Personalized advertising implantation method and system based on AR technology

Abstract

This invention relates to a personalized advertising embedding method and system based on AR technology, belonging to the field of video processing technology. First, an uploaded video is acquired, and scene recognition and planar recognition are performed on the video to obtain scene recognition results and planar recognition results. Based on the scene recognition results and planar recognition results, the optimal advertising position in the video is determined. Then, based on AR technology, a preset AR advertising model is superimposed on the optimal advertising position in the video, thereby embedding advertisements into the video using AR technology. This allows for easy modification later and provides a good viewing experience.

Description

Technical Field

[0001] The present invention relates to the field of video processing technology, and in particular to a personalized advertisement implantation method and system based on AR technology. Background Art

[0002] At present, the way of inserting advertisements in videos is generally that after the brand or video platform reaches a cooperation with the video creator, the video creator inserts the advertisements in the process of making the video. Therefore, when there is a problem with the product or it needs to be modified later, it is difficult to modify the advertisements in the video. It can only be covered by image processing methods such as mosaic, which seriously affects the viewer experience. Therefore, the current way of inserting advertisements in videos has the defects of being unable to be modified later and 2D advertisements being distorted. At the same time, there are also some methods that use image processing technology to put products into the video later based on the completed video, but the advertising products in the video presented in this way appear particularly abrupt and cannot provide the viewer with a good experience.

[0003] Based on this, there is an urgent need for an advertisement placement method and system that is easy to modify and provides a good viewing experience. Summary of the invention

[0004] The purpose of the present invention is to provide a personalized advertising implantation method and system based on AR technology, which utilizes AR technology to implant advertisements into videos, facilitates modification later, and provides a good viewing experience.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A personalized advertising implantation method based on AR technology, the advertising implantation method comprising:

[0007] Get uploaded videos;

[0008] Performing scene recognition and plane recognition on the video to obtain a scene recognition result and a plane recognition result;

[0009] Determine the best advertisement position in the video according to the scene recognition result and the plane recognition result;

[0010] Based on AR technology, a preset AR advertising model is superimposed on the best advertising position of the video.

[0011] A personalized advertising implantation system based on AR technology, the advertising implantation system comprising:

[0012] The best advertisement position determination module is used to obtain an uploaded video; perform scene recognition and plane recognition on the video to obtain a scene recognition result and a plane recognition result; and determine the best advertisement position in the video according to the scene recognition result and the plane recognition result;

[0013] The overlay module is used to overlay a preset AR advertising model on the best advertising position of the video based on AR technology.

[0014] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0015] The present invention is used to provide a personalized advertising implantation method and system based on AR technology. The method first obtains an uploaded video, performs scene recognition and plane recognition on the video to obtain scene recognition results and plane recognition results, and then determines the best advertising position in the video according to the scene recognition results and the plane recognition results. Finally, based on AR technology, a preset AR advertising model is superimposed on the best advertising position in the video, thereby using AR technology to implant advertisements in the video, which is convenient for modification at a later stage and provides a good viewing experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A method flow chart of the advertisement embedding method provided by Embodiment 1 of the present invention;

[0018] Figure 2 A schematic diagram of the flow of the video analysis process provided in Example 1 of the present invention;

[0019] Figure 3 A schematic diagram of the AR processing process provided by Embodiment 1 of the present invention;

[0020] Figure 4 A schematic diagram of the flow of the advertisement recommendation process provided by Embodiment 1 of the present invention;

[0021] Figure 5 This is a system block diagram of the advertisement implantation system provided in Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] The purpose of the present invention is to provide a personalized advertising implantation method and system based on AR technology, which utilizes AR technology to implant advertisements into videos, facilitates modification later, and provides a good viewing experience.

[0024] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] Embodiment 1:

[0026] The current environment for video platform advertising placement solutions is as follows:

[0027] (1) Industry background:

[0028] The current policy support includes Internet advertising policy support and online video development policy support. While making detailed requirements for Internet advertising, the Internet advertising policy encourages and supports the development of the Internet advertising industry. The online video development policy encourages and supports the integration and development of the online video industry with other industries, and promotes the transformation and upgrading of the online video industry.

[0029] (2) Social background:

[0030] 1) Short videos + video creators + product placement to promote consumption

[0031] At present, short video placement has become the main form of digital marketing. According to iMedia Research data, among the digital marketing methods that are likely to arouse consumers' desire to buy in 2021, short video platform celebrity placement topped the list with 64.8%. The combination of high traffic of short videos, high exposure of video creators, and placement of advertisements has greatly promoted consumers' desire to buy.

