Decoration scheme generation system and method based on generative artificial intelligence

Abstract

The invention relates to the technical field of intelligent generation, and particularly discloses a decoration scheme generation system and method based on generative artificial intelligence, which utilize an artificial intelligence technology based on deep learning to analyze the demand of a user, extract the semantic feature of the decoration demand of the user, and meanwhile, generate the decoration scheme according to the semantic feature of the decoration demand. And feature learning is performed on a large number of indoor decoration elements, and an indoor decoration element feature library is established, so that a room decoration effect picture is intelligently generated based on the semantic association interaction features of the indoor decoration element set and the decoration demand of the user. Therefore, the room decoration effect picture can be automatically and intelligently generated for the user to refer to and select according to the personalized requirements of the user, so that the efficiency and the quality of decoration design are improved, and better decoration experience is provided for the user.

Description

Technical Field

[0001] The present application relates to the field of intelligent generation technology, and more specifically, to a decoration plan generation system and method based on generative artificial intelligence. Background Art

[0002] With the rapid development of society and the improvement of people's living standards, people's requirements for living environments are gradually increasing. The design and generation of decoration plans, as an important means to improve the living environment, has become a complex and challenging task. In the design process of decoration plans, the designer's professional knowledge and experience are certainly important. However, due to various factors, such as the designer's personal style, aesthetic concepts, and the diversity of customer needs, it is difficult to ensure that the generated decoration plan can fully meet the customer's expectations. This may lead to poor communication between designers and customers and consume a lot of time and resources. In addition, this traditional design method often finds it difficult to generate multiple feasible design plans for users to choose and compare within a limited time. Therefore, a decoration plan generation system and method based on generative artificial intelligence is desired. Summary of the Invention

[0003] In order to solve the above-mentioned technical problems, the present application is proposed. The embodiments of the present application provide a decoration scheme generation system and method based on generative artificial intelligence. The system utilizes artificial intelligence technology based on deep learning to analyze user needs and extract the semantic features of the user's decoration needs. At the same time, it conducts feature learning on a large number of interior decoration elements and establishes an interior decoration element feature library. Thus, based on the semantic correlation interaction features between the interior decoration element set and the user's decoration needs, room decoration renderings are intelligently generated. In this way, room decoration renderings can be automatically generated based on the user's personalized needs for user reference and selection, thereby improving the efficiency and quality of decoration design and providing users with a better decoration experience.

[0004] Accordingly, according to one aspect of the present application, a system for generating a renovation plan based on generative artificial intelligence is provided, comprising:

[0005] A user demand acquisition module is used to acquire room images and decoration demand information uploaded by users, wherein the decoration demand information includes room type and decoration theme style;

[0006] A user demand semantic understanding module is used to perform semantic understanding and fusion analysis on the room image uploaded by the user and the decoration demand information to obtain a room structure-decoration demand semantic fusion feature vector;

[0007] An interior decoration element acquisition module, used to acquire a collection of interior decoration element images;

[0008] a decoration element semantic feature extraction module, configured to extract semantic features from the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors;

[0009] The room decoration rendering generation module is used to generate a room decoration rendering based on the correlation interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of interior decoration element semantic feature vectors.

[0010] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the user demand semantic understanding module includes: a room structure feature extraction unit, used to extract the room structure features of the room image uploaded by the user to obtain a room structure feature vector; a demand information semantic encoding unit, used to pass the decoration demand information through a decoration demand semantic understander based on the Bert model to obtain a decoration demand semantic encoding feature vector; a semantic fusion unit, used to cascade fuse the decoration demand semantic feature vector and the room structure feature vector to obtain the room structure-decoration demand semantic fusion feature vector.

[0011] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the room structure feature extraction unit is used to: pass the room image uploaded by the user through a room structure feature extractor based on a convolutional neural network model to obtain the room structure feature vector.

[0012] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the decoration element semantic feature extraction module is used to: pass each interior decoration element image in the set of interior decoration element images through a decoration element semantic encoder based on the ViT model to obtain a sequence of interior decoration element semantic feature vectors.

[0013] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the room decoration rendering generation module includes: an association coding unit, which is used to extract the association interaction features between the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector to obtain a room decoration requirement-decoration element association interaction semantic feature vector; a correction unit, which is used to perform feature multi-granularity intrinsic structure correction based on spectral decomposition on the room decoration requirement-decoration element association interaction semantic feature vector to obtain a corrected room decoration requirement-decoration element association interaction semantic feature vector; and an adversarial generation unit, which is used to pass the corrected room decoration requirement-decoration element association interaction semantic feature vector through a decoration plan generator based on an adversarial generation network to obtain the room decoration rendering.

