Furniture layout recommendation method, device, storage medium and program product

Abstract

The application discloses a furniture layout recommendation method and device, a storage medium and a program product, and relates to the technical field of smart home / smart home, and the furniture layout recommendation method comprises the following steps: determining a furniture layout request of a user; determining a to-be-laid area corresponding to the furniture layout request based on house type information, and extracting area structure features corresponding to the to-be-laid area; determining a plurality of target candidate layouts based on the area structure features and a to-be-laid furniture set; generating a target layout scheme of the furniture layout request based on the plurality of target candidate layouts, and sending the target layout scheme to the user. The method solves the problem of incomplete expression of house type structure features in the prior art by constructing a composite feature representation mechanism, effectively improves the retrieval speed under the condition of a large-scale candidate layout library, and improves the screening efficiency of home layout.

Description

Technical Field

[0001] This application relates to the field of smart home / intelligent home, and more specifically, to a method, device, storage medium, and program product for recommending furniture layout. Background Technology

[0002] With the rapid growth in demand for smart home decoration and personalized space design, interior layout recommendation systems have become a key tool for improving users' space planning efficiency. Currently, interior layout recommendation systems typically select layout solutions that meet the user's needs from a candidate layout library based on the floor plan and furniture requirements uploaded by the user.

[0003] However, existing layout schemes do not fully represent the apartment structure and cannot accurately reflect key structural information such as outline and door and window distribution at the same time. Under massive layout candidate data, furniture constraints and structural matching are often processed separately, resulting in a lengthy filtering process and reduced retrieval efficiency.

[0004] Therefore, how to improve the accuracy and efficiency of apartment layout matching in large-scale candidate layout retrieval, while also ensuring the reliability of the results, has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method, device, storage medium, and program product for recommending furniture layouts, in order to solve the technical problem that the screening process for matching apartment layouts in existing large-scale candidate layout searches is lengthy, resulting in insufficient accuracy and low efficiency.

[0006] Firstly, this application provides a method for recommending furniture layout, including:

[0007] Determine the user's furniture layout request; the furniture layout request includes the set of furniture to be placed and the apartment layout information.

[0008] Based on the apartment layout information, the area to be laid out corresponding to the furniture layout request is determined, and the area structure features corresponding to the area to be laid out are extracted; the area to be laid out is used to identify the space range where furniture can be placed; the area structure features are used to characterize the inherent spatial structure information of the area to be laid out.

[0009] Based on the regional structural features and the set of furniture to be deployed, multiple target candidate layouts are identified.

[0010] Based on multiple target candidate layouts, a target layout scheme for the furniture layout request is generated and sent to the user.

[0011] In one possible implementation, based on the apartment layout information, the area to be laid out corresponding to the furniture layout request is determined, and the area structure features corresponding to the area to be laid out are extracted, including:

[0012] Based on the apartment layout information, generate a spatial model corresponding to the area to be laid out.

[0013] Extract readable geometric features, contour features, and door and window location features from the spatial model respectively;

[0014] Feature stitching is performed on contour features, door and window position features, and readable geometric features to generate the regional structural features corresponding to the spatial model;

[0015] Among them, the contour feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the contour of the model; the door and window position feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the door and window; the readable geometric features include at least one of the following: area, aspect ratio, shape factor, convex hull area ratio, number of vertices, number of doors and windows, room type identifier, furniture type identifier, space utilization rate, functional zoning identifier, geometric symmetry index, and boundary curvature index.

[0016] In one possible implementation, based on regional structural features and the set of furniture to be placed, multiple target candidate layouts are determined, including:

[0017] Based on the regional structural features and the set of furniture to be deployed, multiple candidate layouts in the candidate layout library are filtered at multiple levels to obtain multiple target candidate layouts.

[0018] The candidate layout library is used to store multiple searchable and matching candidate layout data.

[0019] In one possible implementation, based on regional structural features and the set of furniture to be placed, multiple candidate layouts in the candidate layout library are filtered at multiple levels to obtain multiple target candidate layouts, including:

[0020] For any candidate layout, based on the set of furniture to be deployed, the candidate layout is filtered by furniture constraints;

[0021] Extract the layout structure features of candidate layouts filtered by furniture constraints, and perform structural constraint filtering on the layout structure features based on readable geometric features;

[0022] Determine the composite similarity between the layout structural features and the regional structural features filtered by structural constraints;

[0023] Based on the composite similarity of multiple candidate layouts, the multiple candidate layouts are sorted in descending order, and the top preset number of candidate layouts in the sorting results are determined as the target candidate layouts.

[0024] In one possible implementation, based on the set of furniture to be deployed, candidate layouts are filtered for furniture constraints, including:

[0025] Based on the set of furniture to be deployed, determine the category constraint features and size constraint features;

[0026] Determine the furniture set corresponding to the candidate layout, and filter the furniture set based on category constraint features;

[0027] After the furniture set is filtered by category constraints, size constraint filtering is performed on the furniture set based on size constraint features;

[0028] Among them, category constraint filtering is used to filter candidate layouts that are not included in the furniture set to be placed; size constraint filtering is used to filter candidate layouts that have at least one piece of furniture whose size does not belong to the preset size range of the corresponding furniture in the furniture set to be placed.

[0029] In one possible implementation, readable geometric features include readable continuous features and readable discrete features; layout structural features include layout continuous features and layout discrete features; based on the readable geometric features, structural constraint filtering of the layout structural features includes:

[0030] Determine the continuous feature tolerance interval corresponding to the readable continuous feature, and determine the discrete feature tolerance interval corresponding to the readable discrete feature;

[0031] Based on the tolerance intervals for continuous features and discrete features, structural constraint filtering is performed on the continuous features and discrete features of the layout, respectively.

[0032] The method also includes:

[0033] When the continuous features of the layout are within the tolerance range of continuous features and the discrete features of the layout are within the tolerance range of discrete features, the layout structural features are determined by filtering through structural constraints.

[0034] In one possible implementation, determining the composite similarity between layout structural features filtered by structural constraints and region structural features includes:

[0035] Determine the cosine similarity and Euclidean distance between the layout structural features and the region structural features, respectively;

[0036] Based on cosine similarity and Euclidean distance, the composite similarity between layout structural features and regional structural features is determined.

[0037] In one possible implementation, the method further includes:

[0038] Determine the room type corresponding to the apartment layout information, and based on the room type, determine the weight ratio of outline features, door and window location features, and readable geometric features;

[0039] Determine the composite similarity between layout structural features and region structural features filtered by structural constraints, including:

[0040] Based on layout structure features, region structure features, and weight ratios, the corresponding weighted cosine similarity and weighted Euclidean distance are determined respectively.

[0041] Based on weighted cosine similarity and weighted Euclidean distance, the composite similarity between layout structural features and regional structural features is determined.

[0042] Secondly, this application provides a furniture layout recommendation device, comprising:

[0043] The processing module is used to determine the user's furniture layout request; the furniture layout request includes the set of furniture to be placed and the apartment layout information.

[0044] The processing module is also used to determine the area to be laid out corresponding to the furniture layout request based on the apartment type information, and to extract the area structure features corresponding to the area to be laid out; the area to be laid out is used to identify the space range where furniture can be placed; the area structure features are used to characterize the inherent spatial structure information of the area to be laid out.

[0045] The processing module is also used to determine multiple target candidate layouts based on regional structural features and the set of furniture to be deployed.

[0046] The processing module is also used to generate a target layout scheme for the furniture layout request based on multiple target candidate layouts, and send the target layout scheme to the user.

[0047] In one possible implementation, the processing module is also used to generate a spatial model corresponding to the area to be laid out based on the apartment type information.

[0048] The processing module is also used to extract readable geometric features, contour features, and door and window position features of the spatial model, respectively.

[0049] The processing module is also used to stitch together contour features, door and window position features and readable geometric features to generate regional structural features corresponding to the spatial model;

[0050] Among them, the contour feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the contour of the model; the door and window position feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the door and window; the readable geometric features include at least one of the following: area, aspect ratio, shape factor, convex hull area ratio, number of vertices, number of doors and windows, room type identifier, furniture type identifier, space utilization rate, functional zoning identifier, geometric symmetry index, and boundary curvature index.

[0051] In one possible implementation, the processing module is further configured to perform multi-level filtering on multiple candidate layouts in the candidate layout library based on regional structural features and the set of furniture to be placed, to obtain multiple target candidate layouts.

[0052] The candidate layout library is used to store multiple searchable and matching candidate layout data.

[0053] In one possible implementation, the processing module is further configured to perform furniture constraint filtering on any candidate layout based on the set of furniture to be placed.

[0054] The processing module is also used to extract the layout structure features of candidate layouts filtered by furniture constraints, and to perform structural constraint filtering on the layout structure features based on readable geometric features.

[0055] The processing module is also used to determine the composite similarity between the layout structural features filtered by structural constraints and the regional structural features.

[0056] The processing module is also used to sort multiple candidate layouts in descending order according to the composite similarity of multiple candidate layouts, and to determine the first preset number of candidate layouts in the sorting result as the target candidate layout.

[0057] In one possible implementation, the processing module is further configured to determine category constraint features and size constraint features based on the set of furniture to be placed.

[0058] The processing module is also used to determine the furniture set corresponding to the candidate layout and to perform category constraint filtering on the furniture set based on category constraint features.

[0059] The processing module is also used to perform size constraint filtering on the furniture set based on size constraint features after the furniture set has been filtered by category constraints.