[0032] 2) Video creators-advertising revenue

[0033] The main ways for video creators to operate include: accepting advertisements, monetizing works, content e-commerce, live broadcast rewards, knowledge payment, etc. According to the Social Media Examiner survey report, >50% of bloggers, video creators and podcasters use advertising to make money. For creators, advertising is the most important way to make money. Video creators need to constantly create content to attract fans. However, the way that brand owners directly find video creators to place ads is basically only the patent of top video creators. For some small and medium-sized video creators, there are very few brand owners who find them for promotion. Only when the number of fans reaches a certain scale and forms a personal brand and model can they have the opportunity to obtain cooperative promotion.

[0034] 3) Current situation of advertising by advertisers: poor effect and high fan base

[0035] ① Advertisers tend to cooperate with top video creators

[0036] At present, advertisers are mainly concerned about the personal exposure of video creators when they seek to place advertisements with video creators. Personal exposure is closely related to the number of fans of video creators. Advertisers tend to place advertisements with top video creators (groups with a large number of fans and high fan activity). In addition, advertisers generally look for partners by searching for keywords of their own brand and searching for similar big brands. The snowball effect of attention is obvious. Advertising resources continue to gather towards top video creators, and traffic is relatively centralized. There are few small and medium-sized video creators with opportunities, and the differentiation of video creators is gradually expanding.

[0037] ② High advertising investment but low benefits

[0038] Advertisers want to find the most mainstream young consumer user groups and establish links, but due to the rapid iteration of the media environment and the rapid shift in user attention, it is difficult to reach the mainstream consumer groups.

[0039] Advertisers want to influence video viewers quickly and on a large scale, but fragmented platforms and self-media make it impossible to select and deliver ads efficiently.

[0040] Advertisers want to target precise audiences and find suitable and effective partners among video creators, but the audience of self-media is unclear, the life cycle is unstable, and the iteration and update are fast, making it difficult to find suitable video creators.

[0041] Advertisers want to achieve high returns with low investment, but after the advertisements are launched, the data is mixed and cannot be scientifically evaluated and optimized.

[0042] Even though advertisers continue to invest in advertising, it is difficult to meet their core needs, reach the mainstream consumer groups, influence video viewers efficiently and on a large scale, and target the exact audience, making it difficult to achieve high efficiency.

[0043] Based on the above environmental background, the current advertising placement solutions for video platforms have the following defects:

[0044] (1) For brands:

[0045] 1) Advertising cannot guarantee exposure and effectiveness

[0046] When brands place ads on video platforms, they will make a comprehensive judgment based on the exposure of the ads, the popularity of the entire network, the quotes of video creators, and other conditions. Since they cannot predict the exposure and popularity of the video in advance, brands cannot guarantee the effect of the placement. Faced with the dilemma of investing a lot of money but not necessarily receiving a high-quality return, they often take a more conservative approach - cooperating with video creators with a large number of fans, but this often means higher costs. At the same time, when faced with a video that suddenly becomes popular, since the creation period has been missed, the video cannot be modified, and the brand cannot use its traffic in exchange for exposure, resulting in missing out on a massive market.

[0047] 2) The advertisements are not differentiated, which reduces the effectiveness of the advertisements

[0048] Brands can only place ads on videos, so all viewers of the video see the same content, which means that ads cannot select video viewers. Therefore, brands spend a lot of money on audiences that are not suitable for their products, and cannot efficiently obtain benefits and returns, resulting in a large number of invalid ads and reduced advertising revenue.

[0049] (2) For video creators: high-quality advertising resources belong to top video creators

[0050] One of the ways for video creators to monetize their traffic is to insert advertisements, either hard or soft, into videos. Through high-quality videos, video creators can gain the attention of fans and thus obtain better advertising resources. Brands can directly find video creators for cooperation, which is basically a privilege enjoyed only by top video creators. For some small and medium-sized video creators, there are very few brands that seek them out for promotion. Only when their number of fans reaches a certain scale can they have the opportunity for cooperative promotion. At the same time, the creation of high-quality video content requires a lot of resources, consuming manpower and material resources. The contradiction between the two will also put some video creators in a survival dilemma.

[0051] (3) For video viewers:

[0052] 1) Forced delivery of advertisements can easily cause discomfort

[0053] Video viewers pay more attention to the experience of watching the video itself. If advertisements are forcibly inserted in the middle of the video, it will cause discomfort to the video viewers, thereby causing resistance to the advertising product itself. However, excessive focus on the video viewer experience will cause the embedded advertising itself to lose a large market. Therefore, it is particularly important to grasp the balance point of advertising presentation.

