Method and System for Generating Reference-Based Interior Space Design Concepts Using Artificial Intelligence

An AI-driven interior design system generates spatial models, extracts area-specific features from reference images, and automates construction drawings, addressing subjective inconsistencies and inefficiencies, ensuring consistent and cost-effective design outcomes.

KR103045199B1Active Publication Date: 2026-11-06DESIGN CUBE CO CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
KR1020260054050
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2026-03-25
Publication Date
2026-11-06
Estimated Expiration
2046-03-25

AI Technical Summary

Technical Problem

Existing interior design processes rely heavily on subjective judgment, are time-consuming, and lack systematic methods to apply reference design features to new spaces, leading to inconsistencies and inefficiencies in design quality and cost estimation.

Method used

An AI-based system that generates a spatial model of a new space, analyzes reference design images to extract area-specific features, and automatically produces 3D interior design results and construction drawings, including material and cost calculations, while allowing for objective selection and combination of design elements based on budget constraints.

Benefits of technology

The system ensures consistent and efficient application of reference design features, reduces variability in design quality, automates the design-to-construction process, and objectively selects replacement materials, enhancing overall design harmony and cost management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112026036465714-PAT00003_ABST
    Figure 112026036465714-PAT00003_ABST
Patent Text Reader

Abstract

The following embodiments relate to a system that utilizes AI to provide interior design results and construction details for a new space based on reference design images.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The following embodiments relate to a system that utilizes AI to generate interior design results and construction detail drawings based on reference design images. More specifically, the following embodiments relate to a technology that generates a space model by receiving spatial information about a new space, extracts reference design feature information by analyzing reference design images obtained by directly receiving input from a user or searching from an external server using an AI model, generates a three-dimensional interior design result by applying the extracted reference design feature information to the space model, and generates construction detail drawings based on the generated design result. Background Technology

[0003] Recently, in the field of interior design, it has become common practice to design new spaces by referencing design images found on various media such as Pinterest, interior design magazines, and social media. Interior designers carry out a series of processes in which they implement a design suitable for the new space based on the reference images preferred by the client, convert it into detailed drawings suitable for construction, and provide them to the construction site.

[0004] However, the existing interior design process has various fundamental limitations and problems, failing to meet the expectations of interior designers and customers.

[0005] The first problem is that the process of extracting design features from reference images and applying them to a new space relies entirely on the designer's subjective judgment. Previously, designers had to visually observe reference images to identify design elements such as color, finishes, and furniture placement, and then manually apply them while considering the size, shape, and structural constraints of the new space. Consequently, the quality of the final product varies significantly depending on the designer's proficiency, and it is difficult to consistently reflect the intent of the reference design in the new space. Furthermore, since it is virtually impossible to systematically separate and analyze design features by area—such as walls, floors, ceilings, and furniture—in the reference images, it is difficult to quantitatively verify the harmony between design elements.

[0006] The second problem is that the entire process requires significant time and cost because each stage, from the generation of 3D design deliverables to the creation of construction detail drawings, is performed in a fragmented manner. Previously, designers had to manually create construction drawings—such as floor plans, ceiling plans, and elevation drawings—separately using tools like CAD, and also manually calculate material quantities and estimate construction costs. This process can lead to discrepancies between the 3D design deliverables and construction drawings, and creates inefficiency as all related drawings and quantity calculations must be redone whenever the design is changed.

[0007] The third problem is that the process of combining multiple reference designs or selecting alternative materials and furniture due to budget constraints is not systematized. When a client wishes to selectively combine elements from different reference designs by area, the designer must manually review the stylistic consistency between each element. Furthermore, when the construction budget is exceeded, the process of searching for alternative materials and furniture and evaluating their similarity to the original design relies entirely on the designer's experience and knowledge, making it difficult to select substitutes objectively and systematically. Prior art literature

[0009] Registered Patent 10-2773091(2025.02.20) The problem to be solved

[0010] The present embodiments aim to overcome the limitations of existing interior design processes and to systematically and consistently apply the characteristics of a reference design to a new space by receiving spatial information about the new space to generate a spatial model, analyzing reference design images obtained by direct input from a user or search from an external server using an AI model to extract design feature information by area, such as wall area, floor area, ceiling area, and furniture area, applying the extracted design feature information to the spatial model to generate a 3D interior design result, and automatically generating construction detailed drawings including construction drawing data, a list of material and furniture quantities, and estimated construction costs based on the generated design result. Furthermore, the present embodiments aim to improve the efficiency and quality of interior design by selectively combining multiple reference design images by area and selecting alternative materials and furniture based on objective criteria according to budget constraints. means of solving the problem

[0012] An electronic device that generates an interior design result and a construction detail drawing of a new space based on a reference design image using AI (Artificial Intelligence) according to one embodiment includes a communication unit that transmits and receives data with an external server, a memory that stores at least one instruction, and a processor that executes said at least one instruction. The processor can generate a spatial model of the new space by receiving spatial information about the new space, obtain at least one reference design image by receiving it directly from a user or by searching from the external server through the communication unit, analyze the obtained reference design image with an AI model to extract reference design feature information, apply the extracted reference design feature information to the generated spatial model to generate a three-dimensional interior design result for the new space, and generate a construction detail drawing based on the three-dimensional interior design result.

[0013] In addition, when the processor generates the spatial model, it receives spatial information including at least one of floor plan data, actual measurement data, and spatial image of the new space, calculates spatial structure data including wall locations, opening locations, ceiling height, and floor area of ​​the new space based on the input spatial information, generates a three-dimensional spatial model based on the calculated spatial structure data, and determines the generated three-dimensional spatial model as a spatial model for the new space.

[0014] In addition, when the processor extracts the reference design feature information, it may use the AI ​​model to divide the acquired reference design image into multiple design areas including a wall area, a floor area, a ceiling area, and a furniture area, extract area-specific feature data including color palette information, pattern information, and finishing material information for each of the divided wall area, floor area, and ceiling area, extract furniture feature data including furniture type, style, size, and arrangement pattern for the divided furniture area, and generate the reference design feature information by integrating the extracted area-specific feature data and furniture feature data.

[0015] In addition, when the processor generates the three-dimensional interior design result, it derives spatial constraints including the size, shape, structural location, and opening location of the new space from the generated space model, matches and applies color palette information, pattern information, and finishing material information corresponding to each of the wall area, floor area, and ceiling area of ​​the space model based on the extracted area-specific feature data, determines the type, size, and placement location of furniture and equipment suitable for the new space based on the extracted furniture feature data and the spatial constraints using the AI ​​model, and integrates the applied area-specific finishing material, color, and pattern information and the determined furniture placement information to generate the three-dimensional interior design result.

[0016] In addition, when the processor generates the construction detail drawing, it may calculate a list of materials and furniture quantities required for construction and an estimated construction cost based on the generated three-dimensional interior design result, generate construction drawing data including at least one of a floor plan, a ceiling plan, an elevation view, and a detailed cross-section based on the generated three-dimensional interior design result, generate priority and process sequence information for each construction process based on the generated construction drawing data, and generate the construction detail drawing including the list of materials and furniture quantities, the estimated construction cost, the construction drawing data, and the priority and process sequence information for each construction process.

[0017] In addition, the processor, in generating the reference design feature information, obtains a plurality of reference design images by receiving input directly from a user or by searching from the external server through the communication unit, divides each of the obtained plurality of reference design images into a plurality of design areas including a wall area, a floor area, a ceiling area, and a furniture area using the AI ​​model, extracts area-specific feature data including at least color palette information, pattern information, and finishing material information for each of the divided wall area, floor area, and ceiling area, and extracts furniture feature data including furniture type, style, size, and arrangement pattern for the divided furniture area, sorts the area-specific feature data and furniture feature data extracted from each of the plurality of reference design images so that they can be compared by reference design image for each of the wall area, floor area, ceiling area, and furniture area, and provides them to the user, receives input from the user selecting a reference design image to be applied to each of the wall area, floor area, ceiling area, and furniture area from the plurality of reference design images, and combines the feature data of the reference design image corresponding to each area based on the selection to produce a new integrated reference design Generate feature information and, using the AI ​​model, determine whether the color scheme, finishing material grade range, and furniture style category match among the feature data by region included in the generated integrated reference design feature information; if, as a result of the determination, there exists a region where the color scheme, finishing material grade range, or furniture style category does not match, use the AI ​​model to automatically correct the feature data of that region to match the color scheme, finishing material grade range, and style category of the adjacent region, and update the integrated reference design feature information by reflecting the automatically corrected feature data.The updated integrated reference design feature information mentioned above can be provided to the user to receive approval input.

[0018] In addition, when the processor generates the construction detailed drawing, it receives budget information from the user, compares the calculated estimated construction cost with the budget information, and if the calculated estimated construction cost exceeds the budget information, it prioritizes selecting items with a high contribution to the cost overrun as replacement targets based on the unit price information of each material and furniture item included in the generated list of material and furniture quantities, and for each selected replacement target item, searches for replaceable candidate materials and candidate furniture belonging to the same category in the material and furniture database of the external server via the communication unit, and uses the AI ​​model to calculate the color similarity, pattern similarity, and finishing material similarity with the extracted reference design feature information for each of the searched candidate materials and candidate furniture, and selects candidate materials and candidate furniture whose calculated color similarity, pattern similarity, and finishing material similarity are each greater than or equal to a preset threshold value as a replacement candidate group, selects combinations that can be constructed within the budget information from the selected replacement candidate group, provides the selected combinations along with comparison information with existing material and furniture items to the user to receive approval input, and based on the approval input, selects the The above-mentioned three-dimensional interior design result can be regenerated by reflecting the combination, and the above-mentioned construction drawing detail can be regenerated by recalculating construction drawing data, a list of material and furniture quantities, and estimated construction costs based on the regenerated three-dimensional interior design result.

[0019] Additionally, an electronic device according to one embodiment may be combined with hardware and controlled by a computer program stored on a medium to execute the method of any one of the methods described above. Effects of the invention

[0021] The present invention can overcome the fundamental limitations of the existing interior design process and dramatically improve the accuracy and efficiency of the process of applying the characteristics of a reference design to a new space. Unlike the existing method that relies on the subjective judgment of the designer, the present invention analyzes reference design images using an AI model to divide them into multiple design areas, such as wall areas, floor areas, ceiling areas, and furniture areas, and systematically extracts color palette information, pattern information, finishing material information, and furniture characteristic data for each area, thereby enabling the intention of the reference design to be reflected in the new space quantitatively and consistently.

[0022] In addition, the present invention receives spatial information of a new space to generate a spatial model, matches and applies reference design feature information to the walls, floors, and ceilings of the new space based on spatial constraints derived from the spatial model, and determines the types, sizes, and placement locations of furniture and equipment using an AI model, thereby automatically generating a 3D interior design result suitable for the size, shape, and structural characteristics of the new space. This is differentiated from existing technologies in that it eliminates variations in the quality of the result due to the designer's proficiency and reduces the time and cost of the design process.

[0023] Furthermore, the present invention can unify the entire process from design results to construction details by automatically generating construction drawing data, such as floor plan layouts, ceiling plans, and elevation drawings, based on the generated 3D interior design results, calculating a list of materials and furniture quantities required for construction and estimated construction costs, and generating construction details that include priority and process sequence information for each construction process. Through this, the problem of discrepancies between 3D design results and construction drawings that previously occurred can be resolved, and all related drawings and quantity calculations can be automatically updated when the design is changed.

[0024] The present invention also selectively combines multiple reference design images by region, and utilizes an AI model to objectively determine whether color schemes, finishing material grade ranges, and furniture style categories match among feature data by region, and automatically corrects inconsistent regions, thereby systematically ensuring stylistic consistency among multiple reference design elements. This region-specific selective combination function supports customers in freely combining the strengths of different reference designs while maintaining overall design harmony.

[0025] Furthermore, the present invention enables the objective and systematic selection of replacement materials that maintain the characteristics of the original design to the greatest extent within the budget by prioritizing replacement targets for items with a high contribution to cost overruns under budget constraints, searching for replacement candidates of the same category in material and furniture databases on an external server, and utilizing an AI model to calculate color similarity, pattern similarity, and finishing material similarity to select only those candidates that meet or exceed preset threshold values. Through this, the replacement material selection process, which previously relied entirely on the designer's experience and knowledge, can be performed based on quantitative criteria.

