume label is used to reflect the type of the first costume object in the video frame of the video. The client sends an advertisement request to the server, and the server obtains the video content costume label of the video from the database according to the video identifier sent by the client. Alternatively, the server directly obtains the video content costume label according to the advertisement request.

[0113] Optionally, the client sends the advertisement request to the server when receiving a request to play a video. The advertisement request carries the video content costume label of the video and / or the video identifier of the video.

[0114] Step 402: receiving a recommended advertisement fed back by the server, the recommended advertisement being determined by the server according to the first similarity between the video content costume label and a candidate advertisement costume label.

[0115] The candidate advertisement costume label is used to reflect the type of the second costume object in the candidate advertisement. The first similarity is used to reflect the similarity between the type of the first costume object and the type of the second costume object. That is, the recommended advertisement is determined by the server according to the similarity between the costume content in the video content and the costume content in the advertisement content.

[0116] The first similarity is calculated by the server according to at least one of the first prediction probability, the second prediction probability, and the label value. The first prediction probability is used to reflect a probability that the first clothing object belongs to the video content clothing label. The second prediction probability is used to reflect a probability that the second clothing object belongs to the candidate advertisement clothing label. The label value is used to reflect a value of the candidate advertisement clothing label.

[0117] In step 403, a playing interface of the video is displayed, and the recommended advertisement is displayed on the playing interface.

[0118] Optionally, the client displays the recommended advertisement immediately after receiving the recommended advertisement. Alternatively, the client displays the recommended advertisement at a set time. The client superimposes a picture of the recommended advertisement on the playing interface of the video.

[0119] Optionally, when the client detects a click operation on the displayed picture of the recommended advertisement, the playing interface of the video is switched to an advertisement detail interface corresponding to the recommended advertisement.

[0120] To sum up, the recommended advertisement display method provided in the embodiments of the present application sends an advertisement request to a server, receives a recommended advertisement fed back by the server, and displays the recommended advertisement. The recommended advertisement is determined according to a similarity between clothing content in video content and clothing content in advertisement content. The recommended advertisement can be matched with clothing content in a video watched by a user, and the possibility that the user is interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved.

[0121] Figure 5 is a flowchart of another advertisement recommendation method provided in the embodiments of the present application. The method can be used in a system as shown in Figure 2 . As shown in Figure 5 , the method comprises the following steps.

[0122] In step 501, the server determines a video content clothing label of a video and a candidate advertisement clothing label of a candidate advertisement.

[0123] Optionally, the video is any video in the server. The candidate advertisement is any advertisement including a picture in the server. The server determines the video content clothing label of the video and the candidate advertisement clothing label of the candidate advertisement at a set time. Alternatively, the server determines the candidate advertisement clothing label of the candidate advertisement at a set time, and determines a video content clothing label of a video corresponding to an advertisement request when the server receives the advertisement request.

[0124] Optionally, as shown in Figure 6 , the implementation process of the server determining the video content clothing label of the video comprises the following steps 5011a to 5011c.

[0125] In step 5011a, p video frames of the video are acquired.

[0126] wherein p is a positive integer. Optionally, the video frames acquired by the server are any video frames in the video. Alternatively, the server acquires key frames of the video according to key frame information of the video.

[0127] For example, the key frame information of the video includes timestamps of 10 key frames, and the server acquires 10 video frames of the video according to the key frame information.

[0128] In step 5011b, a first clothing region including a first clothing object is cropped from the p video frames by an image segmentation model.

[0129] Optionally, the image segmentation model is trained by a target detection algorithm (Single Shot MultiBox Detector, SSD). For example, the image segmentation model is trained by SSD300. The server detects one or more clothing regions possibly including the first clothing object in the p video frames as the first clothing region by the image segmentation model, and crops the first clothing region from the p video frames.

[0130] In step 5011c, the type of the first clothing object is predicted by an image classification model to obtain a video content clothing label.

[0131] Optionally, the image classification model is a convolutional neural network (Convolutional Neural Networks, CNN) model, for example, a ResNet18 model. The server predicts the type of the first clothing object, and can also obtain a first prediction probability corresponding to the video content clothing label. The first prediction probability is used to reflect the probability that the first clothing object belongs to the video content clothing label.

[0132] Optionally, as shown in FIG. 5B, the implementation process of step 5011c includes steps 5011c1 to 5011c2: Figure 7

[0133] In step 5011c1, the type of the first clothing object in the first clothing region is predicted by the image classification model to obtain n candidate video content clothing labels.

[0134] wherein n is a positive integer. Optionally, the first clothing objects included in each of the first clothing regions are the same or different. For example, one of the first clothing regions includes a T-shirt in the first clothing objects, and another of the first clothing regions also includes the T-shirt in the first clothing objects. At this time, at least two candidate video content clothing labels corresponding to the T-shirt are included in the n candidate video content clothing labels.​

[0135] Optionally, the server is capable of obtaining a video clothing prediction probability corresponding to each candidate video content clothing label when predicting the type of the first clothing object. The video clothing prediction probability is used to reflect the probability that the first clothing object corresponding to the candidate video content clothing label belongs to the candidate video content clothing label.

[0136] In step 5011c2, in response to the existence of repeated video labels in the n candidate video content clothing labels, the repeated video labels are merged into de-duplicated video labels to obtain the video content clothing label.

[0137] The server merging the repeated video labels into the de-duplicated video labels means that the server only retains one video label in each group of repeated video labels in the n candidate video content clothing labels, so as to obtain the video content clothing label according to the de-duplicated video labels.

[0138] For example, the candidate video content clothing labels are t-shirt, trousers, sweater, t-shirt, and t-shirt. The server merges the repeated video labels into the de-duplicated video labels to obtain the video content clothing label {t-shirt, trousers, sweater}.

[0139] Optionally, when there are repeated video labels in the n candidate video content clothing labels, the server is also capable of determining the average value of the video clothing prediction probabilities corresponding to all the repeated video labels as the video clothing prediction probability corresponding to the de-duplicated video label.

[0140] Each candidate video content clothing label corresponds to a video clothing prediction probability. For example, the candidate video content clothing labels are t-shirt: 80, trousers: 75, sweater: 92, t-shirt: 78, and t-shirt: 79. The server merges the repeated video labels into the de-duplicated video labels to obtain the video content clothing label {t-shirt: 79, trousers: 75, sweater: 92}. Among them, t-shirt: 79 means that the video clothing prediction probability corresponding to the video label is 79%.

