System and method for advertising through electronic game by using AI

Abstract

The invention discloses a system and method for advertising through an electronic game by using AI, and relates to the technical field of electronic game advertising. The method comprises the following steps: acquiring an image of an interaction area to generate real-time video stream data; performing computer vision analysis on the video stream, extracting non-identity group features of the current interactive group, and forming feature data including dominant age interval labels; taking the label as a query key, and performing matching query according to a dynamic content mapping rule base to determine a target advertisement visual element identifier; positioning a replaceable texture storage area in the game resource package according to the identifier; and on the basis of visual data corresponding to the identifier, the texture content of the region is replaced by an asynchronous rendering updating technology, so that a dynamic advertisement matched with the characteristics of the current crowd is generated and presented in real time in a game picture. According to the method and the device, in an offline anonymous scene, the advertisement in the game is dynamically and accurately updated according to the group characteristic sensed in real time, and the game performance is not influenced.

Description

Technical Field

[0001] This application relates to the field of video game advertising technology, and in particular to a system and method for using AI to advertise through video games. Background Technology

[0002] As a popular digital medium, in-game advertising has become an important marketing channel. However, current methods of implementing in-game advertising face several technical limitations.

[0003] First, static pre-installed advertising solutions, where advertising content (such as images and textures) is directly embedded into specific models of the game scene (such as billboards and walls) during the game development phase, result in completely fixed advertising content after game release. This makes it impossible to adjust to changes in the actual offline environment, operating hours, or real-time audience, lacking flexibility and contextual relevance. Second, user profiling-based online advertising solutions, commonly found in online mobile games, rely on collecting and accumulating players' personal registration information and historical behavioral data to build user profiles and push personalized ads accordingly. However, this method is unsuitable for anonymous interactive game scenarios in offline public places (such as shopping malls, exhibition halls, and arcades) because users in these scenarios are mobile and anonymous, with no traceable personal accounts; forced collection of such data poses serious privacy compliance risks. Finally, simple carousel or random advertising solutions, which switch different advertising materials according to a fixed schedule or random algorithm during game operation, have no correlation between the advertising content switching and the characteristics of the current actual audience, leading to blind advertising, mismatch between content and audience interests, and low advertising efficiency.

[0004] In summary, there is currently a lack of a technical solution that can automatically and in real time perceive the group characteristics of anonymous audiences in offline public interactive game scenarios while strictly protecting personal privacy, and dynamically and seamlessly adjust in-game advertising content accordingly. How to achieve a complete technical loop from anonymous perception to content matching to real-time presentation has become the core technical challenge for improving the accuracy of advertising and user experience in such scenarios. Summary of the Invention

[0005] In order to overcome the above-mentioned deficiencies of the prior art, embodiments of this application provide a method for advertising through video games using AI to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, this application provides a method for advertising through video games using AI, comprising:

[0007] S1. Collect images of the interactive area and generate real-time video stream data;

[0008] S2. Perform computer vision analysis on the real-time video stream data, extract non-identity group characteristics of the current interactive crowd from the image, and form non-identity group characteristic data containing dominant age range labels;

[0009] S3. Use the advantageous age range tag as a query key and input it into the preset dynamic content mapping rule library for matching query. The dynamic content mapping rule library stores the corresponding mapping relationship between different advantageous age range tags and multiple sets of candidate advertising visual elements.

[0010] S4. Based on the matching query results, determine and output the target advertising visual element identifier associated with the advantageous age range tag;

[0011] S5. Based on the target advertising visual element identifier, locate the replaceable texture storage area in the resource pack of the video game;

[0012] S6. Based on the visual data corresponding to the target advertising visual element identifier, the texture content of the replaceable texture storage area is rendered and updated in real time, thereby generating and presenting dynamic advertising content that matches the advantageous age range label in the screen of the video game.

[0013] To address the aforementioned issues, this application also provides a system for advertising through video games using AI, the system comprising:

[0014] The image acquisition and video stream generation module is used to acquire images of the interactive area and generate real-time video stream data;

[0015] The group feature analysis module is used to perform computer vision analysis on the real-time video stream data, extract non-identified group features of the current interactive crowd from the image, and form non-identified group feature data containing dominant age range labels.

[0016] The dynamic content mapping query module is used to input the advantageous age range tag as a query key into a preset dynamic content mapping rule library for matching query. The dynamic content mapping rule library stores the corresponding mapping relationship between different advantageous age range tags and multiple sets of candidate advertising visual elements.

[0017] The advertising element output module is used to determine and output the target advertising visual element identifier associated with the advantageous age range tag based on the matching query results;

[0018] The game texture resource location module is used to locate the replaceable texture storage area in the resource package of the video game based on the target advertising visual element identifier.

[0019] The dynamic advertising rendering module is used to render and update the texture content of the replaceable texture storage area in real time based on the visual data corresponding to the visual element identifier of the target advertisement, thereby generating and presenting dynamic advertising content that matches the advantageous age range label in the screen of the video game.

[0020] Compared with the prior art, this application has the following beneficial effects:

[0021] First, this application constructs a complete technical closed loop from physical space perception to precise matching of virtual content, effectively solving the problem of ineffective advertising in anonymous offline scenarios. By deploying wide-angle cameras to collect images of interactive areas and using computer vision analysis technology to extract non-identified group characteristics (such as dominant age ranges), this method achieves real-time and compliant perception of the mainstream attributes of anonymous audiences. Furthermore, by matching feature tags with a pre-built dynamic content mapping rule library in the cloud, it can intelligently select the target content most relevant to the current audience from multiple candidate advertisements. This technical approach transforms the traditional "one-way broadcast" mode of advertising into an interactive mode of "contextual awareness-intelligent matching," enabling advertising content to adapt to different groups and significantly improving the accuracy and relevance of advertising.