[0054] 2) The ads are short and don’t give you a deep understanding of the product

[0055] The ads that video viewers receive in the video are limited in length and expression, so they cannot have a deep understanding of the product. The product information that can be mentioned in the video is very limited for a complete product, so the ads cannot accurately convey detailed information to the viewers. If the viewer shows interest, he or she needs to go to other platforms to learn more, making the process cumbersome.

[0056] Based on the above defects, this embodiment considers the support of the following technologies and proposes a personalized advertising implantation method for a video platform based on AR technology.

[0057] Technical support includes:

[0058] (1) AR technology

[0059] Augmented Reality is a technology that cleverly integrates virtual information with the real world and is widely used in the multimedia field. Currently, AR technology is in a stage of rapid development. Its products are accepted by the consumer market and have a mature profit model. In the 2020 New Technology Maturity Curve released by Gartner, AR technology has officially left the list, indicating that AR technology has stabilized and has transformed from a technology to be observed to a technology that can be used.

[0060] (2) 5G technology

[0061] The fifth generation of mobile communication technology (5G for short) is a new generation of broadband mobile communication technology with the characteristics of high speed, low latency and large connection. It is the network infrastructure for realizing the interconnection of man, machine and things. By the end of 2021, my country has built and opened 1.425 million 5G base stations, accounting for more than 60% of the world's total, and the number of 5G base stations per 10,000 people has reached 10.1. At present, 5G technology has matured and applied to many fields, providing a strong foundation for the development of new technologies such as AR, VR, and AI.

[0062] Based on the above technical support, this embodiment provides a personalized advertising embedding method for a video platform based on AR technology, such as Figure 1 As shown, the advertisement embedding method includes:

[0063] S1: Obtain an uploaded video; perform scene recognition and plane recognition on the video to obtain a scene recognition result and a plane recognition result; determine the best advertising position in the video according to the scene recognition result and the plane recognition result;

[0064] Before obtaining the uploaded video, the advertising embedding method of this embodiment also includes the step of uploading the video, specifically including: calling the video upload interface to upload the video to the storage warehouse, and receiving the video storage location and video access link returned by the video upload interface, obtaining the uploaded video according to the video storage location, and performing subsequent identification and processing of the best advertising position for the uploaded video.

[0065] As an optional implementation, the advertising implantation method of this embodiment also includes determining a preset AR advertising model, specifically including: performing topic analysis on the uploaded video to obtain topic analysis results, classifying the video according to the scene recognition results and the topic analysis results to obtain the category of the video, and determining the preset AR advertising model according to the category of the video.

[0066] More specifically, the advertisement embedding method of this embodiment is performed by the Vision Advertising Platform, such as Figure 2 As shown, the process of uploading a video, determining the best advertisement position, and classifying the video in this embodiment includes:

[0067] Enable OSS object storage service, smart label service and smart label pipeline. OSS object storage service is used as a storage warehouse for videos uploaded by video viewers. A bucket is created in OSS object storage service to store videos that need to be analyzed. Smart label service provides an interface for video analysis tasks. Smart label pipeline is used as a job queue for video analysis. It provides the function of entering the pipeline and performing video analysis tasks in an orderly manner when concurrent video analysis tasks are submitted. That is, the submitted video analysis tasks are processed in an orderly manner through the pipeline queue.

[0068] Write a video upload interface for uploading video content. The Vision Advertising Platform calls the video upload interface to upload the video to the bucket in the OSS object storage service that stores the video that needs to be analyzed. The video upload interface returns the video storage location and video access link. The video storage location is used as the basic parameter required to establish the smart tag video analysis task interface, that is, the service initiates an analysis task for the video based on the video storage location. The video access link is an access link with public read and write rights, which is used as the video stream input for later AR processing and as a parameter when previewing the video on the playback app.

[0069] Write an interface for submitting smart label jobs, select a parameter template, add three analysis types: scene recognition, plane recognition, and theme analysis in the parameter template, use the video storage location as input, and check the corresponding analysis type according to user needs. Check scene recognition to get scene recognition results, check plane recognition to get plane recognition results, and check theme analysis to get theme analysis results. After checking the corresponding analysis type, the video will be analyzed accordingly and the smart label job query ID will be returned. The video label analysis results can be queried through the ID. The video label analysis results include the scene recognition results, plane recognition results, and theme analysis results of the video.

