A Deep Learning-Based Digital Exhibition Product Display Method

Abstract

This invention discloses a deep learning-based digital exhibition product display method, comprising: S1, constructing a self-evolving dynamic feature optimization network to optimize dynamic aggregation accuracy through feature adaptive iteration; S2, generating a unified dynamic display semantic model based on a holographic information multimodal fusion engine; S3, updating a personalized display intent model in real time based on environmental status and display objectives; S4, adjusting the display path of exhibits in real time based on an exhibition path optimization strategy; S5, generating a comprehensive virtual display effect of exhibit details using quantum detail enhancement generation technology; S6, realizing dynamic push and recommendation of multi-dimensional display content based on a spatiotemporal awareness interaction prediction mechanism; S7, constructing a lightweight quantum enhancement classification model to achieve efficient dynamic display management of exhibits; and S8, constructing an immersive multi-dimensional display effect of exhibits in a virtual exhibition space. This invention has the advantages of superior user interaction experience and high exhibition management efficiency.

Description

Technical Field

[0001] This invention relates to the field of product display technology, and in particular to a digital exhibition product display method based on deep learning. Background Technology

[0002] With the rapid development of digital technology, digital exhibitions have gradually become an important form of showcasing and disseminating product information. In the digital exhibition environment, achieving efficient presentation and dynamic interaction of exhibits through virtual display technology has become key to enhancing user experience. However, existing technologies still face many challenges in terms of the dynamic adaptability, multimodal integration, and personalized recommendations of product displays in digital exhibitions.

[0003] Traditional digital exhibition display methods mostly rely on static content presentation and simple rule-driven recommendations. While these methods can meet basic display needs, they often fall short when faced with complex user interactions, dynamic exhibition environments, and multi-dimensional exhibit display objectives. Specifically, existing technologies have significant shortcomings in the following aspects:

[0004] 1. Lack of dynamic adaptability: Existing display methods typically employ fixed display paths or content layouts, failing to respond in real-time to user interactions and changes in the exhibition environment. This static display model struggles to meet users' personalized needs, resulting in a rather monotonous and unappealing user experience.

[0005] 2. Insufficient Multimodal Fusion: During the exhibit display process, user behavior data, environmental data, and the multimodal features of the exhibits were not effectively integrated. Existing technologies lack efficient multimodal feature aggregation methods, resulting in insufficient semantic consistency and interactive effects in the displayed content, thus failing to achieve an immersive display experience.

[0006] 3. Limited Personalized Recommendation Capabilities: Traditional display methods typically rely on simple rules or statistical models for product recommendations, lacking in-depth modeling of user behavioral intentions. This approach cannot accurately predict user needs, resulting in low relevance and accuracy of recommended content.

[0007] 4. Inefficient display path optimization: Existing technologies rely heavily on static algorithms for planning and dynamically adjusting display paths, which are difficult to adapt to complex changes in exhibition scenarios and real-time user interaction feedback, and cannot provide efficient dynamic path optimization solutions.

[0008] 5. Insufficient Detail Rendering: Virtual detail displays of exhibits typically rely on fixed graphics rendering methods, lacking technical support for dynamically enhancing exhibit details. This deficiency limits the expressive power of the displayed content, making it difficult to fully showcase the overall appearance and value of the exhibits.

[0009] 6. Limitations in data processing capabilities: Existing methods are often limited by data dimensionality, computational complexity, and model generalization ability when processing cross-domain exhibit data, making it impossible to efficiently classify and dynamically manage complex exhibit features.

[0010] Therefore, how to provide a deep learning-based digital exhibition product display method is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0011] One objective of this invention is to propose a deep learning-based digital exhibition product display method. This invention fully utilizes technologies such as dynamic feature aggregation and adaptive display optimization, multimodal deep semantic fusion, personalized intent modeling, reinforcement learning path optimization, generative detail rendering, and virtual simulation mapping. It describes in detail the method for achieving dynamic adaptive display of exhibits, multi-dimensional content optimization, and personalized recommendations. It has the advantages of strong dynamic response capability, rich display content, superior user interaction experience, and high exhibition management efficiency.

[0012] A digital exhibition product display method based on deep learning according to an embodiment of the present invention includes the following steps:

[0013] S1. Construct a self-evolving dynamic feature optimization network, taking multi-dimensional data stream as input, dynamically capturing the changing features of exhibits in the spatiotemporal environment, and optimizing the dynamic aggregation accuracy through feature adaptive iteration.

[0014] S2. Based on the holographic information multimodal fusion engine, visual holographic stream, natural language semantic stream and tactile feedback data are deeply aligned and fused in the multimodal space to generate a unified dynamic display semantic model.

[0015] S3. Through a high-dimensional interactive intent guidance mechanism, based on the multi-dimensional intent analysis and prediction of user interaction behavior, and combined with the environmental status and display objectives, the personalized display intent model is updated in real time.

[0016] S4. Exhibition path optimization strategy based on deep reinforcement learning: By using the deep deterministic policy gradient algorithm, the real-time adjustment of the exhibit display path is achieved through continuous modeling of user behavior and environmental feedback in the state space.

[0017] S5. Employing quantum detail enhancement generation technology, combining exhibit feature information with a dynamic virtual environment, it generates a comprehensive virtual display effect of exhibit details;

[0018] S6. Through a spatiotemporal awareness-based interaction prediction mechanism, the system can dynamically push and recommend multi-dimensional content by modeling the timing of user interactions, the dynamics of exhibition scenes, and individual behavioral intentions in real time.

[0019] S7. Construct a lightweight quantum-enhanced classification model and use self-supervised learning to accurately classify cross-domain exhibit data, thereby achieving efficient dynamic display management of exhibits.

[0020] S8. A virtual simulation mapping-based exhibit reshaping method that uses cross-domain data generation and dynamic environment mapping technology to construct an immersive, multi-dimensional display effect of exhibits in a virtual exhibition space.

[0021] Optionally, S1 specifically includes:

[0022] S11. Distributed acquisition and structured preprocessing of the multidimensional input data of the exhibits. The input data includes the static feature data X of the exhibits. s Dynamic behavioral characteristic data X d The feature data stream is generated from the environmental context-aware data E(t) and the user multimodal interaction data U(t) through feature normalization, dimensionality compression and semantic mapping.

[0023] S12. In the feature extraction module, a hierarchical neural encoder is used to deconstruct the input data into layers, extracting the static feature data X. s Encode as a high-order feature vector F s And process dynamic behavioral feature data X using a time-series-based recurrent neural network. d From the context-aware data E(t) and the user's multimodal interaction data U(t), the corresponding temporal features F are generated. d (t), F e (t) and F u (t);

[0024] S13. Constructing a multi-input multi-output feature optimization network based on a self-evolving feature optimization algorithm:

[0025] F agg (t)=W s ·F s +W d (t)·F d (t)+W e (t)·F e (t)+W u (t)·F u (t);

[0026] Among them, F agg (t) represents the aggregation feature, W s W d (t), W e (t), W u (t) represents the dynamic weight coefficients of static, dynamic, environmental, and interactive features, respectively, and the weights are adjusted in real time through an adaptive evolution strategy combined with the feature importance function;

[0027] S14. An optimization mechanism driven by user feedback signals is used to establish an optimization function based on the real-time prediction error Δ(t) and the user response value R(t), and the feature weights W are updated through gradient updates. i (t+1):

[0028]

[0029] Where η is the dynamic learning rate and L represents the loss function;

[0030] S15. Optimize the feature vector F opt (t) is input to the feature dynamic control module, and processed through a multilayer perceptron network to realize the spatiotemporal dynamic feature modeling and adaptive expression of the exhibits.