[0022] Secondly, this application designs a highly efficient, stable, and non-disruptive dynamic advertising content update mechanism, resolving the conflict between dynamic advertising and game performance assurance. By pre-marking replaceable texture storage areas in the game resource package and establishing a two-layer mapping relationship between advertising element identifiers and game texture resources, rapid and accurate positioning of advertising materials is achieved. Most importantly, by triggering asynchronous texture update operations through the rendering component, the seamless replacement of advertising content is completed in the next frame rendering cycle without interrupting the game's main logic thread. This mechanism ensures real-time presentation and smooth switching of dynamic ads, completely avoiding game stuttering or screen tearing caused by ad updates. While achieving highly dynamic advertising content, it also ensures the smoothness and immersiveness of the game itself. Attached Figure Description

[0023] Figure 1 A flowchart illustrating a method for advertising through video games using AI, provided as an embodiment of this application;

[0024] Figure 2 A functional block diagram of a system for advertising through video games using AI, provided as an embodiment of this application;

[0025] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0026] It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit this application.

[0027] This application provides a method for advertising through video games using AI. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for advertising through video games using AI can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks, and big data and artificial intelligence platforms.

[0028] Reference Figure 1 The diagram shown is a flowchart illustrating a method for advertising through video games using AI, according to an embodiment of this application. In this embodiment, the method for advertising through video games using AI includes:

[0029] S1. Collect images of the interactive area and generate real-time video stream data.

[0030] In some embodiments, capturing images of the interactive area and generating real-time video stream data specifically includes:

[0031] The front area of ​​the multi-faceted transparent box structure of the interactive device is defined as the interactive area;

[0032] The interactive area is continuously captured at a preset frame rate using a wide-angle camera integrated on the top of the housing structure.

[0033] The captured raw image sequence is format-encoded and compressed to generate the real-time video stream data.

[0034] In this embodiment, the interactive area is a clearly defined physical space in front of the video game device, which is the place where the target audience gathers and watches or interacts; the real-time video stream data is a digital data stream containing time-series image frames, which is continuously collected by an image sensor and encoded.

[0035] In this embodiment, this step is the starting point for perceiving the real-world offline audience environment. Specific technical means include clear hardware deployment, physical space definition, and standardized data processing procedures. First, in terms of hardware deployment, the video game device is concretized as an interactive device with a multi-faceted transparent box structure, such as a large interactive display window or experience cabinet in a shopping mall. A wide-angle camera integrated on the top of this box structure is used as the core image acquisition sensor. The wide-angle camera is chosen to ensure it can cover the widest possible field of view in front of the device, thereby capturing a more comprehensive picture of the interactive crowd. Second, in terms of space definition, the fan-shaped or rectangular space directly in front of the multi-faceted transparent box structure is clearly defined as the interactive area. This limits the effective image processing range for subsequent computer vision analysis and avoids interference from irrelevant backgrounds.

[0036] In this embodiment, during the data generation process, the technical implementation follows a standard video acquisition and processing pipeline. The wide-angle camera continuously captures images at a preset frame rate, such as 30 frames per second. This frame rate balances video smoothness and system processing load. The raw signal output by the camera is a series of continuous, uncompressed raw image sequences. Next, the system needs to perform format encoding and compression on this raw image sequence. This is a crucial step in generating real-time video stream data that can be efficiently transmitted and processed in the system. For example, the system calls a video encoding library to compress the raw RGB or YUV format image sequence using efficient video encoding standards such as H.264 or H.265. This compression significantly reduces the data volume, ensuring that the video stream data can be analyzed and transmitted in real-time or near real-time in subsequent steps without causing system delays or blockages due to excessive data volume. Finally, the encoded and compressed data is encapsulated into continuous real-time video stream data that can be read and processed by computer programs.

[0037] In this embodiment, the technical effect of this step is to provide a unique and essential real-world data input source for the entire method. It successfully transforms the anonymous presence and dynamics of crowds in physical space into standardized digital video signals that can be analyzed and processed by a computer system, thus laying the data foundation for subsequent AI-based group characteristic analysis and solving the problem that traditional methods cannot obtain real-time audience information in offline scenarios.

[0038] S2. Perform computer vision analysis on the real-time video stream data, extract non-identity group characteristics of the current interactive crowd from the image, and form non-identity group characteristic data containing dominant age range labels.

[0039] In some embodiments, computer vision analysis is performed on the real-time video stream data to extract non-identified group characteristics of the currently interacting crowd from the image, forming non-identified group characteristic data containing dominant age range labels, specifically including:

[0040] Face detection is performed on consecutive image frames of the real-time video stream data to obtain a set of face position coordinates in each frame;

[0041] Based on the set of face location coordinates, age attribute recognition is performed on each detected face image region to obtain the corresponding estimated age value.

[0042] Statistical distribution analysis is performed on all estimated age values ​​obtained within a continuous preset time period, and the age range with the highest frequency is determined as the dominant age range label.

[0043] The dominant age range labels are combined with statistically obtained population density information and average stay time information to form the non-identity-based group characteristic data.

[0044] This step is the core processing stage of the method, which transforms the raw visual data stream into structured labeled data representing the attributes of the audience group through a series of specific computer vision and statistical algorithms. This implementation process includes a sequentially executed algorithmic pipeline: first, object detection; then, attribute prediction; and finally, data aggregation and statistics. Specifically, the system first performs frame-by-frame face detection on the real-time video stream data from S1. This process is achieved by loading and running a pre-trained face detection model, such as a single-stage detector based on a convolutional neural network. This model infers for each frame of the image and outputs a set containing multiple bounding box coordinates, i.e., a set of face location coordinates. Each bounding box precisely defines a region in the image that is suspected to be a face. This step effectively filters out background interference and locks onto the target region for subsequent analysis.

[0045] After successfully locating a face, the system immediately performs age attribute identification on each detected face image region. The technical approach involves calling another pre-trained deep neural network model, such as a convolutional neural network classifier or regression model specifically for age estimation. This model receives the cropped face region image as input, performs forward propagation calculations, and outputs a predicted age value for the face. It is important to emphasize that this model only predicts age ranges and does not perform any personal identification. This ensures the "non-identity" of feature extraction. For example, the model might predict a face as "25 years old" or categorize it into the "18-25 years old" range.