[0141] Optionally, as shown in Figure 8 The server determines the implementation process of the candidate advertisement clothing label of the candidate advertisement includes the following steps 5012a to 5012c.

[0142] In step 5012a, q pictures of the candidate advertisement are obtained.

[0143] Where q is a positive integer. Optionally, the image obtained by the server can be any image from the candidate advertisements.

[0144] In step 5012b, a second clothing region is extracted from q images using an image segmentation model. The second clothing region includes a second clothing object.

[0145] Optionally, the image segmentation model is the same as that in step 5011b. The server uses the image segmentation model to detect one or more clothing regions in the q images that may contain the second clothing object as the second clothing region, and then extracts the second clothing region from the q images.

[0146] In step 5012c, the type of the second clothing object is predicted by an image classification model to obtain candidate advertising clothing labels.

[0147] Optionally, the image classification model is the same as that in step 5011c. The server predicts the type of the second clothing object and can also obtain a second predicted probability corresponding to the candidate advertising clothing label. The second predicted probability reflects the probability that the second clothing object belongs to the candidate advertising clothing label.

[0148] Optionally, such as Figure 9 As shown, the implementation process of step 5012c includes the following steps 5012c1 to 5012c2:

[0149] In step 5012c1, the type of the second clothing object in the second clothing region is predicted by the image classification model to obtain m candidate advertising clothing labels to be determined.

[0150] Where m is a positive integer. Optionally, each of the second clothing areas may contain the same or different second clothing objects.

[0151] Optionally, when predicting the type of the second clothing object, the server can also obtain the predicted probability of the advertising clothing corresponding to each pending candidate advertising clothing tag. This predicted probability of the advertising clothing reflects the probability that the second clothing object corresponding to the pending candidate advertising clothing tag belongs to that pending candidate advertising clothing tag.

[0152] In step 5012c2, in response to the presence of duplicate advertising tags among the m undetermined candidate advertising clothing tags, the duplicate advertising tags are merged into deduplicated advertising tags to obtain candidate advertising clothing tags.

[0153] The server merges duplicate ad tags into deduplicated ad tags. This means that the server retains only one ad tag from each group of duplicate ad tags among the m undetermined candidate ad clothing tags, and thus obtains candidate ad clothing tags based on the deduplicated ad tags.

[0154] Optionally, when there are duplicate video tags in the m pending candidate advertisement clothing tags, the server can also determine the average of the advertisement clothing prediction probabilities corresponding to all duplicate advertisement tags as the advertisement clothing prediction probability corresponding to the deduplicated advertisement tag.

[0155] Each of the m pending candidate advertisement clothing tags corresponds to a video clothing prediction probability.

[0156] Step 502, the client sends an advertisement request to the server, and the advertisement request carries at least one of the video content clothing tags and the video identifier.

[0157] Optionally, the client sends the advertisement request to the server when receiving a video playing request. Alternatively, the client sends the advertisement request to the server in response to the video played in the client being played for a set duration.

[0158] Step 503, the server acquires the video content clothing tags and the candidate advertisement clothing tags.

[0159] Optionally, the server acquires the video content clothing tags of the video in the database according to the video identifier sent by the client. Alternatively, the server acquires the video content clothing tags in the advertisement request of the client. Alternatively, the server detects the video content clothing tags of the video corresponding to the video identifier according to the video identifier sent by the client, and acquires the video content clothing tags. The server acquires the video content clothing tags and the candidate advertisement clothing tags at the same time, or acquires the video content clothing tags and the candidate advertisement clothing tags at different times.

[0160] Optionally, as shown in FIG. 5B, the implementation process of the server acquiring the candidate advertisement clothing tags includes the following steps 5031a to 5031c: Figure 10

[0161] In step 5031a, the video content clothing tags are mapped to a mapping table to obtain the industry features corresponding to the video content clothing tags.

[0162] The video content clothing tags are the video content clothing tags acquired by the server. The mapping table is used to reflect the corresponding relationship between the video content clothing tags and the industry features. For example, when the video content clothing tags include t-shirt, dress and hat, the industry features corresponding to the video content clothing tags include the clothing industry and the accessory industry. Among them, t-shirt and dress correspond to the clothing industry, and hat corresponds to the accessory industry.

[0163] In step 5031b, the candidate advertisements with the same industry features are determined as a candidate advertisement set.

[0164] ​Optionally, the candidate advertisement corresponds to an industry characteristic. The server filters the candidate advertisements in the server according to the industry characteristic corresponding to the acquired video content costume label. The candidate advertisements with the same industry characteristic as the video content costume label are determined as the candidate advertisement set.

[0165] In step 5031c, the candidate advertisement costume label is acquired according to the candidate advertisement in the candidate advertisement set.

[0166] Optionally, the server acquires the candidate advertisement costume label of all candidate advertisements in the candidate advertisement set. Or the server acquires the candidate advertisement costume label of candidate advertisements in the candidate advertisement set according to a set number at random.

[0167] Step 504, the server calculates the first similarity between the video content costume label and the candidate advertisement costume label according to at least one of the first prediction probability, the second prediction probability and the label value.

[0168] In a possible implementation, the video content costume label corresponds to a first prediction probability, the candidate advertisement costume label corresponds to a second prediction probability and a label value. The first prediction probability is used to reflect the probability that the first costume object belongs to the video content costume label. The first prediction probability is obtained when the server predicts the type of the first costume object. The second prediction probability is used to reflect the probability that the second costume object belongs to the candidate advertisement costume label. The second prediction probability is obtained when the server predicts the type of the second costume object. The label value is used to reflect the value of the candidate advertisement costume label. For example, the server determines the value of the candidate advertisement costume label according to the average selling price of the costume type corresponding to the candidate advertisement costume label.

[0169] As shown in Figure 11 The implementation process of step 504 includes the following steps 5041a and 5041b:

[0170] In step 5041a, the second similarity between the video content costume label and the candidate advertisement costume label is calculated by the text embedding technology.

[0171] Optionally, the server maps the video content costume label to a first vector by the text embedding technology to obtain a first target vector. The server maps the advertisement costume label to a second vector by the text embedding technology to obtain a second target vector. Then the server calculates the cosine similarity between the first target vector and the second target vector to obtain the second similarity.

[0172] In step 5041b, the first similarity is calculated according to the product of the second similarity, the first weight and the second weight, the first weight is determined according to the first prediction probability, and the second weight is determined according to the second prediction probability and the label value.

[0173] Optionally, the first weight is positively correlated with the first predicted probability. The second weight is positively correlated with the product of the second predicted probability and the label value. For example, the server determines the first predicted probability as the first weight and the product of the second predicted probability and the label value as the second weight.