[0031] Optionally, S2 specifically includes:

[0032] S21. Distributed acquisition and multimodal preprocessing are performed on visual holographic stream, natural language semantic stream, and tactile feedback data. Initial visual features V are generated respectively through feature decoder. raw (t), initial semantic features S raw (t) and initial tactile features H raw (t), and use an adaptive noise reduction module to eliminate redundant information to obtain optimized features V(t), S(t) and H(t);

[0033] S22. In the multimodal feature alignment module, a feature projection network based on a self-attention mechanism is constructed to project visual features V(t), semantic features S(t), and tactile features H(t) onto a unified high-dimensional representation space. Feature alignment is achieved through dynamic weight adjustment.

[0034] F align (t)=β v ·V′(t)+β s ·S′(t)+β h ·H′(t);

[0035] Among them, F align (t) represents the aligned multimodal features, and V′(t), S′(t), and H′(t) are the projected feature vectors, β v β s β h The feature alignment weights are dynamically updated in real time by an adaptive optimization strategy.

[0036] S23. Utilize a deep fusion engine based on a multi-layer attention mechanism to process F. align (t) Perform feature fusion and generate fused features F through a multi-layer attention mechanism. fusion (t):

[0037]

[0038] Where, α i (t) represents the dynamic attention weights for the i-th mode, F′ i (t) represents the enhanced representation of multimodal features;

[0039] S24. The fused feature F is processed through a deep semantic parsing network. fusion (t) Perform semantic decomposition and structured modeling to generate semantic hierarchy vector M. semantic (t), extracts semantic logical relationships and dynamic semantic expressions through a multi-layer semantic decoder;

[0040] S25. Establish a semantic consistency verification module to verify the generated semantic hierarchy vector M. semantic (t) Perform semantic integrity verification and logical consistency checks, and output an optimized dynamic presentation semantic model.

[0041] Optionally, S3 specifically includes:

[0042] S31. Collect multimodal user interaction behavior data, including user touch action data T m (t), Voice command data V n (t), gaze tracking data G p (t), pose recognition data P q (t) and interaction context variable C r (t), combined with environmental state data E k (t) and exhibit characteristic data O l (t), the acquired data is preprocessed in a structured manner using a multimodal feature decoder to generate the initial interaction features F. x (t);

[0043] S32. In the multi-dimensional intent parsing module, a multimodal feature embedding network is used to process the initial feature F. x (t) Perform high-dimensional feature projection to generate time series features F. y (t) models historical interaction behavior using a recurrent network based on a self-attention mechanism, generating predictive features F. z (t);

[0044] S33. Define a real-time dynamic intent optimization function based on user behavior characteristics and environmental context information:

[0045] I u (t)=δ1·F y (t)+δ2·E k (t)+δ3·O l (t)+δ4·Cr (t);

[0046] Among them, I u (t) represents the dynamic user intent model, where δ1, δ2, δ3, and δ4 are weight parameters that are dynamically updated using an optimization algorithm based on reinforcement learning.

[0047] S34. Using a multi-objective joint optimization strategy to optimize I u (t) Dynamically adjust and combine intent guidance mechanisms to optimize the matching between user preference expression and display goals, generating a personalized display intent model I. optimal (t):

[0048] I optimal (t)=ψ1·I u (t)+ψ2·Δ(T m (t), V n (t), G p (t));

[0049] Where ψ1 and ψ2 are weighting factors for multi-objective joint optimization, Δ(T) m (t), V n (t), G p (t) represents the dynamic correction function for user interaction behavior characteristics;

[0050] S35. Optimize the personalized display intent model I optimal (t) Input the dynamic display configuration module, and use the dynamic decision engine to determine the display order S of the exhibits. d Layout adjustment parameter L d Interaction Method A d Perform real-time optimization.

[0051] Optionally, S4 specifically includes:

[0052] S41. Construct a state-space model for exhibition path optimization, with the set of state variables Σ(t)={U h (t), E j (t),R k (t)},U h (t) represents the user's current behavior feature vector, E j (t) represents the environmental feedback signal, R k (t) represents the dynamic correlation characteristics between exhibits and user interests;

[0053] S42. Action space Λ(t) = {ξ1, ξ2, ..., ξ3} based on deep reinforcement learning m Each action ξ represents the adjustment method of the exhibit display path, including exhibit position swapping, dynamic insertion, priority reordering, and path expansion;

[0054] S43. Define the reward function Ω(t) for path optimization:

[0055] Ω(t)=κ1·I u (t)+κ2·C q (t)-κ3·T m (t);

[0056] Among them, I u (t) represents the user interest matching score, C q (t) represents the score for the coherence of the display path, T m (t) represents the time consumption for path adjustment, and κ1,κ2,κ3 are the weight factors of the reward function;

[0057] S44. Train the path optimization policy network using the deep deterministic policy gradient algorithm:

[0058]

[0059] Where Θ(t) are the parameters of the policy network, η is the learning rate, and Q(Σ(t),Λ(t)|Θ(t)) is the expected reward value of the state-action combination, which is estimated and updated using a deep neural network;

[0060] S45. In the path optimization decision module, the real-time user behavior data U is combined. real (t) and environmental dynamic feedback E dyn (t), the optimal path adjustment action sequence Λ is generated through the trained policy network. opt (t), and dynamically adjust the display order of exhibits P final (t).

[0061] Optionally, S5 specifically includes:

[0062] S51. Collect multi-dimensional feature data of the exhibits, including geometric morphology information G x Surface material characteristics T y Dynamic Behavior Sequence D z (t) and virtual environment mapping data E w (t), the original data is quantized through the feature quantization module to generate the quantized feature vector F. p (t);

[0063] S52, Utilizing Quantized Details Enhancement Networks for Quantized Features F p (t) Perform detail enhancement, and define the exhibit detail enhancement function:

[0064] F enhanced (t)=α1·G x +α2·T y+α3·Φ(D z (t), E w (t));

[0065] Among them, F enhanced (t) represents the enhanced exhibit features, α1, α2, α3 are weighting parameters, and Φ(D) represents the enhanced exhibit features. z (t),E w (t) represents the joint correction function between dynamic behavior and virtual environment;

[0066] S53, Combining the virtual environment with the enhanced exhibit features F enhanced (t) Perform dynamic rendering, and use a detail enhancement generator to build a dynamic rendering model R. dynamic (t):

[0067] R dynamic (t)=β1·F enhanced (t)+β2·Ψ(E w (t));

[0068] Where β1 and β2 are rendering parameters, Ψ(E w (t) represents the environment mapping optimization function;

[0069] S54. The dynamic rendering model R is verified through a multi-layer detail verification module. dynamic The geometry, surface texture, and dynamic behavior of (t) are verified layer by layer, and the final optimized virtual display model M of the exhibit is generated. optimized (t);

[0070] S55, The optimized virtual display model M optimized (t) Input display control module, through dynamic interaction engine, adjusts the display perspective, detail, and dynamic response of exhibits based on real-time user feedback, generating a real-time updated virtual display effect V. final (t).