[0070] The basic classification of videos can be completed based on the scene recognition results and the theme analysis results. Specifically, the scene recognition results and theme analysis results obtained by the above video analysis method are stored in the corresponding data table in the database as the attributes of the video submitted by the video creator, and the data results are returned to the web front end in the form of an interface. The web front end can divide all creators' videos by scene classification and theme classification, and advertisers can choose suitable videos for investment through different categories. After negotiation between advertisers and video creators, the preset AR advertising model can be determined. This embodiment is based on scene recognition technology, and completes video classification by performing scene and theme analysis on videos, and can make basic classification selections for advertising videos that advertisers need to select.

[0071] At the same time, this embodiment determines the best advertisement position in the video based on the scene recognition results and the plane analysis results.

[0072] After the video platform joins the Vision Advertising Platform of this embodiment, when the video creator releases the video, he can choose to upload the video to the Vision Advertising Platform for the brand to choose. Similarly, he can also choose to upload the video to the Vision Advertising Platform after releasing the video, that is, the uploaded video in this embodiment can be a video that has been released on the video platform. After uploading, the Vision Advertising Platform will analyze the current video and mark the best position (i.e., the best advertising position) where the 3D advertising model can be placed in the video with a black circle. If there is an advertisement in this position, the advertising product content will be displayed. The video viewer can click the "Advertisement Visible" icon in the video platform to change the video viewer's perspective. If the advertisement is visible, the previous advertising product content will be displayed. Otherwise, the preset AR advertising model superimposed by AR technology in this embodiment will be displayed.

[0073] S2: Based on AR technology, a preset AR advertising model is superimposed on the best advertising position of the video.

[0074] S2 may include: obtaining a video according to the video access link, and superimposing a preset AR advertising model on the best advertising position of the video based on AR technology. The superimposition process may include: processing the video using AR technology, tracking the best advertising position of the video using a planar image recognition and motion tracking method, and loading a preset AR advertising model on the best advertising position to obtain a video superimposed with the preset AR advertising model.

[0075] As an optional implementation, after S2, the advertisement embedding method of this embodiment further includes: recognizing the gesture of the video viewer, and adjusting the preset AR advertisement model according to the gesture, for example, enlarging, reducing, rotating, etc. the preset AR advertisement model.

[0076] More specifically, Figure 3 As shown, the superposition process and gesture interaction process include:

[0077] Obtain the network video stream according to the video access link, and process the network video stream based on AR technology on Unity. It should be noted that AR technology uses the image obtained by the camera as the image input source, and provides a Customcamera script. In order to enable AR technology to process the network video stream, this embodiment modifies the Customcamera script and converts the camera image input source into the image input of the network video stream, so that the uploaded video can be processed by AR technology.

[0078] The best ad position set by S1 is tracked by plane image recognition and motion tracking technology, and a preset AR ad model is dragged on the best ad position to obtain a video superimposed with a preset AR ad model. The video superimposed with a preset AR ad model is finally embedded in the Android project, and the AR function is connected when playing the video to superimpose the preset AR ad model on the best ad position in the video. Specifically, this embodiment uses the Unity development tool to develop the relevant AR function, modifies the original VideoCameraDevice component to define the image input source of the camera as the image of the network video stream used in this solution, modifies the input source to the video stream URL through the sample custom Customcamera, and uses the video as input, and then realizes the recognition of the best ad position of the video and superimposes the required AR ad model through the combination of Easyar's ImageTracker and ImageTarget components. The ImageTracker component defines the image input source, as well as the plane image recognition method and the method of displaying the pre-realized model after image recognition. Each AR ad model can be placed under the ImageTarget component, and the AR ad model under the ImageTarget is automatically loaded to the specified ad position after the ImageTracker image recognition is successful.

[0079] This embodiment recognizes the gesture interaction between the video viewer and the screen through the written C# script, such as zooming in the screen with two fingers, so that the AR advertising model can make feedback of zooming in or rotating, so that the video viewer can interact with the AR advertising model to provide product information interaction. Specifically, a related method script for interacting with the model is developed, that is, the gesture recognition between the video viewer and the screen, such as zooming in and out operations, and the AR advertising model can make feedback of zooming in and out. Through interaction with the AR advertising model, more product information can be provided to the video viewer. After the overall AR function is realized, it is packaged and exported as a functional module of the Android project for reference, and the AR advertising effect is presented in real time when the video is played. This embodiment processes network video streams based on AR technology, and uses plane image recognition and motion tracking technology to superimpose 3D advertising models, so that the model changes the angle of the 3D advertising model in the video according to the angle at which the video viewer watches the video, so that the advertising model is completely integrated into the video like a real object. Through gesture recognition technology, users can interact with the AR advertising model, and can perform operations such as rotation, zooming in and out, which is convenient for users to understand the content of the advertisement.