[0174] Optionally, the first similarity sim satisfies: sim = ∑ j c k >*w1(a j )*w2(c k ).

[0175] in, j c k >This is the second similarity calculated by the server using text embedding technology between the j-th tag in the clothing tag of the video content and the k-th tag in the clothing tag of the candidate advertisement. w1(a j ) represents the first weight determined by the server based on the first predicted probability corresponding to the j-th tag in the clothing tag of the video content. w2(c k ) represents the second weight determined by the server based on the second predicted probability and the tag value corresponding to the k-th tag among the candidate clothing tags. Here, m and n are positive integers.

[0176] In another possible implementation, the clothing tag in the video content corresponds to a first prediction probability, and the clothing tag in the candidate advertisement corresponds to a second prediction probability.

[0177] like Figure 12 As shown, the implementation process of step 504 includes the following steps 5042a and 5042b:

[0178] In step 5042a, a second similarity between the clothing tags in the video content and the clothing tags in the candidate advertisements is calculated using text embedding technology.

[0179] The process of the server calculating the second similarity is described in step 5041a, and will not be repeated here.

[0180] In step 5042b, the first similarity is calculated based on the product of the second similarity, the first weight, and the third weight. The first weight is determined based on the first predicted probability, and the third weight is determined based on the second predicted probability.

[0181] Optionally, the third weight is positively correlated with the second predicted probability. For example, the server determines the second predicted probability as the third weight.

[0182] Optionally, the first similarity sim satisfies: sim = ∑ j c k >*w1(a j )*w3(c k ​​​).

[0183] wherein <a j , c k > is the second similarity between the jth label in the video content fashion label and the kth label in the candidate advertisement fashion label calculated by the server through the text embedding technique. w1(a j ) is a first weight determined by the server according to the first prediction probability corresponding to the jth label in the video content fashion label. w3(c k ) is a third weight determined by the server according to the second prediction probability corresponding to the kth label in the candidate advertisement fashion label. Wherein m and n are positive integers.

[0184] In another possible implementation, the candidate advertisement fashion label corresponds to a label value.

[0185] As shown in FIG. 5, the implementation process of step 504 includes the following steps 5043a and 5043b: Figure 13

[0186] In step 5043a, the second similarity between the video content fashion label and the candidate advertisement fashion label is calculated through the text embedding technique.

[0187] The implementation process of the server calculating the second similarity is described in step 5041a, which is not repeated here.

[0188] In step 5043b, the first similarity is calculated according to the product of the second similarity and the fourth weight, and the fourth weight is determined according to the label value.

[0189] Optionally, the fourth weight is positively correlated with the label value. For example, the server determines the label value as the fourth weight.

[0190] Optionally, the first similarity sim satisfies: sim = ∑<a j , c k >*w4(c k ).

[0191] wherein <a j , c k > is the second similarity between the jth label in the video content fashion label and the kth label in the candidate advertisement fashion label calculated by the server through the text embedding technique. w4(c k ) is a fourth weight determined by the server according to the label value corresponding to the kth label in the candidate advertisement fashion label. Wherein m and n are positive integers.

[0192] ​Optionally, the server can further calculate the first similarity between the video content fashion label and the candidate advertisement fashion label according to click data of the candidate advertisement. For example, after a recommended advertisement corresponding to the video is clicked by a user, the server maps the video content fashion label of the video to a third vector to obtain a third target vector. The candidate advertisement fashion label of the recommended advertisement is mapped to a fourth vector to obtain a fourth target vector. The server cross-multiplies the third target vector and the fourth target vector to obtain a combination feature of the video content fashion label and the candidate advertisement fashion label. A coefficient of the combination feature can be obtained through a logistic regression (LR) model, and the first similarity between the video content fashion label and the candidate advertisement fashion label can be obtained according to the coefficient.

[0193] The server calculates the first similarity between the video content fashion label and the candidate advertisement fashion label, and the first similarity is used to reflect the similarity between the type of the first fashion object and the type of the second fashion object.

[0194] In step 505, the server determines the recommended advertisement corresponding to the video from the candidate advertisements according to the first similarity.

[0195] Optionally, the server determines the candidate advertisement with the highest first similarity to the video as the recommended advertisement corresponding to the video. That is, the recommended advertisement corresponding to the video is determined according to the similarity between the fashion content in the video content and the fashion content in the advertisement content.

[0196] Alternatively, the server calculates a recommendation score of the candidate advertisement according to the first similarity and a third similarity, and determines the candidate advertisement with the highest recommendation score as the recommended advertisement. The third similarity is used to reflect the similarity between the first feature of the user account receiving the recommended advertisement and the second feature of the candidate advertisement.

[0197] In step 506, the server sends the recommended advertisement to the client.

[0198] The recommended advertisement is determined by the server according to the first similarity between the video content fashion label and the candidate advertisement fashion label. Optionally, the server sends the recommended advertisement to the client immediately after determining the recommended advertisement.

[0199] In step 507, the client displays a play interface of the video, and the recommended advertisement is displayed on the play interface.

[0200] The steps performed by the server can be implemented as a server-side advertisement recommendation method, and the steps performed by the client can be implemented as a client-side recommended advertisement display method.

[0201] In summary, the advertisement recommendation method provided by the embodiments of the present application includes the following steps: a server determines a video content costume label of a video and a candidate advertisement costume label. The server calculates a first similarity between the video content costume label and the candidate advertisement costume label according to an advertisement request sent by a client, and determines a recommended advertisement corresponding to the video according to the first similarity. The first similarity reflects the similarity between the type of the first costume object and the type of the second costume object, that is, the similarity between the costume content in the video content and the costume content in the advertisement content. The recommended advertisement can be matched with the costume content in the video watched by the user, and the possibility that the user is interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved.

[0202] Optionally, the first similarity is calculated according to the product of the second similarity, the first weight and the second weight. The accuracy when predicting the type of the first costume object and the type of the second costume object, and the value of the recommended advertisement can be referred to for determining the recommended advertisement. The advertisement can be more flexibly recommended to the user.

[0203] Optionally, the first similarity is calculated according to the product of the second similarity, the first weight and the third weight. The deviation existing when predicting the type of the first costume object and the type of the second costume object is reduced, the influence on the accuracy of the determined recommended advertisement is reduced, and the advertisement can be more accurately recommended to the user.

[0204] Optionally, the first similarity is calculated according to the product of the second similarity and the fourth weight. The advertisement can be recommended to the user by referring to the value of the recommended advertisement, and the income of the advertisement provider is improved.