[0071] Optionally, S6 specifically includes:

[0072] S61. Collect multimodal interaction data of users, including user interaction time sequence features T x (t), Dynamic state of the exhibition scene S y (t) and individual behavioral intention B z (t), through the data preprocessing module, the data is normalized and hierarchically encoded to generate a spatiotemporal awareness feature set F. spatio (t)={T x (t),S y (t), B z (t)};

[0073] S62. Using a recurrent neural network based on spatiotemporal awareness to analyze F spatio (t) Perform high-dimensional encoding to construct a user interaction prediction model:

[0074] P interaction (t)=θ1·Λ(T x (t))+θ2·Γ(S y (t))+θ3·Ω(B z (t));

[0075] Among them, P interaction (t) represents the predicted value of user interaction, Λ(T) x (t)), Γ(S) y (t)), Ω(B z (t) are the spatiotemporal encoding functions for temporal features, scene dynamics, and behavioral intent, respectively, and θ1, θ2, θ3 are dynamic weight parameters;

[0076] S63, Combining historical behavioral patterns H m (t) and real-time user interaction features P interaction (t), define the priority scoring function for recommended content:

[0077] R priority (t)=η1·P interaction (t)+η2·H m (t)+η3·C k (t);

[0078] Among them, R priority (t) represents the dynamic priority of the recommended content, η1, η2, η3 are priority weight parameters, and C k (t) represents the matching score of the current exhibit features;

[0079] S64. Dynamically prioritize recommended content using a multi-dimensional dynamic content push engine. priority (t) is sorted and combined with the user preference distribution U pref (t) and scene dynamic feedback E dyn (t), generating a real-time recommended content set D recommend (t);

[0080] S65. Utilize the dynamic display optimization module to optimize D recommend (t) Input display logic controller, which generates a personalized dynamic display scheme by adjusting the display order, display style and interaction method in real time. dynamic (t).

[0081] Optionally, S7 specifically includes:

[0082] S71. Collect cross-domain exhibit data, including the static attribute features G of the exhibits. s Dynamic Behavior Sequence H t (t) and multimodal interaction information I u (t), the original data is normalized, noise filtered, and high-dimensional embedded through the feature preprocessing module to generate the initial feature set F. init (t)={G s H t (t), I u (t)};

[0083] S72. Construct a lightweight quantum-enhanced classification model for F. init (t) Perform feature encoding and quantum enhancement:

[0084] F enh (t)=δ1·Ψ1(G s )+δ2·Ψ2(H t (t))+δ3·Ψ3(I u (t));

[0085] Among them, F enh (t) represents the feature representation after quantum enhancement, Ψ1(G s ), Ψ2(H t (t)), Ψ3(I u (t) represents the quantum enhancement functions for static features, dynamic features, and multimodal interactive data, respectively, and δ1, δ2, and δ3 are the feature enhancement weight parameters;

[0086] S73. Utilizing a self-supervised learning mechanism for F enh (t) Perform feature classification:

[0087]

[0088] Among them, L cls (t) is the classification loss function, Q(y) j |(t)) represents category y j The predicted probability distribution, ξ j Balance the weights for each category;

[0089] S74, L is classified using a lightweight classification network. cls (t) is optimized by using gradient descent combined with a self-supervised learning strategy to adjust the model parameters and generate the optimized classification result R. cls (t);

[0090] S75. Classify the results R cls (t) Input the exhibit dynamic management module, dynamically group and optimize exhibits according to category, and generate a set of display solutions S based on user needs and scenario context.disp (t).

[0091] Optionally, S8 specifically includes:

[0092] S81. Collect cross-domain data of exhibits, including the physical attribute characteristics P of the exhibits. a Dynamic Behavior Sequence D b (t), Material details M c and environmental context information E d (t), through the cross-domain data fusion module, the data is normalized and multimodal encoded to generate a feature set F. cross (t)={P a D b (t), M c E d (t)};

[0093] S82, Based on virtual simulation mapping technology, F cross (t) Perform dynamic feature reshaping:

[0094] F remap (t)=κ1·P a +κ2·Ξ(D b (t), M c )+κ3·Ψ(E d (t));

[0095] Among them, F remap (t) represents the reshaped feature set, κ1,κ2,κ3 are the feature mapping weight parameters, and Ξ(D b (t), M c Ψ(E) is the joint mapping function between dynamic behavior and material properties. d (t) is the dynamic correction function for the environment context;

[0096] S83, Utilizing a multi-dimensional virtual rendering engine for F remap (t) Perform immersive multi-dimensional rendering and construct a virtual display model R. immersive (t):

[0097] R immersive (t)=λ1·F remap (t)+λ2·Φ(E d (t));

[0098] Where λ1, λ2 are rendering parameters, Φ(E d (t) represents the multidimensional rendering function of the virtual environment;

[0099] S84. Verify the virtual display model R through a multi-layered verification module. immersiveThe geometric shape, material details, dynamic behavior, and environmental interaction effects of (t) are hierarchically verified to generate an optimized virtual exhibit model M. optimized (t);

[0100] S85, M optimized (t) Applied to virtual exhibition spaces, using a dynamic scene control module combined with real-time user interaction data U e (t) and changes in the exhibition environment S f (t) Dynamically adjust the display details, interaction logic, and environmental adaptation of virtual exhibits to generate an immersive, multi-dimensional display effect. final (t).

[0101] The beneficial effects of this invention are:

[0102] (1) By combining dynamic feature aggregation network and adaptive display optimization algorithm, this invention realizes efficient feature capture and real-time adjustment of exhibits in dynamic exhibition scenarios, enabling the system to accurately adapt to user behavior and environmental changes, thereby improving the dynamic response capability of the display content and solving the shortcomings of traditional display methods that cannot adapt in real time.

[0103] (2) This invention provides multimodal deep understanding and interactive optimization of exhibit display content through a holographic information multimodal fusion engine, combined with semantic deep alignment and feature dynamic fusion technology, enabling the display system to work efficiently at the multimodal levels of vision, language and touch, thereby enhancing the user’s immersive experience and sense of interaction.

[0104] (3) The path optimization algorithm based on reinforcement learning in this invention uses the deep deterministic policy gradient method to dynamically optimize the display path of exhibits in real time, which effectively shortens the path planning time and improves the display efficiency, and solves the problems of slow response and poor adaptability of traditional path optimization methods.

[0105] (4) This invention generates a high-precision virtual display effect of exhibit details by combining the dynamic features of exhibits with the virtual environment through quantum detail enhancement generation technology, which significantly improves the detail expression and visual quality of exhibits and overcomes the limitations of insufficient detail rendering in traditional methods.

