An exhibition hall interactive display method and system supporting artificial intelligence recognition

Abstract

The application belongs to the technical field of artificial intelligence, and discloses a kind of exhibition hall interactive display method and system supporting artificial intelligence identification, including collecting the multi-modal behavior information of audience in exhibition hall environment, identifying the interactive intention of audience through behavior pattern analysis and semantic understanding, and mapping the interactive intention of audience to exhibition hall virtual scene, forming semantic interaction node space;According to the interactive intention of audience, the association relationship of each node in semantic interaction node space and exhibit or exhibition hall scene element is analyzed, the display content that can be triggered is determined, and the corresponding interactive display request set is generated;When there are different audiences triggering interactive display request at the same time, the interactive display request set is coordinated, and each interactive display request is adjusted and dispatched according to the exhibition hall resource limit and the preset priority rule;Enhance the intelligence and adaptive ability of exhibition hall interaction, improve the fluency and immersion of audience interactive experience.

Description

Technical Field

[0001] This invention relates to the field of deep learning technology, and more specifically, to a method and system for interactive exhibition displays in exhibition halls that supports artificial intelligence recognition. Background Technology

[0002] Existing interactive exhibition methods and systems that support AI recognition mainly suffer from the following problems:

[0003] With the development of artificial intelligence and immersive display technologies, interactive exhibition displays are gradually evolving from static, one-way presentations to dynamic, multimodal human-computer interaction. Although existing interactive exhibition display methods and systems can trigger display content based on the audience's actions, they still have many shortcomings.

[0004] Existing technologies generally use a single action to directly map displayed content, lacking semantic modeling of the relationship between viewer intent and displayed content. Different viewers often trigger the same displayed content, making it impossible to dynamically optimize based on viewers' historical interaction records, behavioral patterns, and personalized needs. This results in a mismatch between the displayed results and the viewer's actual needs, making it difficult to provide an intelligent and immersive interactive experience.

[0005] When multiple visitors simultaneously trigger interaction requests in an exhibition hall, existing systems typically employ queuing mechanisms or fixed priority tables, such as first-come, first-served or preset priority ordering. While this approach avoids device resource conflicts, it struggles to comprehensively consider the match between visitor interaction intentions and the displayed content, and it fails to address the limited resources of the exhibition hall. Consequently, some resources are prone to being idle or overloaded, and the lack of dynamic scheduling and fair optimization mechanisms for multiple visitor requests leads to an overall subpar exhibition experience.

[0006] During the rendering process, existing interactive exhibition methods typically only consider the logical triggering of the displayed content, lacking real-time monitoring of the resource usage of rendering requests on display terminals, computing devices, and rendering channels. Existing methods mostly rely on fixed priority or polling strategies for scheduling. When concurrent rendering requests occur, the overall system resource utilization easily exceeds the threshold, causing frame drops, stuttering, or even exhibition interruptions. Under resource overload, existing technologies lack a unified dynamic adjustment mechanism, often simply discarding some viewer requests, which affects the fairness of interaction and severely reduces the viewer's immersive experience.

[0007] In view of this, the present invention proposes an interactive exhibition method for exhibition halls that supports artificial intelligence recognition to solve the above problems. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: an interactive exhibition method supporting artificial intelligence recognition, comprising:

[0009] S1. Collect multimodal behavioral information of visitors in the exhibition hall environment, identify visitors' interaction intentions through behavioral pattern analysis and semantic understanding, and map visitors' interaction intentions to the virtual scene of the exhibition hall to form a semantic interaction node space;

[0010] S2. Based on the audience's interactive intent, analyze the relationship between each node in the semantic interaction node space and the exhibits or exhibition hall scene elements, determine the triggerable display content, and generate the corresponding interactive display request set.

[0011] S3. When different viewers trigger interactive display requests at the same time, coordinate the set of interactive display requests and adjust and schedule each interactive display request according to the exhibition hall resource limitations and preset priority rules.

[0012] S4. Convert the adjusted and scheduled interactive display requests into corresponding rendering requests, and submit the rendering requests to the exhibition hall interactive terminal. During the rendering process, monitor the amount of resources occupied by each rendering request, obtain the exhibition hall resource utilization rate, generate a rendering strategy based on the exhibition hall resource utilization rate, and dynamically display the exhibition hall content according to the rendering strategy.

[0013] Specifically, the multimodal behavioral information includes visual modal information, auditory modal information, action modal information, and environmental interaction modal information.

[0014] Specifically, the method for identifying the audience's interactive intent includes:

[0015] The acquired multimodal behavioral information is denoised, and multimodal feature vectors are extracted; the multimodal feature vectors include visual modal features, auditory modal features, action modal features, and environmental interaction modal features;

[0016] The multimodal feature vector is input into the pre-built behavior recognition model, which outputs the audience's action category label and emotional state label. The action category label and emotional state label are matched with the preset rule base to form the corresponding behavior pattern label.