[0014] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the association coding unit is used to: perform association coding on the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector based on the Gaussian density map to obtain the room decoration requirement-decoration element association interaction semantic feature vector.

[0015] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the association coding unit includes: a Gaussian density map construction subunit, which is used to construct a Gaussian density map between the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector; a Gaussian discretization subunit, which is used to randomly sample the Gaussian distribution of each position in the Gaussian density map to obtain multiple row vectors, and arrange the multiple row vectors in one dimension to obtain the room decoration requirement-decoration element association interaction semantic feature vector.

[0016] In the above-mentioned decoration plan generation system based on generative artificial intelligence, the correction unit is used to: calculate the full-dimensional fine-grained self-similar structure matrix of the room decoration demand-decoration element associated interaction semantic feature vector; perform spectral decomposition on the full-dimensional fine-grained self-similar structure matrix to obtain a set of multi-granularity feature coding vectors of the room decoration demand-decoration element associated interaction semantic feature; perform information compression on each room decoration demand-decoration element associated interaction semantic feature multi-granularity feature coding vector in the set of multi-granularity feature coding vectors of the room decoration demand-decoration element associated interaction semantic feature to obtain a multi-granularity modulation coding vector of the room decoration demand-decoration element associated interaction semantic feature. The invention relates to a method for calculating a set of quantities; calculating the intrinsic structure significance modulation parameters of each room decoration demand-decoration element associated interaction semantic feature multi-granularity modulation coding vector in the set of the room decoration demand-decoration element associated interaction semantic feature multi-granularity modulation coding vector to obtain a set of intrinsic structure significance modulation parameters; normalizing the set of intrinsic structure significance modulation parameters to obtain a set of intrinsic structure significance modulation coefficients; and based on the set of intrinsic structure significance modulation coefficients, performing multi-granularity integration on the set of the room decoration demand-decoration element associated interaction semantic feature multi-granularity modulation coding vector to obtain a corrected room decoration demand-decoration element associated interaction semantic feature vector.

[0017] According to another aspect of the present application, a method for generating a decoration plan based on generative artificial intelligence is provided, which includes:

[0018] Obtain room images and decoration requirement information uploaded by users, including room type and decoration theme style;

[0019] Performing semantic understanding and fusion analysis on the room image uploaded by the user and the decoration requirement information to obtain a room structure-decoration requirement semantic fusion feature vector;

[0020] Get a collection of interior decoration element images;

[0021] performing semantic feature extraction on the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors;

[0022] Based on the correlation interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of interior decoration element semantic feature vectors, a room decoration rendering is generated.

[0023] Compared to existing technologies, the generative AI-based renovation plan generation system and method provided in this application utilizes deep learning-based AI technology to analyze user needs and extract semantic features of their renovation requirements. Simultaneously, it learns features from a large number of interior decoration elements to establish a feature library of interior decoration elements. This allows for intelligent generation of room decoration renderings based on the semantically correlated interactive features between the collection of interior decoration elements and the user's renovation requirements. This allows for the automatic generation of intelligent room decoration renderings based on the user's personalized needs for reference and selection, thereby improving the efficiency and quality of renovation design and providing users with a better renovation experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 This is a block diagram of a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application.

[0026] Figure 2 This is a schematic diagram of the architecture of a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application.

[0027] Figure 3 This is a block diagram of a user demand semantic understanding module in a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application.

[0028] Figure 4 This is a block diagram of a room decoration rendering generation module in a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application.

[0029] Figure 5 This is a block diagram of an associated coding unit in a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application.

[0030] Figure 6 This is a flowchart of a decoration plan generation method based on generative artificial intelligence according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0032] Figure 1 This is a block diagram of a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application. Figure 2 Schematic diagram of the architecture of the decoration plan generation system based on generative artificial intelligence according to the embodiment of the present application. Figure 1 and Figure 2 As shown, according to an embodiment of the present application, a decoration plan generation system 100 based on generative artificial intelligence includes: a user demand acquisition module 110, which is used to obtain room images and decoration demand information uploaded by users, and the decoration demand information includes room type and decoration theme style; a user demand semantic understanding module 120, which is used to perform semantic understanding and fusion analysis on the room images uploaded by the user and the decoration demand information to obtain a room structure-decoration demand semantic fusion feature vector; an interior decoration element acquisition module 130, which is used to obtain a set of interior decoration element images; a decoration element semantic feature extraction module 140, which is used to perform semantic feature extraction on the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors; a room decoration rendering generation module 150, which is used to generate a room decoration rendering based on the correlation interaction features between the room structure-decoration demand semantic fusion feature vector and the sequence of interior decoration element semantic feature vectors.