[0106] (5) This invention utilizes a lightweight quantum-enhanced classification model and a self-supervised learning mechanism to achieve efficient classification and dynamic management of cross-domain exhibit data, significantly improving the accuracy and processing efficiency of exhibit classification, and providing strong support for accurate recommendation and dynamic management of exhibition content. Attached Figure Description

[0107] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0108] Figure 1 This is a flowchart of a deep learning-based digital exhibition product display method proposed in this invention. Detailed Implementation

[0109] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0110] refer to Figure 1 A deep learning-based method for product display in digital exhibitions includes the following steps:

[0111] S1. Construct a self-evolving dynamic feature optimization network, taking multi-dimensional data stream as input, dynamically capturing the changing features of exhibits in the spatiotemporal environment, and optimizing the dynamic aggregation accuracy through feature adaptive iteration.

[0112] In this embodiment, S1 specifically includes:

[0113] S11. Distributed acquisition and structured preprocessing of the multidimensional input data of the exhibits. The input data includes the static feature data X of the exhibits. s Dynamic behavioral characteristic data X d The feature data stream is generated from the environmental context-aware data E(t) and the user multimodal interaction data U(t) through feature normalization, dimensionality compression and semantic mapping.

[0114] S12. In the feature extraction module, a hierarchical neural encoder is used to deconstruct the input data into layers, extracting the static feature data X. s Encode as a high-order feature vector F s And process dynamic behavioral feature data X using a time-series-based recurrent neural network. d From the context-aware data E(t) and the user's multimodal interaction data U(t), the corresponding temporal features F are generated. d (t), F e (t) and F u (t);

[0115] S13. Constructing a multi-input multi-output feature optimization network based on a self-evolving feature optimization algorithm:

[0116] F agg (t)=W s ·F s +W d (t)·Fd (t)+W e (t)·F e (t)+W u (t)·F u (t);

[0117] Among them, F agg (t) represents the aggregation feature, W s W d (y), W e (t), W u (t) represents the dynamic weight coefficients of static, dynamic, environmental, and interactive features, respectively, and the weights are adjusted in real time through an adaptive evolution strategy combined with the feature importance function;

[0118] S14. An optimization mechanism driven by user feedback signals is used to establish an optimization function based on the real-time prediction error Δ(t) and the user response value R(t), and the feature weights W are updated through gradient updates. i (t+1):

[0119]

[0120] Where η is the dynamic learning rate and L represents the loss function;

[0121] S15. Optimize the feature vector F opt (t) is input to the feature dynamic control module, and processed through a multilayer perceptron network to realize the spatiotemporal dynamic feature modeling and adaptive expression of the exhibits.

[0122] S2. Based on the holographic information multimodal fusion engine, visual holographic stream, natural language semantic stream and tactile feedback data are deeply aligned and fused in the multimodal space to generate a unified dynamic display semantic model.

[0123] In this embodiment, S2 specifically includes:

[0124] S21. Distributed acquisition and multimodal preprocessing are performed on visual holographic stream, natural language semantic stream, and tactile feedback data. Initial visual features V are generated respectively through feature decoder. raw (t), initial semantic features S raw (t) and initial tactile features H raw (t), and use an adaptive noise reduction module to eliminate redundant information to obtain optimized features V(t), S(t) and H(t);

[0125] S22. In the multimodal feature alignment module, a feature projection network based on a self-attention mechanism is constructed to project visual features V(t), semantic features S(t), and tactile features H(t) onto a unified high-dimensional representation space. Feature alignment is achieved through dynamic weight adjustment.

[0126] F align (t)=β v ·V′(t)+β s ·S′(t)+β h ·H′(t);

[0127] Among them, F align (t) represents the aligned multimodal features, and V′(t), S′(t), and H′(t) are the projected feature vectors, β v β s β h The feature alignment weights are dynamically updated in real time by an adaptive optimization strategy.

[0128] S23. Utilize a deep fusion engine based on a multi-layer attention mechanism to process F. align (t) Perform feature fusion and generate fused features F through a multi-layer attention mechanism. fusion (t):

[0129]

[0130] Where, α i (t) represents the dynamic attention weights for the i-th mode, F′ i (t) represents the enhanced representation of multimodal features;

[0131] S24. The fused feature F is processed through a deep semantic parsing network. fusion (t) Perform semantic decomposition and structured modeling to generate semantic hierarchy vector M. semantic (t), extracts semantic logical relationships and dynamic semantic expressions through a multi-layer semantic decoder;

[0132] S25. Establish a semantic consistency verification module to verify the generated semantic hierarchy vector M. semantic (t) Perform semantic integrity verification and logical consistency checks, and output an optimized dynamic presentation semantic model.

[0133] S3. Through a high-dimensional interactive intent guidance mechanism, based on the multi-dimensional intent analysis and prediction of user interaction behavior, and combined with the environmental status and display objectives, the personalized display intent model is updated in real time.

[0134] In this embodiment, S3 specifically includes:

[0135] S31. Collect multimodal user interaction behavior data, including user touch action data T m (t), Voice command data V n (t), gaze tracking data G p (t), pose recognition data P q (t) and interaction context variable C r(t), combined with environmental state data E k (t) and exhibit characteristic data O l (t), the acquired data is preprocessed in a structured manner using a multimodal feature decoder to generate the initial interaction features F. x (t);

[0136] S32. In the multi-dimensional intent parsing module, a multimodal feature embedding network is used to process the initial feature F. x (t) Perform high-dimensional feature projection to generate time series features F. y (t) models historical interaction behavior using a recurrent network based on a self-attention mechanism, generating predictive features F. z (t);

[0137] S33. Define a real-time dynamic intent optimization function based on user behavior characteristics and environmental context information:

[0138] I u (t)=δ1·F y (t)+δ2·E k (t)+δ3·O l (t)+δ4·C r (t);

[0139] Among them, I u (t) represents the dynamic user intent model, where δ1, δ2, δ3, and δ4 are weight parameters that are dynamically updated using an optimization algorithm based on reinforcement learning.

[0140] S34. Using a multi-objective joint optimization strategy to optimize I i (t) Dynamically adjust and combine intent guidance mechanisms to optimize the matching between user preference expression and display goals, generating a personalized display intent model I. optimal (t):

[0141] I optimal (t)=ψ1·I u (t)+ψ2·Δ(T m (t), V n (t), G p (t));

[0142] Where ψ1 and ψ2 are weighting factors for multi-objective joint optimization, Δ(T) m (t), V n (t), G p (t) represents the dynamic correction function for user interaction behavior characteristics;

[0143] S35. Optimize the personalized display intent model I optimal(t) Input the dynamic display configuration module, and use the dynamic decision engine to determine the display order S of the exhibits. d Layout adjustment parameter L d Interaction Method A d Perform real-time optimization.

[0144] S4. Exhibition path optimization strategy based on deep reinforcement learning: By using the deep deterministic policy gradient algorithm, the real-time adjustment of the exhibit display path is achieved through continuous modeling of user behavior and environmental feedback in the state space.