[0017] The behavior recognition model can be either a convolutional neural network model or a recurrent neural network model; based on the knowledge graph, the behavior pattern labels are semantically matched with the elements of the exhibition hall scene to infer the audience's interactive intentions.

[0018] Specifically, the method for forming the semantic interaction node space includes:

[0019] Virtual scenes are constructed in the exhibition hall environment. A corresponding virtual model is established for each exhibit, exhibition area and interactive device. Scene semantic tags are added to the virtual model. At the same time, a semantic annotation relationship table is established to analyze the interactive intentions of the audience and extract the core elements of the interactive intentions. The core elements include intention type, target object, interaction method and trigger time.

[0020] Interactive intentions are bound to scene elements in the virtual model through semantic matching to generate interactive nodes. After the interactive intentions of different viewers are identified and interactive nodes are generated, the interactive nodes are organized according to time sequence and spatial distribution to form a semantic interactive node space.

[0021] Specifically, the method for generating the corresponding interactive display request includes:

[0022] Establish a collection of exhibition content, with each piece of content corresponding to a multimedia explanation of the exhibit, an immersive display of the exhibition area, or an interactive installation experience program, and add semantic tags and relevance weights to the content.

[0023] Collect contextual information from the audience, express the identified audience interaction intentions in a structured way, and estimate the probability value of the audience interaction intentions belonging to the semantic tags of each content through Bayesian inference, based on the audience's contextual information.

[0024] The probability value and the relevance weight of the displayed content are combined to calculate the matching score between the audience's interaction intention and each displayed content. When the matching score exceeds the preset matching score threshold, the corresponding displayed content is determined to be triggerable content, and all displayed content that meets the conditions is aggregated into an interactive display request set.

[0025] Specifically, the method for adjusting and scheduling each interactive display request includes:

[0026] When different viewers trigger interaction at the same time, the set of interactive display requests is coordinated. The total amount of available resources in the exhibition hall is preset. The total amount of available resources is all resources that can be scheduled. The resources that can be scheduled include projection equipment, rendering computing power, display screens and interactive devices. For each display content, the amount of resources required during the display process is preset.

[0027] During the scheduling process, both exhibition hall resource limitations and preset priority rules must be met. Exhibition hall resource limitations include that the total resource consumption required by all assigned display content cannot exceed the total available resources of the exhibition hall, and each visitor can be assigned a maximum of one display content.

[0028] When the exhibition hall resources are limited, when different visitors trigger interaction requests at the same time, the interaction requests of different visitors are sorted according to the preset priority rules. An allocation variable is set for each visitor and display content combination to indicate whether the visitor is assigned to the display content. When the allocation is successful, the allocation variable takes the value of 1, otherwise it takes the value of 0, thereby dynamically adjusting the allocation result of each interactive display request.

[0029] Specifically, the method for converting the adjusted and scheduled interactive display requests into corresponding rendering requests includes:

[0030] The system parses the adjusted and scheduled interactive display requests to obtain the display content, scene location, presentation format, and interaction parameters corresponding to each request. It maps the display content corresponding to each interactive display request to rendering resources, determines the spatial layout of the display content in the virtual scene based on the scene location, and determines the display method based on the presentation format. According to the rendering engine interface specification, it converts the display content, scene location, presentation format, and interaction parameters into rendering instructions. Finally, it encapsulates the rendering instructions to form an executable rendering request.

[0031] Specifically, the method for obtaining the utilization rate of exhibition hall resources includes:

[0032] Rendering requests are submitted to the interactive terminal in the exhibition hall. During the rendering process, the amount of resources used by each rendering request is monitored in real time, including the usage of resources in the display terminal, computing devices, and rendering channels. The resource usage of each rendering request is recorded and associated with the corresponding audience and display content to form resource usage status information. The resource usage of each rendering request is correlated with the allocation status of the rendering request to calculate and obtain the exhibition hall resource utilization rate.

[0033] Specifically, the method for dynamically displaying exhibition hall content according to a rendering strategy includes:

[0034] A preset threshold for exhibition hall resource utilization is set. When the exhibition hall resource utilization is less than or equal to the preset threshold, all rendering requests are allowed to be rendered according to the allocation variables at time point t. When the exhibition hall resource utilization is greater than the preset threshold, a dynamic allocation adjustment mechanism is adopted to adjust the allocation status of rendering requests, generate corresponding rendering strategies, and dynamically display the exhibition hall content according to the rendering strategies.

[0035] An interactive exhibition system for showrooms that supports artificial intelligence recognition includes:

[0036] The audience behavior recognition unit collects multimodal behavior information of the audience in the exhibition hall environment. Through behavior pattern analysis and semantic understanding, it identifies the audience's interaction intentions and maps the audience's interaction intentions to the virtual scene of the exhibition hall to form a semantic interaction node space.

[0037] The content generation unit analyzes the relationship between each node in the semantic interaction node space and the exhibits or exhibition hall scene elements based on the audience's interaction intent, determines the triggerable display content, and generates the corresponding set of interactive display requests.