[0080] This embodiment can also use other ARsdk or Python underlying image processing methods to complete the above-mentioned overlay and gesture interaction methods. At the same time, similar scene analysis methods can be used for scene recognition, and different development tools can be used to complete the core scene analysis and AR rendering steps.

[0081] As an optional implementation, Figure 4 As shown, there are multiple preset AR advertising models, and different preset AR advertising models are superimposed on the best advertising positions using time recommendation method, region recommendation method or user recommendation method. That is, in the same advertising position of the video, different AR advertising models are recommended based on time, region and user portraits provided by different video platforms to achieve personalized AR advertising recommendations.

[0082] Specifically, the time recommendation method is: the advertiser selects a recommended time period to release the advertising model. When the audience watches the video, the server obtains the time information of the audience's viewing, divides the video playback into different time periods, and specifies the advertising models of different advertisers to be played in real time on the advertising position.

[0083] The regional recommendation method is: obtain the user's location permission through the server, obtain the user's regional information, and during non-buyout time, recommend the advertising model of the advertiser in the user's region based on the regional information, or the advertising model in the user's region specified by the advertiser.

[0084] The user recommendation method is as follows: the user recommendation method is a recommendation algorithm of the video platform, which is formulated by the video platform. This embodiment is modified to obtain only its results through user portraits. Through different user portraits of each video platform, a connection is established with a model library composed of a large number of ImageTarget components. The ImageTracker script is modified to collect different user portraits of each video platform. For the video theme type data that users often watch in the user portraits, the ImageTarget of the product type name under the same theme type is loaded. When AR renders the video, the AR advertising model is loaded into the corresponding advertising position in the video according to the corresponding ImageTarget loaded and the predetermined position relationship with the advertising position set in Unity. The advertising model is personalized for the videos seen by different users.

[0085] By combining the above three recommendation methods, the personalized recommendation method for AR advertising models is completed. Of course, this embodiment can also use other recommendation algorithms to realize personalized recommendation of advertising models. The visual advertising platform provided by this embodiment involves three objects: brand, video platform, and video creator. According to a series of problems existing in the current advertising market, a new solution is proposed - scene recognition and plane recognition are performed on the videos in the video platform, and the 3D advertising model is placed on the corresponding plane in the video through AR technology. The video viewers can see the advertising model of the corresponding perspective according to the change of the perspective of the video. At the same time, combined with the user data of the video platform, personalized and targeted advertisements are presented to users. This embodiment can recommend different 3D advertising models at the same coordinate of the video, based on the user portraits obtained through big data in time, region, and different video platforms, to realize personalized recommendation of AR advertisements.

[0086] The advantages of the advertisement embedding method provided in this embodiment are as follows:

[0087] (1) Thanks to the video scene and theme analysis technology used, the massive videos on the video platform can be analyzed quickly and effectively, thereby effectively screening out videos that can use AR technology to embed advertisements, and selecting the best advertising positions in each video. This provides advertisers with more friendly selection suggestions for advertising investment. They can choose videos with corresponding scenes and themes based on their own advertising types, and negotiate with video creators to place advertisements.

[0088] (2) Thanks to EasyAR technology, the original camera input is replaced by network video stream input. Video creators do not have to worry about the advertising effect, and can process a large number of videos and integrate them into Android projects. Using gesture interaction and 3D advertising model interaction technology, users can interact with the advertising model and obtain more product information on the premise that they are interested in the product.

[0089] (3) Personalized advertising and precision marketing. Compared with current advertising types, there is only one advertisement for a type of video. However, by utilizing the big data model of the video platform itself, we can have a better analysis of users' consumption tendencies and show them products that they are more likely to consume in the video.

[0090] In this embodiment, Alibaba Cloud fully provides the required video analysis tasks, and the SDK provided by EasyAR is fully functional and can achieve AR effects. The current video platform's intelligent recommendation algorithm is mature, providing the premise of a personalized recommendation model, and the algorithm implementation platform has a complete and feasible operating process, so the advertising implantation method of this embodiment is feasible.

[0091] On the premise that the advertisement embedding method of this embodiment is feasible, the product advantages of this embodiment include:

[0092] (1) Allowing popular videos to generate secondary profits in the advertising market

[0093] Brands have a larger market. Brands always face the dilemma of being unable to predict the effect of advertising when placing advertisements. However, the Vision Advertising Platform uses a method of adding 3D models to the published videos, so that a large number of existing videos that originally lost the opportunity to monetize advertising traffic can return to the advertising market again, achieving a win-win situation for brands and creators. At the same time, this method of reprocessing the published videos is also conducive to the repeated investment promotion of videos. If the creator and the brand only sign a contract for a period of time, the creator can continue to find other brands to cooperate after the cooperation ends, so that the advertising benefits can continue and the traffic can be continuously utilized.