[0205] In addition, the recommended advertisement is determined from a recommended advertisement set with the same industry characteristic according to the industry characteristic corresponding to the video content costume label, the efficiency of determining the recommended advertisement is improved, and the consumption of computing resources is reduced. When there is a repeated video label in the video content costume label or a repeated advertisement label in the candidate advertisement costume label, the repeated video label and the repeated advertisement label are removed, the number of labels is reduced while ensuring the accuracy of the video content costume label and the candidate advertisement costume label, and the efficiency when calculating the first similarity is improved.

[0206] It should be noted that the order of the steps of the advertisement recommendation method provided by the embodiments of the present application can be appropriately adjusted, and the steps can be appropriately increased or decreased according to the situation. Any person skilled in the art can easily think of changes within the technical range disclosed in the present application, which should be covered within the protection scope of the present application, and therefore will not be described again.

[0207] In a specific example, a user watches a short video in a short video client, the short video has shorts, and the user is characterized by having a certain purchasing power. The user initiates an interest behavior for the short video, such as liking or commenting. The short video server determines a recommended advertisement of a high-priced brand of shorts from the pants industry according to the advertisement request of the short video client, and sends it to the short video client. The short video client displays the recommended advertisement when playing the short video. The clothing content in the recommended advertisement is the same as the clothing content in the short video, so that the user can be interested when watching, and the recommended advertisement meets the user's characteristics. The user is recommended an advertisement related to the user's characteristics and the video content watched. The purchase conversion rate of the advertised goods can be improved.

[0208] Figure 14 is a structural schematic diagram of an advertisement recommendation device provided by an embodiment of the present application. The device can be used in a server in a system as shown in Figure 2 , as shown in Figure 14 , the device 140 comprises:

[0209] The acquisition module 1401 is configured to acquire a video content clothing label and a candidate advertisement clothing label. The video content clothing label is used to reflect the type of the first clothing object in the video frame of the video, and the candidate advertisement clothing label is used to reflect the type of the second clothing object in the candidate advertisement.

[0210] The calculation module 1402 is configured to calculate a first similarity between the video content clothing label and the candidate advertisement clothing label according to at least one of a first prediction probability, a second prediction probability, and a label value. The first similarity is used to reflect the similarity between the type of the first clothing object and the type of the second clothing object, the first prediction probability is used to reflect the probability that the first clothing object belongs to the video content clothing label, the second prediction probability is used to reflect the probability that the second clothing object belongs to the candidate advertisement clothing label, and the label value is used to reflect the value of the candidate advertisement clothing label.

[0211] The first determination module 1403 is configured to determine a recommended advertisement corresponding to the video in the candidate advertisement according to the first similarity.

[0212] In summary, the advertisement recommendation device provided by the embodiments of the present application calculates the first similarity between the video content clothing label and the candidate advertisement clothing label through the calculation module, and determines the recommended advertisement corresponding to the video through the determination module according to the first similarity. The first similarity reflects the similarity between the type of the first clothing object and the type of the second clothing object, i.e. the similarity between the clothing content in the video content and the clothing content in the advertisement content. The recommended advertisement can match the clothing content in the video watched by the user, and the possibility of the user being interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved.

[0213] Optionally, the calculating module 1402 is configured to:

[0214] calculate a second similarity between the video content clothing label and the candidate advertisement clothing label by using a text embedding technique; and calculate the first similarity according to a product of the second similarity, a first weight and a second weight, the first weight being determined according to the first predicted probability, and the second weight being determined according to the second predicted probability and the label value.

[0215] Optionally, the calculating module 1402 is configured to:

[0216] calculate a second similarity between the video content clothing label and the candidate advertisement clothing label by using a text embedding technique; and calculate the first similarity according to a product of the second similarity, a first weight and a third weight, the first weight being determined according to the first predicted probability, and the third weight being determined according to the second predicted probability.

[0217] Optionally, the calculating module 1402 is configured to:

[0218] calculate a second similarity between the video content clothing label and the candidate advertisement clothing label by using a text embedding technique; and calculate the first similarity according to a product of the second similarity and a fourth weight, the fourth weight being determined according to the label value.

[0219] Optionally, the calculating module 1402 is configured to:

[0220] map the video content clothing label to a first vector to obtain a first target vector, and map the advertisement clothing label to a second vector to obtain a second target vector; and calculate a cosine similarity between the first target vector and the second target vector to obtain the second similarity.

[0221] Optionally, the obtaining module 1401 is configured to:

[0222] obtain p video frames of the video, p being a positive integer.

[0223] obtain a first clothing region from the p video frames by using a image segmentation model, the first clothing region including a first clothing object.

[0224] obtain the video content clothing label by predicting a type of the first clothing object by using a image classification model.

[0225] obtain the video content clothing label.

[0226] Optionally, the obtaining module 1401 is configured to:

[0227] obtain n candidate video content clothing labels by predicting a type of the first clothing object in the first clothing region by using a image classification model, n being a positive integer.

[0228] In response to the existence of duplicate video labels in the n candidate video content clothing labels, the duplicate video labels are merged into de-duplicated video labels, to obtain the video content clothing labels.

[0229] Optionally, as shown in the figure, Figure 15 The apparatus 140 further includes:

[0230] The second determination module 1404 is configured to determine an average value of all video clothing prediction probabilities corresponding to the duplicate video labels as a video clothing prediction probability corresponding to the de-duplicated video label.

[0231] Optionally, the acquisition module 1401 is configured to:

[0232] The q pictures of the candidate advertisement are acquired, q being a positive integer.

[0233] The second clothing region is cropped from the q pictures by the image segmentation model, and the second clothing region includes the second clothing object.

[0234] The type of the second clothing object is predicted by the image classification model, to obtain the candidate advertisement clothing label.

[0235] The candidate advertisement clothing label is acquired.

[0236] Optionally, the acquisition module 1401 is configured to:

[0237] The type of the second clothing object in the second clothing region is predicted by the image classification model, to obtain m pending candidate advertisement clothing labels, m being a positive integer.

[0238] In response to the existence of duplicate advertisement labels in the m pending candidate advertisement clothing labels, the duplicate advertisement labels are merged into de-duplicated advertisement labels, to obtain the candidate advertisement clothing label.

[0239] Optionally, the second determination module 1404 is further configured to determine an average value of all advertisement clothing prediction probabilities corresponding to the duplicate advertisement labels as an advertisement clothing prediction probability corresponding to the de-duplicated advertisement label.