[0038] The content scheduling unit coordinates the set of interactive display requests when different viewers trigger interactive display requests simultaneously, and adjusts and schedules each interactive display request according to the exhibition hall resource limitations and preset priority rules.

[0039] The interactive content display unit converts the adjusted and scheduled interactive display requests into corresponding rendering requests and submits the rendering requests to the exhibition hall interactive terminal. During the rendering process, it monitors the amount of resources occupied by each rendering request, obtains the exhibition hall resource utilization rate, generates a rendering strategy based on the exhibition hall resource utilization rate, and dynamically displays the exhibition hall content according to the rendering strategy.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] This invention estimates the probability value of intent by combining Bayesian inference with contextual information, overcoming the recognition bias of the single trigger mode and enabling more precise differentiation of the true intent of different viewers. By comprehensively considering the contextual information of the viewers and the semantic weight of the displayed content through matching scores, different viewers can trigger the displayed content that meets their needs, enhancing the differentiation and immersion of the interactive experience. The weight relationship between the displayed content and the semantic tags of intent establishes a semantic layer bridge, enabling more reasonable content scheduling based on the semantic level. By setting a matching score threshold, the triggering of displayed content can be dynamically determined according to the viewer's behavior and contextual information, enhancing the intelligence and flexibility of the interactive display.

[0042] By monitoring the resource consumption of each rendering request in real time during the rendering process and calculating the exhibition hall resource utilization rate based on the allocation status, a feedback adjustment mechanism capable of dynamically sensing the exhibition hall load was established. This accurately reflects the real-time resource consumption of the system in terms of display terminals, computing devices, and rendering channels. The proposed dynamic allocation adjustment mechanism can proportionally reduce and adjust the allocation status of rendering requests, rather than simply discarding low-priority requests. This ensures that all rendering requests receive the corresponding actual resource quota under resource constraints, achieving fairness and continuity in resource allocation. Based on the exhibition hall resource utilization rate, a corresponding rendering strategy is generated. When resources are sufficient, all requests are supported for normal rendering. When resources are scarce, dynamic adjustment keeps the overall utilization rate below a threshold, thereby ensuring the continuous presentation of the displayed content and the stability of system operation, and enhancing the immersive experience for visitors. Attached Figure Description

[0043] Figure 1This is a schematic diagram of a process for an interactive exhibition hall display method supporting artificial intelligence recognition according to the present invention;

[0044] Figure 2 This is a schematic diagram of the structure of an interactive exhibition system for showrooms that supports artificial intelligence recognition, according to the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0046] Please see Figure 1 As shown, this embodiment provides an interactive exhibition hall display method that supports artificial intelligence recognition, specifically including the following steps:

[0047] S1. Collect multimodal behavioral information of visitors in the exhibition hall environment, identify visitors' interaction intentions through behavioral pattern analysis and semantic understanding, and map visitors' interaction intentions to the virtual scene of the exhibition hall to form a semantic interaction node space;

[0048] S2. Based on the audience's interactive intent, analyze the relationship between each node in the semantic interaction node space and the exhibits or exhibition hall scene elements, determine the triggerable display content, and generate the corresponding interactive display request set.

[0049] S3. When different viewers trigger interactive display requests at the same time, coordinate the set of interactive display requests and adjust and schedule each interactive display request according to the exhibition hall resource limitations and preset priority rules.

[0050] S4. Convert the adjusted and scheduled interactive display requests into corresponding rendering requests, and submit the rendering requests to the exhibition hall interactive terminal. During the rendering process, monitor the amount of resources occupied by each rendering request, obtain the exhibition hall resource utilization rate, generate a rendering strategy based on the exhibition hall resource utilization rate, and dynamically display the exhibition hall content according to the rendering strategy.

[0051] Multimodal behavioral information includes visual modality information, auditory modality information, motor modality information, and environmental interaction modality information.

[0052] Visual modal information includes audience posture, gestures, facial expressions, and gaze direction; auditory modal information includes audience speech content, intonation, and sound source localization; motion modal information includes audience location trajectory within the exhibition hall and distance between the audience and exhibits; environmental interaction modal information includes interactive terminal operation information and audience dwell time; interactive terminal operation information includes touchscreen clicks, swipes, mobile scanning, and APP control.

[0053] Methods for identifying audience interaction intentions include:

[0054] The acquired multimodal behavioral information is denoised, and multimodal feature vectors are extracted; the multimodal feature vectors include visual modal features, auditory modal features, action modal features, and environmental interaction modal features;

[0055] The specific methods for obtaining multimodal feature vectors are as follows: extracting the audience's posture key points, gesture types, facial expression feature vectors, and gaze direction coordinates from visual modal information to form visual modal features; extracting speech keywords, intonation curves, and sound source localization information from auditory modal information to form auditory modal features; extracting the audience's movement path, dwelling area, and spatial distance change information from motion modal information to form motion modal features; and extracting the interaction operation type and dwell time from environmental interaction modal information to form environmental interaction modal features.