[0046] Next, the system enters the crucial statistical decision-making stage, which involves statistically analyzing the distribution of all estimated age values ​​obtained within a consecutive preset time period. Here, the "continuous preset time period" is a configurable sliding time window, such as 30 seconds. The system maintains a queue of samples that have successfully estimated age within this time window. The core of the statistical analysis of these samples is to calculate the frequency of each age range. One way to achieve this is to set fixed age ranges (such as every 5 or 10 years as a range), and then iterate through all estimated age values ​​to count the number of samples falling into each range. Subsequently, the system executes a decision logic to determine the age range with the highest statistical frequency as the dominant age range label within this time window. For example, within a 30-second window, if the statistics show that the number of samples in the "18-25 years old" range is significantly more than other ranges such as "26-35 years old" and "36-45 years old", then the system determines "18-25 years old" as the current dominant age range label. This mechanism based on sliding windows and frequency statistics allows the labels output by the system to smoothly and stably reflect changes in the mainstream characteristics of the population, avoiding frequent flickering of advertising content due to misjudgment of a single frame image or the brief appearance of a few individuals.

[0047] Finally, to generate richer non-identified group characteristic data, the system combines the aforementioned dominant age range labels with other group statistics. This information can be derived from the same processing flow. Crowd density information can be approximated by counting the total number of faces detected in the current frame or time window, while average dwell time information can be estimated by tracking the duration of specific face bounding boxes appearing in consecutive frames and averaging across multiple tracking targets. The system packages the dominant age range labels, the calculated crowd density value, and the average dwell time value into a complete non-identified group characteristic data record. This data record describes the anonymity characteristics of the currently interacting crowd from multiple perspectives, providing a more comprehensive decision-making basis for subsequent accurate content mapping.

[0048] This step successfully distilled valuable group business insights from anonymous video streams. Through purely visual analysis, it identifies group characteristics such as the dominant age group of the audience in real time without touching any personal privacy information. It fundamentally solves the dual dilemma of traditional offline advertising being unable to perceive the audience and online advertising relying on personal identity data, providing a legal, compliant and feasible data foundation for dynamic personalized advertising.

[0049] This step serves as a bridge between initial perception and intelligent decision-making. It directly receives and processes real-time video stream data from S1, and its output, including non-identity-based group characteristic data with dominant age range labels, is directly passed to the subsequent S3 step as the core query basis and decision input. This data is used for matching queries in the rule base, thereby driving the entire dynamic replacement of advertising content.

[0050] In some embodiments, statistical distribution analysis is performed on all estimated age values ​​obtained within a continuous preset time period, and the age range with the highest frequency is determined as the dominant age interval label, specifically including:

[0051] Set a sliding time window with a length of N seconds, and continuously count the estimated age values ​​of all successfully recognized faces within the sliding time window;

[0052] Cluster analysis is performed on the estimated age values ​​within the sliding time window to calculate the frequency distribution of each age interval;

[0053] The age range with the highest frequency exceeding the preset frequency threshold is marked as the dominant age range label at the current moment;

[0054] As the sliding time window slides, the advantageous age range label is updated in real time.

[0055] A sliding time window is a time range used for continuous analysis of streaming data. The window has a fixed length and slides forward over time. Cluster analysis is a method to divide samples in a dataset into different groups, so that samples within the same group have high similarity. Frequency distribution is a statistical result that describes how often different categories or numerical intervals appear in a dataset. A preset frequency threshold is a pre-set value used to determine whether the frequency of a certain category is significant enough.

[0056] This sub-step is a key algorithmic decision-making process for achieving stable and interference-resistant output of dominant age range labels. Its core technical approach lies in designing a statistical decision-making mechanism that combines time smoothing and significance testing. First, the system sets up and maintains a data structure in memory to simulate a sliding time window of length N seconds (e.g., N is set to thirty seconds). This sliding time window can be implemented as a first-in, first-out (FIFO) circular queue. Each element in the queue corresponds to a successful face age recognition event, including its timestamp and estimated age value. The system continuously fills this queue with estimated age values ​​bearing timestamps from previous age attribute recognition steps, while removing older data whose timestamps are earlier than the current time minus N seconds. This ensures that the queue always retains only valid samples from the most recent N seconds. This data structure design ensures that the system always analyzes based on the latest historical data, enabling real-time tracking of changes in population characteristics.

[0057] Next, the system performs cluster analysis on all estimated age values ​​stored in the queue within the current sliding time window to calculate the frequency distribution. This cluster analysis does not refer to complex unsupervised clustering algorithms, but rather to a series of predefined continuous age intervals based on business logic, such as "[0-12]", "[13-17]", "[18-25]", "[26-35]", etc. The system iterates through each estimated age value in the sliding window queue, determines which predefined age interval it falls into, and increments the counter for that interval. After the iteration is complete, each age interval corresponds to a count value. The set of these count values ​​constitutes the frequency distribution of each age interval. For example, within a 30-second window, the system might count 50 times for the "[18-25]" interval, 30 times for the "[26-35]" interval, and a total of 20 times for the other intervals.

[0058] Subsequently, the system executes a creative decision-making logic: instead of simply outputting the highest-frequency interval directly, it introduces a preset frequency threshold for filtering. This preset frequency threshold can be set as an absolute number or a relative proportion, for example, requiring the target interval's frequency to exceed 30% of the total valid samples. The system first filters out all age intervals whose frequency exceeds the preset threshold from the frequency distribution, and then selects the interval with the highest frequency value from these eligible intervals. For example, if the total number of samples is 100 and the preset threshold ratio is 30% (i.e., 30 times), then only the intervals with more than 30 counts (such as "[18-25]": 50 times, "[26-35]": 30 times) will enter the candidates. In the end, the "[18-25]" interval is selected because it has the largest count. The technical significance of this dual judgment mechanism (exceeding the threshold and having the largest value) is that it ensures that the final determined dominant age interval label is not only the most common, but also that its representativeness is statistically significant. This effectively avoids the misjudgment of an age interval that is only slightly higher than other intervals as "dominant" when the population is sparse or the age distribution is extremely dispersed, thereby improving the robustness and accuracy of label decision-making.