[0094] (2) Free cooperation model to reduce trial and error costs

[0095] Compared to the current model where big brands prefer to cooperate with video creators with a large number of fans, the Vision Advertising Platform allows brands to combine the characteristics of the ad content in the video that can be modified, invest a small amount of money in the video to test it, and then deepen cooperation after achieving good results. For example, a brand can first release 30,000 views on a certain video to observe the influence of the video and the revenue of the advertisement. If satisfied with the results, it can continue to release more views; if not satisfied, it can choose not to continue cooperation.

[0096] (3) Creators with small fan bases are more likely to get advertising opportunities

[0097] Creators have more opportunities, which is conducive to improving the quality of videos. Similarly, in response to the above-mentioned free cooperation model, the Vision Advertising Platform enables brands to shift their attention back to the video itself, so that creators with small fan bases can gain attention and obtain more advertising opportunities, so as to better survive in the video market. This model is also conducive to motivating creators to create works that are more recognized by viewers, and strive to obtain long-term influence of videos, so that the traffic value of videos can be continuously converted and reflected.

[0098] (4) Brands combine user portraits to accurately target users and reap high benefits

[0099] The video platform uses big data to select viewers for advertisements. The Vision Advertising Platform uses the big data model of the video platform itself to better analyze the consumption tendencies of users and display products that are more likely to be consumed in the video. For example, Brand A and Brand B respectively placed 30,000 and 40,000 ads in the same ad slot analyzed by the system in Video C. Then the video platform uses user data to analyze the product that the user currently watching Video C is more likely to choose between Brands A and B, and displays this product in the ad slot, and reduces the corresponding playback volume of the brand by one until the playback volume returns to 0, at which point the brand's cooperation with the video ends. This model allows brands to conduct precise advertising targeting the brand's user groups while investing the same amount of money.

[0100] (5) Interactive mode makes the product clearer

[0101] Vision Advertising Platform uses AR technology to allow video viewers to interact with 3D advertising models directly in the video, making the way of conveying information more three-dimensional and no longer limited by time. Viewers can click on the 3D model and use gestures to rotate, zoom in and out of the 3D model. This interactive mode is more interesting, intuitive, and three-dimensional.

[0102] The technical advantages of this embodiment are as follows:

[0103] (1) Computer vision: Video processing using scene recognition technology and plane recognition technology

[0104] Accurately classify and analyze videos. The Vision Advertising Platform uses scene analysis technology to judge the scenes in the video, and combines it with plane recognition technology to analyze the most suitable plane for placing 3D advertising models in the scene, and mark the advertising positions in the video. This technology can quickly and effectively analyze the massive amount of videos on the video platform, thereby effectively screening out videos that can use AR technology to insert advertisements, and selecting the best advertising positions in each video.

[0105] (2) Augmented reality: Using 2D image recognition and motion tracking technology, the model follows the position and perspective changes

[0106] Implicit ad placement allows 3D ads to blend into the scene. Compared with ads in the market where real objects appear in the video, or ads that are added to the video later using image processing technology, there are inevitably defects such as unmodifiable ad content and particularly abrupt 2D images. The Vision Advertising Platform uses the plane image recognition technology in the EasyAR framework to enable 3D models to appear in the corresponding position in the video according to their initial position. At the same time, using motion tracking technology, the model will present the content of the corresponding perspective according to the changes in the perspective of the video content. In other words, the image material added to the video later can be perfectly integrated into the video, and according to the light and shadow conditions and perspective differences in the video, it can show the same characteristics as the real object, while also having the possibility of modification.

[0107] (3) Augmented reality: Change the model based on the user portrait at the same coordinate in the video

[0108] Use AR to assist personalized presentation. Faced with the current situation of a large number of indiscriminate advertisements in the market, combined with the technology of the Vision Advertising Platform that can modify the advertising content, the Vision Advertising Platform proposes to select advertising products that users are more likely to consume based on the user information and big data technology owned by the video platform itself, so as to achieve the purpose of accurate advertising delivery. Combined with the current AR technology, different advertising models can be placed in the same advertising position marked in the video. Brands can place advertisements with a certain amount of playback volume on multiple videos based on the playback volume, and video creators can also accept orders from multiple brands in the same video, which allows brands to obtain greater profits with the same investment.