[0026] The present invention can be utilized in various environments such as interior design offices, interior departments of construction companies, and interior platforms, and can realize automation and objective quality control using AI throughout the entire interior design process based on reference designs. Brief explanation of the drawing

[0028] FIG. 1 is a diagram illustrating the relationship between an external server and an electronic device according to one embodiment. FIG. 2 is a block diagram of an electronic device according to one embodiment. FIG. 3 is a flowchart for explaining the operation of an electronic device according to one embodiment. FIG. 4 is a flowchart illustrating a spatial model generation operation according to one embodiment. FIG. 5 is a flowchart illustrating the operation of extracting reference design feature information according to one embodiment. FIG. 6 is a flowchart illustrating the operation of generating a three-dimensional interior design result according to one embodiment. FIG. 7 is a flowchart illustrating the operation of generating a construction detail drawing according to one embodiment. Specific details for implementing the invention

[0029] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0030] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0031] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0032] Furthermore, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless explicitly and specifically defined otherwise. In certain cases, terms have been selected at the applicant's discretion, and in such cases, their meanings will be described in detail in the relevant explanatory sections. Accordingly, terms used in this disclosure should be defined not merely by their names, but based on their meanings and the content throughout this disclosure.

[0033] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0034] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0035] Throughout this specification, when a part is described as "comprising" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. Furthermore, the singular form used in this specification includes the plural form unless specifically stated otherwise in the text. Additionally, the expression "at least one of a, b, and c" described throughout this specification may encompass 'a alone', 'b alone', 'c alone', 'a and b', 'a and c', 'b and c', or 'a, b, and c all'.

[0036] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0037] Additionally, terms such as “…part,” “…module,” etc., as described in this specification refer to a unit that processes at least one function or operation, which may be implemented in hardware or software, or a combination of hardware and software. Furthermore, embodiments of the present disclosure may be represented by functional block configurations and various processing steps. These functional blocks may be implemented by various numbers of hardware and / or software configurations that execute specific functions. For example, embodiments of the present disclosure may employ integrated circuit configurations such as memory, processing, logic, look-up tables, etc., which can execute various functions under the control of one or more microprocessors or other control devices.

[0038] Each block of the process flow diagrams attached to this specification and combinations of the flow diagrams may be executed by computer program instructions. Since these computer program instructions may be loaded into the processor of a general-purpose computer, a computer for special purposes, or other programmable data processing equipment, the instructions executed through the processor of the computer or other programmable data processing equipment create means for performing the functions described in the flow diagram block(s).

[0039] Since computer program instructions can be loaded onto a computer or other programmable data processing equipment, instructions that perform a series of operation steps on the computer or other programmable data processing equipment to create a process executed by the computer can also provide steps for executing the functions described in the flowchart block(s).

[0040] Additionally, each block may represent a module, segment, or part of code containing one or more executable instructions for executing a specified logical function(s). Furthermore, in some alternative execution examples, the functions mentioned in the blocks may occur out of order. For instance, two blocks described in succession may actually be executed substantially simultaneously, or the blocks may be executed in reverse order according to their corresponding functions.

[0041] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0042] The definitions and acquisition paths of key terms and data used in this specification are clarified below. The following definitions of terms are applied consistently throughout this specification, and the scope of meaning and the source of data for each term are described in detail so that they can be clearly understood by those skilled in the art of this invention.

[0043] "Spatial information" refers to physical information about a new space that is directly input by the user. Spatial information includes at least one of floor plan data, actual measurement data, and spatial images of the new space. Here, floor plan data is two-dimensional drawing data representing the planar shape of the new space and can be input in a CAD (Computer-Aided Design) file format or an image file format. Actual measurement data is numerical data including the length of each wall surface of the new space, ceiling height, and the location and size of openings (windows, doors, etc.), and can be input by the user through direct measurement or can be data automatically measured using a laser distance meter or a LiDAR (Light Detection and Ranging) sensor. Spatial images are photographs or 360-degree panoramic images of the interior of the new space, which can be used by the processor (230) to identify the shape and structure of the space through image analysis.

[0044] "Space structure data" refers to data regarding the structural attributes of a new space calculated by the processor (230) based on input space information. Space structure data includes the location of walls, the location of openings, the ceiling height, and the floor area of ​​the new space. Here, the location of walls is data including the coordinates of the starting and ending points of each wall surface, the length of the wall surface, and the angle between the walls. The location of openings is data including the location, size (width and height), and height from the floor of each window and door. The ceiling height is numerical data representing the vertical distance from the floor surface to the ceiling surface, and if the ceiling height varies depending on the location within the space (e.g., sloped ceiling, stepped ceiling, etc.), the ceiling height for each location can be calculated individually. The floor area is numerical data representing the total area of ​​the floor surface of the new space.

[0045] "Space model" refers to a three-dimensional digital model of a new space generated by a processor (230) based on spatial structure data. The space model may be composed of mesh data or voxel data that define the three-dimensional shapes of the walls, floors, and ceilings of the new space, and may include rendering information such as normal vectors and UV coordinates for each surface. According to one embodiment, the space model may be generated in a Building Information Modeling (BIM) format, in which case each architectural element (wall, floor, ceiling, opening, etc.) is defined as an individual object, making it easy to independently edit and manage each element when applying interior design elements and generating construction drawing data.

[0046] "Reference design image" refers to image data that serves as a reference for interior design, which is obtained by directly inputting from a user or by searching from an external server (100) through a communication unit (210). The reference design image may include photographs of existing interior spaces, portfolio images of interior designers, or 3D rendering images. When inputting directly by a user, the user may input by uploading images of the interior design style they wish to reference to an electronic device (200). When searching from an external server (100), the processor (230) may search the reference design image database of the external server (100) based on keywords (e.g., "modern," "Nordic," "minimal," "classic," etc.) or categories (e.g., "living room," "bedroom," "kitchen," "bathroom," etc.) received from the user. According to one embodiment, the external server (100) can be linked with an online interior platform (e.g., Pinterest, Houzz, etc.) to provide reference design images that reflect the latest interior trends.

[0047] "AI model" means a deep learning-based machine learning model that is stored in memory (220) with pre-learned weight parameters and loaded and executed by a processor (230). In this specification, an AI model may refer to at least one of the following depending on the context: (i) a semantic segmentation model for dividing a reference design image into a plurality of design areas including wall areas, floor areas, ceiling areas, and furniture areas; (ii) a feature extraction model for recognizing features such as color palette information, pattern information, and finishing material information from each divided area; (iii) an object recognition model for identifying the type, style, size, and arrangement pattern of furniture in the furniture area; (iv) a design adaptation model for applying the extracted reference design feature information to the spatial constraints of a new space; (v) a style consistency assessment model for determining whether color series, finishing material grade ranges, and furniture style categories match between feature data extracted from a plurality of reference design images; and (vi) a similarity calculation model for calculating color similarity, pattern similarity, and finishing material similarity with respect to alternative candidate materials and furniture with respect to the reference design feature information. Scoring) model, or (vii) Furniture Layout Optimization model for determining the type, size, and placement location of furniture and fixtures suitable for a new space based on spatial constraints. Each AI model is pre-trained based on a deep learning architecture and learning methodology known to a person with ordinary knowledge in the relevant technical field and stored in memory (220), and can be executed by a processor (230) using a computing device such as a GPU or NPU.

[0048] "Design area" refers to a unit of area generated as a result of the processor (230) semantically segmenting a reference design image using an AI model (semantic segmentation model). The design area consists of multiple areas including a wall area, a floor area, a ceiling area, and a furniture area. Here, the wall area is a set of pixels corresponding to a wall within the reference design image, the floor area is a set of pixels corresponding to a floor, the ceiling area is a set of pixels corresponding to a ceiling, and the furniture area is a set of pixels corresponding to furniture and equipment. Each design area is represented by a pixel-unit segmentation mask, and multiple design areas are mutually exclusive for a single reference design image.

[0049] "Region-specific feature data" refers to data regarding visual features extracted by the processor (230) for each of the divided wall area, floor area, and ceiling area using an AI model (feature extraction model). The region-specific feature data includes color palette information, pattern information, and finishing material information. Here, the color palette information is a list of colors that appear predominantly in the area and can be expressed as RGB (Red, Green, Blue) values, HSV (Hue, Saturation, Value) values, or Pantone color codes. The pattern information is data including the type of visual pattern appearing on the surface of the area (e.g., solid, stripe, check, herringbone, geometric, etc.) and the scale (size) of the pattern. The finishing material information is data including the type of finishing material applied to the area (e.g., paint, wallpaper, tile, wood, stone, fabric, etc.) and texture characteristics.

[0050] "Furniture feature data" refers to feature data regarding furniture and equipment that the processor (230) extracts for a segmented furniture area using an AI model (object recognition model). The furniture feature data includes the type of furniture (e.g., sofa, table, chair, bed, cabinet, etc.), style (e.g., modern, classic, minimal, Nordic, etc.), size (e.g., estimated dimensions of width, length, and height), and arrangement pattern (e.g., wall-mounted arrangement, center arrangement, corner arrangement, symmetrical arrangement, etc.). The furniture feature data is a result value calculated by the processor (230) analyzing the furniture area of ​​a reference design image, and can be extracted along with bounding box coordinates for each individual furniture object.

[0051] "Reference design feature information" refers to a data set generated by the processor (230) by integrating extracted area-specific feature data and furniture feature data. For a single reference design image, the reference design feature information integrates area-specific feature data for the wall area, floor area, and ceiling area, and furniture feature data for the furniture area, into a structured data format (e.g., JSON, XML, etc.). The reference design feature information is subsequently referenced by the processor (230) in the process of generating a three-dimensional interior design result by applying it to the spatial model of the new space.

[0052] "Spatial constraints" refers to data regarding physical constraints of a new space derived from a generated spatial model by the processor (230). Spatial constraints include the size, shape, location of structures, and location of openings of the new space. Here, the size is numerical data of the width, length, and height of the space, the shape is data representing the planar shape of the space (e.g., rectangular, L-shaped, U-shaped, etc.), the location of structures is location and dimension data of structures that impose constraints on furniture placement and the application of finishing materials within the space, such as columns, beams, and wall protrusions, and the location of openings is location, size, and opening / closing direction data of windows, doors, arches, etc. When applying feature information extracted from a reference design image to the new space, spatial constraints are used as a standard for adjusting the size of furniture, changing placement locations, and adjusting the scope of finishing material application by taking into account the physical limits of the space.

[0053] "Three-dimensional interior design result" refers to a three-dimensional digital interior design result for a new space generated by a processor (230) applying extracted reference design feature information to a generated space model. The three-dimensional interior design result includes three-dimensional rendering data with finishing material, color, and pattern information applied to each of the walls, floors, and ceilings of the space model, and three-dimensional model data of furniture and equipment placed within the space. According to one embodiment, the processor (230) may generate a plurality of design candidates and provide them to the user, and determine the final design result based on the user's selection. Additionally, the three-dimensional interior design result may be provided in a form that the user can visually verify in a walkthrough method or a VR (Virtual Reality) environment.

[0054] "Construction drawing data" refers to two-dimensional drawing data required for actual construction, generated by a processor (230) based on a three-dimensional interior design result. The construction drawing data includes at least one of a floor plan, a ceiling plan, an elevation view, and a detailed cross-section. Here, the floor plan is a drawing in the form of a new space viewed from above, including the placement locations of furniture and equipment, and indications of movement paths; the ceiling plan includes lighting placement on the ceiling surface and information on ceiling finishing materials; the elevation view is a drawing in the form of each wall viewed from the front, including wall finishing materials, storage cabinet locations, outlet and switch locations; and the finishing material information is detailed information including the product name, specifications, application location, and application area of ​​each applied finishing material. According to one embodiment, the construction drawing data is generated in a CAD file format or a BIM (Building Information Modeling) format and can be output as a drawing that can be directly utilized at the construction site.

[0055] "Construction detail drawing" refers to a comprehensive construction document that integrates information necessary for the entire actual construction process, generated by the processor (230) based on a three-dimensional interior design result. The construction detail drawing includes a list of material and furniture quantities, estimated construction costs, construction drawing data, and information on the priority and sequence of construction processes. Here, the list of material and furniture quantities is a list that organizes the product name, specifications, quantity, and unit price of each material and furniture required for construction by item, the estimated construction cost is an estimate of the total construction cost calculated based on the list of material and furniture quantities, and the information on the priority and sequence of construction processes is information regarding the sequential relationship and optimal construction order of each construction process (e.g., demolition, plumbing, carpentry, electrical work, tiling, painting, furniture installation, etc.) derived based on the construction drawing data. According to one embodiment, the construction detail drawing may be finally confirmed and output after receiving confirmation input from a user (or construction personnel).

[0056] "Integrated reference design feature information" refers to a new set of feature information data generated by a processor (230) by combining area-specific feature data and furniture feature data, each extracted from multiple reference design images, based on the user's area-specific selection. Specifically, when the user selects feature data of the first reference design image for the wall area, feature data of the second reference design image for the floor area, feature data of the third reference design image for the ceiling area, and feature data of the fourth reference design image for the furniture area, the processor (230) generates integrated reference design feature information by combining feature data corresponding to each area based on these area-specific selections. The integrated reference design feature information is finally determined through an inter-area harmony verification and automatic correction process by the processor's (230) AI model (style matching judgment model).

