Three-dimensional indoor furniture layout generation method, device, electronic device and storage medium

Abstract

The present invention relates to a method, device, electronic device and storage medium for generating a three-dimensional indoor furniture layout. The three-dimensional indoor furniture layout generation method described in the present invention comprises: obtaining three-dimensional indoor boundary data; extracting boundary corner points of the three-dimensional indoor boundary data, and converting the boundary data into vector data form; using a trained BetaVAE model to generate a furniture placement layout diagram represented by a three-dimensional furniture bounding box based on the boundary data in the vector data form; searching for the furniture closest to each furniture bounding box in a preset furniture library, linking the corresponding furniture geometric model into the scene of the furniture placement layout diagram, and obtaining a three-dimensional indoor furniture layout. The three-dimensional indoor furniture layout generation method described in the present invention uses a deep generative network to generate a three-dimensional furniture layout, with a fast generation speed, less error information in the generated scene, and more details.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided design, and in particular to a method, device, electronic equipment and storage medium for generating a three-dimensional indoor furniture layout. Background Art

[0002] In recent years, the online real estate, renovation, and furniture shopping industries have experienced rapid growth and innovation. Intelligent furniture product recommendation technology is gradually being applied to this innovative "Internet Plus" industry and platform. While this technology offers convenience to users, it cannot meet the individual needs of everyone and accommodate individual aesthetic differences. With the advancement of computer technology, computer-aided design tools are becoming increasingly simpler and faster, reducing the computer proficiency requirements for designers and allowing them to focus on the visual expression process.

[0003] Three-dimensional interior layouts can be created manually using computer-aided design software. Designers can adjust the properties of the 3D furniture model and then render it to express the interior furniture space and confirm user needs. However, due to information gaps in communication between users and designers during the real-world design process, designers cannot quickly understand user needs. This requires designers to repeatedly confirm with users and receive feedback before further design steps. By quickly providing users with intuitive visual solutions and simulating real-life home decoration scenarios, users can be helped to specify their needs and further achieve personalized design.

[0004] There are two existing methods for generating 3D interior furniture layouts. The first method loops through the scene, generating only one piece of furniture at a time. This method suffers from low efficiency and image-based prediction, requiring top-down renderings. This method is incapable of generating highly informative furniture, making it lacking research value. Another method involves modeling the general probability distribution of furniture within a space. This method also faces the challenge of designing a perfect model to sample the correct furniture distribution from this probability distribution. Summary of the Invention

[0005] Based on this, the purpose of the present invention is to provide a method, device, electronic device and storage medium for generating a three-dimensional indoor furniture layout, which uses a deep generative network to generate a three-dimensional furniture layout with a fast generation speed, less error information in the generated scene, and more details.

[0006] In a first aspect, the present invention provides a method for generating a three-dimensional indoor furniture layout, comprising the following steps:

[0007] Obtain 3D indoor boundary data;

[0008] Extracting boundary corner points of the three-dimensional indoor boundary data and converting the boundary data into vector data form;

[0009] Using the trained BetaVAE model, generating a furniture layout diagram represented by a three-dimensional furniture bounding box based on the boundary data in the form of vector data;

[0010] The furniture closest to the bounding box of each piece of furniture is found in the preset furniture library, and the corresponding furniture geometric model is linked into the scene of the furniture placement layout diagram to obtain a three-dimensional indoor furniture layout.

[0011] Furthermore, after generating the furniture layout diagram represented by the furniture bounding box, the method further includes the following steps:

[0012] Boundary alignment and normalization processing are performed on the furniture bounding boxes in the furniture placement layout diagram.

[0013] Furthermore, the rules for performing boundary alignment processing on the furniture bounding boxes in the furniture placement layout diagram include:

[0014] When the distance between the furniture bounding box and any boundary in the horizontal direction is less than a first threshold and the furniture bounding box is not facing the boundary, horizontal boundary alignment is performed;

[0015] When the distance between the furniture bounding frame and any boundary in the vertical direction is less than a second threshold and the furniture bounding frame is not facing the boundary, vertical boundary alignment is performed.

[0016] Furthermore, after generating the furniture placement layout diagram represented by the three-dimensional furniture enclosing frame, the method further includes the steps of: saving the current furniture relationship diagram;

[0017] After aligning the boundaries of the furniture enclosing frames in the furniture placement layout diagram, the method further includes the step of adjusting the furniture enclosing frames after the boundaries are aligned according to the furniture relationship diagram to restore the relative positional relationship of the furniture enclosing frames.