[0109] (4) Gesture recognition: Get more information by interacting with 3D advertising models through gestures

[0110] The Vision Advertising Platform uses gesture recognition technology to allow the 3D models in the video to better convey a larger amount of information. It is no longer affected by market restrictions within the video or language expression restrictions, which further demonstrates the superiority of AR technology in the advertising market.

[0111] Embodiment 2:

[0112] This embodiment is used to provide a personalized advertising implantation system for video platforms based on AR technology. Figure 5 As shown, the advertising implantation system includes:

[0113] The best advertisement position determination module M1 is used to obtain an uploaded video; perform scene recognition and plane recognition on the video to obtain a scene recognition result and a plane recognition result; and determine the best advertisement position in the video according to the scene recognition result and the plane recognition result;

[0114] The overlay module M2 is used to overlay a preset AR advertising model on the best advertising position of the video based on AR technology.

[0115] This embodiment can bring a higher sense of experience to users, achieve the marketing effect of advertisers and provide more revenue for video creators. It solves the problem of using AR technology to render video ads and integrates user recommendations from video platforms and advertising models to achieve personalized recommendations.

[0116] Each embodiment in this specification focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0117] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A personalized advertising implantation method based on AR technology. It is characterized in that The advertisement embedding method comprises: Get the uploaded video; Performing scene recognition and plane recognition on the video to obtain a scene recognition result and a plane recognition result; Determine the best advertisement position in the video according to the scene recognition result and the plane recognition result; Based on AR technology, a preset AR advertising model is superimposed on the best advertising position of the video; The preset AR advertising model is a 3D advertising model; There are multiple preset AR advertising models; different preset AR advertising models are superimposed on the optimal advertising position by using a time recommendation method, a region recommendation method or a user recommendation method; The time recommendation method is as follows: the advertiser selects a recommended time period to release the advertising model. When the audience watches the video, the server obtains the time information of the audience's viewing, divides the video playback into different time periods, and specifies the advertising models of different advertisers to be played in real time on the advertising position; the region recommendation method is as follows: the server obtains the user's location authority and the user's region information, and recommends the advertising model of the advertiser in the user's region according to the region information during non-buyout time, or the advertising model in the user's region specified by the advertiser; the user recommendation method is as follows: different user portraits of each video platform are connected to a model library composed of a large number of ImageTarget components, the ImageTracker script is modified, different user portraits of each video platform are collected, and the ImageTarget of the commodity type name under the same theme type is loaded for the video theme type data that the user often watches in the user portrait. When AR renders the video, the AR advertising model is loaded into the corresponding advertising position in the video according to the loaded corresponding ImageTarget and the predetermined position relationship with the advertising position set in Unity; Before obtaining the uploaded video, the advertisement embedding method further includes uploading the video, specifically including: Calling the video upload interface to upload the video to the storage warehouse, and receiving the video storage location and video access link returned by the video upload interface; also includes: opening the OSS object storage service, the smart label service and the smart label pipeline, the OSS object storage service is used as the storage warehouse for video viewers to upload videos, and a bucket for storing videos that need to be analyzed is created in the OSS object storage service; the smart label service provides an interface for video analysis tasks; the smart label pipeline is used as a job queue for video analysis, and provides the function of entering the pipeline to perform video analysis tasks in an orderly manner when concurrent video analysis tasks are submitted, that is, the submitted video analysis tasks are processed in an orderly manner through the pipeline queue; Before superimposing a preset AR advertising model on the best advertising position of the video based on AR technology, the advertising embedding method further includes determining the preset AR advertising model, specifically including: performing a topic analysis on the video to obtain a topic analysis result; classifying the video according to the scene recognition result and the topic analysis result to obtain a category of the video; and determining the preset AR advertising model according to the category of the video; The step of classifying the video according to the scene recognition result and the topic analysis result to obtain the category of the video specifically includes: The scene recognition results and topic analysis results are stored in the corresponding data table in the database as the attributes of the video submitted by the video creator, and the data results are returned to the web front end in the form of an interface. The web front end divides all the creators' videos by scene classification and topic classification; The method of superimposing a preset AR advertising model on the best advertising position of the video based on AR technology specifically includes: processing the video by using AR technology, tracking the best advertising position of the video by using a planar image recognition and motion tracking method, and loading the preset AR advertising model on the best advertising position to obtain a video superimposed with the preset AR advertising model, finally embedding the video superimposed with the preset AR advertising model into an Android project, accessing the AR function when playing the video, and superimposing the preset AR advertising model on the best advertising position in the video; the angle of the preset AR advertising model changes with the change of the viewing angle of the video; After superimposing a preset AR advertising model on the best advertising position of the video based on AR technology, the advertising implantation method further includes: recognizing the gestures of the video viewer, and adjusting the preset AR advertising model according to the gestures, wherein the adjustment includes zooming in, zooming out, and rotating, so that the video viewer can interact with the preset AR advertising model, thereby facilitating the video viewer to understand the advertising content.