[0056] The multimodal feature vector is input into the pre-built behavior recognition model, which outputs the audience's action category label and emotional state label. The action category label and emotional state label are matched with the preset rule base to form the corresponding behavior pattern label.

[0057] The behavior recognition model can be either a convolutional neural network model or a recurrent neural network model; based on the knowledge graph, the behavior pattern labels are semantically matched with the elements of the exhibition hall scene to infer the audience's interactive intentions.

[0058] For example, when a visitor stops in front of a display case of an artifact, points to the case, and says, "Can this be enlarged?", the system first identifies the type of gesture and the coordinates of the gaze direction through visual modal information, then identifies the voice keyword "enlarge" through auditory modal information, obtains the distance between the visitor and the exhibit through action modal information, and finally infers the visitor's interaction intention as "requesting an enlarged display of the exhibit" by combining it with a preset rule base, and generates a corresponding interaction node to trigger the exhibit enlargement demonstration function in the virtual scene.

[0059] Methods for forming a semantic interaction node space include:

[0060] A virtual scene is constructed in the exhibition hall environment. A corresponding virtual model is established for each exhibit, exhibition area and interactive device. Scene semantic tags are added to the virtual model, such as exhibit ID, exhibition area function, interaction method, etc. At the same time, a semantic annotation relationship table is established to support subsequent intent mapping operations. The virtual model is a digital twin model.

[0061] The system analyzes the identified audience interaction intentions and extracts their core elements, including intention type, target object, interaction method, and trigger time. Intention type can be explanation, magnification, comparison, or navigation, while the target object corresponds to exhibits or functional elements in the virtual scene. Through semantic matching, the interaction intentions are bound to scene elements in the virtual model to generate interaction nodes. Each interaction node contains information such as node ID, intention type, triggering audience identity, target element, trigger timestamp, and intention confidence value. These interaction nodes serve as logical trigger points in the virtual scene, driving subsequent content display.

[0062] After the interactive intentions of different viewers are identified and interactive nodes are generated, these interactive nodes are organized according to temporal order and spatial distribution to form a semantic interactive node space. Interactive nodes are connected by relational edges, which are used to represent "continuous interactive intentions of the same viewer", "interactive intentions of different viewers for the same exhibit", or "transfer of interactive intentions across exhibition areas".

[0063] For example, when a visitor asks, "Can this be enlarged?" in front of a display case, the system recognizes the interaction intent as "requesting an enlarged display of the exhibit" and locates the corresponding virtual model node of the exhibit in the virtual scene. The system generates an interaction node, recording the following information: Intent Type: Enlargement Display, Visitor ID: 001, Target Exhibit: Display Case A - Artifact X, Timestamp: 2025-06-21, and adds this interaction node to the semantic interaction node space. Simultaneously, a relationship edge is established between this interaction node and the visitor's previous interaction node "requesting explanation," used to characterize the continuous interaction intent trajectory of the same visitor. Finally, the display scheduling module triggers the enlarged demonstration function based on this interaction node, enabling interaction between the visitor and the exhibition hall scene.

[0064] The methods for generating corresponding interactive display requests include:

[0065] Establish a collection of exhibition content, with each piece of content corresponding to a multimedia explanation of the exhibit, an immersive display of the exhibition area, or an interactive installation experience program, and add semantic tags and relevance weights to the content.

[0066] Collect audience contextual information, structure the identified audience interaction intentions, and combine the audience contextual information to estimate the probability value of the audience interaction intention belonging to each content semantic tag through Bayesian inference; the audience contextual information includes the audience's location, current behavior pattern (the audience's body posture, such as standing, moving, or standing still) and historical interaction records.

[0067] The probability value and the relevance weight of the displayed content are combined to calculate the matching score between the audience's interaction intention and each displayed content. When the matching score exceeds the preset matching score threshold, the corresponding displayed content is determined to be triggerable content, and all displayed content that meets the conditions is aggregated into an interactive display request set.

[0068] Match score: ;in, Representing the audience With the content displayed The matching score; A collection representing all semantic tags for all content; This represents a content semantic tag within a set of content semantic tags; Indicates the content to be displayed With content semantic tags The relevance weights are assigned based on the initial weight values ​​given by the audience's historical interaction data. For example, the weight of "exhibition introduction video" and the intent of "querying information" is labeled as 0.9, while the weight of "triggering interactive experience" is labeled as 0.2. Represents contextual information for a given audience. Under these conditions, audience interaction intent belongs to content semantic tags. The probability value; Representing the audience Interaction intent variables; Indicates contextual information about the audience;

[0069] The following technical problems of existing technologies have been solved: Existing technologies often directly map single actions to displayed content, resulting in a mismatch between interactive content and audience needs; there is a lack of semantic association modeling between intent and content, which may lead to different audiences triggering the same content, resulting in a lack of personalization; existing technologies fail to incorporate information such as audience historical interaction records and behavioral patterns into the inference process, making it impossible to dynamically optimize content matching; and there is a lack of interactive content selection mechanism based on probability inference, making it difficult for the display system to achieve a more intelligent and immersive interactive experience.