[0018] Furthermore, the rules for normalizing the furniture bounding boxes in the furniture placement layout diagram include:

[0019] Calculating the intersection-and-union ratio of the aligned furniture bounding boxes, and deleting furniture bounding boxes whose intersection-and-union ratio is greater than a third threshold;

[0020] Calculate whether the furniture bounding box is within the boundary, and delete the furniture bounding box that is not within the boundary;

[0021] The width of the adjacent channel toward the positive direction of the furniture bounding box is calculated, and when the channel width is less than a fourth threshold, the furniture bounding box is deleted.

[0022] Furthermore, the three-dimensional indoor boundary data and the furniture geometric model are obj files.

[0023] Furthermore, the three-dimensional indoor furniture layout is rendered by a WebGL engine.

[0024] In a second aspect, the present invention further provides a three-dimensional indoor furniture layout generation device, comprising:

[0025] Boundary data acquisition module, used to obtain three-dimensional indoor boundary data;

[0026] A boundary corner point extraction module is used to extract boundary corner points of the three-dimensional indoor boundary data and convert the boundary data into vector data form;

[0027] a furniture placement layout diagram generation module, configured to use the trained BetaVAE model to generate a furniture placement layout diagram represented by a three-dimensional furniture bounding box based on the boundary data in the form of vector data;

[0028] The 3D indoor furniture layout generation module is used to find the furniture that is closest to the bounding box of each furniture in the preset furniture library, link the corresponding furniture geometric model into the scene of the furniture placement layout diagram, and obtain the 3D indoor furniture layout.

[0029] In a third aspect, the present invention further provides an electronic device, comprising:

[0030] at least one memory and at least one processor;

[0031] The memory is used to store one or more programs;

[0032] When the one or more programs are executed by the at least one processor, the at least one processor implements the steps of the method for generating a three-dimensional indoor furniture layout as described in any one of the first aspects of the present invention.

[0033] In a fourth aspect, the present invention further provides a computer-readable storage medium,

[0034] The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for generating a three-dimensional indoor furniture layout as described in any one of the first aspects of the present invention are implemented.

[0035] The present invention provides a three-dimensional indoor furniture layout generation method, device, electronic device and storage medium, which are optimized according to multiple factors such as the boundaries of the furniture layout, the furniture usage space, and the space that can be used by humans. The final furniture layout is more aesthetically pleasing, practical and reasonable than previous methods. Compared with traditional pixel-level prediction networks and modern sequence prediction networks, the advantages are: (1) fast generation speed and good diversity of generated scenes. (2) Combined with the feature disentanglement method, by fixing the dimension of the features, a scene representation with user preferences can be generated. (3) The multi-scale features of the scene information at different encoding stages are utilized, and more small feature information is retained compared to the existing probability distribution generation model.

[0036] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A schematic diagram of the steps of a method for generating a three-dimensional indoor furniture layout provided by the present invention;

[0038] Figure 2 A schematic diagram of a process for preprocessing a home decoration design dataset in one embodiment of the present invention;

[0039] Figure 3 is a schematic diagram of a furniture frame before alignment in one embodiment;

[0040] Figure 4 A schematic diagram of a furniture frame after border alignment processing is performed in one embodiment;

[0041] Figure 5 A schematic diagram of a furniture frame after bounding box alignment processing is performed in one embodiment;

[0042] Figure 6 A schematic diagram of a furniture frame after deleting bounding boxes with relatively large intersections in one embodiment;

[0043] Figure 7 A schematic diagram of the effect of post-processing in one embodiment;

[0044] Figure 8 This is a structural schematic diagram of a three-dimensional indoor furniture layout generation device provided by the present invention. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

[0046] It should be clear that the embodiments described are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the embodiments of the present application.

[0047] The terms used in the embodiments of the present application are for the purpose of describing specific embodiments only and are not intended to limit the embodiments of the present application. The singular forms "a," "the," and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0048] When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0049] In addition, in this application, unless otherwise specified, "plurality" refers to two or more. "And / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0050] In a specific application scenario, the present application provides a method for generating a three-dimensional indoor furniture layout based on a system implementation for automatic layout of three-dimensional indoor furniture.