[0059] Finally, as time passes and new data continuously enters, the sliding time window continues to slide forward, and the aforementioned statistical, filtering, and selection processes are repeatedly executed periodically or triggered. Each execution generates a new frequency distribution based on the latest N-second data and applies the same decision logic, thereby achieving real-time updates to the dominant age range labels. This dynamic update mechanism allows the system to smoothly respond to real changes in population composition. When the mainstream population changes from young people to middle-aged people, the labels output by the system will also switch stably after a short time lag (at most N seconds), rather than experiencing drastic jumps.

[0060] This sub-step endows the system with stable, reliable, and statistically significant group feature identification capabilities. It smooths out instantaneous fluctuations through a sliding time window and filters out insignificant features through preset frequency thresholds, thereby ensuring that the output dominant age range labels accurately and stably reflect the mainstream composition of the current interactive audience. This solves the technical problems of simple real-time analysis being easily affected by instantaneous interference and resulting in unstable output results, laying a solid foundation for the stable and effective matching of subsequent advertising content.

[0061] This sub-step relies heavily on the continuous stream of estimated age values ​​provided by the preceding steps as input. Its stable output of advantageous age range labels is the most critical part of constituting complete non-identity-based group characteristic data. The stability and accuracy of these labels directly determine the effectiveness of content mapping queries in the subsequent S3 step, and are the decision-making cornerstone for the entire method to achieve precise dynamic advertising.

[0062] S3. Use the advantageous age range tag as a query key and input it into a preset dynamic content mapping rule library for matching query. The dynamic content mapping rule library stores the corresponding mapping relationship between different advantageous age range tags and multiple sets of candidate advertising visual elements.

[0063] In some embodiments, the advantageous age range tag is used as a query key and input into a preset dynamic content mapping rule base for matching queries. The dynamic content mapping rule base stores the corresponding mapping relationships between different advantageous age range tags and multiple sets of candidate advertising visual elements, specifically including:

[0064] The dynamic content mapping rule base is pre-installed in the cloud server and contains multiple mapping rule entries. Each mapping rule entry records the correspondence between an advantageous age range tag and a set of candidate advertising visual elements. Each set of candidate advertising visual elements contains multiple advertising materials of different styles.

[0065] The execution of the matching query includes: sending the advantageous age range label to the cloud server via a wireless network, and the cloud server retrieving matching mapping rule entries from the dynamic content mapping rule base it stores, and returning the search results.

[0066] The dominant age range label is an identifier that represents the dominant age group in the current interactive audience; the pre-built dynamic content mapping rule base is a data set that is pre-built and stores the correspondence between different audience characteristics and corresponding advertising content; the candidate advertising visual elements are a set of pre-prepared advertising materials used to match specific audience characteristics.

[0067] This step is a crucial transition from "audience feature recognition" to "advertising content decision-making," and its core technology lies in a pre-built, structured rule query system. The implementation of this step begins with a clearly defined data structure: the dynamic content mapping rule base. This rule base is physically pre-built and runs on a cloud server. This location choice allows the rule base to be centrally managed, uniformly updated, and shared and accessed by multiple interactive gaming devices distributed across different geographical locations. The rule base consists of multiple mapping rule entries, each logically representing a clearly defined correspondence record.

[0068] Specifically, each mapping rule entry precisely records the binding relationship between a specific advantageous age range label and a set of candidate ad visual elements. For example, a rule entry can map the label "18-25" to a set of candidate materials including "sports drink ad A", "fashion brand ad B", and "video game trailer C". This design of "a set of candidate ad visual elements" is important because it provides multiple alternative ads with different styles or themes for the same audience, thereby avoiding the monotony and repetition of ad content and providing the possibility for selection or rotation strategies in subsequent steps.

[0069] The execution process of a matching query is a standard client-server interaction. After the local device (such as an interactive gaming device) generates the current dominant age range label in step S2, the local system initiates a network request. This request involves using the dominant age range label as the query keyword, encapsulated in a data packet via a wireless network module (such as Wi-Fi or 5G), and sending it to the designated cloud server. Upon receiving the query request, the cloud server parses the query key and uses it as an index to search its stored dynamic content mapping rule base. This search can be performed using SQL queries in a relational database or by directly looking up a key-value pair database. The goal is to find the mapping rule entry whose key value exactly matches the input query key. Once a matching entry is found, the cloud server uses the relevant information of "a set of candidate ad visual elements" stored in that entry (e.g., a list of unique identifiers for this set of materials) as the search result and sends it back to the local device via wireless network. This architecture, which deploys the rule base in the cloud and queries it over the network, is a creative design of this step. It allows adjustments to ad placement strategies and updates to ad materials to be made without touching thousands of offline terminal devices. The changes can be made globally simply by modifying the rule base on the cloud server, greatly improving the maintainability, flexibility, and scalability of the system.

[0070] The technical effect of this step is the establishment of a highly efficient and centrally controllable content mapping hub. By querying a pre-set rule base, it instantly transforms abstract group characteristic tags into specific and usable advertising material options, solving the fundamental problem that traditional in-game advertising content is fixed and cannot match the ever-changing audience. This provides core rule-driven capabilities for achieving dynamic and personalized advertising content.

[0071] This step is the decision-making hub in the data processing flow. It receives the dominant age range tags representing the characteristics of the current audience from step S2 as input, performs matching queries on the cloud rule base, and outputs a structured search result, namely the information of candidate ad creative groups that match the current audience. This search result is directly passed to the subsequent step S4 as the basis for the final selection of ad elements, thus closely connecting the feature analysis stage with the content rendering stage.

[0072] S4. Based on the matching query results, determine and output the target advertising visual element identifier associated with the advantageous age range label.