[0070] Compared to existing technologies, the advantages are as follows: By estimating the probability value of intent through Bayesian inference combined with contextual information, the recognition bias of the single trigger mode is overcome, enabling more precise differentiation of the true intent of different viewers; by comprehensively considering the audience's contextual information and the semantic weight of the displayed content through matching scores, different viewers can trigger display content that meets their needs, enhancing the differentiation and immersion of the interactive experience; the weight relationship between the displayed content and the semantic tags of intent establishes a semantic layer bridge, enabling more reasonable content scheduling based on the semantic level; by setting a matching score threshold, the triggering of displayed content can be dynamically determined according to the audience's behavior and contextual information, enhancing the intelligence and flexibility of the interactive display;

[0071] Methods for adjusting and scheduling interactive display requests include:

[0072] When different visitors simultaneously trigger interaction requests in an exhibition hall, existing technologies typically employ simple queuing mechanisms or priority rules, such as processing requests sequentially based on a first-come, first-served principle or a fixed priority table. While this approach avoids conflicts with exhibition hall equipment resources, it has the following shortcomings: it fails to comprehensively consider the match between visitor interaction intentions and exhibition content, easily leading to a mismatch between the displayed content and visitor needs; it ignores the limited resource constraints of the exhibition hall (such as projection equipment bandwidth, rendering computing power, and the number of display screens), potentially resulting in some resources being idle or overloaded; and it lacks a dynamic optimization mechanism, making it difficult to simultaneously satisfy the fairness and overall viewing experience of multiple visitor interaction requests.

[0073] To address the aforementioned issues, when different viewers simultaneously trigger interactions, the set of interactive display requests is coordinated. The total available resources of the exhibition hall are preset, which includes all resources that can be scheduled, such as projection equipment, rendering computing power, display screens, and interactive devices. For each display content, the resource consumption required during the display process is preset.

[0074] During the scheduling process, both exhibition hall resource limitations and preset priority rules must be met. Exhibition hall resource limitations include that the total resource consumption required by all assigned display content cannot exceed the total available resources of the exhibition hall, and each visitor can be assigned a maximum of one display content.

[0075] When the exhibition hall resources are limited, when different visitors trigger interaction requests at the same time, the interaction requests of different visitors are sorted according to the preset priority rules. The preset priority rules include sorting the interaction requests of different visitors according to the matching score between the visitor's interaction intention and each display content.

[0076] For each combination of viewer and displayed content, assign a variable to indicate whether the viewer is assigned to that displayed content; the assignment variable is: When the allocation is successful, the allocation variable takes the value of 1; otherwise, it takes the value of 0, thereby dynamically adjusting the allocation results of each interactive display request.

[0077] Methods for converting adjusted and scheduled interactive display requests into corresponding rendering requests include:

[0078] The system parses the adjusted and scheduled interactive display requests to obtain the display content, scene location, presentation format, and interaction parameters corresponding to each interactive display request; maps the display content corresponding to each interactive display request to rendering resources, determines the spatial layout of the display content in the virtual scene based on the scene location, and determines the display method (e.g., two-dimensional, three-dimensional, animation, or multimedia effects) based on the presentation format.

[0079] According to the rendering engine interface specification, the displayed content, scene location, presentation format, and interaction parameters are transformed into rendering instructions, including resource loading instructions, material binding instructions, animation setting instructions, lighting adjustment instructions, camera view adjustment instructions, and event binding instructions. The rendering instructions are then encapsulated to form executable rendering requests, enabling the exhibition hall interactive terminal to directly read and execute the corresponding rendering instructions.

[0080] Methods for obtaining exhibition hall resource utilization rates include:

[0081] Rendering requests are submitted to the interactive terminal in the exhibition hall. During the rendering process, the amount of resources used by each rendering request is monitored in real time, including the usage of resources in the display terminal, computing devices, and rendering channels. The resource usage of each rendering request is recorded and associated with the corresponding audience and display content to form resource usage status information. The resource usage of each rendering request is correlated with the allocation status of the rendering request to calculate and obtain the exhibition hall resource utilization rate.

[0082] The exhibition hall resource utilization rate is: ;in, Indicates a point in time The utilization rate of exhibition hall resources at that time; Indicates the content to be displayed The resource consumption of the corresponding rendering request; Indicates a point in time The allocation variable at time, that is, at time point Is it for the audience? Assign display content ; This indicates the total amount of available resources in the pre-designed exhibition hall;

[0083] Methods for dynamically displaying exhibition hall content based on rendering strategies include:

[0084] A preset threshold for exhibition hall resource utilization is set. When the exhibition hall resource utilization is less than or equal to the preset threshold, all rendering requests are allowed to be rendered according to the allocation variables at time point t. When the exhibition hall resource utilization is greater than the preset threshold, a dynamic allocation adjustment mechanism is adopted to adjust the allocation status of rendering requests, generate corresponding rendering strategies, and dynamically display the exhibition hall content according to the rendering strategies.