[0060] Among them, category constraint filtering is used to filter candidate layouts that are not included in the furniture set to be placed; size constraint filtering is used to filter candidate layouts that have at least one piece of furniture whose size does not belong to the preset size range of the corresponding furniture in the furniture set to be placed.

[0061] In one possible implementation, the processing module is further configured to determine the continuous feature tolerance interval corresponding to the readable continuous feature, and to determine the discrete feature tolerance interval corresponding to the readable discrete feature.

[0062] The processing module is also used to perform structural constraint filtering on continuous and discrete layout features based on the continuous feature tolerance interval and the discrete feature tolerance interval, respectively.

[0063] The processing module is also used to determine the layout structural features through structural constraint filtering when the continuous features of the layout are within the tolerance range of the continuous features and the discrete features of the layout are within the tolerance range of the discrete features.

[0064] In one possible implementation, the processing module is further configured to determine the cosine similarity and Euclidean distance between the layout structural features and the region structural features, respectively.

[0065] The processing module is also used to determine the composite similarity between layout structural features and region structural features based on cosine similarity and Euclidean distance.

[0066] In one possible implementation, the processing module is further configured to determine the room type corresponding to the apartment type information, and based on the room type, determine the weight ratio of the outline features, door and window location features, and readable geometric features.

[0067] The processing module is also used to determine the corresponding weighted cosine similarity and weighted Euclidean distance based on layout structure features, region structure features, and weight ratios, respectively.

[0068] The processing module is also used to determine the composite similarity between layout structural features and regional structural features based on weighted cosine similarity and weighted Euclidean distance.

[0069] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0070] The memory stores instructions that the computer executes;

[0071] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0072] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0073] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0074] This application provides a method, device, storage medium, and program product for recommending furniture layouts. The method involves: determining a user's furniture layout request; generating a spatial model corresponding to the area to be laid out based on apartment type information; extracting readable geometric features, contour features, and door / window position features from the spatial model; performing feature stitching on the contour features, door / window position features, and readable geometric features to generate the corresponding regional structural features of the spatial model; for any candidate layout, performing furniture constraint filtering on the candidate layouts based on the set of furniture to be placed; extracting the layout structural features of the candidate layouts that have passed the furniture constraint filtering, and performing structural constraint filtering on the layout structural features based on the readable geometric features; determining the composite similarity between the layout structural features that have passed the structural constraint filtering and the regional structural features; arranging multiple candidate layouts in descending order according to the composite similarity of multiple candidate layouts, and determining the top preset number of candidate layouts in the ranking result as target candidate layouts; generating a target layout scheme for the furniture layout request based on multiple target candidate layouts, and sending the target layout scheme to the user. This method addresses the problem of incomplete expression of apartment layout features in existing technologies by constructing a multi-level constraint filtering framework and a composite feature representation mechanism. It achieves collaborative matching of apartment layout and furniture constraints, thereby significantly improving retrieval efficiency and interpretability while ensuring recommendation quality. It also enhances the retrieval efficiency of large-scale candidate layouts and strengthens the matching degree between layout schemes and actual spatial conditions and furniture requirements. This not only effectively narrows the retrieval scope under the condition of a large-scale candidate layout library, but also enhances the adaptability between candidate layouts and actual spatial structures. This alleviates the problems of low retrieval efficiency, matching distortion, and insufficient reliability of recommendation results caused by relying solely on room outlines or fragmented processing of furniture and spatial structure constraints in existing technologies. Attached Figure Description

[0075] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0076] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0077] Figure 1 A schematic diagram of the hardware environment for the recommended furniture layout method provided in this application;

[0078] Figure 2 A flowchart illustrating the recommended furniture layout method provided in this application. Figure 1 ;

[0079] Figure 3A flowchart illustrating the recommended furniture layout method provided in this application. Figure 2 ;

[0080] Figure 4 A flowchart illustrating the recommended furniture layout method provided in this application. Figure 3 ;

[0081] Figure 5 A structural schematic diagram of the recommended device for furniture layout provided in this application;

[0082] Figure 6 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0083] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0084] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0085] According to one aspect of the embodiments of this application, a method for recommending furniture layouts is provided. This furniture layout recommendation method is widely applicable to whole-house intelligent digital control application scenarios such as smart homes, smart home ecosystems, smart home device ecosystems, and intelligence house ecosystems. Optionally, in this embodiment, the above-mentioned furniture layout recommendation method can be applied to, for example... Figure 1 The hardware environment shown consists of terminal device 102 and server 104. For example... Figure 1As shown, server 104 is connected to terminal device 102 via a network and can be used to provide services (such as application services) to the terminal or clients installed on the terminal. A database can be set up on the server or independently of the server to provide data storage services for server 104. Cloud computing and / or edge computing services can be configured on the server or independently of the server to provide data processing services for server 104.

[0086] The aforementioned network may include, but is not limited to, at least one of the following: wired network, wireless network. The aforementioned wired network may include, but is not limited to, at least one of the following: wide area network, metropolitan area network, local area network. The aforementioned wireless network may include, but is not limited to, at least one of the following: Wi-Fi (Wireless Fidelity), Bluetooth. The terminal device 102 may not be limited to PC, mobile phone, tablet computer, smart air conditioner, smart range hood, smart refrigerator, smart oven, smart stove, smart washing machine, smart water heater, smart washing equipment, smart dishwasher, smart projector, smart TV, smart clothes rack, smart curtains, smart audio-visual equipment, smart socket, smart speaker, smart speaker box, smart fresh air equipment, smart kitchen and bathroom equipment, smart bathroom equipment, smart robot vacuum cleaner, smart window cleaning robot, smart mopping robot, smart air purifier, smart steam oven, smart microwave oven, smart water heater, smart air purifier, smart water dispenser, smart door lock, etc.

[0087] With the rapid growth in demand for smart home decoration and personalized space design, interior layout recommendation systems have become a key tool for improving users' space planning efficiency. Currently, interior layout recommendation systems typically select layout solutions that meet the user's needs from a candidate layout library based on the floor plan and furniture requirements uploaded by the user.

[0088] However, existing layout schemes do not fully express the apartment structure and cannot accurately reflect key structural information such as the outline and the distribution of doors and windows. Specifically, the structural expression of existing layout candidate schemes often focuses on the outer outline of the room, while failing to reflect the inherent spatial structural information such as the distribution of doors and windows and the boundary characteristics of the placeable space. This can easily lead to situations where similar-looking schemes are misjudged as similar schemes even though the positions of doors and windows are significantly different, thus affecting the feasibility of furniture placement.

[0089] With massive amounts of candidate layout data, furniture constraints and structural matching are often processed separately, leading to lengthy screening processes and reduced retrieval efficiency. Specifically, as the candidate layout library continues to expand, if the system cannot effectively coordinate the processing of the furniture set to be deployed and the regional structural information, the screening process for candidate layout schemes will become lengthy. The results screened in the previous stage cannot provide sufficient compression for the next stage, thus reducing overall retrieval efficiency and making it difficult to meet the real-time response requirements under large-scale data conditions. In the case of joint retrieval of multiple constraints, a fragmented processing path is usually adopted, with furniture requirements and spatial structure often belonging to different judgment paths. This results in some unsuitable candidate schemes not being eliminated in time in the early stages, which not only increases the subsequent computational burden but also causes insufficient stability of the final recommendation results.

[0090] Furthermore, some high-dimensional similarity calculation methods may result in distorted matching results when there are significant structural differences but similar vector representations, making the recommended layout scheme unreliable in real space.

[0091] Therefore, how to improve the accuracy and efficiency of apartment layout matching in large-scale candidate layout retrieval, while also ensuring the reliability of the results, has become an urgent technical problem to be solved.

[0092] The furniture layout recommendation method provided in this application aims to solve the above-mentioned technical problems of the prior art. When receiving a user's furniture layout request, the method obtains the set of furniture that can be placed and the apartment type information from the furniture layout request; determines the area to be laid out corresponding to the furniture layout request based on the apartment type information, and extracts a composite structural feature vector corresponding to the area to be laid out, which is a fusion of contour ring scan features, door and window position ring scan features and readable geometric features, so that the area to be laid out can represent the spatial range where furniture can be placed, and the regional structural features can represent the inherent spatial structural information of the spatial range; combining the composite structural feature vector and the set of furniture that can be placed, multi-level filtering, including furniture constraint filtering, readable feature filtering and composite similarity filtering, is performed on multiple candidate layouts in the candidate layout library to obtain multiple target candidate layouts; then, a target layout scheme for the furniture layout request is generated based on the multiple target candidate layouts, and the target layout scheme is sent to the user to complete the furniture layout recommendation. This method solves the problem of incomplete expression of apartment structure features in existing technologies by constructing a multi-level constraint filtering framework and a composite feature representation mechanism. It achieves collaborative matching of apartment structure and furniture constraints, thereby significantly improving retrieval efficiency and interpretability while ensuring recommendation quality. It can also improve the retrieval efficiency of large-scale candidate layouts and enhance the matching degree between layout schemes and actual spatial conditions and furniture requirements.

[0093] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0094] Figure 2 A flowchart illustrating the method for recommending furniture layouts provided in the embodiments of this application. Figure 1 The executing entity in this embodiment can be, for example, an indoor layout recommendation system deployed on a cloud server. Figure 2 As shown, the furniture layout recommendation method provided in this embodiment includes:

[0095] S201: Determine the user's furniture layout request.