[0145] In this embodiment, S4 specifically includes:

[0146] S41. Construct a state-space model for exhibition path optimization, with the set of state variables Σ(t)={U h (t), E j (t), R k (t)},U h (t) represents the user's current behavior feature vector, E j (y) represents the environmental feedback signal, R k (t) represents the dynamic correlation characteristics between exhibits and user interests;

[0147] S42. Action space Λ(t) = {ξ1, ξ2, ..., ξ} based on deep reinforcement learning m Each action ξ represents the adjustment method of the exhibit display path, including exhibit position swapping, dynamic insertion, priority reordering, and path expansion;

[0148] S43. Define the reward function Ω(t) for path optimization:

[0149] Ω(t)=κ1·I u (t)+κ2·C q (t)-κ3·T m (t);

[0150] Among them, I u (t) represents the user interest matching score, C q (t) represents the score for the coherence of the display path, T m (t) represents the time consumption for path adjustment, and κ1,κ2,κ3 are the weight factors of the reward function;

[0151] S44. Train the path optimization policy network using the deep deterministic policy gradient algorithm:

[0152]

[0153] Where Θ(t) are the parameters of the policy network, η is the learning rate, and Q(Σ(t), Λ(t)|Θ(t)) is the expected reward value of the state and action combination, which is estimated and updated using a deep neural network;

[0154] S45. In the path optimization decision module, the real-time user behavior data U is combined. real (t) and environmental dynamic feedback E dyn (t), the optimal path adjustment action sequence Λ is generated through the trained policy network. opt (t), and dynamically adjust the display order of exhibits P final (t).

[0155] S5. Employing quantum detail enhancement generation technology, combining exhibit feature information with a dynamic virtual environment, it generates a comprehensive virtual display effect of exhibit details;

[0156] In this embodiment, S5 specifically includes:

[0157] S51. Collect multi-dimensional feature data of the exhibits, including geometric morphology information G x Surface material characteristics T y Dynamic Behavior Sequence D z (t) and virtual environment mapping data E w (t), the original data is quantized through the feature quantization module to generate the quantized feature vector F. p (t);

[0158] S52, Utilizing Quantized Details Enhancement Networks for Quantized Features F p (t) Perform detail enhancement, and define the exhibit detail enhancement function:

[0159] F enhanced (t)=α1·G x +α2·T y +α3·Φ(D z (t),E w (t));

[0160] Among them, F enhanced (t) represents the enhanced exhibit features, α1, α2, α3 are weighting parameters, and Φ(D) represents the enhanced exhibit features. z (t), E w (t) represents the joint correction function between dynamic behavior and virtual environment;

[0161] S53, Combining the virtual environment with the enhanced exhibit features F enhanced (t) Perform dynamic rendering, and use a detail enhancement generator to build a dynamic rendering model R. dynamic (t):

[0162] Rdynamic (t)=β1·F enhanced (t)+β2·Ψ(E w (t));

[0163] Where β1 and β2 are rendering parameters, Ψ(E w (t) represents the environment mapping optimization function;

[0164] S54. The dynamic rendering model R is verified through a multi-layer detail verification module. dynamic The geometry, surface texture, and dynamic behavior of (t) are verified layer by layer, and the final optimized virtual display model M of the exhibit is generated. optimized (t);

[0165] S55, The optimized virtual display model M optimized (t) Input display control module, through dynamic interaction engine, adjusts the display perspective, detail, and dynamic response of exhibits based on real-time user feedback, generating a real-time updated virtual display effect V. final (t).

[0166] S6. Through a spatiotemporal awareness-based interaction prediction mechanism, the system can dynamically push and recommend multi-dimensional content by modeling the timing of user interactions, the dynamics of exhibition scenes, and individual behavioral intentions in real time.

[0167] In this embodiment, S6 specifically includes:

[0168] S61. Collect multimodal interaction data of users, including user interaction time sequence features T x (t), Dynamic state of the exhibition scene S y (t) and individual behavioral intention B z (t), through the data preprocessing module, the data is normalized and hierarchically encoded to generate a spatiotemporal awareness feature set F. spatio (t)={T x (t), S y (t), B z (t)};

[0169] S62. Using a recurrent neural network based on spatiotemporal awareness to analyze F spatio (t) Perform high-dimensional encoding to construct a user interaction prediction model:

[0170] P interaction (t)=θ1·Λ(T x (t))+θ2·Γ(S y (t))+θ3·Ω(B z (t));

[0171] Among them, P interaction(t) represents the predicted value of user interaction, Λ(T) x (t)), Γ(S) y (t)), Ω(B z (t) are the spatiotemporal encoding functions for temporal features, scene dynamics, and behavioral intent, respectively, and θ1, θ2, θ3 are dynamic weight parameters;

[0172] S63, Combining historical behavioral patterns H m (t) and real-time user interaction features P interaction (t), define the priority scoring function for recommended content:

[0173] R priority (t)=η1·P interaction (t)+η2·H m (t)+η3·C k (t);

[0174] Among them, R priority (t) represents the dynamic priority of the recommended content, η1, η2, η3 are priority weight parameters, and C k (t) represents the matching score of the current exhibit features;

[0175] S64. Dynamically prioritize recommended content through a multi-dimensional dynamic content push engine. priority (t) is sorted and combined with the user preference distribution U pref (t) and scene dynamic feedback E dyn (t), generating a real-time recommended content set D recommend (t);

[0176] S65. Utilize the dynamic display optimization module to optimize D recommend (t) Input display logic controller, which generates a personalized dynamic display scheme by adjusting the display order, display style and interaction method in real time. dynamic (t).

[0177] S7. Construct a lightweight quantum-enhanced classification model and use self-supervised learning to accurately classify cross-domain exhibit data, thereby achieving efficient dynamic display management of exhibits.

[0178] In this embodiment, S7 specifically includes:

[0179] S71. Collect cross-domain exhibit data, including the static attribute features G of the exhibits. s Dynamic Behavior Sequence H t (t) and multimodal interaction information I u (t), the original data is normalized, noise filtered, and high-dimensional embedded through the feature preprocessing module to generate the initial feature set F. init (t)={G sH t (t),I u (t)};

[0180] S72. Construct a lightweight quantum-enhanced classification model for F. init (t) Perform feature encoding and quantum enhancement:

[0181] F enh (t)=δ1·Ψ1(G s )+δ2·Ψ2(H t (t))+δ3·Ψ3(I u (t));

[0182] Among them, F enh (t) represents the feature representation after quantum enhancement, Ψ1(G s ), Ψ2(H t (t)), Ψ3(I u (t) represents the quantum enhancement functions for static features, dynamic features, and multimodal interactive data, respectively, and δ1, δ2, and δ3 are the feature enhancement weight parameters;

[0183] S73. Utilizing a self-supervised learning mechanism for F enh (t) Perform feature classification:

[0184]

[0185] Among them, L cls (t) is the classification loss function, Q(y) i |(t)) represents category y j The predicted probability distribution, ξ j Balance the weights for each category;

[0186] S74, L is classified using a lightweight classification network. cls (t) is optimized by using gradient descent combined with a self-supervised learning strategy to adjust the model parameters and generate the optimized classification result R. cls (t);

[0187] S75. Classify the results R cls (t) Input the exhibit dynamic management module, dynamically group and optimize exhibits according to category, and generate a set of display solutions S based on user needs and scenario context. disp (t).