[0085] The dynamic allocation and adjustment mechanism is as follows: ;in, Indicates a point in time The actual resource quota for each rendering request; This indicates the preset threshold for exhibition hall resource utilization. This represents the scaling factor. When the exhibition hall resource utilization rate is greater than the preset exhibition hall resource utilization rate threshold, the resource allocation of all rendering requests needs to be reduced proportionally to ensure that the exhibition hall resource utilization rate is reduced back to below the preset exhibition hall resource utilization rate threshold.

[0086] This solution addresses the following technical problems of existing technologies: Current interactive exhibition hall rendering typically only considers the logical triggering of the displayed content during the rendering process, lacking real-time monitoring of the resource usage of rendering requests on display terminals, computing devices, and rendering channels. This easily leads to uneven resource allocation, causing some displayed content to stutter or delay. Existing methods generally use fixed priority or polling strategies for rendering resource scheduling, causing the overall system resource utilization to exceed the threshold when concurrent rendering requests occur in the exhibition hall, resulting in dropped frames or display interruptions. When resources are overloaded, the lack of a unified dynamic adjustment mechanism makes it impossible to maintain overall performance while ensuring fairness, easily leading to some visitors' requests being completely discarded, resulting in a poor user experience.

[0087] Compared to existing technologies, the advantages are as follows: By monitoring the resource consumption of each rendering request in real time during the rendering process and calculating the exhibition hall resource utilization rate based on the allocation status, a feedback adjustment mechanism capable of dynamically sensing the exhibition hall load is established, accurately reflecting the real-time resource consumption of the system in terms of display terminals, computing devices, and rendering channels. The proposed dynamic allocation adjustment mechanism can proportionally reduce and adjust the allocation status of rendering requests, rather than simply discarding low-priority requests. This ensures that all rendering requests receive the corresponding actual resource quota under resource constraints, achieving fairness and continuity in resource allocation. Based on the exhibition hall resource utilization rate, a corresponding rendering strategy is generated. When resources are sufficient, all requests are supported for normal rendering. When resources are scarce, dynamic adjustment keeps the overall utilization rate below a threshold, thereby ensuring the continuous presentation of the displayed content and the stability of system operation, and enhancing the immersive experience for the audience.

[0088] The preset threshold for exhibition hall resource utilization is set by staff based on historical data analysis results. This historical analysis process includes the system collecting the resource utilization rates of multiple exhibition halls and calculating their average value as a reference to obtain the preset threshold for exhibition hall resource utilization. This threshold can be adjusted by staff during system operation according to the actual situation.

[0089] This embodiment estimates the probability value of intent by combining Bayesian inference with contextual information, overcoming the recognition bias in the single-trigger mode and enabling more precise differentiation of the true intent of different viewers. By comprehensively considering the audience's contextual information and the semantic weight of the displayed content through matching scores, different viewers can trigger display content that meets their needs, enhancing the differentiation and immersion of the interactive experience. The weight relationship between the displayed content and the semantic tags of intent establishes a bridge at the semantic layer, enabling more reasonable content scheduling based on the semantic level. By setting a matching score threshold, the triggering of displayed content can be dynamically determined based on audience behavior and contextual information, enhancing the intelligence and flexibility of the interactive display.

[0090] By monitoring the resource consumption of each rendering request in real time during the rendering process and calculating the exhibition hall resource utilization rate based on the allocation status, a feedback adjustment mechanism capable of dynamically sensing the exhibition hall load was established. This accurately reflects the real-time resource consumption of the system in terms of display terminals, computing devices, and rendering channels. The proposed dynamic allocation adjustment mechanism can proportionally reduce and adjust the allocation status of rendering requests, rather than simply discarding low-priority requests. This ensures that all rendering requests receive the corresponding actual resource quota under resource constraints, achieving fairness and continuity in resource allocation. Based on the exhibition hall resource utilization rate, a corresponding rendering strategy is generated. When resources are sufficient, all requests are supported for normal rendering. When resources are scarce, dynamic adjustment keeps the overall utilization rate below a threshold, thereby ensuring the continuous presentation of the displayed content and the stability of system operation, and enhancing the immersive experience for visitors. Example

[0091] Please see Figure 2 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. This embodiment provides an interactive exhibition system for showrooms that supports artificial intelligence recognition, including:

[0092] The audience behavior recognition unit collects multimodal behavior information of the audience in the exhibition hall environment. Through behavior pattern analysis and semantic understanding, it identifies the audience's interaction intentions and maps the audience's interaction intentions to the virtual scene of the exhibition hall to form a semantic interaction node space.