[0240] Optionally, the acquisition module 1401 is configured to:

[0241] The video content clothing label is mapped to a mapping table, to obtain an industry feature corresponding to the video content clothing label.

[0242] The candidate advertisements with the same industry feature are determined as a candidate advertisement set.

[0243] The candidate advertisement clothing label is acquired according to the candidate advertisements in the candidate advertisement set.

[0244] Optionally, the first determination module 1403 is configured to:

[0245] According to the first similarity and the third similarity, a recommendation score of the candidate advertisement is calculated. The third similarity is used to reflect the similarity between the first feature of the user account receiving the recommended advertisement and the second feature of the candidate advertisement. The candidate advertisement with the highest recommendation score is determined as the recommended advertisement.

[0246] In summary, the advertisement recommendation device provided by the embodiments of the present application calculates the first similarity between the video content clothing label and the candidate advertisement clothing label through the calculation module, and determines the recommended advertisement corresponding to the video according to the first similarity through the determination module. The first similarity reflects the similarity between the types of the first clothing object and the second clothing object, that is, the similarity between the clothing content in the video content and the clothing content in the advertisement content. The recommended advertisement can be matched with the clothing content in the video watched by the user, and the possibility of the user being interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved.

[0247] Optionally, the first similarity is calculated according to the product of the second similarity, the first weight and the second weight. The accuracy in predicting the types of the first clothing object and the second clothing object, and the value of the recommended advertisement can be referred to in determining the recommended advertisement. The advertisement can be more flexibly recommended to the user.

[0248] Optionally, the first similarity is calculated according to the product of the second similarity, the first weight and the third weight. The deviation existing in predicting the types of the first clothing object and the second clothing object is reduced, the influence on the accuracy of the determined recommended advertisement is reduced, and the advertisement can be more accurately recommended to the user.

[0249] Optionally, the first similarity is calculated according to the product of the second similarity and the fourth weight. The value of the recommended advertisement can be referred to in recommending the advertisement to the user, and the income of the advertisement provider is improved.

[0250] In addition, according to the industry characteristics corresponding to the video content clothing label, the recommended advertisement is determined from the recommended advertisement set with the same industry characteristics, the efficiency of determining the recommended advertisement is improved, and the consumption of computing resources is reduced. When there are repeated video labels in the video content clothing label or repeated advertisement labels in the candidate advertisement clothing label, the repeated video labels and the repeated advertisement labels are de-duplicated, the number of labels is reduced while ensuring the accuracy of the video content clothing label and the candidate advertisement clothing label, and the efficiency in calculating the first similarity is improved.

[0251] Figure 16 is a structural schematic diagram of a recommended advertisement display device provided by an embodiment of the present application. The device can be used in a terminal or a client on a terminal as shown in Figure 2 As shown in Figure 16 , the device 160 includes:

[0252] The sending module 1601 is configured to send an advertisement request to a server. The advertisement request carries at least one of a video content costume label and a video identifier. The video content costume label is used to reflect a type of a first costume object in a video frame of a video.

[0253] The receiving module 1602 is configured to receive a recommended advertisement fed back by the server. The recommended advertisement is determined by the server according to a first similarity between the video content costume label and a candidate advertisement costume label. The first similarity is calculated by the server according to at least one of a first prediction probability, a second prediction probability and a label value. The first prediction probability is used to reflect a probability that the first costume object belongs to the video content costume label. The second prediction probability is used to reflect a probability that a second costume object belongs to the candidate advertisement costume label. The label value is used to reflect a value of the candidate advertisement costume label. The candidate advertisement costume label is used to reflect a type of the second costume object in a candidate advertisement. The first similarity is used to reflect a similarity degree between the type of the first costume object and the type of the second costume object. Optionally, the recommended advertisement is determined by the advertisement recommendation apparatus in the method embodiment. Figure 14

[0254] The display module 1603 is configured to display a playing interface of the video, and the playing interface displays the recommended advertisement.

[0255] To sum up, the recommended advertisement display apparatus provided by the embodiment of the present application sends an advertisement request to a server through a sending module, receives a recommended advertisement fed back by the server through a receiving module, and displays the recommended advertisement. The recommended advertisement is determined according to a similarity degree between costume content in a video and costume content in an advertisement. The recommended advertisement can be matched with the costume content in the video watched by a user, and the possibility that the user is interested in the recommended advertisement is improved. The accuracy of the recommended advertisement is improved.

[0256] Those skilled in the art can clearly understand the specific working process of the apparatus and each module described above for the convenience and brevity of description, which can refer to the corresponding process in the foregoing method embodiments, and will not be described herein.

[0257] The embodiment of the present application further provides a computer device, which comprises a processor and a memory. The memory stores at least one instruction, at least one program, a code set or an instruction set. The processor loads and executes the at least one instruction, the at least one program, the code set or the instruction set to implement the advertisement recommendation method or the recommended advertisement display method provided by the foregoing method embodiments.

[0258] Optionally, the computer device is a server. For example, Figure 17 FIG. 1 is a structural schematic diagram of a server provided by an embodiment of the present application.

[0259] ​The server 1700 includes a central processing unit (CPU) 1701, a system memory 1704, including a random access memory (RAM) 1702 and a read-only memory (ROM) 1703, and a system bus 1705 that couples the system memory 1704 to the central processing unit 1701. The computer device 1700 also includes a basic input / output system (I / O) 1706 that helps transfer information between elements within the computer device, and a mass storage device 1707 for storing an operating system 1713, application programs 1714, and other program modules 1715.

[0260] The basic input / output system 1706 includes a display 1708 for displaying information and input devices 1709, such as a mouse, keyboard, etc., for inputting information. Both the display 1708 and input devices 1709 are connected to the central processing unit 1701 through input / output controllers 1710 that are connected to the system bus 1705. The basic input / output system 1706 can also include input / output controllers 1710 for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, the input / output controllers 1710 provide output to a display screen, printer, or other type of output device.

[0261] The mass storage device 1707 is connected to the central processing unit 1701 through a mass storage controller (not shown) that is connected to the system bus 1705. The mass storage device 1707 and its associated computer-readable storage media provide non-volatile storage for the server 1700. That is, the mass storage device 1707 can include a computer-readable storage medium (not shown) such as a hard disk or a compact disc read-only memory (CD-ROM) drive.

[0262] Without loss of generality, the computer readable storage medium can include computer storage medium and communication medium. The computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable storage instructions, data structures, program modules or other data. The computer storage medium includes RAM, ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid state storage technology, CD-ROM, digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. Of course, those skilled in the art can know that the computer storage medium is not limited to the above several. The system memory 1704 and the mass storage device 1707 described above can be collectively referred to as memory.