[0033] As mentioned in the background technology above, traditional methods for generating renovation plans often rely on the designer's creativity and inspiration, as well as repeated communication with the client. However, this method has many limitations. First, the designer's personal style and aesthetic concepts may conflict with the client's needs, resulting in the design plan not fully meeting the client's expectations. Second, during the renovation plan design process, poor communication between the two parties may lead to multiple revisions and adjustments to the design plan, wasting a lot of time and resources. Moreover, this traditional design method often makes it difficult to generate multiple feasible design plans for the client to choose and compare within a limited time.

[0034] To address the above technical issues, the technical concept of this application is to utilize deep learning-based artificial intelligence technology to analyze user needs and extract the semantic features of the user's decoration requirements. At the same time, feature learning is performed on a large number of interior decoration elements to establish an interior decoration element feature library. This allows for the intelligent generation of room decoration renderings based on the semantic association and interaction features between the collection of interior decoration elements and the user's decoration requirements. In this way, room decoration renderings can be automatically generated based on the user's personalized needs for reference and selection, thereby improving the efficiency and quality of decoration design and providing users with a better decoration experience.

[0035] In the aforementioned generative AI-based renovation plan generation system 100, the user requirements acquisition module 110 is used to acquire user-uploaded room images and renovation requirements information, including room type and decoration theme. It should be understood that the room image contains structural information about the room, such as its shape, size, and window and door locations. By analyzing the room image, the physical constraints and available space of the room can be understood, capturing the existing style of the room, such as wall color, floor type, and furniture arrangement. This helps understand the user's current aesthetic preferences and generate a renovation plan that coordinates with the existing decor. Furthermore, the user-provided renovation requirements information includes room type and decoration theme. The room type (e.g., bedroom, living room, kitchen) provides information about the room's intended use and function, guiding the system to generate a renovation plan that suits that specific room type. The decoration theme (e.g., modern, classic, pastoral) reflects the user's aesthetic preferences and provides clear stylistic guidance for the generation of renovation plans.

[0036] In the above-mentioned decoration plan generation system 100 based on generative artificial intelligence, the user demand semantic understanding module 120 is used to perform semantic understanding and fusion analysis on the room image uploaded by the user and the decoration demand information to obtain the room structure-decoration demand semantic fusion feature vector. Specifically, Figure 3 This is a block diagram of a user demand semantic understanding module in a decoration plan generation system based on generative artificial intelligence according to an embodiment of the present application. Figure 3 As shown, the user demand semantic understanding module 120 includes: a room structure feature extraction unit 121, used to extract the room structure features of the room image uploaded by the user to obtain a room structure feature vector; a demand information semantic encoding unit 122, used to pass the decoration demand information through a decoration demand semantic understander based on the Bert model to obtain a decoration demand semantic encoding feature vector; a semantic fusion unit 123, used to cascade fuse the decoration demand semantic feature vector and the room structure feature vector to obtain the room structure-decoration demand semantic fusion feature vector.

[0037] Specifically, the room structure feature extraction unit 121 is used to extract the room structure features of the room image uploaded by the user to obtain a room structure feature vector. In a specific example of the present application, the encoding method for extracting the room structure features of the room image uploaded by the user to obtain the room structure feature vector is to pass the room image uploaded by the user through a room structure feature extractor based on a convolutional neural network model to obtain the room structure feature vector. It should be understood that the convolutional neural network model is a deep learning model that is particularly suitable for processing image data. In the technical solution of the present application, the room image uploaded by the user is input into the room structure feature extractor based on the convolutional neural network model for processing. The room structure feature extractor can capture the contour shape, texture, color and other features in the room image by performing a local sliding convolution operation on the room image, and mine the room's outline size information, the location size information of doors and windows, and the layout information of existing furniture to understand the detailed physical structure and spatial layout information of the room, and provide data support for the subsequent generation of a decoration plan that conforms to the actual structure of the room.

[0038] Specifically, the demand information semantic encoding unit 122 is used to pass the decoration demand information through a decoration demand semantic understander based on the Bert model to obtain a decoration demand semantic encoding feature vector. Those skilled in the art should know that the Bert model (bidirectional encoder model) is a pre-trained language model based on the Transformer architecture, which is good at understanding the semantic meaning of text data. In the technical solution of the present application, the decoration demand information is input into a decoration demand semantic understander based on the Bert model for processing. The decoration demand semantic understander is based on the self-attention mechanism and position encoding in the Transformer architecture. It can learn the dependency and semantic information between the various words in the decoration demand information through bidirectional context modeling, capture the context-related representation of each word, so as to fully understand the user's decoration style preferences and room type requirements, and convert the decoration demand information into a fixed-length feature vector representation to provide support for the subsequent generation of a decoration plan that meets user needs.