[0093] The content generation unit analyzes the relationship between each node in the semantic interaction node space and the exhibits or exhibition hall scene elements based on the audience's interaction intent, determines the triggerable display content, and generates the corresponding set of interactive display requests.

[0094] The content scheduling unit coordinates the set of interactive display requests when different viewers trigger interactive display requests simultaneously, and adjusts and schedules each interactive display request according to the exhibition hall resource limitations and preset priority rules.

[0095] The interactive content display unit converts the adjusted and scheduled interactive display requests into corresponding rendering requests and submits the rendering requests to the exhibition hall interactive terminal. During the rendering process, it monitors the amount of resources occupied by each rendering request, obtains the exhibition hall resource utilization rate, generates a rendering strategy based on the exhibition hall resource utilization rate, and dynamically displays the exhibition hall content according to the rendering strategy.

[0096] Since the electronic device described in this embodiment is the one used in implementing the interactive exhibition method and system supporting artificial intelligence recognition in the embodiments of this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the interactive exhibition method and system supporting artificial intelligence recognition described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art in implementing the interactive exhibition method and system supporting artificial intelligence recognition in the embodiments of this application falls within the scope of protection of this application.

[0097] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0098] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for interactive exhibition displays in a showroom that supports artificial intelligence recognition, characterized in that, include: S1. Collect multimodal behavioral information of visitors in the exhibition hall environment, identify visitors' interaction intentions through behavioral pattern analysis and semantic understanding, and map visitors' interaction intentions to the virtual scene of the exhibition hall to form a semantic interaction node space; The method for forming a semantic interaction node space includes: Virtual scenes are constructed in the exhibition hall environment. A corresponding virtual model is established for each exhibit, exhibition area and interactive device. Scene semantic tags are added to the virtual model. At the same time, a semantic annotation relationship table is established to analyze the interactive intentions of the audience and extract the core elements of the interactive intentions. The core elements include intention type, target object, interaction method and trigger time. Interactive intentions are bound to scene elements in the virtual model through semantic matching to generate interactive nodes; after the interactive intentions of different viewers are identified and interactive nodes are generated, the interactive nodes are organized according to time sequence and spatial distribution to form a semantic interactive node space. S2. Based on the audience's interactive intent, analyze the relationship between each node in the semantic interaction node space and the exhibits or exhibition hall scene elements, determine the triggerable display content, and generate the corresponding interactive display request set. S3. When different viewers trigger interactive display requests at the same time, coordinate the set of interactive display requests and adjust and schedule each interactive display request according to the exhibition hall resource limitations and preset priority rules. S4. Convert the adjusted and scheduled interactive display requests into corresponding rendering requests, and submit the rendering requests to the exhibition hall interactive terminal. During the rendering process, monitor the amount of resources occupied by each rendering request, obtain the exhibition hall resource utilization rate, generate a rendering strategy based on the exhibition hall resource utilization rate, and dynamically display the exhibition hall content according to the rendering strategy. The method for dynamically displaying exhibition hall content based on a rendering strategy includes: A preset threshold for exhibition hall resource utilization is set. When the exhibition hall resource utilization is less than or equal to the preset threshold, all rendering requests are allowed to be rendered according to the allocation variables at time point t. When the exhibition hall resource utilization is greater than the preset threshold, a dynamic allocation adjustment mechanism is adopted to adjust the allocation status of rendering requests, generate corresponding rendering strategies, and dynamically display the exhibition hall content according to the rendering strategies.

2. The interactive exhibition method supporting artificial intelligence recognition according to claim 1, characterized in that, The multimodal behavioral information includes visual modal information, auditory modal information, motor modal information, and environmental interaction modal information.

3. The exhibition hall interactive display method supporting artificial intelligence recognition according to claim 2, characterized in that, The method for identifying the audience's interactive intent includes: The acquired multimodal behavioral information is denoised, and multimodal feature vectors are extracted; the multimodal feature vectors include visual modal features, auditory modal features, action modal features, and environmental interaction modal features; The multimodal feature vector is input into the pre-built behavior recognition model, which outputs the audience's action category label and emotional state label. The action category label and emotional state label are matched with the preset rule base to form the corresponding behavior pattern label. The behavior recognition model can be either a convolutional neural network model or a recurrent neural network model; based on the knowledge graph, the behavior pattern labels are semantically matched with the elements of the exhibition hall scene to infer the audience's interactive intentions.

4. The interactive exhibition method supporting artificial intelligence recognition according to claim 3, characterized in that, The method for generating the corresponding interactive display request includes: Establish a collection of exhibition content, with each piece of content corresponding to a multimedia explanation of the exhibit, an immersive display of the exhibition area, or an interactive installation experience program, and add semantic tags and relevance weights to the content. Collect contextual information from the audience, express the identified audience interaction intentions in a structured way, and estimate the probability value of the audience interaction intentions belonging to the semantic tags of each content through Bayesian inference, based on the audience's contextual information. The probability value and the relevance weight of the displayed content are combined to calculate the matching score between the audience's interaction intention and each displayed content. When the matching score exceeds the preset matching score threshold, the corresponding displayed content is determined to be triggerable content, and all displayed content that meets the conditions is aggregated into an interactive display request set.