[0263] The memory stores one or more programs configured to be executed by the one or more central processing units 1701, and the one or more programs contain instructions for implementing the above method embodiments, and the central processing unit 1701 executes the one or more programs to implement the method provided by each method embodiment.

[0264] According to various embodiments of the present application, the server 1700 can also be connected to a remote server on a network through a network such as the Internet. That is, the server 1700 can be connected to the network 1712 through the network interface unit 1711 connected to the system bus 1705, or can be connected to other types of network or remote server system (not shown) using the network interface unit 1711.

[0265] The memory also includes one or more programs stored in the memory, and the one or more programs contain steps performed by the server in the method provided by the embodiments of the present application.

[0266] Optionally, a client is installed on the computer device, and the computer device is a terminal. For example, Figure 18 is a structural schematic diagram of a terminal provided by an embodiment of the present application.

[0267] Generally, the terminal 1800 includes a processor 1801 and a memory 1802.

[0268] The processor 1801 can include one or more processing cores, such as a 4-core processor, an 8-core processor, and the like. The processor 1801 can be implemented in at least one of a hardware form of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), a PLA (Programmable Logic Array). The processor 1801 can also include a main processor and a co-processor, the main processor being a processor for processing data in an awake state, also referred to as a CPU (Central Processing Unit), and the co-processor being a low-power processor for processing data in a standby state. In some embodiments, the processor 1801 can be integrated with a GPU (Graphics Processing Unit) that is responsible for rendering and drawing of content to be displayed by the display screen. In some embodiments, the processor 1801 can further include an AI (Artificial Intelligence) processor for processing computing operations related to machine learning.

[0269] The memory 1802 can include one or more computer-readable storage media that can be non-transitory. The memory 1802 can also include a high-speed random access memory, and a nonvolatile memory such as one or more disk storage devices, flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 1802 is used to store at least one instruction for being executed by the processor 1801 to implement the recommended advertisement display method provided by the method embodiment of the present application.

[0270] In some embodiments, the terminal 1800 can also optionally include a peripheral device interface 1803 and at least one peripheral device. The processor 1801, the memory 1802, and the peripheral device interface 1803 can be connected through a bus or a signal line. Each peripheral device can be connected to the peripheral device interface 1803 through a bus, a signal line, or a circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 1804, a display screen 1805, a camera assembly 1806, an audio circuit 1807, a positioning assembly 1808, and a power supply 1809.

[0271] The peripheral interface 1803 can be used to connect at least one I / O (Input / Output) related peripheral device to the processor 1801 and the memory 1802. In some embodiments, the processor 1801, the memory 1802 and the peripheral interface 1803 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 1801, the memory 1802 and the peripheral interface 1803 can be implemented on a separate chip or circuit board, and the embodiments of the present application do not limit this.

[0272] The radio frequency circuit 1804 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 1804 communicates with a communication network and other communication devices through electromagnetic signals. The radio frequency circuit 1804 converts electrical signals into electromagnetic signals for transmission, or converts received electromagnetic signals into electrical signals. Optionally, the radio frequency circuit 1804 includes an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and the like. The radio frequency circuit 1804 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes but is not limited to: the World Wide Web, a metropolitan area network, an intranet, various generations of mobile communication networks (2G, 3G, 4G and 5G), a wireless local area network and / or a WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 1804 can also include NFC (Near Field Communication) related circuitry, and the present application does not limit this.

[0273] The display screen 1805 is configured to display a UI (User Interface). The UI can include graphics, text, icons, video, and any combination thereof. When the display screen 1805 is a touch display screen, the display screen 1805 is also capable of capturing touch signals on or above the surface of the display screen 1805. The touch signals can be input to the processor 1801 as control signals for processing. At this time, the display screen 1805 can also be configured to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen 1805 can be one, arranged on the front panel of the terminal 1800; in other embodiments, the display screen 1805 can be at least two, arranged on different surfaces of the terminal 1800 or in a folding design; in still other embodiments, the display screen 1805 can be a flexible display screen, arranged on a curved surface or a folding surface of the terminal 1800. Even, the display screen 1805 can also be arranged in an irregular shape other than a rectangle, i.e., a special-shaped screen. The display screen 1805 can be made of materials such as LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), etc.

[0274] The camera assembly 1806 is configured to capture images or videos. Optionally, the camera assembly 1806 includes a front-facing camera and a rear-facing camera. Typically, the front-facing camera is arranged on the front panel of the terminal 1800, and the rear-facing camera is arranged on the back of the terminal. In some embodiments, the rear-facing camera is at least two, which are any one of a main camera, a depth-of-field camera, a wide-angle camera, and a telephoto camera, to realize the background blur function by fusing the main camera and the depth-of-field camera, the panoramic shooting and VR (Virtual Reality) shooting function by fusing the main camera and the wide-angle camera, or other fusion shooting functions. In some embodiments, the camera assembly 1806 can also include a flash. The flash can be a single-color-temperature flash or a dual-color-temperature flash. The dual-color-temperature flash refers to a combination of a warm light flash and a cold light flash, which can be used for light compensation under different color temperatures.

[0275] The audio circuit 1807 can include a microphone and a speaker. The microphone is used to collect sound waves of a user and an environment, and convert the sound waves into an electrical signal input to the processor 1801 for processing, or input to the radio frequency circuit 1804 to realize voice communication. For the purpose of stereo sound collection or noise reduction, the microphone can be multiple, respectively arranged at different parts of the terminal 1800. The microphone can also be an array microphone or an omnidirectional collection type microphone. The speaker is used to convert the electrical signal from the processor 1801 or the radio frequency circuit 1804 into sound waves. The speaker can be a traditional diaphragm speaker, or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, not only can the electrical signal be converted into a sound wave that a human can hear, but also can be converted into a sound wave that a human cannot hear for the purpose of ranging, etc. In some embodiments, the audio circuit 1807 can also include a headphone jack.

[0276] The positioning component 1808 is used to position the current geographic position of the terminal 1800 to realize navigation or LBS (Location Based Service). The positioning component 1808 can be a positioning component based on the GPS (Global Positioning System) of the United States, the Beidou system of China or the Galileo system of Russia.