[0096] Understandably, with the rapid growth in demand for smart home decoration and personalized space design, interior layout recommendation systems have become a key tool for improving users' space planning efficiency. Among them, furniture layout recommendation systems are usually deployed in home decoration design platforms, real estate digital display platforms, home furnishing retail guide platforms, and space design tools for individual users. They are used to automatically generate layout schemes that match the space conditions based on the floor plan and furniture requirements submitted by the user.

[0097] In actual business processes, users typically initiate furniture layout requests via mobile terminals, web pages, or design workstations. The requests include the furniture set to be arranged and the apartment layout information, which usually reflects the room outline, door and window locations, and space boundaries.

[0098] The platform typically has a candidate layout library, which pre-stores a large number of historical layout schemes, template schemes or designer-generated schemes, and also saves the corresponding spatial attributes, furniture configurations and structural descriptions for each scheme.

[0099] When the system receives a furniture layout request, it first needs to identify the area to be laid out corresponding to the user's request, then search for available layout schemes in the candidate layout library, and finally generate a recommendation result and return it to the user.

[0100] The operation of such systems typically involves multiple functional units, including a front-end request access module, a floor plan parsing module, a candidate layout retrieval module, and a result generation and display module. The entire chain of processing from request parsing to solution output needs to be completed within a short response time.

[0101] Understandably, the furniture layout request serves as the input to the entire recommendation process, representing the user's specific needs for furniture placement in the target space, as well as the spatial constraints corresponding to those needs; the furniture layout request includes the set of furniture to be placed and the apartment layout information.

[0102] Specifically, the furniture set to be arranged is used to represent the set of furniture objects that the user expects to arrange in the target space. The furniture set to be arranged may include at least one of the following: furniture name, furniture category, quantity, external dimensions, orientation restrictions, combination relationship, and functional attributes. The room type information is used to represent the spatial structure input for the furniture arrangement. For example, it may include at least one of the following: room outline, wall boundaries, door and window positions, opening direction, beam and column positions, room type, floor height information, and drawing scale.

[0103] In this embodiment, the execution entity can be an interior layout recommendation system deployed on a cloud server, or a local recommendation module installed on a design workstation or user terminal. This application does not impose any special restrictions on the configuration of the interior layout recommendation system or the local recommendation module.

[0104] When a user initiates a layout request via mobile terminal, web page, or design client, the front-end interaction module receives the user's input of furniture requirements and apartment layout information, and generates a standardized furniture layout request message, which is the furniture layout request initiated by the user at that time. If the user has already uploaded the apartment layout, measurement results, or historical design files to the platform in advance, the system can also automatically assemble and generate the corresponding furniture layout request by reading the data resources associated with the user's account.

[0105] In one possible implementation, to ensure the consistency of subsequent data processing, the interior layout recommendation system can also perform format validation, field completion, outlier filtering, and unit unification on the request content after determining the furniture layout request; for example, converting size data from different sources into millimeters or centimeters, converting the scale in the floor plan image into the actual space scale, and mapping furniture names to the standard furniture category codes within the platform.

[0106] In one possible implementation, the apartment information can be uploaded in the form of a two-dimensional floor plan. The interior layout recommendation system extracts wall lines, corner points, door and window symbols, and room names through image recognition, vectorization parsing, or structured file reading, and then converts the parsing results into spatial topology data.

[0107] In another possible implementation, the apartment layout information can be provided directly from engineering drawings, digital building models, 3D scanning results, or structured building data in the design platform, from which the system can directly read the room outline and door and window distribution information.

[0108] In addition, the furniture set to be arranged can be selected manually by the user, including items such as sofas, coffee tables, TV cabinets, beds, and wardrobes, or the system can automatically generate a basic furniture list based on the room type, which the user can then confirm or modify.

[0109] Understandably, by unifying the user's placement requirements with the structural information of the target space during the request determination stage, the identification of the area to be laid out, the extraction of regional structural features, and the retrieval of candidate layouts can have a common data foundation, thereby avoiding distortion of the subsequent matching link due to scattered input sources and inconsistent expressions.

[0110] S202: Based on the apartment layout information, determine the area to be laid out corresponding to the furniture layout request, and extract the area structure features corresponding to the area to be laid out.

[0111] The area to be laid out refers to the space within the overall building space reflected in the apartment information that is used to mark the area where furniture can be placed, that is, the actual space where the furniture set to be laid out can be placed. This area to be laid out is neither the simple outline of the room nor the unprocessed original floor plan area, but a space model that can be laid out after spatial analysis, boundary correction and obstacle removal.

[0112] Regional structural features refer to the structural expression data extracted from the deployable spatial model, which is used to characterize the inherent spatial structural information of the area to be laid out, such as the shape of the area, the orientation of the boundaries, the distribution of doors and windows, the location of openings, accessibility, corner points, length-to-width ratio, local concave-convex relationships, and geometric constraints related to furniture placement.

[0113] Based on the floor plan information obtained from the furniture layout request, determine the room or functional space corresponding to the request, such as living room, bedroom, dining room, study or multi-functional room; then extract the closed outline of the space from the floor plan, identify wall boundaries, door openings, window openings, balcony connection areas, column protrusion areas and areas where obstacles cannot be placed, and then perform trimming and correction processing on the original room area according to the furniture placement rules to form the area to be laid out.

[0114] For example, when the opening range of a door, the area reserved in front of a window, the passageway, or the location of an air conditioning unit occupies part of the space, the system removes or downgrades these areas from the available space for furniture placement, so that the area to be laid out can more accurately identify the space available for furniture placement.

[0115] When extracting regional structural features, a unified spatial model is first constructed for the region to be laid out, and then structural features that can be used for retrieval and matching are extracted from the spatial model.

[0116] For example, the currently acquired room outline can be converted into a polygon point set, a sequence of boundary line segments, or a grid occupancy map. Alternatively, a three-dimensional spatial grid model can be constructed for the entire house area, which can be further decomposed into sub-models such as wall outlines, door and window locations, and functional zoning boundaries. Then, structural features such as area size, boundary orientation, spatial connectivity, and relative positions of components can be extracted from them.

[0117] Understandably, by converting the originally scattered outlines, doors, windows, and boundary conditions in the apartment information into regional structural features that can be used for machine retrieval and similarity calculation, the inherent structural attributes of the space can participate in the subsequent candidate layout screening process. This can simultaneously reflect the differences in the distribution of available space boundaries and doors and windows, reduce the probability of mistakenly retaining candidate layouts that are similar in appearance but have high actual usability, and provide a discriminative input basis for subsequent multi-level filtering.

[0118] S203: Based on the regional structural features and the set of furniture to be deployed, determine multiple target candidate layouts.

[0119] In this embodiment of the application, the interior layout recommendation system pre-stores a large number of candidate feasible furniture arrangement schemes, and these pre-stored arrangement schemes are standardized layout data that have undergone compliance verification and can be directly adapted to conventional interior spaces. Each pre-stored feasible candidate furniture arrangement scheme is completely bound to the corresponding set of furniture configuration data and the adapted space structure data. The space structure data includes, for example, all furniture categories, furniture quantity, furniture dimensions, furniture combination topology, furniture passage spacing parameters, and structural feature parameters such as the space area, aspect ratio, door and window distribution, main spatial direction, and prohibited placement areas that the layout is adapted to.

[0120] The candidate furniture arrangement schemes (i.e. candidate layouts) pre-stored in the aforementioned indoor layout recommendation system are valid layout samples generated by standardized arrangement rules, without spatial interference, and meeting normal home use standards. They do not have invalid issues such as layout conflicts, size exceeding limits, or obstructed passage. Therefore, these pre-stored candidate furniture arrangement schemes can be used as the basic sample pool for this target candidate layout screening.

[0121] By utilizing the regional structural features of the area to be laid out and the set of furniture objects that the user expects to place in the target space (i.e., the furniture set to be laid out), the system filters and identifies multiple candidate layouts that conform to the regional structural features and the furniture set to be laid out from a large number of candidate feasible furniture arrangement schemes (i.e., multiple candidate layout data) pre-stored in the interior layout recommendation system, and determines them as target candidate layouts.

[0122] For example, the interior layout recommendation system establishes a furniture category index, a quantity index, and a size range index for each candidate layout in the candidate layout library. When a furniture layout request is received, the system first locates the initial subset of candidate layouts based on the furniture category set in the furniture set to be laid out, and then determines whether the furniture composition of the candidate layouts covers the furniture items required by the furniture set to be laid out.

[0123] Understandably, by incorporating the set of furniture to be deployed along with regional structural features into the candidate layout screening process, the screening efficiency and reliability of the screening results are improved, and invalid calculations in the subsequent layout scheme generation stage are reduced.

[0124] S204: Based on multiple target candidate layouts, generate a target layout scheme for the furniture layout request and send the target layout scheme to the user.

[0125] Understandably, the target candidate layout refers to the set of candidate layouts that have been retained after multi-level filtering, which already meet the basic requirements of the current request in terms of both furniture composition and regional structure adaptation.

[0126] The target layout scheme is the final recommendation result output to the user. It can be a single layout scheme selected from multiple target candidate layouts, or one or more recommendation schemes generated based on multiple target candidate layouts.