[0057] "Budget information" refers to information regarding construction costs that is directly input by the user. Budget information includes the total budget amount that the user can invest in the interior construction of a new space, and may additionally include budget allocation ratios by item (e.g., ratio of material costs to labor costs, ratio of furniture costs to finishing material costs, etc.). Budget information is used as a standard for the processor (230) to compare with the estimated construction costs and to perform the process of selecting alternative materials and alternative furniture in case of cost overrun.

[0058] "Replacement candidate group" refers to a set of candidate materials and candidate furniture in which the color similarity, pattern similarity, and finishing material similarity calculated for each of the searched candidate materials and candidate furniture by the processor (230) using an AI model (similarity calculation model) are each greater than or equal to a preset standard value. The replacement candidate group consists of alternative candidates that can reduce costs while maintaining visual similarity to the original materials and furniture, and the processor (230) selects a combination that can be constructed within the budget information from the replacement candidate group and provides it to the user.

[0059] "Style Category" refers to classification information assigned by the processor (230) to classify the overall design style of each design area or furniture area. The style category may include, but is not limited to, modern, classic, minimal, industrial, Scandinavian / Nordic, Japanese / Wa-style, vintage, rustic, bohemian, and mid-century modern. The style category can be determined by the processor (230) using a style classification AI model based on a combination of finishing material types, color information, pattern information, and texture information for each design area. The style category is subsequently used as a criterion for evaluating style consistency between areas during the process of selectively combining multiple reference design images, and is also referenced as a criterion for determining whether the style matches the original design when selecting alternative materials and alternative furniture.

[0060] "Design intent description" refers to text data in the form of natural language in which the processor (230) describes the visual characteristics of each design area from a semantic perspective using a Vision-Language Multimodal AI model. The design intent description includes sentences that comprehensively describe the visual impression of the finishing material applied to the design area, the emotional characteristics of the color, the spatial effect of the pattern, and the direction of the overall design concept. The processor (230) can convert the design intent description into a vector representation using a text embedding model and add the converted vector representation to the feature data for each design area to include it in the reference design feature information. The vector representation of the design intent description is used as a criterion for evaluating semantic style consistency between design areas during the process of selectively combining multiple reference design images, and can also be used to calculate semantic similarity with the original design intent when selecting alternative materials and alternative furniture.

[0061] "Harmony score" refers to a numerical value calculated by the processor (230) by quantitatively evaluating the visual consistency and aesthetic balance between design areas combined from different reference design images. The harmony score can be calculated by calculating the color harmony between the combined design areas, the interoperability score between style categories, and the consistency between finishing material grades, respectively, and then weighted summing them. Here, the color harmony is calculated based on color harmony theory, the interoperability score between style categories is calculated based on a predefined compatibility matrix between design styles, and the consistency between finishing material grades is calculated based on the grade difference of the finishing material applied to each design area. If the calculated harmony score is greater than or equal to a preset threshold value, the processor (230) determines that the integrated reference design feature information is appropriate, and if it is less than the threshold value, it can perform automatic correction for the discrepancies between the design areas. Meanwhile, the harmony score can also be adjusted according to the physical environmental conditions of the space to be applied (direction of natural light incidence, ratio of width to length of the space, ratio of ceiling height to floor area, etc.).

[0062] "Style similarity score" refers to an indicator calculated by the processor (230) by quantifying the degree of style consistency between each design area within a single reference design image. The style similarity score is calculated by comparing the style category, color tone, and finish grade of each design area, and a higher value indicates that the design elements within the reference design image are harmoniously composed. The style similarity score is used as a criterion for judging the suitability of a combination during the process of selectively combining multiple reference design images.

[0063] "Style reflection score" refers to an indicator calculated by quantitatively evaluating how faithfully the three-dimensional interior design result generated by the processor (230) reflects the stylistic characteristics of the reference design image. The style reflection score can be calculated by weighted summing the degree of agreement for each item of the reference design characteristic information (type of finishing material, color, pattern, texture, type of furniture, furniture style, etc.). When the processor (230) generates multiple design candidates, it can calculate a style reflection score for each design candidate and rank the design candidates in order of high style reflection and present them to the user.

[0064] "Construction standard data" refers to data regarding standard specifications and construction methods required for actual construction, which is stored in memory (220) and referenced by the processor (230) during the process of generating construction detailed drawings. The construction standard data may include information on construction methods by finishing material (e.g., grout spacing when installing tiles, pattern matching method when installing wallpaper, etc.), information on criteria for calculating the required quantity of each material (e.g., number of tiles required per square meter, amount of paint required per square meter, etc.), information on criteria for minimum spacing distance when arranging furniture, information on the standard duration required for each construction process, and information on dependency on the construction sequence (e.g., electrical wiring work → wall finishing → floor installation order). The construction standard data is used as basic reference data when the processor (230) calculates construction drawing data, material quantity lists, and priorities for each construction process.

[0065] "Similarity judgment criteria data" is stored in memory (220) and refers to reference data that the processor (230) refers to in order to calculate color similarity, pattern similarity, and finishing material similarity during the process of selecting replacement materials and replacement furniture. The similarity judgment criteria data may include color space conversion parameters for determining color similarity (e.g., RGB-HSV conversion matrix, CIE Lab color distance standard, etc.), a pattern classification system for determining pattern similarity (e.g., solid, stripe, check, floral, geometric, etc.), a material category classification system for determining finishing material similarity (e.g., wood-solid wood / plywood / MDF, stone-marble / granite / artificial stone, etc.), and preset reference values ​​for each similarity.

[0066] "Spatial metadata" refers to additional information regarding the attributes of a spatial model that the processor (230) stores in memory (220) in addition to the determined spatial model. Spatial metadata may include the creation date and time of the spatial model, the type of spatial information used (floor plan data, actual measurement data, spatial image, LiDAR scan data, etc.), the coordinate system information of the spatial model, the total floor area, the classification of the use of each room, and the version information of the spatial model. Spatial metadata is subsequently utilized to efficiently reference the attribute information of the spatial model during the process of generating design results and detailed drawings for construction.

[0067] "Color similarity" refers to a numerical value between 0 and 1 calculated by the processor (230) using an AI model (similarity calculation model) to quantify the degree of visual similarity between the color of the candidate material or candidate furniture and the color palette of the corresponding area included in the reference design feature information. Color similarity can be calculated by calculating the color distance between the color value of the candidate material or candidate furniture and the color value included in the color palette of the reference design feature information based on the CIEDE2000 color difference in the CIE Lab color space or the Euclidean distance in the HSV color space, and by normalizing the calculated color distance. A value closer to 1 indicates a higher degree of color similarity with the reference design feature information. "Pattern similarity" refers to a numerical value between 0 and 1 calculated by quantifying the degree of visual similarity between the surface pattern of the candidate material or candidate furniture and the pattern information of the corresponding area included in the reference design feature information, and is calculated based on the degree of agreement of the pattern type, scale, direction, and density. "Finishing material similarity" refers to a numerical value between 0 and 1 calculated by quantifying the degree of similarity in material characteristics between the material of a candidate material or candidate furniture and the finishing material information of the corresponding area included in the reference design feature information, and is calculated based on whether the material category is of the same or similar series, the degree of similarity in surface texture characteristics, and the degree of similarity in material grade.

[0068] "Critical path" refers to a sequence of construction processes that, if delayed in the overall construction schedule, directly affects the overall construction completion schedule, which is calculated by the processor (230) based on the dependency relationships between construction processes and the duration of each construction process. A construction process corresponding to the critical path is a process in which there is no parallel alternative path or the float / slack is zero, and a delay in such a process leads directly to an increase in the overall construction period. The processor (230) identifies the construction process corresponding to the critical path and marks it separately on the construction detail drawing, and may include precautions to prevent delays in such processes (e.g., pre-ordering materials, securing skilled personnel, strict adherence to curing periods, etc.).

[0069] "User acceptance probability" refers to a value between 0 and 1 calculated by the processor (230) using a user preference learning model to predict the likelihood that each candidate material or candidate furniture included in the alternative candidate group will be approved by the user. The user acceptance probability is predicted by a user preference learning model built using the user's history of material and furniture selection, history of alternative candidate approval, history of finishing material changes, and history of feedback on design results from past design projects as training data. The closer the value is to 1, the higher the likelihood that the candidate material or candidate furniture will be approved as it matches the user's taste.

[0070] The "Overall Recommendation Score" refers to a value calculated by the processor (230) for each candidate material or candidate furniture included in the alternative candidate group by weighting the user acceptance probability to the calculated color similarity, pattern similarity, and finishing material similarity. The Overall Recommendation Score is an indicator that comprehensively reflects visual similarity with reference design feature information and the user's personal preferences, and the processor (230) can rank alternative candidates in order of highest Overall Recommendation Score and suggest them to the user.

[0072] FIG. 1 is a diagram illustrating the relationship between an external server and an electronic device according to one embodiment.

[0073] Referring to FIG. 1, an interior design system according to one embodiment of the present disclosure may include an external server (100) and an electronic device (200). In this case, the interior design system according to one embodiment of the present disclosure may be a system that generates an interior design result of a new space and a construction detail drawing based on a reference design image utilizing AI (Artificial Intelligence) by using communication between the external server (100) and the electronic device (200).

[0074] The external server (100) performs the role of storing and providing reference design images. Specifically, the external server (100) stores reference design image data of various interior design styles, a database of materials and furniture, unit price information by material and furniture, construction case data, etc. The external server (100) can search for and provide reference design images corresponding to a specific style, space type, or keyword in response to a request from the electronic device (200), or can provide catalog information of materials and furniture.

[0075] Additionally, the external server (100) may additionally store interior design portfolio data, 3D model data of materials and furniture, specification information by manufacturer of materials and furniture, construction company information, and market price fluctuation data. According to one embodiment, the external server (100) may be linked with an online interior platform (e.g., Pinterest, Houzz, etc.) to collect and store reference design images reflecting the latest interior trends in real time. These information may be referenced in the process in which the electronic device (200) uses an AI model to extract reference design feature information and generate detailed drawings for construction.

[0076] These external servers (100) may be implemented as an FTP server (File Transfer Protocol Server), a web server, a database server, a cloud server, etc., but are not limited thereto. The external servers (100) may be implemented as a single server, but may also be implemented as a set or combination of two or more servers.

[0077] The electronic device (200) is a device that uses AI to generate interior design results and construction details for a new space based on reference design images. For example, the electronic device (200) may be a device that performs operations such as generating a space model of a new space, acquiring reference design images, extracting AI-based reference design feature information, generating three-dimensional interior design results, and generating construction details by performing data communication with an external server (100).

[0078] An electronic device (200) according to one embodiment of the present disclosure may be a server, a terminal, or a device that performs various other operations and functions. For example, if the electronic device (200) is a server, the electronic device may be an FTP server, a web server, a database server, or a cloud-type server, but is not limited thereto.

[0079] As another example, if the electronic device (200) is a terminal, the electronic device may be a smartphone, tablet PC, laptop PC, desktop PC, etc., but is not limited thereto.

[0080] Meanwhile, the electronic device (200) can directly receive spatial information about a new space from the user. Additionally, the electronic device (200) can directly receive reference design images from the user or obtain reference design images by searching for them from an external server (100) through the communication unit (210). Furthermore, the electronic device (200) can output the generated three-dimensional interior design results and construction details to the user through the display unit or transmit them to an external device through the communication unit (210).

[0081] Hereinafter, an electronic device (200) according to one embodiment of the present disclosure will be described in more detail with reference to FIG. 2.

[0083] FIG. 2 is a block diagram of an electronic device according to one embodiment.

[0085] Referring to FIGS. 1 and 2, an electronic device (200) according to one embodiment of the present disclosure may include a communication unit (210), a memory (220), and a processor (230). However, the embodiments of the present disclosure are not limited thereto, and some of the communication unit (210), memory (220), and processor (230) shown in FIG. 2 may be omitted or additional components may be included. For example, the electronic device (200) may further include a display unit for visually displaying the generated three-dimensional interior design result and an input unit for receiving selection input of spatial information or reference design images from a user. These components included in the electronic device (200) are interconnected via a bus or network interface, so that the entire process from generating a spatial model to acquiring a reference design image, extracting AI-based reference design feature information, generating a three-dimensional interior design result, and generating a detailed drawing for construction can be performed organically.

[0086] The communication unit (210) is configured to transmit and receive data with the external server (100). The communication unit (210) receives a reference design image from the external server (100) and performs the role of querying material and furniture information from the material and furniture database of the external server (100).

[0087] The communication unit (210) may support wired communication or wireless communication. Wireless communication may include, for example, Wi-Fi (Wireless Fidelity), Bluetooth, LTE (Long Term Evolution), 5G, NFC (Near Field Communication), but is not limited thereto. The communication unit (210) may communicate with an external server (100) using protocols such as HTTP, HTTPS, WebSocket, and may exchange data using a RESTful API method.