[0188] S8. A virtual simulation mapping-based exhibit reshaping method that uses cross-domain data generation and dynamic environment mapping technology to construct an immersive, multi-dimensional display effect of exhibits in a virtual exhibition space.

[0189] In this embodiment, S8 specifically includes:

[0190] S81. Collect cross-domain data of exhibits, including the physical attribute characteristics P of the exhibits. a Dynamic Behavior Sequence D b (t), Material details M c and environmental context information E d (t), through the cross-domain data fusion module, the data is normalized and multimodal encoded to generate a feature set F. cross (t)={P a D b (t), M c E d (t)};

[0191] S82, Based on virtual simulation mapping technology, F cross (t) Perform dynamic feature reshaping:

[0192] F remap (t)=κ1·P a +κ2·Ξ(D b (t), M c )+κ3·Ψ(E d (t));

[0193] Among them, F remap (t) represents the reshaped feature set, κ1,κ2,κ3 are the feature mapping weight parameters, and Ξ(D b (t), M c Ψ(E) is the joint mapping function between dynamic behavior and material properties. d (t) is the dynamic correction function for the environment context;

[0194] S83, Utilizing a multi-dimensional virtual rendering engine for F remap (t) Perform immersive multi-dimensional rendering and construct a virtual display model R. immersive (t):

[0195] R immersive (t)=λ1·F remap (t)+λ2·Φ(E d (t));

[0196] Where λ1, λ2 are rendering parameters, Φ(E d (t) represents the multidimensional rendering function of the virtual environment;

[0197] S84. Verify the virtual display model R through a multi-layered verification module. immersive The geometric shape, material details, dynamic behavior, and environmental interaction effects of (t) are hierarchically verified to generate an optimized virtual exhibit model M. optimized (t);

[0198] S85, Moptimized (t) Applied to virtual exhibition spaces, using a dynamic scene control module combined with real-time user interaction data U e (t) and changes in the exhibition environment S f (t) Dynamically adjust the display details, interaction logic, and environmental adaptation of virtual exhibits to generate an immersive, multi-dimensional display effect. final (t).

[0199] Example 1:

[0200] To verify the feasibility of this invention, the deep learning-based digital exhibition product display method was applied to an international digital technology exhibition. The exhibition aimed to showcase various technological products, including smart home devices, virtual reality devices, and related innovative technologies, through digital means. The exhibition attracted over 200 exhibitors and more than 5,000 visitors from around the world. The exhibition setting was complex, visitor needs were diverse, and the variety of exhibits was extensive; traditional static display methods could no longer meet the requirements for personalized needs and real-time response.

[0201] In this scenario, the digital exhibition system is deployed based on the method of this invention, encompassing dynamic feature aggregation, path optimization, detail enhancement, multimodal fusion, and virtual simulation mapping technologies. The system comprehensively applies this method throughout the entire process, from visitor behavior collection to exhibit display optimization. The specific implementation steps and actual results are as follows:

[0202] The system first collects real-time data on visitor interaction behavior, exhibit environment data, and dynamic changes in the exhibition scene through a dynamic feature aggregation module. This data includes visitor gaze patterns, dwell time, and touch responses to specific exhibits. In initial testing, the system processed over 50GB of data daily and completed feature extraction and optimization within seconds. This feature significantly improves the exhibits' dynamic adaptability, overcoming the limitations of traditional display models that cannot quickly respond to user needs.

[0203] Regarding path optimization, the system uses a reinforcement learning algorithm to plan dynamically adjusted visitor paths. During peak hours, visitor flow is dense, and traditional methods for path planning are time-consuming and cannot be dynamically adjusted. Test results show that after adopting the path optimization module of this invention, the response time for path planning is reduced from an average of 45 seconds to 3 seconds, and the proportion of visitors in congested areas is reduced.

[0204] The system also provides high-precision rendering of virtual display content through quantum detail enhancement generation technology. Taking a smart home device as an example, its virtual display, supported by a multi-dimensional rendering engine, presents realistic surface textures, dynamic behavior simulations, and lighting effects. Visitors' dwell time on the exhibit has increased from an average of 30 seconds in traditional displays to 85 seconds, significantly enhancing the exhibit's appeal.

[0205] In terms of personalized recommendations, the system uses a spatiotemporally aware interaction prediction module and a lightweight quantum-enhanced classification model to push exhibits that match visitors' interests. On one particular day, a visitor showed a high level of interest in virtual reality equipment. The system captured the visitor's behavior and interaction patterns in real time and accurately recommended five relevant devices, achieving a recommendation accuracy of 94%. Visitor feedback on the recommended content indicated that the relevance of the recommendations was significantly better than traditional rule-based recommendation methods.

[0206] Furthermore, the virtual simulation mapping module of this invention supports multi-dimensional display of exhibits in a virtual exhibition space. In testing, a drone product achieved flight dynamic simulation and environment-adaptive display through this module. User interaction frequency in the virtual scene increased by 120%, and the intended purchase volume of exhibits increased by 38%. Specific performance data is shown in Table 1 below:

[0207] Table 1. Analysis of the Deployment Effect of the Digital Exhibition System

[0208]

[0209] As demonstrated by the above tests and practical applications, this invention significantly improves the dynamic adaptability of exhibit display, the efficiency of display path optimization, and the accuracy of personalized content recommendations in complex digital exhibition scenarios. The system exhibits superior performance in key technologies such as multimodal fusion, real-time response, and virtual simulation display, overcoming many shortcomings of traditional display methods in terms of dynamic adaptability, data processing efficiency, and user experience optimization.

[0210] Specifically, this invention achieves efficient adaptation of exhibits to environmental changes through dynamic feature aggregation technology, significantly shortening response time and providing a guarantee for intelligent management in complex exhibition scenarios. Through the path optimization module, the efficiency of visitor path planning during peak hours at exhibitions is improved by 15 times, while significantly reducing the proportion of congested areas, thereby enhancing visitor mobility and satisfaction.

[0211] In terms of multimodal fusion and recommendation, this invention accurately identifies visitors' interests based on a deep learning prediction mechanism and provides personalized exhibit recommendations by combining quantum-enhanced classification technology, improving the recommendation accuracy to 94% and effectively meeting visitors' personalized needs.

[0212] Furthermore, the application of this invention in the field of virtual display has comprehensively enhanced the display effect of exhibits. Through generative detail enhancement and virtual simulation mapping technology, the detail expression and interactive experience of exhibits have reached industry-leading levels. User interaction frequency and purchase intention rates have significantly increased, further validating the practical value of this invention in optimizing exhibition effects and driving commercial conversion.