[0127] For example, the system first reads the sorting results, furniture coordinate parameters, orientation parameters, adjacency relationships and adaptation scores of multiple target candidate layouts, and then maps the candidate layouts to the user's target spatial coordinate system in combination with the actual spatial dimensions of the area to be laid out.

[0128] For the direct reuse generation method, the system can select one or more target candidate layouts with the highest composite similarity as the target layout scheme, and perform geometric scaling, position translation and orientation correction processing according to the size ratio of the area to be laid out, so that the furniture placement result is consistent with the current space.

[0129] For the comprehensive generation method, the system can also extract the core furniture placement patterns that appear stably from multiple target candidate layouts, such as placing the main sofa against the main wall, placing the TV cabinet facing the sofa, and placing the headboard against a solid wall. Then, it can combine the furniture size of the furniture to be placed and the current door and window conditions of the room to perform local rearrangement and generate a new comprehensive layout result.

[0130] In one possible implementation, during the process of generating the target layout scheme, the indoor layout recommendation system can also perform collision verification, access verification, and visual rule verification to ensure that the recommended scheme can be actually implemented.

[0131] Among them, collision check is used to check whether the outer boundary of furniture overlaps with walls, doors, windows, columns or other furniture; passage check is used to determine whether there is sufficient passage width between furniture and between furniture and door openings; visual rule check is used to confirm whether the orientation of large furniture, the relationship of furniture central axis and the organization of functional areas conform to conventional interior design rules.

[0132] For example, when a local conflict is detected after a target candidate layout is mapped to the current layout area, the system can perform local fine-tuning according to a preset adjustment strategy, such as moving along the wall, angle correction, replacing furniture models of similar size, or redistributing the positions of secondary furniture, so as to obtain a feasible target layout scheme while ensuring that the core layout structure is not destroyed.

[0133] After the target layout scheme is formed, the system encapsulates it into a data structure suitable for terminal display. This data structure may include a two-dimensional floor plan, a three-dimensional preview scene, a furniture list, furniture coordinates, size descriptions, reasons for recommendation, and compatibility ratings, and sends it to the user terminal through a network interface. After receiving the data, the user terminal can display the recommendation results on mobile devices or the Web (World Wide Web) in the form of floor plan overlay, interactive drag-and-drop view, or three-dimensional walkthrough view.

[0134] Understandably, this step transforms the target candidate layout into a layout result that is perceptible to the end user, enabling the retrieval results generated by the aforementioned regional structure recognition and multi-level filtering to truly become a layout scheme that can be displayed, interacted with, and further edited. By introducing spatial mapping, verification, and fine-tuning mechanisms in the generation stage, the problem of candidate layouts being locally unsuitable for placement in the current actual space is avoided, thereby improving the actual usability and stability of the recommendation results.

[0135] The furniture layout recommendation control method provided in this embodiment determines the user's furniture layout request; based on the apartment type information, it determines the area to be laid out corresponding to the furniture layout request and extracts the area structure features corresponding to the area to be laid out; based on the area structure features and the set of furniture to be placed, it performs multi-level filtering on multiple candidate layouts in the candidate layout library to obtain multiple target candidate layouts; based on the multiple target candidate layouts, it generates the target layout scheme for the furniture layout request and sends the target layout scheme to the user. This method solves the problem of incomplete expression of apartment type structure features in the prior art by constructing a multi-level constraint filtering framework and a composite feature representation mechanism. It not only effectively narrows the retrieval range under the condition of a large-scale candidate layout library, but also enhances the adaptability between candidate layouts and actual spatial structures. This alleviates the problems of low retrieval efficiency, matching distortion, and insufficient reliability of recommendation results caused by relying only on the room outline or fragmented processing of furniture constraints and spatial structure constraints in the prior art.

[0136] Figure 3 A flowchart illustrating the method for recommending furniture layouts provided in the embodiments of this application. Figure 2 .like Figure 3 As shown, in this embodiment... Figure 2 Based on the embodiments, the method for recommending furniture layout is described in detail. The method for recommending furniture layout shown in this embodiment includes:

[0137] S301: Determine the user's furniture layout request.

[0138] Step S301 is similar to step S201 above, and will not be repeated here.

[0139] S302: Based on the apartment layout information, generate a spatial model corresponding to the area to be laid out.

[0140] Among them, the spatial model is used to structurally represent the area to be laid out. When it is generated, the room boundaries, door and window outlines and spatial connectivity in the apartment information are used as inputs to map the two-dimensional plane boundary into a computable geometric expression so that various structural features can be extracted in a unified manner later.

[0141] In this embodiment of the application, the spatial reference point of the spatial model is the outline reference center point, which is the basic positioning parameter built into the spatial model; wherein, the outline reference center point can be used as the encoding origin of the outline features and the door and window position features. It is usually taken as the geometric centroid or the modified center point to ensure that the ray can cover the entire area to be laid out and reduce feature drift caused by eccentricity.

[0142] S303: Extract the readable geometric features, contour features, and door and window position features of the spatial model respectively.

[0143] S304: Perform feature stitching on contour features, door and window position features, and readable geometric features to generate the regional structural features corresponding to the spatial model.

[0144] Among them, the contour feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the contour of the model; the door and window position feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the door and window.

[0145] Readable geometric features are used to characterize directly interpretable geometric attributes, including at least one of the following: area, aspect ratio, shape factor, convex hull area ratio, number of vertices, number of doors and windows, room type identifier, furniture type identifier, space utilization rate, functional zoning identifier, geometric symmetry index, and boundary curvature index. Among these, area, aspect ratio, and shape factor reflect spatial scale and shape compactness; convex hull area ratio and number of vertices reflect contour complexity; number of doors and windows, room type identifier, and furniture type identifier reflect functional attributes; and space utilization rate, functional zoning identifier, geometric symmetry index, and boundary curvature index are used to characterize spatial adaptability and structural stability.

[0146] For example, the contour ring scan feature can perform angular scanning along the boundary starting from the centroid of the region or a reference corner point, recording the boundary distance, inflection point distribution, and side length changes in each scanning direction to characterize the overall shape and local convex-concave relationship of the area to be laid out; the door and window position ring scan feature can record the relative position, number, length, and orientation of doors and windows on the boundary under the same ring scan coordinate system to reflect the constraints of structural openings on furniture placement; readable geometric features include room length and width, area, perimeter, aspect ratio, main direction, effective placement area, number of corner points, number of right angles, number of concave angles, and statistical values ​​related to the continuous length of the wall, used to form a structural description that is easy to filter quickly.

[0147] The obtained contour features, door and window position features, and readable geometric features are concatenated to generate the region structure features corresponding to the spatial model. These region structure features are expressed, for example, as vectors. The feature obtained after concatenating the contour features and door and window position features is... ,and ;feature The corresponding region structure features are obtained by concatenating the readable geometric features with the readable geometric features. ;in, For contour feature sub-vectors, For the gate position feature vector, For window position feature vectors, These are readable geometric feature vectors.

[0148] In one possible implementation, in order to improve the comparability of apartment type data from different sources, the interior layout recommendation system can also perform scale normalization, rotation alignment, coordinate resampling and noise smoothing before extracting regional structural features, so that the structural features of different apartment types are in a unified representation space.

[0149] In the specific implementation, the floor plan outline, door and window positions and room semantic tags of the corresponding area to be laid out are parsed based on the floor plan information in the furniture layout request. A spatial model is then built based on the parsing results.

[0150] Then, taking the center point of the contour reference as the origin, multiple rays are evenly emitted along the circumference at preset angle intervals (such as 6°). The distance between the first intersection point of the ray and the model contour is calculated to form the contour feature vector, which is to obtain the contour feature. At the same time, rays are emitted again with the same origin and angle interval, and the distance between the intersection point of the ray and the door and window boundary is calculated to form the door and window position feature vector, which is to obtain the door and window position feature.

[0151] Readable geometric features can be directly calculated from the geometric properties of the spatial model. For example, the area can be obtained by using the polygon area formula, the aspect ratio can be obtained by using the ratio of the long side to the short side of the minimum bounding rectangle, the shape factor can be obtained by using the combination relationship between perimeter and area, and the convex hull area ratio can be obtained by using the ratio of the convex hull area to the original contour area. The corresponding identifiers can be generated by combining the results of door and window statistics, room semantic recognition, and functional zoning recognition.

[0152] Finally, the contour features, door and window position features, and readable geometric features are concatenated according to a unified dimension rule to obtain a region structure feature vector that represents the area to be laid out.

[0153] In one possible implementation, multiple rays can be uniformly emitted circumferentially at a preset angle interval (e.g., 6°) with the center point of the contour reference as the origin. Based on the results of the same ray scan, the distance between the first intersection point of the ray and the model contour, and the distance between the ray and the door / window boundary are calculated respectively. Then, the contour feature vector and the door / window position feature vector are generated synchronously, that is, the contour features and door / window position features corresponding to the spatial model are generated synchronously.

[0154] Understandably, the structural features of this area serve as a unified representation of the area to be laid out. Among them, the contour features reflect the spatial shape, the door and window position features reflect spatial constraints, and the readable geometric features reflect spatial semantics and geometric attributes. When these three are combined, they can simultaneously depict the form, constraints, and functional attributes of the area to be laid out. This allows for a more accurate assessment of the suitability of candidate solutions to the current apartment type during subsequent layout retrieval or matching, improves the completeness of the representation of the area to be laid out, enhances the influence of door and window constraints on layout judgment, reduces the probability of false matching caused by relying solely on similar outer contours, and improves the accuracy and stability of candidate layout selection. Ultimately, this improves the feasibility and space utilization of furniture layout recommendations.