[0073] In some embodiments, based on the matching query results, the target advertising visual element identifier associated with the advantageous age range tag is determined and output, specifically including:

[0074] Based on the search results, obtain a group of candidate advertising visual elements associated with the advantageous age range tag;

[0075] When the candidate ad visual element group contains multiple alternative elements, a target ad visual element is determined from the multiple alternative elements;

[0076] The unique code of the identified target advertising visual element in the dynamic content mapping rule base is output as the target advertising visual element identifier.

[0077] A target ad visual element identifier is a string or code used to uniquely identify the specific ad creative that is ultimately selected and will be displayed.

[0078] This step is crucial for making the final decision on advertising content. Its core technology lies in designing a decision-making logic to determine a unique target from a candidate set. This logic ensures the diversity and rationality of ad placement. First, the system receives and parses the matching query results from step S3, i.e., the search results returned by the cloud server. Based on these results, the system can clearly obtain a group of candidate ad visual elements precisely associated with the current advantageous age range tag. This group is essentially an ordered or unordered list of ad creative identifiers. For example, for the tag "18-25", the search results might return a list containing three identifiers: "adv_sports_001", "adv_music_002", and "adv_tech_003".

[0079] Next, the system needs to determine the number of candidate elements within the candidate ad visual element group, which is the starting point for branching logic. When the system detects that the group contains multiple candidate elements, it must execute a deterministic selection algorithm to filter out a target ad visual element. This selection process is not random but can be implemented based on various preset strategies. A common implementation is a carousel strategy, where the system maintains a carousel index counter locally for each advantageous age range tag. Each time a selection is needed, the corresponding element identifier is retrieved from the candidate list based on the current index, and then the index is incremented (reset if it exceeds the list length). For example, for the above list, the first selection is "adv_sports_001", the second is "adv_music_002", and so on. Another implementation is a weighted random strategy, where the system presets a weight value for each element in the candidate list. This weight can be calculated based on factors such as the advertiser's delivery priority, the freshness of the creative, or historical click performance. During selection, the system calculates the probability distribution based on the weight of each element and performs a random sampling to determine the final target. The implementation of these strategies ensures that, under the same audience characteristics, advertising content can be varied in an orderly or probabilistic manner, effectively avoiding the repetitive fatigue of displaying a single advertisement and improving the coverage of advertising reach and user experience.

[0080] After identifying a specific target ad visual element using the aforementioned strategies, the system needs to generate a clear output instruction. This target ad visual element has a pre-assigned unique code in the cloud-based dynamic content mapping rule base. The system extracts this unique code as the final output of this step: the target ad visual element identifier. This identifier is a crucial data token that uniquely points to a specific ad creative file (such as an image or video file) stored in the cloud or locally, providing a precise index for resource location and loading in subsequent steps.

[0081] This step enables more refined and controllable advertising decisions. By introducing intelligent selection strategies into the matched candidate content, it ensures that the final ads are not only relevant to group characteristics but also optimized for display under the guidance of these strategies. This solves the technical problems of monotonous and repetitive ads and simplistic delivery strategies that may result from simple mapping, thereby improving the overall effectiveness and user acceptance of advertising campaigns.

[0082] This step is the end of the content decision chain and the beginning of the execution chain. It directly receives and processes the retrieval results from step S3. Its output, the target ad visual element identifier, is a crucial piece of data that connects the preceding and following steps. This identifier concludes the content matching and decision-making process based on audience characteristics and is then sent to steps S5 and S6 as the sole input for the next stage of resource positioning and rendering updates, thereby driving the physical rendering action.

[0083] S5. Based on the target advertising visual element identifier, locate the replaceable texture storage area in the resource pack of the video game.

[0084] In some embodiments, locating a replaceable texture storage area in the resource pack of a video game based on the target advertising visual element identifier specifically includes:

[0085] Assign a unique texture resource identifier to each pre-tagged replaceable advertisement display location, and establish an address mapping relationship between each texture resource identifier and the physical storage address in the resource package of the video game;

[0086] A mapping relationship between the target advertisement visual element identifier and the texture resource identifier is established in advance;

[0087] Based on the visual element identifier of the target advertisement, query the element mapping relationship to obtain the corresponding texture resource identifier;

[0088] The address mapping relationship is queried based on the texture resource identifier to obtain the corresponding physical storage address, thereby locating the replaceable texture storage area.

[0089] Replaceable texture storage areas are specific data blocks reserved in runtime memory or resource files by video game applications. The texture image data stored in this area is designed to be dynamically replaced during game execution. Texture resource identifiers are strings or integer codes used internally by the game engine to uniquely identify and manage a texture resource. Address mapping is a data structure that records the correspondence between logical identifiers and the specific physical locations of resources in memory or storage media. Element mapping is another data structure that records the correspondence between ad content identifiers and in-game texture resource identifiers.

[0090] This step is the core of implementing "physical addressing" for virtual billboards. The technical means lies in building and utilizing a two-layer mapping query system to precisely link abstract advertising decisions to resource locations within the game rendering engine. This implementation relies on pre-planning during game development and rapid querying during system runtime. First, during game development and resource packaging, engineers pre-mark alternative advertising display locations in the game scene, such as billboards on virtual buildings, banners in stadiums, or paint schemes on vehicles. For each such display location, the game engine assigns it a unique texture resource identifier, such as "AdBoard_Texture_01". Simultaneously, the engine establishes and maintains an address mapping table. This table uses the texture resource identifier as the key and the physical storage address (such as memory offset or file path) of the texture resource within the game resource package as the value. For example, the table records that "AdBoard_Texture_01" maps to the memory address pointer "0x7FAA1100".

[0091] Secondly, the system needs to establish an independent element mapping relationship in advance. This mapping relationship logically connects the advertising content management system and the game resource management system. Specifically, it maps the target advertising visual element identifier (e.g., "adv_energy_drink_002") from the cloud rule base to the specific texture resource identifier (e.g., "AdBoard_Texture_01") in the game. This mapping relationship can be stored in a local configuration file or database, and it defines "which advertisement should be displayed on which billboard".