[0088] Additionally, the communication unit (210) can receive reference design image data, 3D model data of materials and furniture, unit price information of materials and furniture, and catalog information from an external server (100). Furthermore, the communication unit (210) can perform the role of collecting the latest material market prices, furniture inventory status, and construction company information through data transmission and reception with external APIs (e.g., interior platform API, material distribution API, BIM (Building Information Modeling) server API, etc.).

[0089] The memory (220) is configured to store at least one instruction. Additionally, the memory (220) can store various information required for the operation, processing, or control operation of the processor (230) in an electronic format.

[0090] The memory (220) may include, for example, at least one of a main memory and an auxiliary memory. The main memory may be implemented using a semiconductor storage medium such as ROM and / or RAM. The auxiliary memory may include forms such as a Hard Disk Drive (HDD) or a Solid State Drive (SSD). Additionally, the memory (220) may be implemented as a Flash Memory, Magnetic Storage Media, Optical Storage Media, etc., but is not limited thereto.

[0091] An AI model may be stored in the memory (220). The AI ​​model may include a deep learning-based image analysis model for analyzing reference design images to extract reference design feature information. Specifically, the AI ​​model may include at least one of a Semantic Segmentation model for dividing a reference design image into multiple design areas including a wall area, a floor area, a ceiling area, and a furniture area; a Feature Extraction model for recognizing features such as color, pattern, and finishing material from each divided area; an Object Recognition model for identifying the type, style, and size of furniture; a Design Adaptation model for applying the extracted feature information to the spatial constraints of a new space; and a Style Consistency Assessment model for determining the degree of harmony between feature data extracted from multiple reference design images. This AI model may be stored in the memory (220) along with pre-learned weight parameters and loaded and executed by the processor (230).

[0092] Additionally, the memory (220) may additionally store a 3D modeling engine for generating a 3D spatial model from spatial information, a drawing generation engine for generating construction drawing data from 3D interior design results, and a quantity calculation engine for calculating the quantity and cost of materials and furniture. These multiple engines and AI models can be executed sequentially or in parallel in a pipeline form by the processor (230) to contribute to automating the entire process from generating a spatial model to generating a detailed drawing for construction.

[0093] Furthermore, the memory (220) may store construction standard data required for generating construction detailed drawings. The construction standard data may include information on construction methods for each finishing material (e.g., grout spacing when installing tiles, pattern matching method when installing wallpaper, etc.), information on criteria for calculating the required quantity of each material (e.g., number of tiles required per square meter, amount of paint required per square meter, etc.), information on criteria for minimum spacing distance when arranging furniture, information on the standard duration required for each construction process, and information on dependency on the construction sequence (e.g., electrical wiring work → wall finishing → floor installation order). By utilizing this construction standard data, the processor (230) can generate construction detailed drawings that are applicable to actual construction.

[0094] Additionally, the memory (220) may store similarity judgment criteria data for selecting replacement materials and replacement furniture. The similarity judgment criteria data may include color space conversion parameters for determining color similarity (e.g., RGB-HSV conversion matrix, CIE Lab color distance standard, etc.), a pattern classification system for determining pattern similarity (e.g., solid, stripe, check, floral, geometric, etc.), a material category classification system for determining finishing material similarity (e.g., wood-solid wood / plywood / MDF, stone-marble / granite / artificial stone, etc.), and preset reference values ​​for each similarity.

[0095] The processor (230) can be implemented in various ways. For example, one or more processors (230) can be implemented as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), NPU (Neural Processing Unit), Microcontroller (MCU, Micro Controller Unit), DSP (Digital Signal Processor), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), or a combination thereof. In particular, a GPU or NPU can be utilized for the execution of an AI model, thereby improving the speed of analysis of reference design images and the speed of rendering of 3D interior design results.

[0096] One or more processors (230) may control one or any combination of other components of the electronic device (200) and may perform operations or data processing related to communication. One or more processors (230) may execute one or more programs or instructions stored in memory (220). For example, one or more processors (230) may perform a method according to one embodiment of the present disclosure by executing one or more instructions stored in memory (220).

[0097] In embodiments of the present disclosure, the processor (230) may mean a system-on-chip (SoC) in which one or more processors (230) and other electronic components are integrated, a single-core processor, a multi-core processor, or a core included in a single-core processor or a multi-core processor, wherein the core may be implemented as a CPU, GPU, APU, MIC, DSP, NPU, hardware accelerator, or machine learning accelerator, but the embodiments of the present disclosure are not limited thereto.

[0098] In the case where the method according to one embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by a single processor (230) or by a plurality of processors (230). For example, when a first operation, a second operation, and a third operation are performed by the method according to one embodiment, the first operation, the second operation, and the third operation may all be performed by a first processor, or the first operation and the second operation may be performed by a first processor (e.g., a general-purpose processor) and the third operation may be performed by a second processor (e.g., an artificial intelligence dedicated processor).

[0099] One or more processors (230) may be implemented as a single-core processor including one core, or as one or more multicore processors including multiple cores (e.g., homogeneous multicore or heterogeneous multicore). When one or more processors (230) are implemented as multicore processors, each of the multiple cores included in the multicore processor may include internal processor memory (220), such as on-chip memory (220), and a common cache shared by the multiple cores may be included in the multicore processor (230). Additionally, each of the multiple cores (or some of the multiple cores) included in the multicore processor (230) may independently read and execute program instructions for implementing a method according to one embodiment of the present disclosure, or all (or some) of the multiple cores may be linked together to read and execute program instructions for implementing a method according to one embodiment of the present disclosure.

[0100] When a method according to one embodiment of the present disclosure includes a plurality of operations, the plurality of operations may be performed by one of the plurality of cores included in a multi-core processor, or may be performed by a plurality of cores. For example, when a first operation, a second operation, and a third operation are performed by a method according to one embodiment, the first operation, the second operation, and the third operation may all be performed by a first core included in a multi-core processor, or the first operation and the second operation may be performed by a first core included in a multi-core processor and the third operation may be performed by a second core included in a multi-core processor.

[0101] Meanwhile, the processor (230) can receive spatial information about a new space to create a spatial model, obtain at least one reference design image by receiving it directly from a user or by searching from an external server (100) through a communication unit (210), analyze the obtained reference design image with an AI model to extract reference design feature information, apply the extracted reference design feature information to the created spatial model to create a three-dimensional interior design result for the new space, and create a construction detail drawing based on the three-dimensional interior design result. Each operation of the processor (230) can be performed by referring to the AI ​​model, three-dimensional modeling engine, drawing generation engine, quantity calculation engine, and construction standard data stored in the memory (220).

[0102] Hereinafter, with reference to FIG. 3, the process by which the processor (230) generates an interior design result and a construction detail drawing based on a reference design image will be explained in detail.

[0104] FIG. 3 is a flowchart illustrating the operation of an electronic device according to one embodiment. Each step (S100 to S500) of FIG. 3 can be performed by the processor (230) of FIG. 2.

[0105] Referring to FIGS. 1 to 3, a processor (230) of an electronic device (200) according to one embodiment of the present disclosure may sequentially perform a spatial model generation step (S100), a reference design image acquisition step (S200), an AI-based reference design feature information extraction step (S300), a three-dimensional interior design result generation step (S400), and a construction detail drawing generation step (S500). Data generated or extracted in each step may be used as input for subsequent steps, and processed in the form of a series of pipelines from the input of spatial information for a new space to the generation of a final construction detail drawing.

[0106] First, the processor (230) can receive spatial information about the new space and generate a spatial model of the new space (S100). The user can input spatial information including at least one of floor plan data, actual measurement data, and a spatial image of the new space through the input section or input interface of the electronic device (200). Based on the input spatial information, the processor (230) calculates spatial structure data including wall locations, opening locations, ceiling height, and floor area of ​​the new space, generates a three-dimensional spatial model based on the calculated spatial structure data, and determines the generated three-dimensional spatial model as the spatial model for the new space. The detailed operation of this step is explained in more detail with reference to FIG. 4.

[0107] Next, the processor (230) can obtain at least one reference design image by receiving input directly from the user or by searching from the external server (100) through the communication unit (210) (S200). When the user inputs directly, the user can upload an image of the interior design style they wish to reference to the electronic device (200). When searching from the external server (100), the processor (230) can search the reference design image database of the external server (100) based on keywords (e.g., "modern," "Nordic," "minimal," "classic," etc.) or categories (e.g., "living room," "bedroom," "kitchen," "bathroom," etc.) received from the user. Through this, the user can obtain reference images for the desired interior design style through various channels.

[0108] Specifically, the reference design image database of the external server (100) may store metadata tagged for each image, such as style tags (e.g., modern, Nordic, minimal, classic, industrial, etc.), space type tags (e.g., living room, bedroom, kitchen, bathroom, study, etc.), color tone tags (e.g., warm tone, cool tone, neutral, etc.), and main finishing material tags (e.g., wood, stone, tile, fabric, etc.). The processor (230) can perform a search by matching these metadata with keywords or categories received from the user.

[0109] According to one embodiment, the processor (230) may obtain a plurality of reference design images from a user. When the user provides a plurality of reference design images, the processor (230) may extract design feature information for each image in step S300 described below, and generate integrated reference design feature information by combining the features of the plurality of images based on the user's selection by region. Detailed operations regarding this will be explained in more detail with reference to FIG. 5.

[0110] Additionally, according to one embodiment, the external server (100) may be linked with an online interior platform (e.g., Pinterest, Houzz, etc.) to collect and store reference design images reflecting the latest interior trends in real time, and the processor (230) may include these latest trend images in search results in priority when a user makes a search request. Furthermore, the processor (230) may recommend and provide reference design images of a similar style that match the user's preferences based on a style analysis of the reference design images previously selected by the user.

[0111] Next, the processor (230) can analyze the acquired reference design image using an AI model to extract reference design feature information (S300). The processor (230) can use an AI model (semantic segmentation model) to divide the reference design image into multiple design areas including wall areas, floor areas, ceiling areas, and furniture areas, extract area-specific feature data and furniture feature data for each divided area, and integrate the extracted feature data to generate reference design feature information. Through this step, the reference design analysis previously performed visually by a designer is replaced by an AI-based automated analysis, thereby improving the accuracy and consistency of design feature extraction. The detailed operation of this step is explained in more detail with reference to FIG. 5.

[0112] Next, the processor (230) can apply the extracted reference design feature information to the generated space model to generate a three-dimensional interior design result for the new space (S400). The processor (230) derives spatial constraints from the space model, including the size, shape, location of structures, and location of openings of the new space, and matches and applies color palette information, pattern information, and finishing material information corresponding to the walls, floors, and ceilings of the space model based on the extracted area-specific feature data. It also determines the type, size, and placement location of furniture and equipment using an AI model and integrates them to generate a three-dimensional interior design result. Through this, the stylistic features of the reference design are automatically applied considering the physical constraints of the new space, thereby generating a customized three-dimensional design result suitable for actual construction. The detailed operation of this step is explained in more detail with reference to FIG. 6.

[0113] Next, the processor (230) can generate a construction detail drawing based on a three-dimensional interior design result (S500). Based on the three-dimensional interior design result, the processor (230) calculates a list of materials and furniture quantities required for construction and an estimated construction cost, generates construction drawing data including at least one of a floor plan, a ceiling plan, an elevation view, and a detailed cross-section, generates priority and process sequence information for each construction process, and generates a construction detail drawing including this information. Through this, processes such as creating construction drawings, calculating quantities, estimating costs, and establishing process plans, which were previously performed manually by a designer, are automated, thereby resolving discrepancies between the design result and the construction drawing, and allowing all related construction information to be automatically updated when the design is changed. The detailed operation of this step is explained in more detail with reference to FIG. 7.

[0114] Accordingly, an electronic device (200) according to one embodiment of the present disclosure performs a series of processing steps, such as acquiring a reference design image in S200 based on a spatial model generated in S100, analyzing the acquired reference design image with an AI model in S300 to extract design feature information, applying the extracted feature information to a spatial model in S400 to generate a three-dimensional interior design result, and generating a construction detail drawing based on the design result generated in S500. In particular, data generated at each step from S100 to S500 is sequentially transmitted to subsequent steps, so that the entire process from the input of spatial information to the output of a construction detail drawing is seamlessly linked, thereby enabling the automation and unification of the reference design-based interior design process.

[0116] FIG. 4 is a detailed flowchart of S100 of FIG. 3, and is a flowchart for explaining a spatial model generation operation according to one embodiment. Each step (S110 to S140) of FIG. 4 can be performed by the processor (230) of FIG. 2.

[0117] Referring to FIG. 4, a processor (230) according to one embodiment can receive spatial information regarding an indoor space to be designed from a user terminal or an external input device through a communication unit (210) (S110). Here, the spatial information may include at least one of floor plan data, actual measurement data, and a spatial image.