[0155] S305: Based on the regional structural features and the set of furniture to be placed, multiple candidate layouts in the candidate layout library are filtered at multiple levels to obtain multiple target candidate layouts.

[0156] The candidate layout library is used to store multiple searchable and matching candidate layout data.

[0157] In this embodiment of the application, the candidate layout library is used to store a large amount of candidate layout data that can be retrieved and matched. That is to say, the data in the candidate layout library is a large number of candidate feasible furniture arrangement schemes pre-stored in the indoor layout recommendation system. In addition to the furniture placement result itself, each candidate layout can also be associated with and stored additional information such as the furniture set description, applicable room type, layout area structural features, design source, historical click feedback and adaptation score.

[0158] Multi-level filtering refers to the fact that the interior layout recommendation system does not directly perform a one-time full high-precision matching on the candidate layout library. Instead, it performs layered screening of candidate layouts and gradually narrows down the candidate range according to the processing order from coarse to fine and from low cost to high precision. This allows candidate layouts that obviously do not meet the current furniture requirements or spatial structure constraints to be eliminated in the early stage, thereby reducing the processing burden of subsequent high-dimensional similarity calculation and ranking operations.

[0159] Based on the set of furniture to be deployed, furniture constraint filtering is performed to determine whether the type, quantity, and size of the furniture in the candidate layout are compatible with the furniture in the current furniture layout request. After completing the furniture constraint filtering, structural constraint filtering is performed on the candidate layouts that pass the furniture constraint filtering based on the regional structural features. For the candidate layouts that pass the structural constraint filtering, the similarity between the regional structural features of the corresponding area to be deployed and the pre-stored regional structural features of each candidate layout is calculated. Based on the similarity of the obtained candidate layouts, the candidate layout with the highest similarity is selected as the target candidate layout for recommendation in this iteration.

[0160] For example, the interior layout recommendation system establishes a furniture category index, a quantity index, and a size range index for each candidate layout in the candidate layout library. When a furniture layout request is received, the system first locates the initial subset of candidate layouts based on the furniture category set in the furniture set to be laid out, and then determines whether the furniture composition of the candidate layouts covers the furniture items required by the furniture set to be laid out.

[0161] Specifically, for furniture with a combination relationship, such as a bed and bedside table, a dining table and dining chair, a sofa and coffee table, the system can also use the furniture relationship graph to perform subgraph isomorphism detection to determine whether the furniture topological relationship in the candidate layout satisfies the furniture layout request.

[0162] If the user further inputs specific dimensions, minimum channel width, or placement restrictions in the request, the system can compare the corresponding parameters in the candidate layouts item by item and retain only the layout results that meet the category and size conditions.

[0163] For candidate layouts filtered by furniture constraints, the inherent spatial structure information of the area to be laid out is compared item by item with the spatial structure information corresponding to the candidate layout. For example, the differences in area, aspect ratio, number of corner points, number of doors and windows, and main direction between the area to be laid out and the corresponding space of the candidate layout are compared to eliminate candidate layouts with obvious differences. Then, for candidate layouts filtered by structural constraints, the similarity between the regional structural features of the area to be laid out and the pre-stored regional structural features of each candidate layout is calculated. In order to reflect both the consistency of feature direction and the difference in amplitude, the cosine similarity and Euclidean distance of the corresponding features can be calculated respectively, and a composite similarity can be generated based on the two. After sorting the composite similarity from high to low, the system retains the candidate layouts that rank in the top set number or exceed the set threshold as the target candidate layouts for recommendation in this time.

[0164] Understandably, by incorporating the set of furniture to be deployed and the regional structural features into the candidate layout screening process, the search space can still be effectively narrowed even when the candidate layout library is large. Among them, furniture constraint filtering ensures that the candidate layout meets the request requirements at the furniture content level, while structural constraint filtering ensures that the candidate layout matches the area to be deployed at the spatial adaptation level. This improves the efficiency of candidate screening and the reliability of the retained results, and reduces invalid calculations in the subsequent scheme generation stage.

[0165] S306: Based on multiple target candidate layouts, generate a target layout scheme for the furniture layout request and send the target layout scheme to the user.

[0166] Step S306 is similar to step S204 above, and will not be described again here.

[0167] The furniture layout recommendation control method provided in this embodiment determines the user's furniture layout request; generates a spatial model corresponding to the area to be laid out based on the apartment type information; extracts readable geometric features, contour features, and door and window position features from the spatial model; performs feature splicing on the contour features, door and window position features, and readable geometric features to generate the area structure features corresponding to the spatial model; performs multi-level filtering on multiple candidate layouts in the candidate layout library based on the area structure features and the set of furniture to be placed, obtaining multiple target candidate layouts; generates the target layout scheme for the furniture layout request based on the multiple target candidate layouts, and sends the target layout scheme to the user. This method solves the problem of incomplete expression of apartment type structure features in existing technologies by constructing a multi-level constraint filtering framework and a composite feature representation mechanism. It not only effectively narrows the retrieval range under the condition of a large-scale candidate layout library, but also enhances the adaptability between candidate layouts and actual spatial structures, thereby alleviating the problems of low retrieval efficiency, matching distortion, and insufficient reliability of recommendation results caused by relying solely on the room's outer contour or fragmented processing of furniture constraints and spatial structure constraints in existing technologies.

[0168] Figure 4 A flowchart illustrating the method for recommending furniture layouts provided in the embodiments of this application. Figure 3 .like Figure 4 As shown, in this embodiment... Figure 3 Based on the embodiments, the multi-level filtering of candidate layouts in the furniture layout recommendation method is described in detail. The furniture layout recommendation method shown in this embodiment includes:

[0169] S401: Determine the user's furniture layout request.

[0170] S402: Based on the apartment layout information, generate a spatial model corresponding to the area to be laid out.

[0171] S403: Extract the readable geometric features, contour features, and door and window position features of the spatial model respectively.

[0172] S404: Perform feature stitching on contour features, door and window position features, and readable geometric features to generate the regional structural features corresponding to the spatial model.

[0173] Steps S401-S404 are similar to steps S301-S304 above, and will not be repeated here.

[0174] S405: For any candidate layout, perform furniture constraint filtering on the candidate layout based on the set of furniture to be deployed.

[0175] S406: Extract the layout structure features of candidate layouts filtered by furniture constraints, and perform structural constraint filtering on the layout structure features based on readable geometric features.

[0176] S407: Determine the composite similarity between the layout structural features and the regional structural features filtered by structural constraints.

[0177] S408: Based on the composite similarity of multiple candidate layouts, sort the multiple candidate layouts in descending order, and determine the first preset number of candidate layouts in the sorting results as the target candidate layouts.

[0178] Among them, furniture constraint filtering is used to first eliminate candidate layouts that do not meet the furniture requirements. The furniture set in the candidate layout is matched with the furniture set to be placed by category and size to filter out candidate layouts that do not contain the required furniture or whose furniture size is not within the preset range.

[0179] Layout structure features are used to characterize the inherent spatial structure information of candidate layouts. They are usually composed of continuous and discrete structure features and are used for subsequent structural constraint filtering and similarity calculation.

[0180] Structural constraint filtering is used to further filter the layout structure features of candidate layouts based on readable geometric features. Specifically, it compares the continuous and discrete structural features of candidate layouts with the features corresponding to the current home layout request, eliminates schemes with obvious deviations in the apartment layout, and retains candidate layouts with high structural similarity.

[0181] Composite similarity is used to measure the overall matching degree between candidate layouts and regional structural features. It usually calculates cosine similarity and Euclidean distance together to form a comprehensive score and serve as the basis for ranking.

[0182] Understandably, each candidate layout in the candidate layout library has pre-stored furniture configuration data, outline structure data, and door and window position data, which are used to compare with the currently acquired set of furniture to be placed and the readable geometric features corresponding to the area to be placed, so as to select candidate layouts that pass the furniture constraint filtering and structural constraint filtering.

[0183] For each candidate layout in the candidate layout library, the pre-stored furniture configuration data of the corresponding candidate layout is compared one by one with the furniture configuration data of the current furniture set to be deployed. Specifically, the furniture category is compared one by one with the furniture categories in the candidate layout, and the length, width and height of the furniture are combined with the preset size thresholds to determine whether it meets the space accommodation conditions, thus obtaining candidate layouts that pass the furniture constraint filtering. For the candidate layouts that pass the furniture constraint filtering, continuous structural parameters and discrete structural parameters are extracted. Continuous structural parameters include, for example, aspect ratio, space utilization rate and boundary curvature index. Discrete structural parameters include, for example, number of vertices, number of doors and windows and functional area identifiers. These are input as layout structural features into the structural constraint filtering module to obtain candidate layouts that pass the structural constraint filtering. The structural constraint filtering module generates a tolerance interval based on the readable geometric features corresponding to the regional structural features, and candidate layouts that do not fall into the tolerance interval are eliminated.

[0184] After filtering by structural constraints, the system vectorizes the retained layout structural features and regional structural features, and calculates cosine similarity and Euclidean distance based on preset weights, and then fuses the two to obtain composite similarity. The higher the composite similarity, the more consistent the candidate layout is with the area to be laid out in terms of spatial structure and furniture adaptation. The system then sorts the multiple candidate layouts in descending order according to the calculated composite similarity, and selects the candidate layouts within the first preset number as the target candidate layout output.