[0039] Specifically, the semantic fusion unit 123 is used to cascade fuse the decoration requirement semantic feature vector and the room structure feature vector to obtain the room structure-decoration requirement semantic fusion feature vector. It should be understood that the decoration requirement semantic feature vector and the room structure feature vector respectively reflect the user's decoration requirements and the actual physical structure of the room. By cascading and fusing the two, the room structure features and the decoration requirement semantic features can be integrated to obtain a more comprehensive feature representation, thereby fully understanding the specific context information of the room, so that the subsequent decoration plan generation can take into account the physical structure of the room and the user's personalized needs at the same time, and generate a decoration plan that better meets the user's expectations. For example, if the user specifies the room type as "bedroom" and the decoration requirements include semantic features such as "comfortable" and "warm", a decoration plan that is consistent with the bedroom function, comfortable and warm can be generated, such as using soft colors, comfortable furniture and warm lighting design.

[0040] In the above-mentioned decoration scheme generation system 100 based on generative artificial intelligence, the interior decoration element acquisition module 130 is used to obtain a collection of interior decoration element images. It should be understood that interior decoration elements are the basic units that constitute a decoration scheme, and their types and styles are numerous, including but not limited to various decorative materials and decorations such as walls, floors, ceilings, furniture, lamps, curtains, etc. In the technical solution of the present application, in order to generate a rich and diverse decoration scheme, a collection of interior decoration element images is constructed by acquiring a large number of interior decoration element images as a basis for reference and selection. The collection of interior decoration element images is a rich element library that contains decoration element images of various styles, materials and colors, providing a rich selection of materials for subsequent decoration scheme generation. In specific implementation, interior decoration element images can be collected and organized from various sources (such as online databases, public work libraries provided by designers or public resources on the Internet, custom elements uploaded by users, etc.) to construct the collection of interior decoration element images.

[0041] In the above-mentioned decoration scheme generation system 100 based on generative artificial intelligence, the decorative element semantic feature extraction module 140 is used to extract semantic features from the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors. In a specific example of the present application, the encoding method for extracting semantic features from the set of interior decoration element images is to pass each interior decoration element image in the set of interior decoration element images through a decorative element semantic encoder based on the ViT model to obtain a sequence of interior decoration element semantic feature vectors. It should be understood that the ViT model (Vision Transformer) is an image processing model based on the Transformer architecture, which divides the image into a series of small image blocks and inputs these small image blocks as sequence data into the Transformer for context-associated encoding to extract the semantic feature representation of the image. In the technical solution of the present application, the decorative element semantic encoder based on the ViT model is used to process each interior decoration element image separately. The decorative element semantic encoder divides the interior decorative element image into a series of image blocks of a fixed number and size, maps each image block into an embedding vector, and uses an attention mechanism to model the contextual association relationship between the embedding vectors of each image block to capture information about the shape, texture, color and overall semantic meaning of the decorative elements in the interior decorative element image, such as the category of the decorative elements (such as furniture, lamps or textiles), style (such as modern, traditional or bohemian style) and function (such as seating, lighting or decoration), etc., providing a semantic description of the interior decorative elements for subsequent decoration plan generation.

[0042] In the above-mentioned decoration plan generation system 100 based on generative artificial intelligence, the room decoration rendering generation module 150 is used to generate a room decoration rendering based on the correlation interaction features between the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector. Specifically, Figure 4 This is a block diagram of a room decoration rendering generation module in a decoration scheme generation system based on generative artificial intelligence according to an embodiment of the present application. Figure 4As shown, the room decoration rendering generation module 150 includes: an association coding unit 151, which is used to extract the association interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of the interior decoration element semantic feature vector to obtain a room decoration requirement-decoration element association interaction semantic feature vector; a correction unit 152, which is used to perform feature multi-granularity intrinsic structure correction based on spectral decomposition on the room decoration requirement-decoration element association interaction semantic feature vector to obtain a corrected room decoration requirement-decoration element association interaction semantic feature vector; and an adversarial generation unit 153, which is used to pass the corrected room decoration requirement-decoration element association interaction semantic feature vector through a decoration scheme generator based on a generative adversarial network to obtain the room decoration rendering.