[0051] The algorithm platform of this system is realized by combining BetaVAE de-entanglement network, UNet3Plus feature extraction and discrimination tool and post-processing algorithm. The application platform can be designed by the front-end framework that supports WebGL lightweight rendering engine.

[0052] In view of the problems in the background technology, the embodiment of the present application provides a method for generating a three-dimensional indoor furniture layout, such as Figure 1 As shown, the method includes the following steps:

[0053] S01: Acquire 3D indoor boundary data.

[0054] In a preferred embodiment, the three-dimensional indoor boundary data is an obj file.

[0055] S02: Extracting boundary corner points of the three-dimensional indoor boundary data, and converting the boundary data into vector data.

[0056] In a preferred embodiment, the obj file containing the boundary information is read based on the trimesh package, and the boundary vertices are rearranged based on the boundary scan line algorithm to solve the problem of disordered model vertices.

[0057] S03: Using the trained BetaVAE model, generate a furniture layout diagram represented by a three-dimensional furniture bounding box according to the boundary data in the form of vector data.

[0058] In a specific embodiment, the training of the BetaVAE model is divided into data preprocessing and network training.

[0059] Among them, the data preprocessing of the home decoration design dataset is carried out, and the main process is as follows:

[0060] (1) Load the original dataset, which is the scene described by the json file and the model described by the obj file.

[0061] (2) The entire apartment in the .json file is divided into four categories according to the theme of the room: bedroom, living room, dining room, and office.

[0062] (3) Design the Room class and the Furniture class. Based on the scene-room-model hierarchy, read the data and instantiate the classes. Save different types of methods as separate .pkl files.

[0063] (4) Load the pkl file and construct each room into a matrix data of size (C*N,16), where C represents the total number of furniture types contained in the data set, and N represents the maximum number of furniture in the scene. The 16 columns of data are furniture coordinates (3 columns), furniture size (3 columns), furniture orientation (9 columns), and whether the furniture appears in the scene (1 column). For each piece of furniture in the room, extract its relationship graph to form an adjacency matrix of size (C*N,C*N,12). Among them, the 16 columns of data are the distance between the furniture in the XOY plane (7 columns), the furniture z coordinate (1 column), the furniture orientation (1 column), the furniture size difference (2 columns), and whether two pieces of furniture appear together (1 column). The steps of data preprocessing are as follows: Figure 2 shown.

[0064] The preprocessed data is input into the layout generation network for training. The main training process is as follows:

[0065] (1) Set the model optimizer to Adam, batch_size to 64, learning rate to 0.001, learning rate decreases by 0.1 every 300 epochs, and saves the model every 100 epochs.

[0066] (2) Input the indoor furniture layout information of a specific theme with a batch_size of 64 into the BetaVAE network proposed in this patent after data form pruning. After three encoder layers, the obtained z is the distribution of latent variable features. After three decoder layers, the corresponding furniture data is generated from z. Each encoding layer is composed of a sparse connection layer, a channel attention module, and a downsampling layer in sequence. Each decoding layer is composed of an upsampling layer, a channel attention module, and a sparse connection layer in sequence.

[0067] i. The sparsely connected linear connection layer performs multiple random connections, and only linearly calculates the data of any N channels each time. By using a small number of multiple times strategy in sparse connections, it achieves computing performance comparable to that of the fully connected layer with less computation, and has higher computational efficiency.

[0068] ii. The channel attention module consists of a fully connected layer, a batch normalization layer, a Swish activation function, and a Sequencing Encoding (SE) layer. The SE layer improves the encoder's attention to salient features by reducing the data dimension.

[0069] iii. The downsampling layer is composed of a fully connected layer, a batch normalization layer, and a Swish activation function in sequence. It mainly reduces the dimensionality of the data and reduces the complexity of subsequent operations.

[0070] iv. The upsampling layer has a similar structure to the downsampling layer and is mainly used to increase the dimension of the data and ultimately restore it to the same data structure as the input.

[0071] (3) Using the reconstructed data, an adjacency matrix of size (batch_size, C*N, C*N, 12) is constructed, which establishes a connection between each reconstructed piece of furniture. The adjacency matrix is ​​then transformed to a matrix of size (batch_size, 12, C*N, C*N), which is then input into the UNet3Plus discriminant network proposed in this patent, which has been pruned based on the data characteristics, to extract the relative features of the furniture.