[0277] The power supply 1809 is used to supply power to each component in the terminal 1800. The power supply 1809 can be alternating current, direct current, disposable battery or rechargeable battery. When the power supply 1809 includes a rechargeable battery, the rechargeable battery can be a wired charging battery or a wireless charging battery. The wired charging battery is a battery charged through a wired line, and the wireless charging battery is a battery charged through a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0278] In some embodiments, the terminal 1800 further includes one or more sensors 1810. The one or more sensors 1810 include but are not limited to: an acceleration sensor 1811, a gyroscope sensor 1812, a pressure sensor 1813, a fingerprint sensor 1814, an optical sensor 1815 and a proximity sensor 1816.

[0279] The acceleration sensor 1811 can detect the acceleration size in three coordinate axes of the coordinate system established by the terminal 1800. For example, the acceleration sensor 1811 can be used to detect the components of the gravitational acceleration in three coordinate axes. The processor 1801 can control the touch display screen 1805 to display the user interface in a landscape view or a portrait view according to the gravitational acceleration signal collected by the acceleration sensor 1811. The acceleration sensor 1811 can also be used for game or user motion data collection.

[0280] The gyroscope sensor 1812 can detect the body direction and rotation angle of the terminal 1800, and can collect 3D motions of a user on the terminal 1800 in cooperation with the acceleration sensor 1811. The processor 1801 can implement the following functions according to data collected by the gyroscope sensor 1812: motion sensing (e.g., changing a UI according to a tilt operation of a user), image stabilization when photographing, game control, and inertial navigation.

[0281] The pressure sensor 1813 can be disposed at a side frame of the terminal 1800 and / or under the touch display screen 1805. When the pressure sensor 1813 is disposed at the side frame of the terminal 1800, a grip signal of a user on the terminal 1800 can be detected, and left / right hand recognition or a shortcut operation can be performed by the processor 1801 according to the grip signal collected by the pressure sensor 1813. When the pressure sensor 1813 is disposed under the touch display screen 1805, operable controls on a UI can be controlled by the processor 1801 according to a pressure operation of a user on the touch display screen 1805. The operable controls include at least one of a button control, a scroll bar control, an icon control, and a menu control.

[0282] The fingerprint sensor 1814 is used to collect a fingerprint of a user, and the identity of the user can be recognized by the processor 1801 according to the fingerprint collected by the fingerprint sensor 1814 or by the fingerprint sensor 1814. When the identity of the user is recognized as a trusted identity, the processor 1801 authorizes the user to perform a related sensitive operation, which includes unlocking a screen, viewing encrypted information, downloading software, payment, and changing a setting, etc. The fingerprint sensor 1814 can be disposed at a front surface, a back surface, or a side surface of the terminal 1800. When a physical button or a manufacturer's logo is disposed on the terminal 1800, the fingerprint sensor 1814 can be integrated with the physical button or the manufacturer's logo.

[0283] The optical sensor 1815 is used to collect ambient light intensity. In an embodiment, the processor 1801 can control the display brightness of the touch display screen 1805 according to the ambient light intensity collected by the optical sensor 1815. Specifically, when the ambient light intensity is high, the display brightness of the touch display screen 1805 is increased, and when the ambient light intensity is low, the display brightness of the touch display screen 1805 is decreased. In another embodiment, the processor 1801 can also dynamically adjust the photographing parameters of the camera assembly 1806 according to the ambient light intensity collected by the optical sensor 1815.

[0284] The proximity sensor 1816, also called distance sensor, is usually arranged on the front panel of the terminal 1800. The proximity sensor 1816 is used to collect the distance between the user and the front of the terminal 1800. In one embodiment, when the proximity sensor 1816 detects that the distance between the user and the front of the terminal 1800 gradually becomes smaller, the touch display screen 1805 is switched from the bright screen state to the screen-off state under the control of the processor 1801; when the proximity sensor 1816 detects that the distance between the user and the front of the terminal 1800 gradually becomes larger, the touch display screen 1805 is switched from the screen-off state to the bright screen state under the control of the processor 1801.

[0285] Those skilled in the art can understand that the structure shown in the foregoing embodiments is not a limitation on the terminal 1800, and the terminal 1800 can include more or fewer components than those shown in the drawings, or combine certain components, or adopt a different component arrangement. Figure 18

[0286] The embodiments of the present application further provide a computer storage medium, which can store at least one instruction, at least one program, a code set or an instruction set. When the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor of a computer device, the advertisement recommendation method or the method for displaying recommended advertisements provided by the above-mentioned method embodiments is implemented.

[0287] Those skilled in the art can understand that all or part of the steps of the above-mentioned embodiments can be completed by hardware, or by a program instructing related hardware, and the program can be stored in a computer readable storage medium, such as a read-only memory, a magnetic disk or an optical disk.

[0288] The above-mentioned only is the optional embodiment of the present application, and does not limit the present application. Any modification, equivalent switching, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.​

Claims

1. An advertisement recommendation method characterized by comprising: The method comprises: obtaining a video content clothing label and a candidate advertisement clothing label, the video content clothing label being used to reflect a type of a first clothing object in a video frame of a video, and the candidate advertisement clothing label being used to reflect a type of a second clothing object in a candidate advertisement; mapping the video content clothing label to a first vector by a text embedding technology to obtain a first target vector, and mapping the candidate advertisement clothing label to a second vector to obtain a second target vector; and calculating a cosine similarity of the first target vector and the second target vector to obtain a second similarity; calculating a first similarity according to a product of the second similarity, a first weight and a second weight, the first weight being positively correlated with a first prediction probability, and the second weight being positively correlated with a product of a second prediction probability and a label value; or calculating the first similarity according to a product of the second similarity, the first weight and a third weight, the third weight being positively correlated with the second prediction probability; or calculating the first similarity according to a product of the second similarity and a fourth weight, the fourth weight being positively correlated with the label value; the first similarity being used to reflect a similarity degree of the type of the first clothing object and the type of the second clothing object, the first prediction probability being used to reflect a probability that the first clothing object belongs to the video content clothing label, the second prediction probability being used to reflect a probability that the second clothing object belongs to the candidate advertisement clothing label, and the label value being used to reflect a value of the candidate advertisement clothing label, the label value comprising an average selling price of a clothing type corresponding to the candidate advertisement clothing label; determining a recommended advertisement corresponding to the video in the candidate advertisement according to the first similarity.

2. The method of claim 1, wherein, The method comprises: obtaining p video frames of the video, p being a positive integer; obtaining a first clothing region from the p video frames by an image segmentation model, the first clothing region comprising the first clothing object; predicting a type of the first clothing object by an image classification model to obtain the video content clothing label; obtaining the video content clothing label.