[0092] When the system runs, the localization process begins after receiving the target ad visual element identifier from step S4. The first step is to query the element mapping relationship. The system uses the received target ad visual element identifier as the query key to search the element mapping relationship table and retrieve the corresponding texture resource identifier. For example, inputting "adv_energy_drink_002" will retrieve "AdBoard_Texture_01". Next, the system performs the second query, which is to query the address mapping relationship based on the texture resource identifier just obtained. The system uses "AdBoard_Texture_01" as the key to search the address mapping relationship table maintained by the game engine and finally obtains the specific physical storage address of the texture resource in memory, such as the pointer "0x7FAA1100". This physical storage address is the finally located replaceable texture storage area. This two-layer mapping architecture decouples the ad content logic and game rendering resource management, allowing the allocation and changes of ad materials (by updating the element mapping relationship) to be independent of the game program itself, greatly enhancing the system's flexibility and configurability.

[0093] The technical effect of this step is to establish an efficient and precise positioning channel from advertising decisions to rendering resources. Through a pre-established two-layer mapping relationship, it quickly converts the advertising identifiers of the business layer into physical memory addresses that the underlying rendering system can directly manipulate. This solves the core technical challenge of how dynamic content can accurately replace specific areas in a static game screen, providing the necessary addressing mechanism for real-time, lossless advertising content updates.

[0094] This step is the linchpin connecting ad content decisions and graphics rendering execution. It receives the target ad visual element identifier from step S4 as input, performs a two-stage mapping query, and outputs a specific physical storage address. This physical storage address is the direct object of operation for the subsequent real-time texture rendering update in step S6, ensuring that the correct ad content data can be accurately filled into the predetermined position in the game screen.

[0095] S6. Based on the visual data corresponding to the target advertising visual element identifier, the texture content of the replaceable texture storage area is rendered and updated in real time, thereby generating and presenting dynamic advertising content that matches the advantageous age range label in the screen of the video game.

[0096] In some embodiments, based on the visual data corresponding to the target advertising visual element identifier, the texture content of the replaceable texture storage area is rendered and updated in real time, thereby generating and presenting dynamic advertising content matching the advantageous age range label in the screen of the video game, specifically including:

[0097] Obtain the corresponding visual data file based on the target advertisement visual element identifier;

[0098] The visual data file is parsed into a texture map format to obtain the target texture data;

[0099] The target texture data is updated to the replaceable texture storage area by the rendering component of the video game;

[0100] Without interrupting the main logic thread of the game, the rendering component uses the target texture data to render the screen, thereby generating and presenting the dynamic advertising content that matches the advantageous age range label on the display screen of the video game.

[0101] This step is the final execution stage of the entire method. Its technical approach aims to seamlessly, efficiently, and without impacting user experience, integrate and present the decided advertising content within the game screen, completing a closed loop from data analysis to visual presentation. This implementation process is a standard graphics resource loading and rendering pipeline. First, the system needs to obtain the corresponding entity visual data file based on the target advertising visual element identifier obtained from step S4. Technically, the system maintains an advertising material resource library locally or in the cloud, where each material file is named or indexed by its unique identifier. For example, based on the identifier "adv_energy_drink_002", the system locates and loads the image file named "energy_drink_002.png" into the memory buffer via a file system interface or network download interface.

[0102] Next, the system must parse the acquired raw visual data file into a texture map format that the graphics rendering engine can directly use. This process is accomplished by calling image decoding libraries. For example, the system uses the libpng or stb_image library to decode the PNG format file data in memory, converting it from a compressed byte stream into a two-dimensional array containing RGB or RGBA pixel data. This data is the target texture data that can be directly processed by the graphics card. The decoding process may also include preprocessing operations such as color space conversion and resolution scaling (to match the UV coordinates of the in-game billboard model).

[0103] Subsequently, the system performs the core texture update operation through the game's rendering component. The rendering component is the low-level module in the game engine responsible for interacting with graphics APIs (such as OpenGL, DirectX, or Vulkan). The system passes two key parameters to this component: one is the physical storage address of the replaceable texture storage area located in step S5 (such as the texture object ID in OpenGL), and the other is a pointer to the parsed target texture data in memory. The rendering component then calls the texture update function of the graphics API, such as OpenGL's glTexSubImage2D, to asynchronously upload the target texture data to the graphics card's video memory, replacing the original pixel data in the specified texture object. This "asynchronous" characteristic is crucial, meaning that after the data upload command is submitted to the GPU, the main game logic thread on the CPU (responsible for gameplay, physics simulation, etc.) can continue execution without waiting for the upload to complete, thus ensuring the smoothness of the game and avoiding stuttering caused by ad updates.

[0104] Finally, without interrupting the main game logic thread, the rendering component uses the newly updated texture data in the video memory to draw the screen in the next frame or the most recent rendering cycle. When the game engine draws the billboard model that is bound to the replaceable texture, the graphics card samples pixels from the updated texture, thereby generating and presenting dynamic advertising content that precisely matches the current dominant age group label on the final synthesized video game display screen. For example, when the dominant group is young people, the billboard renders an advertisement for sports drinks in real time, realizing the real-time and dynamic response of virtual world advertising content to the composition of the real world population.

[0105] This step enables the final visualization of dynamically updated ad content without affecting the core gameplay experience. Through efficient resource loading, asynchronous graphics API calls, and deep integration with the game rendering loop, it ensures that new ad content can smoothly and instantly replace old content. This ultimately solves the core technical problem of traditional in-game ads being static and unable to adapt to changes in scenarios and audiences, achieving real-time and context-relevant ad delivery.

[0106] This step marks the end of the data flow and the point of physical effect realization in the method. It directly receives the target advertising visual element identifier from step S4 to obtain data, and receives the address of the replaceable texture storage area located from step S5 to determine the update location. By performing the rendering update operation in this step, it transforms the analysis, decision-making, and addressing results of all previous steps into a dynamic advertising screen that is visible to the user and matches the target age range tag, completing a complete technical loop from physical perception to virtual presentation.