[0072] i. The first three layers of the downsampling path use two-dimensional convolution operations to increase the input data from 12 dimensions to 64, 128, and 256 dimensions, respectively, resulting in a matrix of size (batch_size, 256, C*N, C*N).

[0073] ii. In the middle two layers of the downsampling path, we first perform max pooling with a kernel size of N, reducing the matrix size to (batch_size, 256, C, C), followed by two convolutions.

[0074] iii. For the last two layers of the downsampling path, we first perform max pooling with a kernel size of C, changing the matrix size to (batch_size, 256, 1, 1), followed by two convolutions.

[0075] iv. For the first two layers of the upsampling path, resize the matrix back to (batch_size, 256, C, C)

[0076] v. In the middle two layers of the upsampling path, the matrix obtained in step i is directly pooled into a matrix of size (batch_size, 256, C, C) and convolved into a matrix of size (batch_size, 128, C, C). The matrices generated in steps ii and iv are convolved into matrices of size (batch_size, 128, C, C) respectively. The three are concatenated in the second dimension to obtain a matrix of size (batch_size, 384C, C). After one convolution, the matrix is ​​convolved into a matrix of size (batch_size, 128, C, C), realizing the first multi-scale information fusion.

[0077] vi. In the last two layers of the upsampling path, the matrix obtained in i is convolved into a matrix of size (batch_size, 128, C*N, C*N), the matrix obtained in iv is upsampled by two-line average pooling and convolved once to obtain a matrix of size (batch_size, 128, C*N, C*N), the matrix obtained in v is upsampled by two-line average pooling and convolved once to obtain a matrix of size (batch_size, 128, C*N, C*N), the three are concatenated in the second dimension to obtain a matrix of size (batch_size, 384C*N, C*N), which is convolved once to become a matrix of size (batch_size, 128, C*N, C*N), realizing the second multi-scale information fusion.

[0078] vii. Perform the final convolution to obtain a matrix of size (batch_size, 53, C*N, C*N). Perform matrix conversion to obtain a matrix of size (batch_size, C*N, C*N, 53). Rearrange the data, calculate the loss with the ground-truth data, perform gradient backpropagation, and update the BetaVAE generator and UNet3Plus discriminator.

[0079] S04: Finding the furniture closest to the bounding box of each piece of furniture in a preset furniture library, linking the corresponding furniture geometric model into the scene of the furniture placement layout diagram, and obtaining a three-dimensional indoor furniture layout.

[0080] In a preferred embodiment, Figure 3-Figure 5 As shown, before linking the corresponding furniture geometric model into the scene of the furniture layout diagram, the following steps are also included:

[0081] S11: performing boundary alignment and normalization processing on the furniture bounding boxes in the furniture placement layout diagram.

[0082] The rules for performing boundary alignment processing on the furniture bounding boxes in the furniture placement layout diagram include:

[0083] When the distance between the furniture bounding box and any boundary in the horizontal direction is less than a first threshold and the furniture bounding box is not facing the boundary, horizontal boundary alignment is performed;

[0084] When the distance between the furniture bounding frame and any boundary in the vertical direction is less than a second threshold and the furniture bounding frame is not facing the boundary, vertical boundary alignment is performed.

[0085] Preferably, after generating the furniture placement layout diagram represented by the three-dimensional furniture enclosing frame, the method further includes the step of saving the current furniture relationship diagram.

[0086] After aligning the boundaries of the furniture enclosing frames in the furniture placement layout diagram, the method further includes the step of adjusting the furniture enclosing frames after the boundaries are aligned according to the furniture relationship diagram to restore the relative positional relationship of the furniture enclosing frames.

[0087] The rules for normalizing the furniture bounding boxes in the furniture placement layout diagram include:

[0088] like Figure 6 As shown, the intersection-and-union ratio of the aligned furniture bounding frames is calculated, and the furniture bounding frames whose intersection-and-union ratio is greater than a third threshold are deleted;

[0089] Calculate whether the furniture bounding box is within the boundary, and delete the furniture bounding box that is not within the boundary;

[0090] The width of the adjacent channel toward the positive direction of the furniture bounding box is calculated, and when the channel width is less than a fourth threshold, the furniture bounding box is deleted.

[0091] Considering that the bounding box volume of some storage furniture changes during operation and that a certain amount of space needs to be reserved for human passage, this patent proposes a post-processing method that considers the maximum width change of furniture during operation, operate_space, to be 1 / 3 of the width when the furniture is facing in the positive direction, and the passage size to be 0.35m. A threshold is set to max(operate_space,0.35). If the threshold is not met, the unusable furniture is removed.