3. The method of claim 2, wherein, The method comprises: predicting the type of the first clothing object in the first clothing region by the image classification model to obtain n candidate video content clothing labels, n being a positive integer; in response to a repeated video label existing in the n candidate video content clothing labels, merging the repeated video label into a de-duplicated video label to obtain the video content clothing label.

4. The method of claim 1, wherein, The method comprises: mapping the video content clothing label to a mapping table to obtain an industry feature corresponding to the video content clothing label; determining the candidate advertisements with the same industry feature as a candidate advertisement set; obtaining the candidate advertisement clothing label according to the candidate advertisements in the candidate advertisement set.

5. The method of claim 1, wherein, The determining the recommended advertisement corresponding to the video from the candidate advertisements according to the first similarity degree comprises: calculating a recommendation score of the candidate advertisement according to the first similarity degree and a third similarity degree, the third similarity degree reflecting a similarity between a first feature of a user account receiving the recommended advertisement and a second feature of the candidate advertisement; determining the candidate advertisement with the highest recommendation score as the recommended advertisement.

6. A method of recommending an advertisement display, characterized by, The method comprises: sending an advertisement request to a server, the advertisement request carrying at least one of a video content clothing label and a video identifier, the video content clothing label reflecting a type of a first clothing object in a video frame of a video; receiving a recommended advertisement fed back by the server, the recommended advertisement being determined by the server according to a first similarity degree between the video content clothing label and a candidate advertisement clothing label, the first similarity degree being calculated by the server in the following manner: mapping the video content clothing label to a first vector to obtain a first target vector and mapping the candidate advertisement clothing label to a second vector to obtain a second target vector by using a text embedding technology; calculating a cosine similarity between the first target vector and the second target vector to obtain a second similarity degree; calculating the first similarity degree according to a product of the second similarity degree, a first weight and a second weight, the first weight being positively correlated with a first prediction probability, and the second weight being positively correlated with a product of a second prediction probability and a label value; or, calculating the first similarity degree according to a product of the second similarity degree, the first weight and a third weight, the third weight being positively correlated with the second prediction probability; or, calculating the first similarity degree according to a product of the second similarity degree and a fourth weight, the fourth weight being positively correlated with the label value; the first prediction probability reflecting a probability that the first clothing object belongs to the video content clothing label, the second prediction probability reflecting a probability that the second clothing object belongs to the candidate advertisement clothing label, and the label value reflecting a value of the candidate advertisement clothing label, the label value comprising an average selling price of a clothing type corresponding to the candidate advertisement clothing label, the first similarity degree reflecting a similarity between the type of the first clothing object and the type of the second clothing object; displaying a play interface of the video, the play interface displaying the recommended advertisement.

7. An advertisement recommendation device characterized by comprising: The device comprises: an acquisition module configured to acquire a video content clothing label and a candidate advertisement clothing label, the video content clothing label reflecting a type of a first clothing object in a video frame of a video, and the candidate advertisement clothing label reflecting a type of a second clothing object in a candidate advertisement; The computing module is configured to map the video content costume label to a first vector by a text embedding technology to obtain a first target vector, and map the candidate advertisement costume label to a second vector to obtain a second target vector; calculate a cosine similarity between the first target vector and the second target vector to obtain a second similarity; calculate a first similarity according to a product of the second similarity, a first weight and a second weight, the first weight being positively correlated with a first prediction probability, and the second weight being positively correlated with a product of a second prediction probability and a label value; or, calculate the first similarity according to a product of the second similarity, the first weight and a third weight, the third weight being positively correlated with the second prediction probability; or, calculate the first similarity according to a product of the second similarity and a fourth weight, the fourth weight being positively correlated with the label value; the first similarity is used to reflect a similarity degree between a type of the first costume object and a type of the second costume object, the first prediction probability is used to reflect a probability that the first costume object belongs to the video content costume label, the second prediction probability is used to reflect a probability that the second costume object belongs to the candidate advertisement costume label, and the label value is used to reflect a value of the candidate advertisement costume label, the label value including an average selling price of a costume type corresponding to the candidate advertisement costume label; The determining module is configured to determine a recommended advertisement corresponding to the video in the candidate advertisements according to the first similarity.

8. A recommended advertisement display device characterized by comprising: The apparatus includes: The sending module is configured to send an advertisement request to a server, the advertisement request carrying at least one of a video content costume label and a video identifier, the video content costume label being used to reflect a type of a first costume object in a video frame of a video. receive a recommended advertisement fed back by the server, the recommended advertisement being determined by the server according to a first similarity between the video content clothing label and a candidate advertisement clothing label, the first similarity being calculated by the server in the following manner: mapping the video content clothing label to a first vector to obtain a first target vector and mapping the candidate advertisement clothing label to a second vector to obtain a second target vector by using a text embedding technology; calculating a cosine similarity between the first target vector and the second target vector to obtain a second similarity; calculating the first similarity according to a product of the second similarity, a first weight and a second weight, the first weight being positively correlated with a first prediction probability, and the second weight being positively correlated with a product of a second prediction probability and a label value; or calculating the first similarity according to a product of the second similarity, the first weight and a third weight, the third weight being positively correlated with the second prediction probability; or calculating the first similarity according to a product of the second similarity and a fourth weight, the fourth weight being positively correlated with the label value; the first prediction probability being used to reflect a probability that the first clothing object belongs to the video content clothing label, the second prediction probability being used to reflect a probability that the second clothing object belongs to the candidate advertisement clothing label, and the label value being used to reflect a value of the candidate advertisement clothing label, the label value including an average selling price of a clothing type corresponding to the candidate advertisement clothing label, the candidate advertisement clothing label being used to reflect a type of a second clothing object in a candidate advertisement, and the first similarity being used to reflect a similarity degree between a type of the first clothing object and a type of the second clothing object; a display module configured to display a playing interface of the video, the playing interface displaying the recommended advertisement.

9. A computer device, comprising: The computer device includes a processor and a memory, and the memory stores at least one program, which is loaded and executed by the processor to implement the advertisement recommendation method of any one of claims 1 to 5 or the recommended advertisement display method of claim 6.

10. A computer storage medium, characterized in that, The computer storage medium stores at least one program, and when the at least one program is loaded and executed by the processor of the computer device, the advertisement recommendation method of any one of claims 1 to 5 or the recommended advertisement display method of claim 6 is implemented.

11. A computer program product, characterised in that, The computer program product includes computer instructions stored in a computer readable storage medium; the processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to implement the advertisement recommendation method of any one of claims 1 to 5 or the recommended advertisement display method of claim 6.

Images