[0107] In some embodiments, updating the target texture data to the replaceable texture storage area via the rendering component of the video game specifically includes:

[0108] Pass the memory address pointer of the replaceable texture storage area and the physical storage address of the target texture data to the rendering component;

[0109] An asynchronous texture update operation is triggered by the rendering component, so that the texture binding state of the replaceable texture storage area is updated to point to the target texture data;

[0110] During the rendering of the next frame of the video game, a replaceable texture storage area is drawn using the target texture data it points to.

[0111] This sub-step details the core technology for achieving high-performance dynamic texture replacement. It specifically describes how the rendering component works in conjunction with the graphics processor to update texture data without affecting game smoothness. First, the system needs to prepare and pass in the necessary parameters to initiate the update operation. These parameters include two key addresses: one is a pointer to the memory address of the replaceable texture storage area in the graphics processor's video memory, which was obtained in step S5 through address mapping, such as an OpenGL texture object ID; the other is the physical storage address of the target texture data in the system's main memory, pointing to the RGB pixel array obtained after the pre-parse step S6. The system passes these two address information to the game's rendering component through function calls.

[0112] Subsequently, the rendering component triggers an asynchronous texture update operation, which is the most innovative technical step in this process. Specifically, the rendering component calls a specific function provided by the underlying graphics API, such as glTexSubImage2D in OpenGL or UpdateSubresource in DirectX. The key is that this function call is asynchronous. After the CPU issues this update instruction, the instruction is placed in the GPU's command queue. The CPU can immediately return without waiting for the GPU to actually complete the transfer and filling of texture data, and continue executing other tasks in the game's main logic thread, such as physics simulation and game state updates. This asynchronous mechanism ensures the stability of the game's frame rate. The result of this instruction is that the texture binding state of the replaceable texture storage area on the GPU is atomically updated to point to the newly passed target texture data. This means that the GPU will read pixels from this new data source during subsequent sampling.

[0113] Finally, during the rendering of the next frame of the video game, the rendering system uses the updated binding state to draw the image. When the rendering pipeline processes a graphical element (such as a billboard model) that requires the use of the replaceable texture storage area, the GPU's texture sampling unit reads the updated target texture data residing in video memory and applies it to the corresponding model surface, thereby completing the visual update of the replaceable texture storage area in the final output image. This process is completely synchronized with the game's main rendering loop, so the changes in advertising content perceived by the player are smooth and instantaneous, without any screen tearing or logical interruptions.

[0114] This step achieves zero-latency dynamic ad content switching and ensures absolute smoothness of the core game experience. It decouples the time-consuming texture upload operation from the CPU main thread through asynchronous graphics instructions, fundamentally solving the technical problem that dynamic content updates may cause game frame rate drops, stuttering or freezing, so that the dynamic characteristics of ads will not cause any performance interference to the main functions of the video game.

[0115] This sub-step is a precise graphical implementation of the "real-time rendering update" action in step S6. It directly relies on the texture region pointers and texture data addresses provided by the preceding steps, translating these inputs into a specific, efficient GPU operation. Its successful execution is the direct technical reason why "dynamic advertising content" is accurately rendered "in the next frame's rendering process," and it is the final execution step that translates data update instructions into the final screen pixels.

[0116] like Figure 2 The diagram shown is a functional block diagram of a system that uses AI to advertise through video games, according to an embodiment of this application.

[0117] The system 100 described in this application, which uses AI to advertise through video games, can be installed in an electronic device. Depending on its functionality, the system 100 may include an image acquisition and video stream generation module 101, a group feature analysis module 102, a dynamic content mapping query module 103, an advertising element output module 104, a game texture resource positioning module 105, and a dynamic advertising rendering module 106. The modules described in this application can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0118] In this embodiment, the functions of each module / unit are as follows:

[0119] The image acquisition and video stream generation module 101 is used to acquire images of the interactive area and generate real-time video stream data.

[0120] The group feature analysis module 102 is used to perform computer vision analysis on the real-time video stream data, extract non-identity group features of the current interactive crowd from the image, and form non-identity group feature data containing dominant age range labels.

[0121] The dynamic content mapping query module 103 is used to input the advantageous age range tag as a query key into a preset dynamic content mapping rule library for matching query. The dynamic content mapping rule library stores the corresponding mapping relationship between different advantageous age range tags and multiple sets of candidate advertising visual elements.

[0122] The advertising element output module 104 is used to determine and output the target advertising visual element identifier associated with the advantageous age range tag based on the matching query results;

[0123] The game texture resource positioning module 105 is used to locate the replaceable texture storage area in the resource package of the video game according to the target advertising visual element identifier.

[0124] The dynamic advertising rendering module 106 is used to render and update the texture content of the replaceable texture storage area in real time based on the visual data corresponding to the target advertising visual element identifier, thereby generating and presenting dynamic advertising content that matches the advantageous age range label in the screen of the video game.

[0125] In the embodiments provided in this application, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0126] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0128] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application.

[0129] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of this application without departing from the spirit and scope of the technical solutions of this application.

Claims

1. A method for advertising through video games using AI, characterized in that, The method includes: S1. Collect images of the interactive area and generate real-time video stream data; S2. Perform computer vision analysis on the real-time video stream data, extract non-identity group characteristics of the current interactive crowd from the image, and form non-identity group characteristic data containing dominant age range labels; S3. Use the advantageous age range tag as a query key and input it into the preset dynamic content mapping rule library for matching query. The dynamic content mapping rule library stores the corresponding mapping relationship between different advantageous age range tags and multiple sets of candidate advertising visual elements. S4. Based on the matching query results, determine and output the target advertising visual element identifier associated with the advantageous age range tag; S5. Based on the target advertising visual element identifier, locate the replaceable texture storage area in the resource pack of the video game; S6. Based on the visual data corresponding to the target advertising visual element identifier, the texture content of the replaceable texture storage area is rendered and updated in real time, thereby generating and presenting dynamic advertising content that matches the advantageous age range label in the screen of the video game.