[0185] In this embodiment, by first screening furniture, then screening structures, and finally ranking by similarity, the candidate range can be significantly narrowed and the accuracy and stability of layout recommendations can be improved. Cosine similarity is used to measure whether the direction of feature vectors is consistent, and Euclidean distance is used to measure whether the difference in feature amplitude is small. The combined use of the two can alleviate the "identical differentiator" problem that may occur in high-dimensional space by relying on a single similarity index, that is, the mismatch problem caused by obvious structural differences but local similarity of vector representations.

[0186] In one possible implementation, based on the set of furniture to be arranged, category constraint features and size constraint features are determined; the furniture set corresponding to the candidate layout is determined, and the furniture set is filtered by category constraint features; after the furniture set passes the category constraint filtering, the furniture set is filtered by size constraint features.

[0187] Among them, category constraint filtering is used to filter candidate layouts that are not included in the furniture set to be placed; size constraint filtering is used to filter candidate layouts that have at least one piece of furniture whose size does not belong to the preset size range of the corresponding furniture in the furniture set to be placed.

[0188] Understandably, category constraint features are used to characterize the category requirements of the furniture set to be arranged. They are usually composed of the names, type identifiers and their combination relationships of the furniture to be arranged, and are used to indicate the furniture categories that should be included in the candidate layout.

[0189] Size constraint features are used to characterize the size requirements of the furniture set to be arranged. They typically correspond to the length, width, and height range of each piece of furniture to be arranged, and can be combined with allowable installation tolerances to form a preset size range.

[0190] The furniture set is the actual collection of furniture extracted from the candidate layout. It includes the name, category identifier, and size information of each piece of furniture in the candidate layout, and is used to compare it item by item with the furniture set to be placed.

[0191] Category constraint filtering is used to first determine whether the furniture set of the candidate layout completely covers the furniture categories required by the furniture set to be placed. If any furniture to be placed does not appear in the furniture set, the candidate layout is removed.

[0192] Size constraint filtering is used to further verify whether the size of the corresponding furniture in the candidate layout falls within the preset size range after the category constraint is satisfied. If the size of any furniture does not meet the requirement, the corresponding candidate layout is filtered out.

[0193] After determining the set of furniture to be deployed corresponding to the current furniture layout request, a category constraint feature is generated for each piece of furniture in the set. This category constraint feature can be obtained, for example, through a furniture coding table, a category mapping table, or a rule base. At the same time, a size constraint feature is generated based on the furniture's nominal specifications, installation allowance, and space adaptation margin.

[0194] The corresponding furniture set is parsed from the layout description file, 3D model file or structured configuration table of the candidate layout in the candidate layout library, and then matched with the furniture set to be placed by name and category to complete the furniture constraint filtering.

[0195] For any candidate layout that undergoes furniture constraint filtering, when the furniture set in the candidate layout can cover all the furniture to be placed, the size verification stage (i.e., size constraint filtering) is entered. The system reads the size parameters of the corresponding furniture and compares them with the preset size range. If the measured size of any furniture exceeds the range, the candidate layout is determined not to meet the size constraint; if the measured size of any furniture does not exceed the range, the candidate layout is determined to meet the size constraint. At this time, the candidate layout meets both the category constraint and the size constraint, that is, the candidate layout passes the furniture constraint filtering.

[0196] In this embodiment, the furniture constraint filtering process can be completed by a constraint verification module on the server side. This module can be implemented using vector comparison, hash indexing, or rule-based judgment to improve retrieval efficiency. This embodiment does not impose any special restrictions on the constraint verification module.

[0197] Understandably, this furniture constraint filtering method, through a sequential verification approach prioritizing category and then size, eliminates candidate layouts that clearly do not meet furniture requirements in advance. Then, it performs precise size verification on candidate layouts that pass the category check. This allows for faster elimination of schemes with missing furniture categories or incompatible sizes from the candidate layout library, reducing the computational load of subsequent structure matching and similarity calculations, thereby minimizing invalid calculations and improving the matching accuracy between candidate layouts and the set of furniture to be placed. Simultaneously, it enhances the deployability and practical usability of the recommended results, thereby improving the response efficiency and result stability of furniture layout recommendations. Since category constraints ensure the completeness of furniture configuration and size constraints ensure the usability of furniture specifications, the results of this furniture constraint filtering can provide a more reliable candidate basis for subsequent layout recommendations.

[0198] In one possible implementation, the readable geometric features include readable continuous features and readable discrete features; the layout structural features include layout continuous features and layout discrete features; a continuous feature tolerance interval corresponding to the readable continuous features is determined, and a discrete feature tolerance interval corresponding to the readable discrete features is determined; based on the continuous feature tolerance interval and the discrete feature tolerance interval, structural constraint filtering is performed on the layout continuous features and the layout discrete features respectively; if the layout continuous features are within the continuous feature tolerance interval and the layout discrete features are within the discrete feature tolerance interval, the layout structural features are determined to pass the structural constraint filtering.

[0199] Among them, readable continuous features are used to characterize continuous readable geometric attributes, which typically include numerical attributes such as area and aspect ratio; readable discrete features are used to characterize discrete readable geometric attributes, which typically include integer attributes such as the number of doors and windows.

[0200] Continuous layout features are used to characterize continuous structural attributes in candidate layouts, while discrete layout features are used to characterize discrete structural attributes in candidate layouts.

[0201] The continuous feature tolerance interval is used to limit the acceptable range of variation for readable continuous features. For example, it can represent the range of allowed differences for continuous variables such as area, side length, and ratio. The discrete feature tolerance interval is used to limit the acceptable integer neighborhood range for readable discrete features. For example, it can represent the allowed proximity relationships for discrete variables such as the number of doors and windows, functional area types, and corner point categories. All of the above features and intervals can be generated jointly from the candidate layout structure description, the apartment type parsing results, and the rule base.

[0202] In this embodiment, the tolerance intervals for continuous features and discrete features can be preset thresholds, or the thresholds for each feature item can be dynamically adjusted based on historical successful matching data to balance accuracy and recall.

[0203] After determining the regional structural features of the area to be laid out, continuous geometric quantities such as area and aspect ratio are first extracted from these features. Corresponding continuous feature tolerance intervals are then constructed based on preset deviation coefficients. For example, a symmetrical interval is generated centered on the target area, and an allowable vertical fluctuation range is generated based on the target aspect ratio. Simultaneously, discrete geometric quantities such as the number of doors and windows are extracted, and discrete feature tolerance intervals are generated according to allowable discrete offsets. Corresponding continuous and discrete layout features are extracted from the layout structural features of the candidate layout and compared with the two tolerance intervals mentioned above. When both the continuous and discrete layout features fall within the continuous feature tolerance interval and the discrete layout features fall within the discrete feature tolerance interval, the candidate layout is determined to have passed the structural constraint filtering.

[0204] For example, continuous features may include area, aspect ratio, and shape factor. For this type of feature, a symmetry tolerance interval can be used for structural constraint filtering. The specific tolerance interval for this continuous feature is shown in Formula 1 below:

[0205]

[0206] in, For any continuous feature, the actual measured value. A fault tolerance index with continuous characteristics. This represents the tolerance interval for the corresponding continuous feature.

[0207] The tolerance interval for discrete features can include the number of doors, windows, vertices, etc. For this type of feature, integer neighborhood constraints can be used for structural constraint filtering. The specific tolerance interval for discrete features is shown in Formula 2 below:

[0208]

[0209] in, Let be the actual quantity of any discrete feature. This represents the maximum permissible deviation of the discrete feature. Let be the number of discrete features corresponding to the candidate layout, and It is an integer; This is the integer neighborhood constraint (i.e., the tolerance interval) corresponding to the discrete features.

[0210] In this embodiment, the structural constraint filtering process can be completed by the feature matching module on the server side. The feature matching module can be implemented using a rule engine, a threshold judgment unit, or a vectorized comparison unit. In practical applications, the module can also be implemented in other ways, and this embodiment does not limit this.

[0211] Understandably, by separating continuous geometric constraints from discrete structural constraints and performing joint verification on the same candidate layout, the candidate layout can satisfy both the approximate consistency of spatial scale and the consistency of inherent structures such as doors and windows, thereby avoiding misjudgment as a usable layout based solely on similar outlines.

[0212] Since the filtering conditions directly correspond to the regional structural features, obviously unsuitable schemes can be quickly eliminated from the candidate layout library, reducing the number of objects for subsequent similarity calculations and improving retrieval efficiency. At the same time, setting tolerance intervals for continuous and discrete features is beneficial to balancing the flexibility of spatial adaptation with the strictness of structural constraints, making the final output layout scheme more stable and reliable in terms of placement and structural matching.

[0213] In one possible implementation, the cosine similarity and Euclidean distance of the layout structural features and the region structural features are determined respectively; based on the cosine similarity and Euclidean distance, the composite similarity of the layout structural features and the region structural features is determined.

[0214] In this embodiment of the application, in order to reflect both the consistency of feature direction and the difference in amplitude, the cosine similarity and Euclidean distance of the corresponding features can be calculated respectively, and a composite similarity can be generated based on the two. After sorting the composite similarity from high to low, the system retains the candidate layouts that rank at the top of a set number or exceed a set threshold as the target candidate layouts.