5. The interactive exhibition method for supporting artificial intelligence recognition according to claim 4, characterized in that, The method for adjusting and scheduling each interactive display request includes: When different viewers trigger interaction at the same time, the set of interactive display requests is coordinated. The total amount of available resources in the exhibition hall is preset. The total amount of available resources is all resources that can be scheduled. The resources that can be scheduled include projection equipment, rendering computing power, display screens and interactive devices. For each display content, the amount of resources required during the display process is preset. During the scheduling process, both exhibition hall resource limitations and preset priority rules must be met. Exhibition hall resource limitations include that the total resource consumption required by all assigned display content cannot exceed the total available resources of the exhibition hall, and each visitor can be assigned a maximum of one display content. When the exhibition hall resources are limited, when different visitors trigger interaction requests at the same time, the interaction requests of different visitors are sorted according to the preset priority rules. An allocation variable is set for each visitor and display content combination to indicate whether the visitor is assigned to the display content. When the allocation is successful, the allocation variable takes the value of 1, otherwise it takes the value of 0, thereby dynamically adjusting the allocation result of each interactive display request.

6. The interactive exhibition method for supporting artificial intelligence recognition according to claim 5, characterized in that, The method for converting adjusted and scheduled interactive display requests into corresponding rendering requests includes: The system parses the adjusted and scheduled interactive display requests to obtain the display content, scene location, presentation format, and interaction parameters corresponding to each request. It maps the display content corresponding to each interactive display request to rendering resources, determines the spatial layout of the display content in the virtual scene based on the scene location, and determines the display method based on the presentation format. According to the rendering engine interface specification, it converts the display content, scene location, presentation format, and interaction parameters into rendering instructions, encapsulates the rendering instructions, and forms an executable rendering request.

7. The interactive exhibition method supporting artificial intelligence recognition according to claim 6, characterized in that, The methods for obtaining the utilization rate of exhibition hall resources include: Rendering requests are submitted to the interactive terminal in the exhibition hall. During the rendering process, the amount of resources used by each rendering request is monitored in real time, including the usage of resources in the display terminal, computing devices, and rendering channels. The resource usage of each rendering request is recorded and associated with the corresponding audience and display content to form resource usage status information. The resource usage of each rendering request is correlated with the allocation status of the rendering request to calculate and obtain the exhibition hall resource utilization rate.

8. An interactive exhibition system supporting artificial intelligence recognition, used to implement the interactive exhibition method supporting artificial intelligence recognition as described in any one of claims 1 to 7, characterized in that, include: The audience behavior recognition unit collects multimodal behavior information of the audience in the exhibition hall environment. Through behavior pattern analysis and semantic understanding, it identifies the audience's interaction intentions and maps the audience's interaction intentions to the virtual scene of the exhibition hall to form a semantic interaction node space. The method for forming a semantic interaction node space includes: Virtual scenes are constructed in the exhibition hall environment. A corresponding virtual model is established for each exhibit, exhibition area and interactive device. Scene semantic tags are added to the virtual model. At the same time, a semantic annotation relationship table is established to analyze the interactive intentions of the audience and extract the core elements of the interactive intentions. The core elements include intention type, target object, interaction method and trigger time. Interactive intentions are bound to scene elements in the virtual model through semantic matching to generate interactive nodes; after the interactive intentions of different viewers are identified and interactive nodes are generated, the interactive nodes are organized according to time sequence and spatial distribution to form a semantic interactive node space. The content generation unit analyzes the relationship between each node in the semantic interaction node space and the exhibits or exhibition hall scene elements based on the audience's interaction intent, determines the triggerable display content, and generates the corresponding set of interactive display requests. The content scheduling unit coordinates the set of interactive display requests when different viewers trigger interactive display requests simultaneously, and adjusts and schedules each interactive display request according to the exhibition hall resource limitations and preset priority rules. The interactive content display unit converts the adjusted and scheduled interactive display requests into corresponding rendering requests and submits the rendering requests to the exhibition hall interactive terminal. During the rendering process, it monitors the amount of resources occupied by each rendering request, obtains the exhibition hall resource utilization rate, generates a rendering strategy based on the exhibition hall resource utilization rate, and dynamically displays the exhibition hall content according to the rendering strategy. The method for dynamically displaying exhibition hall content based on a rendering strategy includes: A preset threshold for exhibition hall resource utilization is set. When the exhibition hall resource utilization is less than or equal to the preset threshold, all rendering requests are allowed to be rendered according to the allocation variables at time point t. When the exhibition hall resource utilization is greater than the preset threshold, a dynamic allocation adjustment mechanism is adopted to adjust the allocation status of rendering requests, generate corresponding rendering strategies, and dynamically display the exhibition hall content according to the rendering strategies.

Images