[0043] Specifically, the association coding unit 151 is used to extract the association interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of the interior decoration element semantic feature vector to obtain the room decoration requirement-decoration element association interaction semantic feature vector. In a specific example of the present application, the encoding method for extracting the association interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of the interior decoration element semantic feature vector is to perform association coding on the room structure-decoration requirement semantic fusion feature vector and the sequence of the interior decoration element semantic feature vector based on a Gaussian density map to obtain the room decoration requirement-decoration element association interaction semantic feature vector. It should be understood that a Gaussian density map is a visualization method for describing data distribution, by mapping the data onto a two-dimensional plane and weighting each data in the form of a Gaussian function to obtain an image representing the data distribution density. In the technical solution of the present application, the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector is mapped into the Gaussian density space based on the Gaussian function, and the semantic similarity between each feature vector is measured by calculating the mean and variance between each feature vector to construct a Gaussian density map for representing the correlation and interaction relationship between the room structure-decoration requirement semantic fusion feature vector and the sequence of the interior decoration element semantic feature vector, thereby capturing the intrinsic correlation and interaction between the user's decoration requirements and each interior decoration element to guide the generation of decoration plans, thereby improving the overall quality of the generated decoration plans.

[0044] Figure 5 FIG. 1 is a block diagram of an associated coding unit in a decoration scheme generation system based on generative artificial intelligence according to an embodiment of the present application. Figure 5As shown, the association coding unit 151 includes: a Gaussian density map construction subunit 1511, which is used to construct a Gaussian density map between the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector; a Gaussian discretization subunit 1512, which is used to randomly sample the Gaussian distribution of each position in the Gaussian density map to obtain multiple row vectors, and arrange the multiple row vectors in one dimension to obtain the room decoration requirement-decoration element association interaction semantic feature vector.

[0045] Specifically, the correction unit 152 is used to perform a feature multi-granularity intrinsic structure correction on the room decoration demand-decoration element association interaction semantic feature vector based on spectral decomposition to obtain a corrected room decoration demand-decoration element association interaction semantic feature vector. In particular, in the technical solution of the present application, the room decoration demand-decoration element association interaction semantic feature vector may present a semantic confusion problem due to the non-hierarchical integration of the feature structure. The spatial attributes contained in the user demand and the style features corresponding to the decorative elements may be mixed in a linear superposition manner in the vector space, resulting in a lack of clear hierarchical structural division of semantic information of different categories, making it difficult for the model to accurately identify the independent contribution and cross-correlation logic between spatial functional requirements and decorative aesthetic features, and thus the core requirements may be interfered with by edge features when generating decoration plans. In addition, the traditional feature fusion mechanism may only stay at a single or limited granularity level, and fail to perform multi-scale deconstruction of the semantic information of decoration requirements and decorative elements. At the same time, the nonlinear semantic dependencies between decoration requirements and decorative elements may only be reflected as surface co-occurrence relationships in the room decoration requirement-decoration element interaction semantic feature vector. This lacks structured modeling of deep semantic constraints, which can lead to the generated decoration scheme violating functional logic or stylistic consistency, impacting the scheme's practicality and user experience. To address this technical issue, the technical solution of this application performs a multi-granularity intrinsic structure correction based on spectral decomposition on the room decoration requirement-decoration element interaction semantic feature vector to obtain a corrected room decoration requirement-decoration element interaction semantic feature vector.

[0046] Specifically, the correction unit 152 is used to calculate the full-dimensional fine-grained self-similar structure matrix of the room decoration requirement-decoration element association interaction semantic feature vector, which is expressed as follows:

[0047]

[0048] M=D1⊙D2

[0049] v i ,v j ∈V

[0050] Among them, V represents the semantic feature vector of the interaction between room decoration requirements and decorative elements, v i and v j denote the i-th and j-th eigenvalues of the semantic feature vector of the interaction between room decoration requirements and decorative elements, respectively. w1, w2, w3, and w4 denote different weight hyperparameters, ⊙ denotes matrix dot product, D1 denotes the forward weighted matrix of the semantic feature vector of the interaction between room decoration requirements and decorative elements, and D2 denotes the reverse weighted matrix of the semantic feature vector of the interaction between room decoration requirements and decorative elements. Represents the value of the (i, j)th position in the forward weighted matrix of the semantic features of the interaction between room decoration requirements and decorative elements, It represents the value of the (i, j)th position of the inverse weighted matrix of the semantic features of the interaction between room decoration requirements and decorative elements, and M represents the full-dimensional fine-grained self-similar structure matrix.

[0051] Specifically, through full-dimensional, fine-grained analysis, the nonlinear coupling relationship between the two in terms of spatial distribution, attribute compatibility, and stylistic coordination is captured. This structured representation can transform implicit cross-modal interactions into mathematically interpretable tensor structures, providing topologically aware prior constraints for subsequent multi-granular feature modulation. Specifically, at the feature expression level, self-similarity metrics are used to explore potential isomorphic regions between decorative element features and spatial requirement features, enhancing the complementary expression of cross-dimensional features and enabling the system to simultaneously capture the degree of matching between macro-layout patterns and micro-material details.