[0215] Understandably, layout structural features typically include continuous layout features and discrete layout features retained after filtering by structural constraints. Regional structural features are formed by splicing together the geometric, boundary, and door and window distribution information of the region to be laid out. Both are encoded as vectors of the same dimension to facilitate similarity calculation.

[0216] Euclidean distance quantifies the magnitude difference between layout structural features and region structural features, and is a component of composite similarity; cosine similarity quantifies the directional consistency between layout structural features and region structural features, and is also a component of composite similarity. In actual calculations, the feature vectors can be normalized first, and then the dot product, magnitude, and sum of squared differences can be calculated separately to reduce the influence of different feature dimensions on the results.

[0217] For example, the layout structure features can be denoted as the first feature vector, and the region structure features as the second feature vector. The cosine similarity between the two can be determined by the ratio of the vector dot product to the product of the magnitudes, and the Euclidean distance can be obtained by taking the square root of the sum of the squares of the differences in the corresponding dimensions. Then, the cosine similarity and the Euclidean distance are input into the fusion function. The fusion function can construct a composite similarity by using a weighted normalization method. Here, the cosine similarity corresponds to the direction matching term, and the Euclidean distance corresponds to the deviation penalty term. The larger the output value of the composite similarity, the closer the candidate layout is to the region to be laid out in terms of spatial shape, boundary distribution, and structural relationship.

[0218] For example, the composite similarity between layout structural features and region structural features is determined based on cosine similarity and Euclidean distance; this composite similarity can be calculated using the following formula 3:

[0219]

[0220] in, This is a vector representation of the regional structural features corresponding to the area to be laid out. For the first Vector representation of the layout structure features of each candidate layout For layout and structural features With regional structural characteristics Composite similarity, For layout and structural features With regional structural characteristics cosine similarity, For layout and structural features With regional structural characteristics Euclidean distance; It is a very small constant used to avoid the denominator being zero and to ensure computational stability. It is an L2 norm.

[0221] In this embodiment, the fusion function can be implemented with fixed weights in the candidate layout retrieval module, or its parameters can be calibrated based on layout type, spatial scale, or historical matching results. This embodiment does not impose any special limitations on the fusion function.

[0222] Understandably, by calculating both the layout structural features and regional structural features of the structural constraint filter, and then ranking the candidate layouts based on the composite similarity, the candidate layouts that satisfy both the structural constraints and have a high degree of matching with the target region can be given higher priority. Since cosine similarity can reflect the consistency of feature distribution trends, while Euclidean distance can supplement the reflection of structural size and local offset differences, the joint calculation of the two can reduce the misjudgment caused by relying on only a single indicator, so that the screening results can take into account both the overall morphological consistency and the local structural fit.

[0223] Meanwhile, based on the structural constraint filtering that has eliminated obviously unsuitable solutions, the matching accuracy between candidate layouts and the area to be laid out can be further improved, reducing false recommendations for layouts that are similar in appearance but inconsistent in spatial structure. Composite similarity can provide a stable numerical basis for subsequent ranking, enabling the system to converge to highly relevant solutions relatively quickly in a large-scale candidate layout library, thereby improving the reliability of recommendation results and retrieval efficiency.

[0224] In one possible implementation, the room type corresponding to the apartment type information is determined, and based on the room type, the weight ratios of the outline features, door and window location features, and readable geometric features are determined; based on the layout structure features, regional structure features, and weight ratios, the corresponding weighted cosine similarity and weighted Euclidean distance are determined respectively; based on the weighted cosine similarity and weighted Euclidean distance, the composite similarity between the layout structure features and the regional structure features is determined.

[0225] Among them, room type is used to characterize the spatial use of the apartment information. It can be determined by the functional labels in the apartment plan, the spatial boundary shape and historical identification results, and serves as the basis for subsequent weight allocation.

[0226] The weight ratio is used to adjust the influence of contour features, door and window position features and readable geometric features in similarity calculation. It can pre-establish a mapping table between room type and weight combination, and directly retrieve the corresponding parameters after the corresponding room type is identified.

[0227] Weighted cosine similarity is used to measure the degree of alignment between layout structural features and regional structural features in terms of directional consistency, while weighted Euclidean distance is used to measure the degree of closeness between the two in terms of amplitude difference. Combining the two can avoid misjudgment caused by a single metric.

[0228] After completing the parsing of the apartment information, the interior layout recommendation system first identifies the room type of the area to be laid out, such as bedroom, living room, study, or dining room, and inputs the room type into the weight configuration module. The weight configuration module outputs the weight ratio of contour features, door and window position features, and readable geometric features according to preset rules. For example, the weight of door and window position features is increased for bedroom scenes to enhance the ability to judge the bed's avoidance relationship, and the weight of contour features is increased for living room scenes to enhance the ability to adapt to the placement boundaries of large furniture such as sofas and TV walls.

[0229] The layout structure features and regional structure features filtered by structural constraints are then input into the similarity calculation module. First, the sub-feature vectors of the two types of features are weighted and normalized based on the weight ratio. Then, the weighted cosine similarity and weighted Euclidean distance are calculated, and the two are combined according to the preset fusion relationship to generate a composite similarity.

[0230] In this embodiment, the fusion relationship can employ linear weighting, piecewise mapping, or normalized combination to ensure that the similarity result simultaneously reflects both structural orientation consistency and numerical difference. In practical applications, this module can be deployed on a server-side feature computation unit, using a GPU (Graphics Processing Unit) to perform vectorized operations; this embodiment does not impose any special restrictions on this.

[0231] Understandably, room type first determines the focus of feature attention, and then the weight ratios are used to give different influences to structural elements in different spatial scenarios. Then, weighted cosine similarity and weighted Euclidean distance are used to jointly represent the similarity between layout structural features and regional structural features, so that the ranking of candidate layouts in different room scenarios is more in line with actual placement needs. After adopting this method, the positions of doors and windows, outline constraints and geometric readability can be differentiated in similarity calculation, reducing the probability of candidate schemes with similar shapes but mismatched spatial constraints entering the top ranks, and improving the ability of composite similarity to represent the actual layout feasibility, thereby improving the accuracy, stability and scene adaptability of the recommendation results.

[0232] S409: Based on multiple target candidate layouts, generate a target layout scheme for the furniture layout request and send the target layout scheme to the user.

[0233] Step S409 is similar to step S306 above, and will not be described again here.

[0234] In one possible implementation, during the process of generating the target layout scheme, the interior layout recommendation system further executes a furniture mapping algorithm.

[0235] For example, the furniture mapping algorithm is based on a constrained bipartite graph matching mechanism to match the furniture set requested by the user with the furniture set in the candidate layout. Specifically, the system first constructs a mapping graph of furniture names and functional attributes, and matches the furniture in the user request with the furniture in the candidate scheme through constraints (such as furniture size and placement).

[0236] For example, if the user requests a "workstation" size of 1.2m × 0.6m, while the "workstation" size in the candidate layout is 1.22m × 0.6m, the system automatically marks the size deviation as 2%. This mapping result is directly overlaid on the floor plan of the candidate layout and used as a visual output of the recommendation result.

[0237] The furniture layout recommendation method provided in this implementation determines the user's furniture layout request; generates a spatial model corresponding to the area to be laid out based on the apartment type information; extracts readable geometric features, contour features, and door and window position features from the spatial model; performs feature stitching on the contour features, door and window position features, and readable geometric features to generate the area structure features corresponding to the spatial model; for any candidate layout, performs furniture constraint filtering on the candidate layout based on the set of furniture to be placed; extracts the layout structure features of the candidate layouts that have passed the furniture constraint filtering, and performs structural constraint filtering on the layout structure features based on the readable geometric features; determines the composite similarity between the layout structure features that have passed the structural constraint filtering and the area structure features; sorts the multiple candidate layouts in descending order according to the composite similarity of the multiple candidate layouts, and determines the top preset number of candidate layouts in the sorting results as target candidate layouts; generates a target layout scheme for the furniture layout request based on the multiple target candidate layouts, and sends the target layout scheme to the user. This method addresses the problem of incomplete expression of apartment layout features in existing technologies by constructing a multi-level constraint filtering framework and a composite feature representation mechanism. It achieves collaborative matching of apartment layout and furniture constraints, thereby significantly improving retrieval efficiency and interpretability while ensuring recommendation quality. It also enhances the retrieval efficiency of large-scale candidate layouts and strengthens the matching degree between layout schemes and actual spatial conditions and furniture requirements. This not only effectively narrows the retrieval scope under the condition of a large-scale candidate layout library, but also enhances the adaptability between candidate layouts and actual spatial structures. This alleviates the problems of low retrieval efficiency, matching distortion, and insufficient reliability of recommendation results caused by relying solely on room outlines or fragmented processing of furniture and spatial structure constraints in existing technologies.

[0238] Figure 5 A structural diagram of the recommended device for the furniture layout provided in this application. (See diagram below.) Figure 5 As shown, this application provides a furniture layout recommendation device 500, which includes:

[0239] The processing module 501 is used to determine the user's furniture layout request; the furniture layout request includes the set of furniture to be arranged and the apartment layout information.

[0240] The processing module 501 is also used to determine the area to be laid out corresponding to the furniture layout request based on the apartment type information, and extract the area structure features corresponding to the area to be laid out; the area to be laid out is used to identify the space range where furniture can be placed; the area structure features are used to characterize the inherent spatial structure information of the area to be laid out.

[0241] The processing module 501 is also used to determine multiple target candidate layouts based on the regional structural features and the set of furniture to be deployed.