[0092] In a preferred embodiment, a greedy strategy is used to optimize the layout of furniture from the perspectives of the relationship between furniture and scene boundaries, the relationship between furniture and furniture, and the relationship between furniture and people. Considering that not all furniture in interior design touches walls, it is recommended to set a threshold of 0.2m. If the distance between the generated furniture and the wall is within the threshold, boundary alignment is performed. In addition, considering that when laying out in a small space, the distance between large furniture and multiple walls may be less than the threshold, the direction and space of use of the furniture are taken into account when performing boundary alignment. Finally, considering that the generated furniture layout should meet people's living needs, it is recommended to set a threshold of 0.35m when aligning furniture to ensure that the generated layout is comfortable and not too crowded.

[0093] After the above processing, based on the processed bounding box output, the furniture database is searched for the furniture closest to the bounding box attributes, and the corresponding furniture geometric model is linked into the scene to replace the bounding box. The furniture geometric model is an obj file.

[0094] Through the above method, a complete three-dimensional indoor residential scene that can be rendered by the WebGL engine is finally output.

[0095] With the help of the system proposed in this application, any front-end framework combined with the WebGL rendering engine can realize the automatic display of three-dimensional indoor scenes.

[0096] The present application also provides a three-dimensional indoor furniture layout generation device. Figure 8 As shown, the three-dimensional indoor furniture layout generating device 400 includes:

[0097] Boundary data acquisition module 401, used to acquire three-dimensional indoor boundary data;

[0098] Boundary corner point extraction module 402, used to extract boundary corner points of the three-dimensional indoor boundary data and convert the boundary data into vector data form;

[0099] The furniture placement layout diagram generating module 403 is configured to use the trained BetaVAE model to generate a furniture placement layout diagram represented by a three-dimensional furniture bounding box according to the boundary data in the form of vector data;

[0100] The 3D indoor furniture layout generation module 404 is used to find the furniture closest to the bounding box of each furniture in the preset furniture library, link the corresponding furniture geometric model into the scene of the furniture placement layout diagram, and obtain the 3D indoor furniture layout.

[0101] Preferably, it also includes:

[0102] The post-processing module is used to perform boundary alignment and normalization processing on the furniture bounding boxes in the furniture placement layout diagram.

[0103] Preferably, the rules for performing boundary alignment processing on the furniture bounding boxes in the furniture placement layout diagram include:

[0104] When the distance between the furniture bounding box and any boundary in the horizontal direction is less than a first threshold and the furniture bounding box is not facing the boundary, horizontal boundary alignment is performed;

[0105] When the distance between the furniture bounding frame and any boundary in the vertical direction is less than a second threshold and the furniture bounding frame is not facing the boundary, vertical boundary alignment is performed.

[0106] Preferably, it also includes:

[0107] A furniture relationship diagram saving unit, used to save the current furniture relationship diagram;

[0108] The furniture bounding frame alignment unit is used to adjust the furniture bounding frames after boundary alignment according to the furniture relationship diagram, so as to restore the relative position relationship of the furniture bounding frames.

[0109] Preferably, the rules for normalizing the furniture bounding boxes in the furniture placement layout diagram include:

[0110] Calculating the intersection-and-union ratio of the aligned furniture bounding boxes, and deleting furniture bounding boxes whose intersection-and-union ratio is greater than a third threshold;

[0111] Calculate whether the furniture bounding box is within the boundary, and delete the furniture bounding box that is not within the boundary;

[0112] The width of the adjacent channel toward the positive direction of the furniture bounding box is calculated, and when the channel width is less than a fourth threshold, the furniture bounding box is deleted.

[0113] Preferably, it also includes:

[0114] Interface information acquisition unit, used to obtain interface information created by platform operation and maintenance personnel;

[0115] An interface configuration unit is used to configure the associated interface of the data product according to the interface information.

[0116] Preferably, the three-dimensional indoor boundary data and the furniture geometric model are obj files.

[0117] Preferably, the three-dimensional indoor furniture layout is rendered by a WebGL engine.

[0118] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0119] An embodiment of the present application further provides an electronic device, including:

[0120] at least one memory and at least one processor;

[0121] The memory is used to store one or more programs;

[0122] When the one or more programs are executed by the at least one processor, the at least one processor implements the steps of the method for generating a three-dimensional indoor furniture layout as described above.