2. The method for inserting advertisement according to claim 1, It is characterized in that The uploaded video is a video that has been published on the video platform.

3. The method for inserting advertisement according to claim 1, It is characterized in that The obtaining of the uploaded video specifically includes: obtaining the uploaded video according to the video storage location.

4. The method for inserting advertisement according to claim 1, It is characterized in that The method of superimposing a preset AR advertising model on the best advertising position of the video based on AR technology specifically includes: acquiring the video according to the video access link, and superimposing a preset AR advertising model on the best advertising position of the video based on AR technology.

5. A personalized advertising placement system based on AR technology, It is characterized in that The advertisement implantation system comprises: The best advertisement position determination module is used to obtain an uploaded video; perform scene recognition and plane recognition on the video to obtain a scene recognition result and a plane recognition result; and determine the best advertisement position in the video according to the scene recognition result and the plane recognition result; Before obtaining the uploaded video, the advertisement implantation system further includes uploading the video, specifically including: Calling the video upload interface to upload the video to the storage warehouse, and receiving the video storage location and video access link returned by the video upload interface; also includes: opening the OSS object storage service, the smart label service and the smart label pipeline, the OSS object storage service is used as the storage warehouse for video viewers to upload videos, and a bucket for storing videos that need to be analyzed is created in the OSS object storage service; the smart label service provides an interface for video analysis tasks; the smart label pipeline is used as a job queue for video analysis, and provides the function of entering the pipeline to perform video analysis tasks in an orderly manner when concurrent video analysis tasks are submitted, that is, the submitted video analysis tasks are processed in an orderly manner through the pipeline queue; An overlay module, used for overlaying a preset AR advertising model on the best advertising position of the video based on AR technology; The preset AR advertising model is a 3D advertising model; There are multiple preset AR advertising models; different preset AR advertising models are superimposed on the optimal advertising position by using a time recommendation method, a region recommendation method or a user recommendation method; The time recommendation method is as follows: the advertiser selects a recommended time period to release the advertising model. When the audience watches the video, the server obtains the time information of the audience's viewing, divides the video playback into different time periods, and specifies the advertising models of different advertisers to be played in real time on the advertising position; the region recommendation method is as follows: the server obtains the user's location authority and the user's region information, and recommends the advertising model of the advertiser in the user's region according to the region information during non-buyout time, or the advertising model in the user's region specified by the advertiser; the user recommendation method is as follows: different user portraits of each video platform are connected to a model library composed of a large number of ImageTarget components, the ImageTracker script is modified, different user portraits of each video platform are collected, and the ImageTarget of the commodity type name under the same theme type is loaded for the video theme type data that the user often watches in the user portrait. When AR renders the video, the AR advertising model is loaded into the corresponding advertising position in the video according to the loaded corresponding ImageTarget and the predetermined position relationship with the advertising position set in Unity; Before overlaying a preset AR advertising model on the best advertising position of the video based on AR technology, the method further includes determining the preset AR advertising model, specifically including: performing a topic analysis on the video to obtain a topic analysis result; classifying the video according to the scene recognition result and the topic analysis result to obtain a category of the video; and determining the preset AR advertising model according to the category of the video; The step of classifying the video according to the scene recognition result and the topic analysis result to obtain the category of the video specifically includes: The scene recognition results and topic analysis results are stored in the corresponding data table in the database as the attributes of the video submitted by the video creator, and the data results are returned to the web front end in the form of an interface. The web front end divides all the creators' videos by scene classification and topic classification; The method of superimposing a preset AR advertising model on the best advertising position of the video based on AR technology specifically includes: processing the video by using AR technology, tracking the best advertising position of the video by using a planar image recognition and motion tracking method, and loading the preset AR advertising model on the best advertising position to obtain a video superimposed with the preset AR advertising model, finally embedding the video superimposed with the preset AR advertising model into an Android project, accessing the AR function when playing the video, and superimposing the preset AR advertising model on the best advertising position in the video; the angle of the preset AR advertising model changes with the change of the viewing angle of the video; After superimposing a preset AR advertising model on the best advertising position of the video based on AR technology, it also includes: recognizing the gestures of the video viewer, and adjusting the preset AR advertising model according to the gestures, wherein the adjustment includes zooming in, zooming out, and rotating, so that the video viewer can interact with the preset AR advertising model, making it easier for the video viewer to understand the advertising content.

Images