2. The method for advertising through video games using AI as described in claim 1, characterized in that, Capture images from the interactive area and generate real-time video stream data, specifically including: The front area of ​​the multi-faceted transparent box structure of the interactive device is defined as the interactive area; The interactive area is continuously captured at a preset frame rate using a wide-angle camera integrated on the top of the housing structure. The captured raw image sequence is format-encoded and compressed to generate the real-time video stream data.

3. The method for advertising through video games using AI as described in claim 1, characterized in that, The real-time video stream data is subjected to computer vision analysis to extract non-identified group features of the currently interacting crowd from the images, forming non-identified group feature data containing dominant age range labels, specifically including: Face detection is performed on consecutive image frames of the real-time video stream data to obtain a set of face position coordinates in each frame; Based on the set of face location coordinates, age attribute recognition is performed on each detected face image region to obtain the corresponding estimated age value. Statistical distribution analysis is performed on all estimated age values ​​obtained within a continuous preset time period, and the age range with the highest frequency is determined as the dominant age range label. The dominant age range labels are combined with statistically obtained population density information and average stay time information to form the non-identity-based group characteristic data.

4. The method for advertising through video games using AI as described in claim 3, characterized in that, A statistical distribution analysis is performed on all estimated age values ​​obtained within a consecutive preset time period, and the age range with the highest frequency is determined as the dominant age range label, specifically including: Set a sliding time window with a length of N seconds, and continuously count the estimated age values ​​of all successfully recognized faces within the sliding time window; Cluster analysis is performed on the estimated age values ​​within the sliding time window to calculate the frequency distribution of each age interval; The age range with the highest frequency exceeding the preset frequency threshold is marked as the dominant age range label at the current moment; As the sliding time window slides, the advantageous age range label is updated in real time.

5. A method for advertising through video games using AI as described in claim 1, characterized in that, The advantageous age range tag is used as the query key and input into a pre-set dynamic content mapping rule base for matching. The dynamic content mapping rule base stores the corresponding mapping relationships between different advantageous age range tags and multiple sets of candidate ad visual elements, specifically including: The dynamic content mapping rule base is pre-installed in the cloud server and contains multiple mapping rule entries. Each mapping rule entry records the correspondence between an advantageous age range tag and a set of candidate advertising visual elements. Each set of candidate advertising visual elements contains multiple advertising materials of different styles. The execution of the matching query includes: sending the advantageous age range label to the cloud server via a wireless network, and the cloud server retrieving matching mapping rule entries from the dynamic content mapping rule base it stores, and returning the search results.

6. The method for advertising through video games using AI as described in claim 5, characterized in that, Based on the matching query results, the target advertising visual element identifiers associated with the advantageous age range tags are determined and output, specifically including: Based on the search results, obtain a group of candidate advertising visual elements associated with the advantageous age range tag; When the candidate ad visual element group contains multiple alternative elements, a target ad visual element is determined from the multiple alternative elements; The unique code of the identified target advertising visual element in the dynamic content mapping rule base is output as the target advertising visual element identifier.

7. A method for advertising through video games using AI as described in claim 1, characterized in that, Based on the target advertisement visual element identifier, the replaceable texture storage area is located in the resource pack of the video game, specifically including: Assign a unique texture resource identifier to each pre-tagged replaceable advertisement display location, and establish an address mapping relationship between each texture resource identifier and the physical storage address in the resource package of the video game; A mapping relationship between the target advertisement visual element identifier and the texture resource identifier is established in advance; Based on the visual element identifier of the target advertisement, query the element mapping relationship to obtain the corresponding texture resource identifier; The address mapping relationship is queried based on the texture resource identifier to obtain the corresponding physical storage address, thereby locating the replaceable texture storage area.

8. A method for advertising through video games using AI as described in claim 1, characterized in that, Based on the visual data corresponding to the target advertising visual element identifier, the texture content of the replaceable texture storage area is rendered and updated in real time, thereby generating and presenting dynamic advertising content that matches the advantageous age range tag in the screen of the video game, specifically including: Obtain the corresponding visual data file based on the target advertisement visual element identifier; The visual data file is parsed into a texture map format to obtain the target texture data; The target texture data is updated to the replaceable texture storage area by the rendering component of the video game; Without interrupting the main logic thread of the game, the rendering component uses the target texture data to render the screen, thereby generating and presenting the dynamic advertising content that matches the advantageous age range label on the display screen of the video game.

9. A method for advertising through video games using AI as described in claim 8, characterized in that, The target texture data is updated to the replaceable texture storage area using the rendering component of the video game, specifically including: Pass the memory address pointer of the replaceable texture storage area and the physical storage address of the target texture data to the rendering component; An asynchronous texture update operation is triggered by the rendering component, so that the texture binding state of the replaceable texture storage area is updated to point to the target texture data; During the rendering of the next frame of the video game, a replaceable texture storage area is drawn using the target texture data it points to.

10. A system for advertising through video games using AI, for implementing the method for advertising through video games using AI as described in any one of claims 1-9, characterized in that, The system includes: The image acquisition and video stream generation module is used to acquire images of the interactive area and generate real-time video stream data; The group feature analysis module is used to perform computer vision analysis on the real-time video stream data, extract non-identified group features of the current interactive crowd from the image, and form non-identified group feature data containing dominant age range labels. The dynamic content mapping query module is used to input the advantageous age range tag as a query key into a preset dynamic content mapping rule library for matching query. The dynamic content mapping rule library stores the corresponding mapping relationship between different advantageous age range tags and multiple sets of candidate advertising visual elements. The advertising element output module is used to determine and output the target advertising visual element identifier associated with the advantageous age range tag based on the matching query results; The game texture resource location module is used to locate the replaceable texture storage area in the resource package of the video game based on the target advertising visual element identifier. The dynamic advertising rendering module is used to render and update the texture content of the replaceable texture storage area in real time based on the visual data corresponding to the visual element identifier of the target advertisement, thereby generating and presenting dynamic advertising content that matches the advantageous age range label in the screen of the video game.

Images