[0118] Specifically, floor plan data is data representing the two-dimensional structure of an indoor space to be designed, and may include a CAD (Computer-Aided Design) file format (e.g., DWG, DXF, etc.), an architectural drawing in PDF format, or a floor plan in a digital image format. The floor plan data may include the arrangement of walls in the indoor space, the location of windows and doors, the classification of the use of each room (e.g., living room, bedroom, kitchen, bathroom, etc.), and key dimension information. When the processor (230) receives floor plan data in a CAD file format, it can automatically extract vector-based drawing elements (wall lines, opening symbols, dimension lines, etc.) through a CAD parser. Additionally, when the processor (230) receives floor plan data in a PDF format or a digital image format, it can utilize an AI model (e.g., image segmentation model, OCR model, etc.) to automatically recognize wall lines, opening locations, and dimension text from the floor plan image and convert them into digital drawing data.

[0119] Actual measurement data refers to quantitative dimensional information obtained by actually measuring the indoor space being designed. Actual measurement data may include the length of each wall surface, distances between walls, ceiling height, width and height of windows and doors, location and dimensions of structures such as columns or beams, and floor level differences (e.g., steps). Actual measurement data can be entered by the user through manual measurement, or it can be automatically acquired through a mobile device equipped with a laser distance meter, an ultrasonic distance meter, or a LiDAR (Light Detection And Ranging) sensor.

[0120] According to one embodiment, the processor (230) can receive actual measurement data from a user terminal (e.g., smartphone, tablet, etc.) equipped with a LiDAR sensor. In this case, when the user scans an indoor space using a mobile device equipped with a LiDAR sensor, the mobile device can generate three-dimensional point cloud data based on the time of reflection by irradiating an infrared laser onto the surface of the indoor space. The processor (230) can estimate the plane equation of each wall surface from the point cloud data received through the communication unit (210) and calculate the intersection line between adjacent walls to automatically calculate the position and length of the wall. Additionally, the processor (230) can detect areas where the point density changes rapidly or areas where points are absent in the point cloud data, and automatically identify the corresponding areas as openings such as windows or doors. Through this, the user can obtain precise actual measurement data without separate manual measurement.

[0121] A space-capturing image refers to photographic or video data of an indoor space to be designed. The space-capturing image may include a 2D photograph taken with a standard camera, a panoramic image taken with a 360-degree camera (e.g., Ricoh Theta, Insta360, etc.), or a walk-through video of the indoor space in the form of a video. The processor (230) can analyze the space-capturing image using an AI model to determine the current state of the indoor space, the types and locations of existing furniture and finishing materials, the direction of natural light inflow and illumination distribution, and the overall atmosphere of the space.

[0122] Meanwhile, according to one embodiment, the processor (230) can restore the three-dimensional shape of an indoor space by applying a Structure from Motion (SfM) or Multi-View Stereo (MVS) algorithm to a plurality of spatial images. Specifically, the processor (230) can extract feature points from a plurality of spatial images taken at different locations and angles, estimate the relative position and pose of the camera through feature point matching between images, and then perform triangulation to generate a three-dimensional point cloud. Through this, the three-dimensional structure of an indoor space can be obtained using only a standard camera without a LiDAR sensor or a smartphone, thereby improving accessibility for creating a spatial model.

[0123] Additionally, the processor (230) may provide an input interface to the user terminal upon receiving spatial information. The input interface may include a file upload area for uploading a floor plan file, a numerical input field for inputting actual measurement data, a shooting interface for uploading or transmitting spatial images or shooting in real time, and a scan button for starting a LiDAR scan. Through this, the user can flexibly input one or more spatial information depending on the type of data they possess.

[0124] Next, the processor (230) can calculate spatial structure data of the indoor space to be designed based on the received spatial information (S120). The spatial structure data may include at least one of wall locations, opening (window and door) locations, ceiling height, and floor area.

[0125] Specifically, the processor (230) can calculate wall location data by extracting information such as the starting and ending point coordinates of the wall, the thickness of the wall, and the material of the wall (e.g., concrete, lightweight wall, glass wall, etc.) from floor plan data or actual measurement data. The wall location data defines the position and direction of each wall on a two-dimensional coordinate plane and is subsequently used as a basic framework for creating a three-dimensional space model.

[0126] The opening location data may include the center coordinates, width, height, sill height, and opening / closing direction of windows and doors located in each wall. The processor (230) may automatically calculate the opening location by recognizing the symbols of windows and doors included in the floor plan data, or detect the location and size of windows and doors by analyzing the spatial image using an AI model (e.g., an object detection model). The opening location data can subsequently be used for analyzing the direction of natural light inflow, securing movement paths when arranging furniture, and generating construction drawing data.

[0127] Ceiling height data represents the vertical distance from the floor surface to the ceiling surface and may be uniform throughout the space or vary in some areas (e.g., multi-story structure, sloped ceiling, down ceiling, etc.). The processor (230) can generate ceiling height data by obtaining the ceiling height from actual measurement data or by calculating the height difference between the floor surface and the ceiling surface from the point cloud of LiDAR scan data. If the ceiling height is not uniform within the space, the processor (230) can individually calculate and store the ceiling height for each area.

[0128] Floor area data represents the floor surface area of ​​the indoor space to be designed and may include the total area and the area for each room. The processor (230) can generate floor area data by calculating the area of ​​a closed polygon defined by wall location data. Additionally, the processor (230) can separately calculate the net floor area by excluding the area of ​​structures occupying the floor, such as columns and equipment boxes, from the floor area data. The floor area data can subsequently be used to calculate furniture placement density, calculate the required amount of finishing materials, and estimate construction costs.

[0129] Meanwhile, the processor (230) can verify consistency between multiple spatial information sources (plan data, actual measurement data, spatial image, LiDAR point cloud, etc.) when calculating spatial structure data. Specifically, the processor (230) compares the wall location and dimensions calculated from the plan data with the wall location and dimensions calculated from the actual measurement data or LiDAR scan data, and if a discrepancy exceeding a preset tolerance (e.g., 5cm, 10cm, etc.) occurs, it can notify the user of the discrepancy area and request additional verification. Through this, the accuracy of the spatial structure data is guaranteed, and the reliability of the 3D design results and construction detail drawings generated in subsequent stages can be improved.

[0130] Additionally, according to one embodiment, the processor (230) can automatically identify structural constraints of the indoor space when calculating spatial structure data. Structural constraints may include at least one of the locations of load-bearing walls, buried locations of water supply and drainage pipes and gas pipes, locations of electrical outlets and switches, distribution of floor heating pipes, locations of air conditioner outdoor unit connection pipes, and locations of ventilation openings. The processor (230) can generate structural constraint data by analyzing the equipment drawing layer included in the floor plan data or by automatically detecting the locations of outlets, switches, ventilation openings, etc., in the space capture image through an AI model. This structural constraint data can be referenced as part of the spatial constraint conditions during the subsequent process of generating a 3D design result (S400) and generating a construction detail drawing (S500), thereby ensuring that a design result with guaranteed constructability is generated.

[0131] Next, the processor (230) can generate a three-dimensional spatial model of the indoor space to be designed based on the generated spatial structure data (S130). The three-dimensional spatial model is a digital model that represents the physical structure of the indoor space to be designed in three dimensions, and is a model in which architectural elements such as walls, ceilings, floors, windows, and doors are placed in a three-dimensional space at a scale corresponding to actual dimensions.

[0132] Specifically, the processor (230) can create a three-dimensional wall object by extruding a two-dimensional wall line included in the wall location data in a vertical direction up to the ceiling height. Additionally, the processor (230) can create window and door openings at corresponding locations of the wall object based on the opening location data and insert predefined window and door objects into the corresponding openings. The processor (230) can form the lower and upper parts of the three-dimensional space model by creating the floor surface and the ceiling surface as surface objects, respectively.

[0133] According to one embodiment, the processor (230) can generate a three-dimensional spatial model from LiDAR point cloud data. In this case, the processor (230) can convert the point cloud into a mesh-shaped three-dimensional surface model by performing noise filtering, downsampling, and normal vector estimation on the point cloud, and then applying a Poisson surface reconstruction or ball pivoting algorithm. The processor (230) can generate a structured three-dimensional spatial model by detecting planar regions (walls, floors, ceilings) in the mesh model using a Random Sample Consensus (RANSAC) algorithm or the like, and classifying the detected planar regions into wall, floor, and ceiling objects. Through this, a three-dimensional spatial model close to the actual shape can be precisely generated even for indoor spaces with complex shapes (e.g., curved walls, arched ceilings, irregular structures, etc.).

[0134] According to another embodiment, the processor (230) can generate a 3D spatial model by applying a monocular depth estimation AI model or a NeRF (Neural Radiance Fields)-based 3D reconstruction model to a spatially captured image. The monocular depth estimation model generates a 2.5-dimensional (2.5D) depth map by estimating the depth value of each pixel from a single 2D image and can reconstruct a 3D spatial structure by registering multiple depth maps. The NeRF-based model learns a 3D volumetric representation from images captured at multiple viewpoints and can generate a high-quality 3D model capable of rendering at any viewpoint. This AI-based 3D reconstruction method has a wide range of applications in that it can generate a 3D spatial model using only images captured with a standard camera without the need for separate 3D scanning equipment.

[0135] Additionally, the processor (230) can extract existing finishing material information from a space-capturing image when creating a three-dimensional space model and reflect it in the three-dimensional space model. Specifically, the processor (230) can analyze the space-capturing image using a semantic segmentation AI model to identify the current finishing material type (e.g., wallpaper, paint, tile, flooring, carpet, etc.) and color of each wall surface, floor surface, and ceiling surface. The processor (230) can map the identified finishing material information as a texture to the corresponding surface of the three-dimensional space model. Through this, the three-dimensional space model can visually and accurately reflect the current state of the indoor space to be designed, and can subsequently be used to evaluate the contrast or harmony between the existing finishing material and the new finishing material during the process of generating a design result based on a reference design image.

[0136] Furthermore, according to one embodiment, the processor (230) can convert and store the generated three-dimensional spatial model into a BIM (Building Information Modeling) compatible format (e.g., IFC (Industry Foundation Classes), Revit, etc.). The three-dimensional spatial model in BIM format may include not only simple geometric shape information but also metadata such as the material, grade, manufacturer, and construction method of each building element, thereby maximizing usability in the subsequent process of generating construction drawing data and construction detail drawings.

[0137] Next, the processor (230) can determine the generated three-dimensional space model as the space model of the indoor space to be designed (S140). The processor (230) stores the determined space model in memory (220) and can subsequently refer to it continuously during the process of generating a design result based on a reference design image (S400) and generating a detailed drawing for construction (S500).

[0138] According to one embodiment, before determining the spatial model, the processor (230) may output the generated three-dimensional spatial model in the form of a three-dimensional viewer on the display of the user terminal and request the user to verify it. The user can rotate, enlarge, and reduce the spatial model through the three-dimensional viewer to view it from various perspectives and examine whether the wall positions, opening sizes, ceiling heights, etc., match the actual space. If the user determines that modifications are necessary, the processor (230) may update the three-dimensional spatial model by receiving the user's modification input (e.g., moving wall positions, changing opening sizes, adjusting ceiling heights, etc.). Once the user's verification and approval are complete, the processor (230) may determine the three-dimensional spatial model as the final spatial model and store it in memory (220).

[0139] Meanwhile, according to another embodiment, the processor (230) may automatically determine the spatial model without separate confirmation from the user. In this case, the processor (230) may automatically verify the completeness of the generated three-dimensional spatial model. Specifically, the processor (230) may verify integrity conditions such as whether the connection between walls in the three-dimensional spatial model forms a complete closed structure, whether openings are properly placed within the wall range, whether the ceiling height meets legal minimum standards (e.g., minimum ceiling height of a living room under the Building Act of 2.1m, etc.), and whether the floor area is a positive value. If all integrity conditions are satisfied, the processor (230) may automatically determine the corresponding three-dimensional spatial model as the spatial model, and if even one of the integrity conditions is not satisfied, it may notify the user of the discrepancy and request supplementary input of spatial information.

[0140] Additionally, the processor (230) may add spatial metadata to the determined spatial model and store it in memory (220). The spatial metadata may include the creation date and time of the spatial model, the type of spatial information used (floor plan data, actual measurement data, spatial image, LiDAR scan data, etc.), the coordinate system information of the spatial model, the total floor area, the classification of the use of each room, and the version information of the spatial model. This spatial metadata can be utilized to efficiently reference the attribute information of the spatial model during the subsequent process of generating design results and detailed drawings for construction.

[0141] Through this, the processor (230) can accurately model the physical structure of the indoor space to be designed in three dimensions, thereby securing a spatial model that serves as the basis for generating an interior design result based on a reference design image. In particular, since the spatial model can be generated by flexibly utilizing at least one of floor plan data, actual measurement data, spatial image capture and LiDAR scan data, it is possible to generate a three-dimensional spatial model regardless of the type of data the user possesses.

[0143] FIG. 5 is a detailed flowchart of S300 of FIG. 3, and is a flowchart for explaining the operation of extracting reference design feature information according to one embodiment. Each step (S310 to S340) of FIG. 5 can be performed by the processor (230) of FIG. 2.