[0052] Specifically, the correction unit 152 is further configured to perform spectral decomposition on the full-dimensional fine-grained self-similar structure matrix to obtain a set of multi-granularity feature encoding vectors of room decoration requirements-decoration element associated interaction semantic features, which can be expressed as follows:

[0053]

[0054] Where Λ represents a diagonal matrix, λ1 and λ m denote the first and mth eigenvalues of the diagonal matrix, respectively, (·) T represents the transpose of the vector, U represents the set of multi-granularity feature encoding vectors of the semantic features of the room decoration requirements-decoration elements association interaction, x1, x2 and x m They respectively represent the first, second and mth room decoration requirement-decoration element associated interaction semantic feature multi-granularity feature encoding vectors in the set of room decoration requirement-decoration element associated interaction semantic feature multi-granularity feature encoding vectors.

[0055] Specifically, through the mathematical framework of spectral decomposition, while maintaining global structural coherence, the nonlinearly interwoven cross-modal interaction features are projected onto an orthogonal basis coordinate system composed of eigenvectors, thereby separating independent semantic components representing different structural scales. This decomposition mechanism not only achieves a hierarchical deconstruction of complex interaction features but also constructs a quantitative assessment system for structural importance through the eigenvalue spectrum, enabling the system to identify progressive structural associations from macroscopic spatial layout to microscopic material texture.

[0056] Specifically, the correction unit 152 is further configured to perform information compression on each of the multi-granularity feature coding vectors of the room decoration requirement-decoration element associated interaction semantic feature to obtain a set of multi-granularity modulation coding vectors of the room decoration requirement-decoration element associated interaction semantic feature, which is expressed as follows:

[0057]

[0058] Among them, x i represents the multi-granularity feature encoding vector of the semantic features of the interaction between the decoration requirements and decoration elements of the i-th room, ||·|| represents the Euclidean norm, and y i Represents the multi-granularity modulation coding vector of the semantic features of the interaction between the decoration requirements and decoration elements of the i-th room.

[0059] That is, by performing information compression processing on the multi-granularity feature encoding vectors of the semantic features of the interactive interaction between room decoration requirements and decorative elements, on the basis of retaining the key semantic association features, the feature representation form is targetedly optimized to adapt to the structural requirements of the feature vector in the subsequent multi-granularity integration process. At the same time, with the help of compression operations, a nonlinear transformation mechanism is introduced to explore high-order semantic dependencies beyond linear decomposition, enhance the discrimination ability and structural significance differences between features of different granularity, and provide more discriminative and representational input data for subsequent modulation coding and feature integration.

[0060] Specifically, the correction unit 152 is further configured to calculate an intrinsic structural saliency modulation parameter of each room decoration requirement-decoration element associated interactive semantic feature multi-granularity modulation coding vector in the set of room decoration requirement-decoration element associated interactive semantic feature multi-granularity modulation coding vectors to obtain a set of intrinsic structural saliency modulation parameters, which is expressed as follows:

[0061]

[0062] Among them, α and β represent different weight parameters, ||·||1 represents the first norm, ||·||2 represents the second norm, L represents the length of the multi-granularity modulation coding vector of the semantic features of the room decoration requirements-decoration elements association interaction, a i represents y i The corresponding intrinsic structure saliency modulation parameters.

[0063] That is, by quantitatively evaluating the intrinsic structural significance of the multi-granularity modulation coding vector of the semantic features of the room decoration requirements-decorative elements association interaction, an intrinsic structural significance modulation parameter that can reflect the value of its own structural information is generated for each multi-granularity modulation coding vector of the semantic features of the room decoration requirements-decorative elements association interaction. In this way, a dynamic importance measurement system is constructed, which provides a basis for weight distribution based on the actual semantic value of the features for the subsequent adaptive integration of multi-granularity features, solves the problem of differentiated contribution evaluation of features of different granularity in the association interaction process, and ensures that the integration process can focus on the feature components that carry core semantic information.

[0064] Specifically, the correction unit 152 is further configured to perform normalization processing on the set of intrinsic structure significance modulation parameters to obtain a set of intrinsic structure significance modulation coefficients, which can be expressed as:

[0065] w i =Softmax(a i )

[0066] Among them, Softmax(·) represents the normalization function, w i Indicates a i The corresponding intrinsic structure significance modulation coefficient.

[0067] That is, the normalization technique is applied to the set of intrinsic structure saliency modulation parameters to convert them into a set of intrinsic structure saliency modulation coefficients whose sum is 1, so as to standardize the parameter scale. At the same time, the sharpness of the distribution of intrinsic structure saliency modulation coefficients is controlled through appropriate normalization methods, providing a stable and reasonable weight distribution basis for subsequent multi-granularity feature integration.