[0123] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units 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 can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0124] The present application also provides a computer-readable storage medium.

[0125] The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for generating a three-dimensional indoor furniture layout as described above are implemented.

[0126] Computer-usable storage media includes permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0127] The present invention provides a method, device, electronic device, and storage medium for generating 3D interior furniture layouts. Any contractually involved party can complete data input or data query requests. This eliminates the need for specific platform users to identify responsible roles and generate 3D interior furniture layouts. Actual business records can be converted into 3D interior furniture layout generation rules, and the validity period of permission rules can be configured based on business order information. Through the product's interface configuration, the system can clearly define data input, output, storage, and query targets, facilitating the efficient distribution and transmission of data internally and externally.

[0128] The above-described embodiments merely illustrate several implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A method for generating a three-dimensional indoor furniture layout, characterized in that: The following steps are involved: Obtain 3D indoor boundary data; Extracting boundary corner points of the three-dimensional indoor boundary data and converting the boundary data into vector data form; Using the trained BetaVAE model, generating a furniture layout diagram represented by a three-dimensional furniture bounding box based on the boundary data in the form of vector data; Performing boundary alignment and normalization processing on the furniture bounding boxes in the furniture placement layout diagram; Searching for furniture that is closest to the bounding box of each piece of furniture in a preset furniture library, linking the corresponding furniture geometric model into the scene of the furniture layout diagram, and obtaining a three-dimensional indoor furniture layout; The rules for normalizing the furniture bounding boxes in the furniture placement layout diagram include: Calculating the intersection-and-union ratio of the aligned furniture bounding boxes, and deleting furniture bounding boxes whose intersection-and-union ratio is greater than a third threshold; Calculate whether the furniture bounding box is within the boundary, and delete the furniture bounding box that is not within the boundary; The width of the adjacent channel toward the positive direction of the furniture bounding box is calculated, and when the channel width is less than a fourth threshold, the furniture bounding box is deleted.

2. A three-dimensional indoor furniture layout generation method according to claim 1, characterized in that: Rules for performing boundary alignment processing on the furniture bounding boxes in the furniture placement layout diagram include: When the distance between the furniture bounding box and any boundary in the horizontal direction is less than a first threshold and the furniture bounding box is not facing the boundary, horizontal boundary alignment is performed; When the distance between the furniture bounding frame and any boundary in the vertical direction is less than a second threshold and the furniture bounding frame is not facing the boundary, vertical boundary alignment is performed.

3. The method for generating a three-dimensional indoor furniture layout according to claim 2, characterized in that: After generating the furniture placement layout diagram represented by the three-dimensional furniture bounding box, the method further includes the steps of: saving the current furniture relationship diagram; After aligning the boundaries of the furniture enclosing frames in the furniture placement layout diagram, the method further includes the step of adjusting the furniture enclosing frames after the boundaries are aligned according to the furniture relationship diagram to restore the relative positional relationship of the furniture enclosing frames.

4. The method for generating a three-dimensional indoor furniture layout according to claim 1, wherein: The three-dimensional indoor boundary data and the furniture geometric model are in the form of obj files.

5. The method for generating a three-dimensional indoor furniture layout according to claim 1, wherein: The three-dimensional interior furniture layout is rendered by a WebGL engine.

6. A 3D indoor furniture layout generation device for executing the 3D indoor furniture layout generation method according to any one of claims 1 to 5, characterized in that: include: Boundary data acquisition module, used to obtain three-dimensional indoor boundary data; A boundary corner point extraction module is used to extract boundary corner points of the three-dimensional indoor boundary data and convert the boundary data into vector data form; a furniture placement layout diagram generation module, configured to use the trained BetaVAE model to generate a furniture placement layout diagram represented by a three-dimensional furniture bounding box based on the boundary data in the form of vector data; The 3D indoor furniture layout generation module is used to find the furniture that is closest to the bounding box of each furniture in the preset furniture library, link the corresponding furniture geometric model into the scene of the furniture placement layout diagram, and obtain the 3D indoor furniture layout.

7. An electronic device, characterized in that: include: at least one memory and at least one processor; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the at least one processor implements the steps of the method for generating a three-dimensional indoor furniture layout as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for generating a three-dimensional indoor furniture layout according to any one of claims 1 to 5 are implemented.

Images