[0144] Referring to FIG. 5, a processor (230) according to one embodiment can divide at least one reference design image received in step S200 into a plurality of design regions using a semantic segmentation AI model (S310). Here, the plurality of design regions may include a wall region, a floor region, a ceiling region, and a furniture region.

[0145] Specifically, the semantic segmentation AI model is a deep learning-based image segmentation model that can classify each pixel of a reference design image into one of predefined design area categories (wall, floor, ceiling, furniture, window, door, lighting fixture, decoration, etc.). The semantic segmentation AI model can be implemented as a deep neural network with an encoder-decoder structure (e.g., DeepLabV3+, U-Net, SegFormer, etc.) and can achieve high segmentation accuracy after being pre-trained on an interior design image dataset and then fine-tuned on indoor space images.

[0146] The processor (230) can obtain a pixel-unit segmentation mask corresponding to each design area as the output of a semantic segmentation AI model. The segmentation mask is a two-dimensional map that assigns a label to the design area to which the pixel belongs for each pixel of a reference design image, and is subsequently used as a basis for extracting feature data by area.

[0147] Meanwhile, the processor (230) can individually separate multiple wall surfaces included in the reference design image when dividing the wall surface area. For example, if three wall surfaces of different directions are visible in the reference design image, the processor (230) can individually divide each wall surface into a first wall surface area, a second wall surface area, and a third wall surface area. Through this, even if different finishing materials (e.g., paint on the first wall surface, wallpaper on the second wall surface, tile on the third wall surface, etc.) are applied to each wall surface, independent feature data can be extracted for each wall surface.

[0148] Additionally, the processor (230) may perform instance segmentation in addition to semantic segmentation when dividing the furniture area. Instance segmentation means separating multiple objects belonging to the same category (e.g., multiple chairs, multiple cushions, etc.) individually and assigning a unique instance identifier (instance ID) to each furniture object. The processor (230) may apply an instance segmentation AI model (e.g., Mask R-CNN, YOLACT, etc.) to individually obtain the bounding box and pixel-unit mask of each furniture object included in the reference design image. Through this, the type, style, color, size, etc. of each furniture can be individually analyzed during the subsequent furniture feature data extraction (S330) process.

[0149] According to one embodiment, the processor (230) can refine the boundaries of the segmentation mask. Specifically, the processor (230) can apply a post-processing technique, such as a Conditional Random Field (CRF), to the initial output of the segmentation AI model to precisely adjust the boundaries between design areas. Through this, segmentation errors that may occur at the boundaries between walls and floors, furniture and walls, etc., are minimized, and the accuracy of the feature data extracted thereafter can be improved.

[0150] Furthermore, according to one embodiment, the processor (230) may additionally divide areas of additional elements such as lighting fixtures, decorative items (e.g., picture frames, flower pots, clocks, etc.) and curtains / blinds when dividing design areas in a reference design image. These additional element areas may subsequently be referenced as elements that determine the overall atmosphere and style of the space during the process of generating design results (S400).

[0151] Next, the processor (230) can extract region-specific feature data for each of the divided multiple design regions, including at least one of the type of finishing material, color information, pattern information, and texture information of the design region (S320).

[0152] Specifically, the processor (230) can apply a segmentation mask corresponding to each design area to extract only the pixel data of the corresponding area from the reference design image, and analyze the extracted pixel data with an AI model to produce feature data for each area.

[0153] The type of finishing material is information indicating the type of finishing material applied to the corresponding design area, and in the case of the wall area, it may include paint, wallpaper, tile, marble, solid wood panel, exposed concrete, brick, etc., and in the case of the floor area, it may include laminate flooring, solid wood flooring, tile, marble, polished tile, carpet, epoxy, etc., and in the case of the ceiling area, it may include paint, gypsum board, wood panel, exposed ceiling (exposed concrete or duct), etc. The processor (230) can identify the type of finishing material of each design area using a finishing material classification AI model. The finishing material classification AI model is an image classification model trained with various finishing material image datasets, and can classify the type of finishing material from the pixel data of the corresponding design area and output a classification confidence score.

[0154] Color information is information representing the main color characteristics of the design area and may include a dominant color, a secondary color, a color palette, and a color distribution. The processor (230) may extract the dominant color by applying a K-means clustering or median cut algorithm to the pixel data of the design area. The extracted color may be expressed as RGB (Red, Green, Blue) values ​​and HSV (Hue, Saturation, Value) values, and is subsequently used for color harmony analysis when generating design results. Additionally, the processor (230) may classify the colors of the design area in terms of lightness (bright / dark), saturation (vivid / dull), and color temperature (warm / cool) using a color analysis AI model.

[0155] Pattern information is information indicating the presence or absence and type of a visual pattern applied to the corresponding design area. The types of patterns may include solid, stripe, check / plaid, geometric, floral, marble, wood grain, brick, herringbone, etc. The processor (230) can identify the presence or absence and type of a pattern from the pixel data of the corresponding design area using a pattern recognition AI model. The pattern recognition AI model is a classification model based on texture analysis, and can output the type of pattern, the direction of the pattern (vertical, horizontal, diagonal, etc.), the scale (size) of the pattern, and the density of the pattern by analyzing repetitive visual elements of the corresponding area.

[0156] Texture information is information representing the surface texture characteristics of the design area and may include glossiness, roughness, transparency, and reflectivity. The processor (230) can extract texture information by analyzing the distribution of highlights, micro-texture characteristics, and light reflection patterns from the pixel data of the design area. Texture information can be used to determine the physical properties (PBR, Physically-Based Rendering Parameters) of the material applied to the surface of the 3D space model when generating the design result.

[0157] Meanwhile, according to one embodiment, the processor (230) may additionally assign style category information to the feature data of each design area. The style category is classification information representing the overall design style of the design area and may include modern, classic, minimal, industrial, Scandinavian / Nordic, Japanese / Wa-style, vintage, rustic, bohemian, mid-century modern, etc. The processor (230) may determine the style category of the design area based on a combination of finishing material type, color information, pattern information, and texture information using a style classification AI model. This style category information may subsequently be used as a criterion for evaluating style consistency between areas during the process of selectively combining multiple reference design images.

[0158] Next, the processor (230) can extract furniture feature data for the furniture area, including at least one of the type, style, color, size information and placement location of the furniture (S330).

[0159] Specifically, the processor (230) can apply a furniture recognition AI model to each furniture instance in the divided furniture area in step S310 to individually extract furniture feature data.

[0160] The type of furniture is information representing the functional classification of the furniture object and may include sofas, tables (dining tables, coffee tables, side tables, etc.), chairs (dining chairs, lounge chairs, office chairs, etc.), beds, storage units (wardrobes, chests of drawers, bookshelves, sideboards, etc.), lighting fixtures (floor lamps, pendant lamps, table lamps, etc.), and rugs, etc. The processor (230) can classify the type of each furniture instance using a furniture recognition AI model. The furniture recognition AI model is an object classification model trained on various furniture image datasets and can classify the type of furniture with high accuracy based on the shape, proportions, and visual features of the furniture.

[0161] The style of the furniture is information representing the design style of the furniture object and may include modern, classic, minimal, Nordic, industrial, vintage, mid-century modern, etc. The processor (230) can classify the style of the furniture based on the shape features of the furniture (e.g., straight shape vs. curved shape, angled legs vs. tapered legs, etc.), material characteristics (e.g., metal frame vs. solid wood frame) and the presence or absence of decorative elements using a style classification AI model.

[0162] The color of the furniture is information representing the dominant color and accent color of the furniture object. The processor (230) can calculate the dominant color and accent color by applying a color extraction algorithm to pixel data within a segmentation mask area corresponding to the furniture instance.

[0163] The size information of the furniture is information indicating the approximate dimensions (width, depth, height) of the furniture object. Since the reference design image is a two-dimensional image, it is difficult to directly measure the exact dimensions, but the processor (230) can estimate the approximate size of the furniture based on the relative size ratio between the furniture object and other design areas (walls, floors, doors, etc.) within the reference design image. For example, the height of the furniture can be calculated proportionally based on the height of a standard door (approx. 2,100 mm), or the three-dimensional size of the furniture can be estimated by applying a monocular depth estimation AI model.

[0164] The placement location of the furniture is information indicating the position where the furniture object is placed within the indoor space. The processor (230) can determine the placement location by estimating the center point of the bottom of the bounding box of the furniture instance in the reference design image as a point of contact with the floor surface and calculating the relative coordinates of the point of contact within the image. Additionally, the processor (230) can extract a placement type (e.g., placement close to the wall, placement in the center of the room, placement next to the window, etc.) by analyzing the relative positional relationship between the furniture and architectural elements such as walls, windows, and doors (e.g., placement adjacent to a wall, placement in the center of the room, placement next to a window, etc.). This placement location information can be referenced to place the furniture within a three-dimensional space model when generating design results later.

[0165] Meanwhile, the processor (230) may additionally include material information of the furniture in the furniture feature data. The materials of the furniture may include leather, fabric, solid wood, plywood, MDF, metal, glass, plastic, rattan, etc. The processor (230) can identify the material by analyzing the surface texture characteristics of the furniture instance, and this material information can be used to calculate the material requirements and estimate costs when generating detailed drawings for construction later.

[0166] Additionally, according to one embodiment, the processor (230) can search for furniture product information corresponding to the type and style of the recognized furniture in a database based on the classification result of a furniture recognition AI model. The furniture product information may include the product name, brand, price range, and availability of purchase of ready-made products similar to the furniture. This furniture product information can be used to calculate the furniture purchase list and costs when generating detailed drawings for construction later.

[0167] Next, the processor (230) can determine reference design feature information including region-specific feature data extracted in step S320 and furniture feature data extracted in step S330 and store it in memory (220) (S340).

[0168] Specifically, the reference design feature information integrates area-specific feature data (type of finishing material, color information, pattern information, texture information, style category, etc.) for each of multiple design areas and furniture feature data (type of furniture, style, color, size information, placement location, material, etc.) for each furniture instance into a structured data format. The processor (230) can generate the reference design feature information in a structured data format such as JSON or XML and store it in memory (220). Through this, the reference design feature information can be efficiently referenced during the subsequent design result generation (S400) process.

[0169] Meanwhile, the processor (230) may additionally include information on the overall mood of the reference design image in the reference design feature information. The mood information may include multidimensional classifications such as warm / cool, bright / dark, cozy / open, luxurious / restrained. The processor (230) may calculate the overall mood of the reference design image by comprehensively analyzing the color tone, texture characteristics, lighting distribution, etc. of the feature data by region, and include this in the reference design feature information. This mood information can subsequently be used to selectively combine multiple reference design images or to adjust the overall mood of the design result.

[0170] Meanwhile, according to one embodiment, the processor (230) can generate a design intent description for each design area in the form of natural language by using a Vision-Language Multimodal AI model in addition to numerical feature data when generating reference design feature information. The design intent description is text data that describes the visual features of the design area from a semantic perspective, and may include sentences that comprehensively describe color, texture, style, and spatial atmosphere, such as, for example, regarding a wall area, "a luxury modern style with a warm beige tone slate texture stone finish that reveals fine golden grain under natural light." Specifically, the Vision-Language Multimodal AI model can be implemented as a structure combining an image encoder and a language decoder (e.g., BLIP-2, LLaVA, GPT-4 Vision, etc.), and can receive pixel data with a segmentation mask applied to each design area as input and output a natural language sentence describing the visual features of the area. The processor (230) inputs a predefined prompt into a vision-language multimodal AI model to induce the design intent description to include the visual impression of the finishing material, the emotional characteristics of the color, the spatial effect of the pattern, and the direction of the overall design concept. The processor (230) converts the generated design intent description into a vector representation using a text embedding model and adds the converted vector representation to the feature data of each design area to include it in the reference design feature information. This vector representation of the design intent description can subsequently be used as a criterion for evaluating semantic style consistency between design areas during the process of selectively combining multiple reference design images.For example, the processor (230) can calculate the cosine similarity between the design intent description vector for the wall area of ​​the first reference design image and the design intent description vector for the floor area of ​​the second reference design image to determine whether the two areas can form a semantically harmonious space when combined. Furthermore, the processor (230) can also utilize the design intent description when selecting alternative materials and alternative furniture during the subsequent process of generating detailed drawings for construction (S500). Specifically, the processor (230) inputs product images of alternative candidate materials or candidate furniture into a vision-language multimodal AI model to generate the design intent description of the candidate, and can calculate the semantic similarity between the vector representation of the generated design intent description and the vector representation of the original design intent description included in the reference design feature information. Through this, it is possible to quantitatively evaluate whether the consistency of the design intent—that is, the overall atmosphere and emotional direction that the space intends to convey—is maintained, which is difficult to capture solely by individual numerical similarities such as color, pattern, and material.