[0068] Specifically, the correction unit 152 is further configured to perform multi-granularity integration on the set of multi-granularity modulation coding vectors of the room decoration requirement-decoration element association interaction semantic feature based on the set of intrinsic structure significance modulation coefficients to obtain a corrected room decoration requirement-decoration element association interaction semantic feature vector, which is expressed as follows:

[0069]

[0070] Among them, V′ represents the corrected semantic feature vector of the room decoration demand-decoration element association interaction.

[0071] That is, by assigning specific weights to the multi-granularity modulation coding vectors of the semantic features of the interaction between the decoration requirements and decorative elements of each room and performing weighted combination, the feature information of each granularity is effectively integrated, the structural components containing key information are strengthened, and the secondary or noise components are suppressed, so as to obtain a corrected semantic feature vector of the interaction between the decoration requirements and decorative elements that concentrates key structural information and improves the representation ability, providing a more accurate, robust and discriminative feature basis for the subsequent generation of room decoration renderings.

[0072] Specifically, the adversarial generation unit 153 is configured to process the corrected semantic feature vector of the interaction between room decoration requirements and decorative elements through a decoration scheme generator based on a generative adversarial network to obtain the room decoration rendering. It should be understood that a generative adversarial network (GAN) is a generative model based on deep learning, consisting of two parts: a generator and a discriminator. The generator's task is to generate image data by learning the underlying data distribution, while the discriminator's task is to determine whether the input data is realistic. In other words, the generator model attempts to generate realistic images, while the discriminator model attempts to distinguish the generated images from real images, forcing the generator model to generate higher-quality images. In the technical solution of the present application, a decoration scheme generator based on a generative adversarial network is used to process the corrected semantic feature vector of the interaction between room decoration requirements and decorative elements. The generator module of the decoration scheme generator based on the generative adversarial network is responsible for generating a room decoration rendering based on the feature information in the corrected semantic feature vector of the interaction between room decoration requirements and decorative elements, while the discriminator is responsible for determining whether the decoration rendering generated by the generator meets the user's decoration requirements. During the training process, the generator and the discriminator compete and confront each other, and optimize performance by continuously adjusting parameters to improve the quality and accuracy of the generated decoration renderings.

[0073] In summary, the generative AI-based renovation plan generation system according to the embodiments of the present application is described. It utilizes deep learning-based AI technology to analyze user needs, extracting semantic features of the user's renovation requirements. Simultaneously, it learns features of a large number of interior decoration elements to establish an interior decoration element feature library. This allows for intelligent generation of room decoration renderings based on the semantically correlated interactive features between the collection of interior decoration elements and the user's renovation requirements. This allows for the automatic generation of intelligent room decoration renderings based on the user's personalized needs for reference and selection, thereby improving the efficiency and quality of renovation design and providing users with a better renovation experience.

[0074] Figure 6Flowchart of a method for generating a decoration plan based on generative artificial intelligence according to an embodiment of the present application. Figure 6 As shown, according to the embodiment of the present application, the decoration plan generation method based on generative artificial intelligence includes the following steps: S110, obtaining the room image and decoration requirement information uploaded by the user, wherein the decoration requirement information includes the room type and decoration theme style; S120, performing semantic understanding and fusion analysis on the room image uploaded by the user and the decoration requirement information to obtain a room structure-decoration requirement semantic fusion feature vector; S130, obtaining a set of interior decoration element images; S140, performing semantic feature extraction on the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors; S150, generating a room decoration rendering based on the correlation interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of interior decoration element semantic feature vectors.

[0075] Here, those skilled in the art will understand that the specific operations of each step in the above-mentioned decoration plan generation method based on generative artificial intelligence have been referred to above. Figures 1 to 5 The description of the decoration plan generation system based on generative artificial intelligence has been introduced in detail, and therefore, its repeated description will be omitted.

[0076] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0077] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0078] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be encompassed therein. Any reference to a figure in a claim should not be construed as limiting the claim to which it relates.

[0079] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A decoration plan generation system based on generative artificial intelligence, characterized in that: include: A user demand acquisition module is used to acquire room images and decoration demand information uploaded by users, wherein the decoration demand information includes room type and decoration theme style; A user demand semantic understanding module is used to perform semantic understanding and fusion analysis on the room image uploaded by the user and the decoration demand information to obtain a room structure-decoration demand semantic fusion feature vector; An interior decoration element acquisition module, used to acquire a collection of interior decoration element images; a decoration element semantic feature extraction module, configured to extract semantic features from the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors; The room decoration rendering generation module is used to generate a room decoration rendering based on the correlation interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of interior decoration element semantic feature vectors.