[0242] The processing module 501 is also used to generate a target layout scheme for the furniture layout request based on multiple target candidate layouts, and send the target layout scheme to the user.

[0243] In one possible implementation, the processing module 501 is further configured to generate a spatial model corresponding to the area to be laid out based on the apartment type information.

[0244] The processing module 501 is also used to extract readable geometric features, contour features and door and window position features of the spatial model respectively.

[0245] The processing module 501 is also used to perform feature splicing on contour features, door and window position features and readable geometric features to generate regional structure features corresponding to the spatial model;

[0246] Among them, the contour feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the contour of the model; the door and window position feature is generated by emitting rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection point of the ray and the door and window; the readable geometric features include at least one of the following: area, aspect ratio, shape factor, convex hull area ratio, number of vertices, number of doors and windows, room type identifier, furniture type identifier, space utilization rate, functional zoning identifier, geometric symmetry index, and boundary curvature index.

[0247] In one possible implementation, the processing module 501 is further configured to perform multi-level filtering on multiple candidate layouts in the candidate layout library based on regional structural features and the set of furniture to be placed, to obtain multiple target candidate layouts.

[0248] The candidate layout library is used to store multiple searchable and matching candidate layout data.

[0249] In one possible implementation, the processing module 501 is further configured to perform furniture constraint filtering on any candidate layout based on the set of furniture to be placed.

[0250] The processing module 501 is also used to extract the layout structure features of the candidate layouts filtered by furniture constraints, and to perform structural constraint filtering on the layout structure features based on readable geometric features.

[0251] The processing module 501 is also used to determine the composite similarity between the layout structural features filtered by structural constraints and the regional structural features.

[0252] The processing module 501 is also used to sort the multiple candidate layouts in descending order according to the composite similarity of the multiple candidate layouts, and to determine the first preset number of candidate layouts in the sorting result as the target candidate layout.

[0253] In one possible implementation, the processing module 501 is further configured to determine category constraint features and size constraint features based on the set of furniture to be placed.

[0254] The processing module 501 is also used to determine the furniture set corresponding to the candidate layout, and to perform category constraint filtering on the furniture set based on category constraint features.

[0255] The processing module 501 is also used to perform size constraint filtering on the furniture set based on size constraint features after the furniture set has been filtered by category constraints;

[0256] Among them, category constraint filtering is used to filter candidate layouts that are not included in the furniture set to be placed; size constraint filtering is used to filter candidate layouts that have at least one piece of furniture whose size does not belong to the preset size range of the corresponding furniture in the furniture set to be placed.

[0257] In one possible implementation, the processing module 501 is further configured to determine the continuous feature tolerance interval corresponding to the readable continuous feature, and to determine the discrete feature tolerance interval corresponding to the readable discrete feature.

[0258] The processing module 501 is also used to perform structural constraint filtering on the continuous feature and the discrete feature of the layout based on the continuous feature tolerance interval and the discrete feature tolerance interval, respectively.

[0259] The processing module 501 is also used to determine the layout structural features through structural constraint filtering when the continuous features of the layout are within the tolerance range of the continuous features and the discrete features of the layout are within the tolerance range of the discrete features.

[0260] In one possible implementation, the processing module 501 is further configured to determine the cosine similarity and Euclidean distance between the layout structural features and the region structural features, respectively.

[0261] The processing module 501 is also used to determine the composite similarity between layout structural features and region structural features based on cosine similarity and Euclidean distance.

[0262] In one possible implementation, the processing module 501 is further configured to determine the room type corresponding to the apartment type information, and based on the room type, determine the weight ratio of the outline features, door and window location features, and readable geometric features.

[0263] The processing module 501 is also used to determine the corresponding weighted cosine similarity and weighted Euclidean distance based on the layout structure features, region structure features and weight ratios, respectively.

[0264] The processing module 501 is also used to determine the composite similarity between layout structural features and regional structural features based on weighted cosine similarity and weighted Euclidean distance.

[0265] The furniture layout recommendation device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0266] Figure 6 A schematic diagram of the structure of the electronic device provided in this application. Figure 6 As shown, the electronic device 600 provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device 600 further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0267] In a specific implementation, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to perform the above-described method.

[0268] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0269] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0270] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0271] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0272] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0273] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0274] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0275] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0276] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0277] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0278] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0279] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0280] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0281] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for recommending furniture layout, characterized in that, The method includes: Determine the user's furniture layout request; the furniture layout request includes the set of furniture to be placed and the apartment layout information; Based on the apartment layout information, the area to be laid out corresponding to the furniture layout request is determined, and the area structure features corresponding to the area to be laid out are extracted; the area to be laid out is used to identify the space range where furniture can be placed; the area structure features are used to characterize the inherent spatial structure information of the area to be laid out. Based on the regional structural features and the set of furniture to be deployed, multiple target candidate layouts are determined. Based on the multiple target candidate layouts, a target layout scheme for the furniture layout request is generated, and the target layout scheme is sent to the user.

2. The method for recommending furniture layout according to claim 1, characterized in that, The step of determining the area to be laid out corresponding to the furniture layout request based on the apartment layout information, and extracting the area structure features corresponding to the area to be laid out, includes: Based on the apartment layout information, a spatial model corresponding to the area to be laid out is generated; The readable geometric features, contour features, and door and window position features of the spatial model are extracted respectively. The contour features, the door and window position features, and the readable geometric features are spliced ​​together to generate the region structure features corresponding to the spatial model; The contour feature is generated by radiating rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection points of the rays and the model contour; the door and window position feature is generated by radiating rays uniformly in the circumferential direction from the center point of the contour reference of the spatial model as the origin and recording the distance between the intersection points of the rays and the doors and windows; the readable geometric features include at least one of the following: area, aspect ratio, shape factor, convex hull area ratio, number of vertices, number of doors and windows, room type identifier, furniture type identifier, space utilization rate, functional zoning identifier, geometric symmetry index, and boundary curvature index.

3. The method for recommending furniture layout according to claim 2, characterized in that, Based on the regional structural features and the set of furniture to be placed, multiple target candidate layouts are determined, including: Based on the regional structural features and the set of furniture to be placed, multiple candidate layouts in the candidate layout library are filtered at multiple levels to obtain multiple target candidate layouts. The candidate layout library is used to store multiple searchable and matching candidate layout data.

4. The method for recommending furniture layout according to claim 3, characterized in that, Based on the regional structural features and the set of furniture to be placed, multiple candidate layouts in the candidate layout library are filtered at multiple levels to obtain multiple target candidate layouts, including: For any of the candidate layouts, furniture constraint filtering is performed on the candidate layouts based on the set of furniture to be deployed; Extract the layout structure features of the candidate layouts filtered by the furniture constraints, and perform structural constraint filtering on the layout structure features based on the readable geometric features; Determine the composite similarity between the layout structural features filtered by the structural constraints and the region structural features; Based on the composite similarity of the multiple candidate layouts, the multiple candidate layouts are sorted in descending order, and the first preset number of candidate layouts in the sorting result are determined as the target candidate layout.

5. The method for recommending furniture layout according to claim 4, characterized in that, The step of filtering the candidate layouts based on the set of furniture to be deployed includes: Based on the set of furniture to be deployed, determine the category constraint features and size constraint features; Determine the furniture set corresponding to the candidate layout, and perform category constraint filtering on the furniture set based on the category constraint features; After the furniture set is filtered by the category constraint, the furniture set is then filtered by size constraint based on the size constraint feature; The category constraint filtering is used to filter candidate layouts that are not included in the furniture set to be placed; the size constraint filtering is used to filter candidate layouts in the furniture set where at least one piece of furniture does not belong to the preset size range of the corresponding furniture in the furniture set to be placed.

6. The method for recommending furniture layout according to claim 4, characterized in that, The readable geometric features include readable continuous features and readable discrete features; the layout structural features include layout continuous features and layout discrete features; the structural constraint filtering of the layout structural features based on the readable geometric features includes: Determine the continuous feature tolerance interval corresponding to the readable continuous feature, and determine the discrete feature tolerance interval corresponding to the readable discrete feature; Based on the continuous feature tolerance interval and the discrete feature tolerance interval, structural constraint filtering is performed on the continuous layout features and the discrete layout features, respectively. The method further includes: If the continuous feature of the layout is within the tolerance range of the continuous feature and the discrete feature of the layout is within the tolerance range of the discrete feature, then the layout structural feature is determined to pass the structural constraint filtering.

7. The method for recommending furniture layout according to claim 4, characterized in that, Determining the composite similarity between the layout structural features filtered by the structural constraints and the region structural features includes: Determine the cosine similarity and Euclidean distance between the layout structural features and the region structural features, respectively; Based on the cosine similarity and the Euclidean distance, the composite similarity between the layout structural features and the region structural features is determined.

8. The method for recommending furniture layout according to claim 4, characterized in that, The method further includes: Determine the room type corresponding to the apartment layout information, and based on the room type, determine the weight ratio of the outline feature, the door and window position feature, and the readable geometric feature; Determining the composite similarity between the layout structural features filtered by the structural constraints and the region structural features includes: Based on the layout structure features, the region structure features, and the weight ratio, the corresponding weighted cosine similarity and weighted Euclidean distance are determined respectively. Based on the weighted cosine similarity and the weighted Euclidean distance, the composite similarity between the layout structural features and the region structural features is determined.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-8.

Images