[0171] Additionally, the processor (230) can calculate and assign a style similarity score to the reference design feature information. The style similarity score is an indicator that quantifies the degree of style consistency between each design area of ​​the reference design image. The processor (230) can calculate the style similarity score between areas by comparing the style category, color tone, and finishing grade of each design area, and a higher style similarity score indicates that the design elements within the reference design image are harmoniously composed. This style similarity score can be used as a criterion for judging the suitability of a combination during the process of selectively combining multiple reference design images thereafter.

[0172] Furthermore, according to one embodiment, when a plurality of reference design images are received, the processor (230) may individually perform the steps S310 to S340 described above for each reference design image to extract reference design feature information for each reference design image. In this case, the processor (230) may generate integrated reference design feature information by selectively combining feature data of different reference design images for each design area from the feature information of the plurality of reference design images.

[0173] Specifically, the processor (230) can output the area-specific feature data and furniture feature data extracted from each of the multiple reference design images to the display of the user terminal by arranging them so that they can be compared for each of the wall area, floor area, ceiling area, and furniture area. For example, the processor (230) can arrange the wall finish / color / pattern of the first reference design image and the wall finish / color / pattern of the second reference design image in parallel for the wall area so that the user can visually compare them. In addition, the processor (230) can receive input from the user to select the reference design image to be applied for each design area. For example, if the user wishes to combine the wall area feature data of the first reference design image, the floor area feature data of the second reference design image, and the furniture feature data of the third reference design image, the processor (230) can generate a single integrated reference design feature information by extracting the feature data of the design area selected by the user from each reference design image. Through this, users can freely combine their favorite design elements from different reference design images to create a customized interior design concept that is not limited to a single reference design image.

[0174] Meanwhile, the processor (230) can generate integrated reference design feature information and then verify the overall harmony of the integrated reference design feature information using an AI model. Harmony verification means evaluating the visual consistency and aesthetic balance between design areas combined from different reference design images. The processor (230) can calculate a harmony score by using a harmony evaluation AI model to comprehensively evaluate the consistency of color tones between the combined design areas, the interoperability of style categories, the balance of finishing material grades, and the unity of the overall spatial atmosphere.

[0175] Meanwhile, according to one embodiment, when evaluating the harmony of integrated reference design feature information, the processor (230) may additionally consider not only visual consistency between design areas but also physical environmental conditions of the new space derived from the space model generated in step S100. Physical environmental conditions may include the direction and amount of natural light incident based on the window location and size of the new space, the aspect ratio of the space, the ceiling height to floor area ratio, and the openness of the space (e.g., open plan vs. partitioned plan). Specifically, the processor (230) can simulate the main direction of natural light incident and the illuminance distribution by time period based on the window location and size data of the space model, and predict what color shift the finishing material of each design area exhibits under natural light based on the simulated illuminance distribution. For example, natural light with a cool color temperature entering from a north-facing window may visually neutralize warm-toned wall finishing materials, whereas natural light with a warm color temperature from a south-facing window may further emphasize warm-toned wall finishing materials. The processor (230) can incorporate the predicted color variation results due to natural light into the harmony evaluation so that even with the same combination of finishing materials, the harmony score can be calculated differently depending on the natural light incidence conditions. Additionally, the processor (230) can evaluate the scale suitability of the pattern based on the size and ratio of the new space. For example, if a large pattern (e.g., large pattern tiles of 600 mm or more) is applied to a small space (e.g., floor area of ​​less than 10 m²), the processor (230) can determine that the pattern is excessively emphasized relative to the space, which may cause visual imbalance, and thus lower the harmony score.Conversely, if a small pattern (e.g., mosaic tiles of 100mm or less) is applied to a large space (e.g., floor area of ​​30m² or more), the harmony score may be lowered on the grounds that the pattern may be visually distracting. Furthermore, the processor (230) may evaluate the suitability of the lightness and color tone distribution of the finishing material based on the ratio of ceiling height to floor area. For example, if a dark-colored finishing material is applied to the ceiling and upper wall surfaces in a space with a relatively low ceiling height (e.g., 2.3m or less), the processor (230) may lower the harmony score on the grounds that the visual pressure of the space may increase and suggest automatically correcting the color of the ceiling area to a bright tone. Through this, even with the same combination of feature data for each area, the harmony evaluation and correction optimized for the physical environmental conditions of the space to be applied can be performed.

[0176] According to one embodiment, a harmony evaluation AI model can calculate color harmony based on color harmony theory, a score for interoperability between design styles (e.g., high compatibility between modern and minimal, low compatibility between modern and classic, etc.), and consistency between finishing material grades (e.g., grade mismatch between high-end marble flooring and economy plywood furniture, etc.), and calculate an overall harmony score by weighted summing these values.

[0177] If the calculated harmony score is greater than or equal to a preset threshold value, the processor (230) may determine that the integrated reference design feature information is appropriate and store it in memory (220). On the other hand, if the harmony score is less than a preset threshold value, the processor (230) may use an AI model to automatically correct the integrated reference design feature information to maintain consistency in color tone, style category, or finish grade between design areas.

[0178] Specifically, the processor (230) can analyze the cause of low color harmony to identify the design area with the greatest discrepancy and adjust the feature data of that area. For example, if the color harmony is low because the color tone of the wall area is warm tone and the color tone of the floor area is cool tone, the processor (230) can automatically correct the color tone of the floor area toward a warm tone. Additionally, if the furniture style is modern and the wall finish is classic-style wallpaper, the processor (230) can automatically correct the wall finish to paint or exposed concrete that matches the modern style. Furthermore, if high-end marble tiles are applied to the floor but the furniture is made of economical plywood, the processor (230) can automatically correct the material grade of the furniture to a level that matches the grade of the finish material, such as solid wood or metal.

[0179] The processor (230) may output the integrated reference design feature information, for which automatic correction has been performed, to the display of the user terminal and request confirmation from the user. If the user approves the automatic correction result, the processor (230) may determine the corrected integrated reference design feature information as the final reference design feature information and store it in memory (220). If the user rejects the automatic correction result, the processor (230) may maintain the original integrated reference design feature information or provide an interface to allow the user to manually change the combination.

[0180] Through this, the processor (230) can extract systematic feature information including finishing materials, colors, patterns, textures, and types, styles, colors, sizes, and placement locations of furniture for each design area from a reference design image, and use this as a basis for generating a three-dimensional interior design result. In particular, by selectively combining feature data by area from multiple reference design images and verifying and automatically correcting the degree of harmony using an AI model, a design concept that reflects the diverse tastes of users while maintaining visual unity can be derived.

[0182] FIG. 6 is a detailed flowchart of S400 of FIG. 3, and is a flowchart for explaining the operation of generating a three-dimensional interior design result according to one embodiment. Each step (S410 to S440) of FIG. 6 can be performed by the processor (230) of FIG. 2.

[0183] Referring to FIG. 6, a processor (230) according to one embodiment may apply a finishing material corresponding to the area-specific feature data of the reference design feature information extracted in step S300 to each design area of ​​the spatial model determined in step S100 (S410). Here, the application of the finishing material means mapping a 3D finishing material texture corresponding to the type, color, pattern, and texture of the finishing material extracted from the reference design image onto surfaces such as walls, floors, and ceilings of the 3D spatial model.

[0184] Specifically, the processor (230) can search for a corresponding 3D finish material texture from a finish material texture database stored in memory (220) or an external server (100) based on the finish material type information of each design area included in the reference design feature information. The finish material texture database may store PBR (Physically-Based Rendering) texture data, such as high-resolution texture images, normal maps, roughness maps, and displacement maps, for each type of finish material. The processor (230) can select the texture most similar to the color information, pattern information, and texture information included in the reference design feature information from among the searched finish material textures.

[0185] According to one embodiment, the processor (230) may calculate color similarity, pattern similarity, and texture similarity, respectively, when selecting a finishing material texture, and select the finishing material texture with the highest overall similarity by weighted summing these. Color similarity may be calculated based on the CIEDE2000 color difference between the color palette included in the reference design feature information and the color palette of the finishing material texture. Pattern similarity may be calculated based on the degree of agreement of the type, direction, and scale of the pattern. Texture similarity may be calculated based on the difference in glossiness and roughness values.

[0186] The processor (230) can apply the selected finish material texture to the surface of the corresponding design area of ​​the 3D space model using UV mapping. At this time, the processor (230) can automatically adjust the tiling ratio of the texture according to the size of the wall surface so that the finish material pattern corresponds to the scale of the actual finish material. For example, in the case of a 600mm × 600mm tile finish material, the processor (230) can calculate the number of repetitions of the tile based on the actual dimensions of the wall surface and apply a texture tiling ratio corresponding thereto.

[0187] Meanwhile, the processor (230) can individually apply a finishing material texture corresponding to each wall surface when different finishing materials are applied to multiple wall surfaces (e.g., paint on the first wall surface, accent wallpaper on the second wall surface, tile on the third wall surface, etc.). Through this, the finishing material characteristics for each wall surface extracted from the reference design image can be precisely reflected in the three-dimensional space model.

[0188] Additionally, according to one embodiment, the processor (230) can fine-tune the color of the finishing material texture based on color information extracted from a reference design image when the finishing material is applied. Specifically, if the color of the texture retrieved from the finishing material texture database does not exactly match the color of the reference design image, the processor (230) can apply a color correction algorithm to adjust the HSV value of the texture to approximate the color of the reference design image. Through this, even if there is no texture that exactly matches the finishing material texture database, the color tone of the reference design image can be reproduced as faithfully as possible.

[0189] Next, the processor (230) can place a three-dimensional furniture object corresponding to furniture feature data included in the reference design feature information within the spatial model (S420). The three-dimensional furniture object is three-dimensional model data corresponding to the type, style, color, and size of the furniture.

[0190] Specifically, the processor (230) can search for a corresponding 3D furniture object in a 3D furniture model database stored in memory (220) or an external server (100) based on the type and style information of the furniture included in the furniture feature data. The 3D furniture model database may store 3D model data for various types of furniture, such as sofas, tables, chairs, beds, cabinets, and lighting fixtures, according to the style of each piece of furniture (modern, classic, minimal, Nordic, etc.). The 3D furniture model data may include geometric shape data in the form of a mesh, texture data, material data (PBR parameters), and physical dimension information of the furniture (width, depth, height).

[0191] The processor (230) can adjust the color of the retrieved 3D furniture object to correspond to the color information of the furniture included in the furniture feature data. For example, if the main color of the sofa is extracted as gray in the furniture feature data and the base color of the sofa object retrieved from the 3D furniture model database is beige, the processor (230) can convert and apply the texture color of the sofa object to a gray tone.

[0192] Additionally, the processor (230) can scale the size of the three-dimensional furniture object to correspond to the size information included in the furniture feature data. The processor (230) can calculate a scale factor by comparing the approximate dimensions of the furniture estimated from the furniture feature data with the basic dimensions of the furniture object stored in the three-dimensional furniture model database, and can adjust the size of the furniture object by applying the scale factor.

[0193] The processor (230) can place three-dimensional furniture objects in appropriate locations within a spatial model based on placement location information included in furniture feature data. The processor (230) can determine the three-dimensional coordinates and rotation angles of each furniture object within a three-dimensional spatial model by referring to the relative placement relationships of the furniture extracted from a reference design image (e.g., the sofa is adjacent to the wall, the coffee table is in front of the sofa, the dining table is in the center of the room, etc.).

[0194] According to one embodiment, the processor (230) can perform collision detection to prevent collision between furniture when furniture is placed. The processor (230) can detect whether there is physical overlap between furniture or between furniture and a wall using the bounding box of each furniture object, and if overlap occurs, it can automatically adjust the position of the furniture to resolve the interference.

[0195] Meanwhile, the processor (230) can consider securing a circulation path when arranging furniture. The processor (230) calculates a movement path from an entrance (door) to each functional area (living room, kitchen, bedroom, etc.) within the space model and can adjust the position of the furniture so that the effective width of the movement path does not narrow to less than a preset minimum width (e.g., 600mm, 900mm, etc.) due to the furniture arrangement. Through this, the design result can be made to have a space configuration suitable not only in terms of aesthetics but also in terms of practicality.

[0196] Subsequently, the processor (230) verifies the suitability of the finishing material applied in step S410 and the furniture placed in step S420 based on spatial constraints, and can adjust the application of the finishing material and the placement of the furniture if necessary (S430). Spatial constraints may include structural constraints included in the spatial structure data calculated in step S100, the use of the indoor space to be designed, relevant regulations, and physical limit conditions.

[0197] Specifically, the processor (230) can verify whether furniture is placed in the area required for opening and closing windows and doors based on location data of openings (windows, doors) included in the space model. The processor (230) can detect interference between the door's swing radius and the bounding box of the furniture object, and if interference occurs, automatically move the position of the furniture outside the door's swing radius. Additionally, the processor (230) can secure an appropriate clearance around the window so that there is no hindrance to the installation and operation of curtains or blinds.