2. The decoration plan generation system based on generative artificial intelligence according to claim 1 is characterized in that: The user demand semantic understanding module includes: a room structure feature extraction unit, configured to extract room structure features of the room image uploaded by the user to obtain a room structure feature vector; A demand information semantic encoding unit, configured to pass the decoration demand information through a decoration demand semantic understander based on a Bert model to obtain a decoration demand semantic encoding feature vector; The semantic fusion unit is used to perform cascade fusion on the decoration requirement semantic feature vector and the room structure feature vector to obtain the room structure-decoration requirement semantic fusion feature vector.

3. The decoration plan generation system based on generative artificial intelligence according to claim 2 is characterized in that: The room structure feature extraction unit is used to: The room image uploaded by the user is passed through a room structure feature extractor based on a convolutional neural network model to obtain the room structure feature vector.

4. The decoration plan generation system based on generative artificial intelligence according to claim 3 is characterized in that: The decorative element semantic feature extraction module is used to: Each interior decoration element image in the set of interior decoration element images is passed through a decoration element semantic encoder based on a ViT model to obtain a sequence of interior decoration element semantic feature vectors.

5. The decoration plan generation system based on generative artificial intelligence according to claim 4 is characterized in that: The room decoration effect diagram generation module includes: an association coding unit, configured to extract association interaction features between the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector to obtain a room decoration requirement-decoration element association interaction semantic feature vector; a correction unit, configured to perform a feature multi-granularity intrinsic structure correction on the room decoration requirement-decoration element association interaction semantic feature vector based on spectral decomposition to obtain a corrected room decoration requirement-decoration element association interaction semantic feature vector; The adversarial generation unit is used to pass the corrected room decoration requirement-decoration element association interaction semantic feature vector through a decoration scheme generator based on a generative adversarial network to obtain the room decoration rendering.

6. The decoration plan generation system based on generative artificial intelligence according to claim 5 is characterized in that: The associated coding unit is used to: Based on the Gaussian density map, the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector is associated and encoded to obtain the room decoration requirement-decoration element associated interaction semantic feature vector.

7. The decoration plan generation system based on generative artificial intelligence according to claim 6 is characterized in that: The associated coding unit includes: A Gaussian density map construction subunit, configured to construct a Gaussian density map between the sequence of the room structure-decoration requirement semantic fusion feature vector and the interior decoration element semantic feature vector; The Gaussian discretization subunit is used to randomly sample the Gaussian distribution of each position in the Gaussian density map to obtain multiple row vectors, and to arrange the multiple row vectors in one dimension to obtain the room decoration requirement-decoration element association interaction semantic feature vector.

8. The decoration plan generation system based on generative artificial intelligence according to claim 7 is characterized in that: The correction unit is used to: Calculating a full-dimensional fine-grained self-similar structure matrix of the semantic feature vector of the room decoration demand-decoration element association interaction; Performing spectral decomposition on the full-dimensional fine-grained self-similar structure matrix to obtain a set of multi-granularity feature encoding vectors of semantic features of room decoration requirements-decoration elements association interaction; performing information compression on each of the multi-granularity feature coding vectors of the room decoration requirement-decoration element associated interaction semantic feature to obtain a set of multi-granularity modulation coding vectors of the room decoration requirement-decoration element associated interaction semantic feature; Calculating the intrinsic structure saliency modulation parameter of each room decoration requirement-decoration element associated interactive semantic feature multi-granularity modulation coding vector in the set of room decoration requirement-decoration element associated interactive semantic feature multi-granularity modulation coding vectors to obtain a set of intrinsic structure saliency modulation parameters; Normalizing the set of intrinsic structure saliency modulation parameters to obtain a set of intrinsic structure saliency modulation coefficients; Based on the set of intrinsic structure significance modulation coefficients, multi-granularity integration is performed on the set of multi-granularity modulation coding vectors of the room decoration requirement-decoration element association interaction semantic feature to obtain a corrected room decoration requirement-decoration element association interaction semantic feature vector.

9. A decoration plan generation method based on generative artificial intelligence, characterized in that: include: Obtain room images and decoration requirement information uploaded by users, including room type and decoration theme style; Performing semantic understanding and fusion analysis on the room image uploaded by the user and the decoration requirement information to obtain a room structure-decoration requirement semantic fusion feature vector; Get a collection of interior decoration element images; performing semantic feature extraction on the set of interior decoration element images to obtain a sequence of interior decoration element semantic feature vectors; Based on the correlation interaction features between the room structure-decoration requirement semantic fusion feature vector and the sequence of interior decoration element semantic feature vectors, a room decoration rendering is generated.

Images