[0213] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A deep learning-based method for product display in digital exhibitions, characterized in that, Includes the following steps: S1. Construct a self-evolving dynamic feature optimization network, taking multi-dimensional data stream as input, dynamically capturing the changing features of exhibits in the spatiotemporal environment, and optimizing the dynamic aggregation accuracy through feature adaptive iteration. S2. Based on the holographic information multimodal fusion engine, visual holographic stream, natural language semantic stream and tactile feedback data are deeply aligned and fused in the multimodal space to generate a unified dynamic display semantic model. S3. Through a high-dimensional interactive intent guidance mechanism, based on the multi-dimensional intent analysis and prediction of user interaction behavior, and combined with the environmental status and display objectives, the personalized display intent model is updated in real time. S4. Exhibition path optimization strategy based on deep reinforcement learning: By using the deep deterministic policy gradient algorithm, the real-time adjustment of the exhibit display path is achieved through continuous modeling of user behavior and environmental feedback in the state space. S5. Employing quantum detail enhancement generation technology, combining exhibit feature information with a dynamic virtual environment, it generates a comprehensive virtual display effect of exhibit details; S6. Through a spatiotemporal awareness-based interactive prediction mechanism, the system can dynamically push and recommend multi-dimensional content by modeling the timing of user interactions, the dynamics of exhibition scenes, and individual behavioral intentions in real time. S7. Construct a lightweight quantum-enhanced classification model and use self-supervised learning to accurately classify cross-domain exhibit data, thereby achieving efficient dynamic display management of exhibits. S8. A virtual simulation mapping-based exhibit reshaping method that uses cross-domain data generation and dynamic environment mapping technology to construct an immersive, multi-dimensional display effect of exhibits in a virtual exhibition space.

2. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S1 specifically includes: S11. Distributed acquisition and structured preprocessing of the multidimensional input data of the exhibits. The input data includes the static feature data X of the exhibits. s Dynamic behavioral characteristic data X d The feature data stream is generated from the environmental context-aware data E(t) and the user multimodal interaction data U(t) through feature normalization, dimensionality compression and semantic mapping. S12. In the feature extraction module, a hierarchical neural encoder is used to deconstruct the input data into layers, extracting the static feature data X. s Encode as a high-order feature vector F s And process dynamic behavioral feature data X using a time-series-based recurrent neural network. d From the context-aware data E(t) and the user's multimodal interaction data U(t), the corresponding temporal features F are generated. d (t), F e (t) and F u (t); S13. Constructing a multi-input multi-output feature optimization network based on a self-evolving feature optimization algorithm: F agg (t)=W s ·F s +W d (t)·F d (t)+W e (t)·F e (t)+W u (t)·F u (t); Among them, F agg (t) represents the aggregation feature, W s W d (t), W e (t), W u (t) represents the dynamic weight coefficients of static, dynamic, environmental, and interactive features, respectively, and the weights are adjusted in real time through an adaptive evolution strategy combined with the feature importance function; S14. An optimization mechanism driven by user feedback signals is used to establish an optimization function based on the real-time prediction error Δ(t) and the user response value R(t), and the feature weights W are updated through gradient updates. i (t+1): Where η is the dynamic learning rate and L represents the loss function; S15. Optimize the feature vector F opt (t) is input to the feature dynamic control module, and processed through a multilayer perceptron network to realize the spatiotemporal dynamic feature modeling and adaptive expression of the exhibits.

3. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S2 specifically includes: S21. Distributed acquisition and multimodal preprocessing are performed on visual holographic stream, natural language semantic stream, and tactile feedback data. Initial visual features V are generated respectively through feature decoder. raw (t), initial semantic features S raw (t) and initial tactile features H raw (t), and use an adaptive noise reduction module to eliminate redundant information to obtain optimized features V(t), S(t) and H(t); S22. In the multimodal feature alignment module, a feature projection network based on a self-attention mechanism is constructed to project visual features V(t), semantic features S(t), and tactile features H(t) onto a unified high-dimensional representation space. Feature alignment is achieved through dynamic weight adjustment. F align (t)=β v ·V′(t)+β s ·S′(t)+β h ·H′(t); Among them, F align (t) represents the aligned multimodal features, and V′(t), S′(t), and H′(t) are the projected feature vectors, β v β s β h The feature alignment weights are dynamically updated in real time by an adaptive optimization strategy. S23. Utilize a deep fusion engine based on a multi-layer attention mechanism to process F. align (t) Perform feature fusion and generate fused features F through a multi-layer attention mechanism. fusion (t): Where, α i (t) represents the dynamic attention weights for the i-th mode, F′ i (t) represents the enhanced representation of multimodal features; S24. The fused feature F is processed through a deep semantic parsing network. fusion (t) Perform semantic decomposition and structured modeling to generate semantic hierarchy vector M. semantic (t), extracts semantic logical relationships and dynamic semantic expressions through a multi-layer semantic decoder; S25. Establish a semantic consistency verification module to verify the generated semantic hierarchy vector M. semantic (t) Perform semantic integrity verification and logical consistency checks, and output an optimized dynamic presentation semantic model.

4. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S3 specifically includes: S31. Collect multimodal user interaction behavior data, including user touch action data T m (t), Voice command data V n (t), gaze tracking data G p (t), pose recognition data P q (t) and interaction context variable C r (t), combined with environmental state data E k (t) and exhibit characteristic data O l (t), the acquired data is preprocessed in a structured manner using a multimodal feature decoder to generate the initial interaction features F. x (t); S32. In the multi-dimensional intent parsing module, a multimodal feature embedding network is used to process the initial feature F. x (t) Perform high-dimensional feature projection to generate time series features F y (t) models historical interaction behavior using a recurrent network based on a self-attention mechanism, generating predictive features F. z (t); S33. Define a real-time dynamic intent optimization function based on user behavior characteristics and environmental context information: I u (t)=δ1·F y (t)+δ2·E k (t)+δ3·O l (t)+δ4·C r (t); Among them, I u (t) represents the dynamic user intent model, where δ1, δ2, δ3, and δ4 are weight parameters that are dynamically updated using an optimization algorithm based on reinforcement learning. S34. Using a multi-objective joint optimization strategy to optimize I u (t) Dynamically adjust and combine intent guidance mechanisms to optimize the matching between user preference expression and display goals, generating a personalized display intent model I. optimal (t): I optimal (t)=ψ1·I u (t)+ψ2·Δ(T m (t),V n (t),G p (t)); Where ψ1 and ψ2 are weighting factors for multi-objective joint optimization, Δ(T) m (t), V n (t), G p (t) represents the dynamic correction function for user interaction behavior characteristics; S35. Optimize the personalized display intent model I optimal (t) Input the dynamic display configuration module, and use the dynamic decision engine to determine the display order S of the exhibits. d Layout adjustment parameter L d Interaction Method A d Perform real-time optimization.

5. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S4 specifically includes: S41. Construct a state-space model for exhibition path optimization, with the set of state variables Σ(t)={U h (t), E j (t), R k (t)},U h (t) represents the user's current behavior feature vector, E j (t) represents the environmental feedback signal, R k (t) represents the dynamic correlation characteristics between exhibits and user interests; S42. Action space Λ(t) = {ξ1, ξ2, ..., ξ} based on deep reinforcement learning m }, each action ξ i This indicates how the exhibit display path can be adjusted, including exhibit position swapping, dynamic insertion, priority reordering, and path expansion; S43. Define the reward function Ω(t) for path optimization: Ω(t)=κ1·I u (t)+κ2·C q (t)-κ3·T m (t); Among them, I u (t) represents the user interest matching score, C q (t) represents the score for the coherence of the display path, T m (t) represents the time consumption for path adjustment, and κ1,κ2,κ3 are the weight factors of the reward function; S44. Train the path optimization policy network using the deep deterministic policy gradient algorithm: in, Here are the parameters of the policy network, and η is the learning rate. The expected reward value of the combination of state and action is estimated and updated using a deep neural network; S45. In the path optimization decision module, the real-time user behavior data U is combined. real (t) and environmental dynamic feedback E dyn (t), the optimal path adjustment action sequence Λ is generated through the trained policy network. opt (t), and dynamically adjust the display order of exhibits P final (t).

6. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S5 specifically includes: S51. Collect multi-dimensional feature data of the exhibits, including geometric morphology information G x Surface material characteristics T y Dynamic Behavior Sequence D z (t) and virtual environment mapping data E w (t), the original data is quantized through the feature quantization module to generate the quantized feature vector F. p (t); S52, Utilizing Quantized Details Enhancement Networks for Quantized Features F p (t) Perform detail enhancement, and define the exhibit detail enhancement function: f enhanced (t)=α1·G x +α2·T y +α3·Φ(D z (t),E w (t)); Among them, F enhanced (t) represents the enhanced exhibit features, α1, α2, α3 are weighting parameters, and Φ(D) represents the enhanced exhibit features. z (t),E w (t) represents the joint correction function between dynamic behavior and virtual environment; S53, Combining the virtual environment with the enhanced exhibit features F enhanced (t) Perform dynamic rendering, and use a detail enhancement generator to build a dynamic rendering model R. dynamic (t): R dynamic (t)=β1·F enhanced (t)+β2·Ψ(E w (t)); Where β1 and β2 are rendering parameters, Ψ(E w (t) represents the environment mapping optimization function; S54. The dynamic rendering model R is verified through a multi-layer detail verification module. dynamic The geometry, surface texture, and dynamic behavior of (t) are verified layer by layer, and the final optimized virtual display model M of the exhibit is generated. optimized (t); S55, The optimized virtual display model M optimized (t) Input display control module, through dynamic interaction engine, adjusts the display perspective, detail, and dynamic response of exhibits based on real-time user feedback, generating a real-time updated virtual display effect V. final (t).

7. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S6 specifically includes: S61. Collect multimodal interaction data of users, including user interaction time sequence features T x (t), Dynamic state of the exhibition scene S y (t) and individual behavioral intention B z (t), through the data preprocessing module, the data is normalized and hierarchically encoded to generate a spatiotemporal awareness feature set F. spatio (t)={T x (t), S y (t), B z (t)}; S62. Using a recurrent neural network based on spatiotemporal awareness to analyze F spatio (t) Perform high-dimensional encoding to construct a user interaction prediction model: Among them, P interaction (t) represents the predicted value of user interaction, Λ(T) x (t)), Ω(B z (t) are the spatiotemporal encoding functions for temporal features, scene dynamics, and behavioral intent, respectively, and θ1, θ2, θ3 are dynamic weight parameters; S63, Combining historical behavioral patterns H m (t) and real-time user interaction features P interaction (t), define the priority scoring function for recommended content: R priority (t)=η1·P interaction (t)+η2·H m (t)+η3·C k (t); Among them, R priority (t) represents the dynamic priority of the recommended content, η1, η2, η3 are priority weight parameters, and C k (t) represents the matching score of the current exhibit features; S64. Dynamically prioritize recommended content using a multi-dimensional dynamic content push engine. priority (t) is sorted and combined with the user preference distribution U pref (t) and scene dynamic feedback E dyn (t), generating a real-time recommended content set D recommend (t); S65. Utilize the dynamic display optimization module to optimize D recommend (t) Input display logic controller, which generates a personalized dynamic display scheme by adjusting the display order, display style and interaction method in real time. dynamic (t).

8. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, Specifically, S7 includes: S71. Collect cross-domain exhibit data, including the static attribute features G of the exhibits. s Dynamic Behavior Sequence H t (t) and multimodal interaction information I u (t), the original data is normalized, noise filtered, and high-dimensional embedded through the feature preprocessing module to generate the initial feature set F. init (t)={G s H t (t), I u (t)}; S72. Construct a lightweight quantum-enhanced classification model for F. init (t) Perform feature encoding and quantum enhancement: F enh (t)=δ1·Ψ1(G s )+δ2·Ψ2(H t (t))+δ3·Ψ3(I u (t)); Among them, F enh (t) represents the feature representation after quantum enhancement, Ψ1(G s ), Ψ2(H t (t)), Ψ3(I u (t) represents the quantum enhancement functions for static features, dynamic features, and multimodal interactive data, respectively, and δ1, δ2, and δ3 are the feature enhancement weight parameters; S73. Utilizing a self-supervised learning mechanism for F enh (t) Perform feature classification: Among them, L cls (t) is the classification loss function, Q(y) j |(t)) represents category y j The predicted probability distribution, ξ j Balance the weights for each category; S74, L is classified using a lightweight classification network. cls (t) is optimized by using gradient descent combined with a self-supervised learning strategy to adjust the model parameters and generate the optimized classification result R. cls (t); S75. Classify the results R cls (t) Input the exhibit dynamic management module, dynamically group and optimize exhibits according to category, and generate a set of display solutions S based on user needs and scenario context. disp (t).

9. The digital exhibition product display method based on deep learning according to claim 1, characterized in that, S8 specifically includes: S81. Collect cross-domain data of exhibits, including the physical attribute characteristics P of the exhibits. a Dynamic Behavior Sequence D b (t), Material details M c and environmental context information E d (t), through the cross-domain data fusion module, the data is normalized and multimodal encoded to generate a feature set F. cross (t)={P a D b (t), M c E d (t)}; S82, Based on virtual simulation mapping technology, F cross (t) Perform dynamic feature reshaping: F remap (t)=κ1·P a +κ2·Ξ(D b (t),M c )+κ3·Ψ(E d (t)); Among them, F remap (t) represents the reshaped feature set, κ1,κ2,κ3 are the feature mapping weight parameters, and Ξ(D b (t),M c Ψ(E) is the joint mapping function between dynamic behavior and material properties. d (t) is the dynamic correction function for the environment context; S83, Utilizing a multi-dimensional virtual rendering engine for F remap (t) Perform immersive multi-dimensional rendering and construct a virtual display model R. immersive (t): R immersive (t)=λ1·F remap (t)+λ2·Φ(E d (t)); Where λ1, λ2 are rendering parameters, Φ(E d (t) represents the multidimensional rendering function of the virtual environment; S84. Verify the virtual display model R through a multi-layered verification module. immersive The geometric shape, material details, dynamic behavior, and environmental interaction effects of (t) are hierarchically verified to generate an optimized virtual exhibit model M. optimized (t); S85, M optimized (t) Applied to virtual exhibition spaces, using a dynamic scene control module combined with real-time user interaction data U e (t) and changes in the exhibition environment S f (t) Dynamically adjust the display details, interaction logic, and environmental adaptation of virtual exhibits to generate an immersive, multi-dimensional display effect. final (t).

Images