[0198] According to one embodiment, the processor (230) can refer to the location information of electrical outlets, switches, and vents included in the spatial structure data to detect cases where furniture is placed at a location that makes access impossible and can adjust the location of the furniture. For example, if a storage cabinet is placed in front of a wall outlet and access to the outlet is blocked, the processor (230) can move the location of the storage cabinet by a preset minimum distance (e.g., 100 mm) or more from the outlet.

[0199] Additionally, the processor (230) can verify the suitability of the application of the finishing material. The processor (230) can determine whether the type of finishing material is suitable according to the use of the interior space to be designed (e.g., bathroom, kitchen, living room, etc.). For example, if a solid wood floor without waterproofing is applied to the bathroom floor, the processor (230) can detect the unsuitability of the finishing material and suggest replacing it with waterproof tiles or polished tiles. Additionally, if wallpaper is applied to the area around the gas stove on the kitchen wall, the processor (230) can suggest replacing it with tiles that have heat resistance and stain resistance.

[0200] Additionally, according to one embodiment, the processor (230) can simulate the acoustic environment of a new space based on the finishing material applied in step S410 and the furniture placed in step S420, and verify whether acoustic performance suitable for the use of the space has been achieved. The acoustic environment simulation can be performed based on shape data of the space model, acoustic characteristic data of the finishing material applied to each surface, and acoustic shielding characteristic data of the placed furniture. Specifically, the processor (230) can query the absorption coefficient and acoustic reflectivity for each finishing material by frequency band from the finishing material acoustic characteristic database stored in memory (220) or an external server (100). The finishing material acoustic characteristic database may store absorption coefficients at frequency bands of 125Hz, 250Hz, 500Hz, 1kHz, 2kHz, and 4kHz for each type of finishing material, such as paint, wallpaper, tile, wood panel, carpet, marble, and exposed concrete. Additionally, the processor (230) can query the acoustic shielding characteristics of each piece of furniture, such as sofas, curtains, bookshelves, and rugs, from the furniture acoustic characteristic database (e.g., mid-to-high frequency sound absorption effect of a fabric sofa, acoustic scattering effect of a large bookshelf, etc.). Based on the three-dimensional shape data of the space model, the sound absorption rate data of the finishing material applied to each surface, and the acoustic shielding characteristic data of the placed furniture, the processor (230) can calculate the reverberation time (RT60, Reverberation Time 60) of the space. Reverberation time refers to the time required for the sound pressure level to attenuate by 60 dB after a sound source is generated within the space, and the processor (230) can calculate the reverberation time by applying the Sabine equation or the Eyring equation. The processor (230) can use the calculated reverberation time as the recommended reverberation time for each space use (e.g., living room 0.4–0.6 seconds, bedroom 0.3–0.5 seconds, home theater 0.5–0.5 seconds).Acoustic inadequacy can be detected when the reverberation time exceeds the recommended range compared to (e.g., 8 seconds). When acoustic inadequacy is detected, the processor (230) may suggest changes to finishing materials or furniture to adjust the reverberation time to within the recommended range. For example, if the reverberation time is excessively long (e.g., the reverberation time of a living room is 1.0 second or longer), the processor (230) may suggest replacing a part of the area where a hard finishing material (e.g., marble floor, glass wall, etc.) is applied with a finishing material with a high sound absorption rate (e.g., fabric acoustic panel, carpet, etc.) or adding furniture with sound absorption effects (e.g., fabric sofa, thick curtain, rug, etc.). At this time, the processor (230) may calculate the change in similarity with reference design feature information resulting from the replacement of finishing materials or the addition of furniture, and provide the user with a change proposal that optimizes the balance between improving acoustic performance and maintaining design consistency.

[0201] Meanwhile, the processor (230) can detect the application of finishing materials that require the removal or alteration of the wall (e.g., installation of niches, installation of shelves penetrating the wall, etc.) based on location information of the load-bearing wall, and notify the user that such alteration to the load-bearing wall is structurally impossible.

[0202] Furthermore, the processor (230) can verify whether the location of water usage facilities (washbasin, sink, etc.) is within the range where the water supply and drainage pipe can be connected based on the buried location of the water supply and drainage pipe, and if pipe extension is required, notify the user of the relevant details.

[0203] According to one embodiment, the processor (230) can automatically adjust the application of finishing materials and furniture placement based on the results of verifying spatial constraints. When automatically adjusting, the processor (230) can search for an optimal placement that satisfies spatial constraints while maintaining maximum similarity with reference design feature information. For example, when adjusting furniture positions, the processor (230) can determine the furniture positions in a direction that minimizes the amount of position adjustment due to spatial constraints while maintaining maximum proportions of the relative positional relationships between furniture in the reference design image (e.g., distance between a sofa and a coffee table, distance between a dining table and a wall, etc.).

[0204] Additionally, according to one embodiment, the processor (230) may generate a plurality of design candidates based on the results of S410 and S420. When there are multiple combinations of finishing material textures and furniture objects corresponding to reference design feature information, the processor (230) may generate a design candidate for each combination and calculate a style reflection score with respect to the reference design image for each design candidate, thereby ranking the design candidates in order of highest style reflection. The style reflection score may be calculated by weighted summing the degree of agreement for each item of the reference design feature information (finishing material type, color, pattern, texture, furniture type, furniture style, etc.). The processor (230) may present the ranked plurality of design candidates to the user, allowing the user to select a preferred design candidate.

[0205] Meanwhile, the processor (230) can use an AI model to perform automatic adjustment of associated elements in response to changes in specific design elements within the design result. For example, if a user changes the floor finish from light-toned hardwood flooring to dark-toned marble, the processor (230) can automatically adjust the color tone of the wall paint, the color tone of the furniture, and the color temperature of the lighting to maintain overall color balance in response to the change in floor color. This automatic adjustment of associated elements can be performed by an AI model based on color harmony theory and interior design principles, and can automatically maintain the harmony of the entire space whenever the user changes individual design elements.

[0206] Additionally, according to one embodiment, the processor (230) can receive feedback on the design result from the user and iteratively modify the design result. The processor (230) provides a 3D viewer interface of the design result to the user terminal and can receive modification inputs, such as changing the finishing material, changing the color, replacing furniture, or moving the furniture position, by selecting a specific design area. The processor (230) updates the design result based on the received modification inputs and re-performs spatial constraint verification (S430) on the updated design result to check whether the modified design result satisfies the spatial constraints. Through this feedback-based iterative modification process, an optimal design result that reflects both the user's preferences and actual spatial conditions can be derived.

[0207] Furthermore, the processor (230) can provide the user with the results of an analysis of the degree of style reflection relative to the reference for the design result. The processor (230) can calculate the degree of matching of finishing materials, color, pattern, texture, and furniture style between each design area of ​​the design result and the corresponding design area of ​​the reference design image, and provide the user with the degree of style reflection for each area in the form of a visualized report. Through this, the user can quantitatively verify to what extent the design result reflects the style of the reference design image and decide whether to make additional modifications to areas with low reflection.

[0208] Next, the processor (230) can determine the 3D space model, which includes the finishing material applied in step S410 and the furniture placed in step S420 and is adjusted based on the space constraints in step S430, as a 3D interior design result and store it in memory (220) (S440).

[0209] Specifically, the three-dimensional interior design result is three-dimensional space model data with finishing materials applied and furniture placed, and may include finishing material texture information for each design area, three-dimensional model data and placement coordinates of each furniture object, lighting setting information, and spatial metadata. The processor (230) stores the determined three-dimensional interior design result in memory (220) and can subsequently refer to it during the process of generating a detailed drawing for construction (S500).

[0210] According to one embodiment, the processor (230) can render and output a three-dimensional interior design result on the display of a user terminal. The processor (230) can generate a high-quality rendering image that reflects the reflection, shadow, and indirect lighting effects of the finishing material according to the lighting by applying a ray tracing or global illumination-based rendering technique for realistic visualization. Additionally, the processor (230) can provide an interactive viewer interface so that the user can view the design result by rotating, zooming in, and zooming out from various viewpoints through a three-dimensional viewer.

[0211] Meanwhile, the processor (230) can reflect lighting simulation results in a three-dimensional interior design result. The processor (230) can perform a simulation of natural light inflow based on the location and size data of windows included in the space model, and can simulate changes in the distribution of natural light by time of day (morning, afternoon, evening, etc.) and reflect them in the design result. In addition, the processor (230) can perform an artificial lighting simulation based on the location, type, and light intensity information of lighting fixtures included in the design result, and can visually provide the user with changes in the atmosphere of the space according to the lighting.

[0212] Through this, the processor (230) can apply the finishing materials, colors, patterns, textures, and furniture styles and arrangements extracted from the reference design image to a 3D model of the interior space to be designed, and generate a 3D interior design result that guarantees constructability based on spatial constraints. In particular, through the generation and ranking of multiple design candidates, iterative modification based on user feedback, AI-based automatic adjustment of related elements, and analysis of style reflection relative to the reference, a design result that optimally reflects the user's taste and actual space conditions can be produced.

[0214] FIG. 7 is a detailed flowchart of S500 of FIG. 3, and is a flowchart for explaining the operation of generating a detailed drawing for construction according to one embodiment. Each step (S510 to S540) of FIG. 7 can be performed by the processor (230) of FIG. 2.

[0215] Referring to FIG. 7, a processor (230) according to one embodiment can calculate a list of materials and furniture quantities required for construction and an estimated construction cost based on the three-dimensional interior design result determined in step S400 (S510).

[0216] Specifically, the processor (230) can calculate the required amount of materials based on the coverage area (area or length) of each finishing material included in the three-dimensional interior design result and the unit specifications of the finishing material. For example, in the case of floor tiles, the processor (230) can calculate the required quantity of tiles by dividing the floor area (e.g., 25 m²) by the unit area of ​​the tile (e.g., 0.36 m², 600 mm × 600 mm tile) and calculate the final quantity of materials by adding the cutting loss rate (e.g., 10%) that occurs during construction.

[0217] In the case of wall finishing materials, the processor (230) can calculate the finishing area based on each wall surface data of the three-dimensional interior design result, and can calculate the pure finishing area by subtracting the area of ​​openings such as windows and doors. In the case of wallpaper, the processor (230) can calculate the required number of rolls based on the specifications (width and length) of the wallpaper roll and can reflect the additional amount required due to pattern matching. In the case of paint, the processor (230) can calculate the required amount of paint based on the application area and the application area per unit of paint (e.g., 8 to 10 m² per 1 L) and can reflect the number of applications (e.g., 2 to 3 times).

[0218] In the case of ceiling finishing materials, the processor (230) can calculate the ceiling finishing area, the length of the lightweight steel frame, the number of hanger bolts, the number of finishing material panels, etc. based on the ceiling surface data of the three-dimensional interior design result.

[0219] The processor (230) can automatically aggregate the type, style, specifications, and quantity of each furniture object placed in the 3D design result in the case of furniture quantity. The processor (230) can organize the furniture list by type and generate furniture quantity data including the product name (or similar product name), specifications, quantity, and unit price of each piece of furniture.

[0220] Meanwhile, the processor (230) can additionally calculate the required amount of auxiliary materials (e.g., adhesive, mortar, grout, silicone, nails, screws, etc.) when calculating the quantity of materials and furniture. The processor (230) can calculate the type and quantity of necessary auxiliary materials by referencing a construction standard database according to the standard construction method corresponding to each type of finishing material.

[0221] Next, the processor (230) can calculate the estimated construction cost based on the calculated quantity of materials and the quantity of furniture.

[0222] Specifically, the processor (230) can calculate material costs by matching the type and quantity of each material included in the material quantity data with the unit price information of the material unit price database stored in the memory (220) or external server (100). The material unit price database may store product name, specifications, brand, unit price, and supplier information for each finishing material, and the processor (230) can expand the user's range of choices by providing multiple brand and grade unit prices for the same finishing material.

[0223] Additionally, the processor (230) can calculate the cost of furniture by searching the furniture unit price database for the furniture unit price corresponding to the type, style, and specifications of each piece of furniture included in the furniture quantity data. The furniture unit price database may store market price information by type, style, and specifications of furniture.

[0224] The processor (230) can additionally calculate construction costs. Construction costs may include labor costs, equipment usage fees, transportation costs, waste disposal costs, etc., required for the installation of finishing materials. The processor (230) can calculate construction costs by multiplying the construction area by the construction unit price corresponding to each type of finishing material (e.g., unit price per m² for tile installation, unit price per m² for wallpaper installation, unit price per m² for paint installation, etc.). Construction unit prices may be stored by region in a construction unit price database, and the processor (230) can apply the construction unit price of the corresponding region based on the location information of the indoor space to be designed.

[0225] The processor (230) can calculate the total estimated construction cost by summing the material costs, furniture costs, and construction costs, and can generate a detailed statement of items (material costs, furniture costs, construction costs). In addition, the processor (230) can calculate a contingency cost (e.g., 10 to 15% of the total cost) and reflect it in the estimated construction cost.

[0226] Furthermore, according to one embodiment, the processor (230) may receive budget information from the user. The budget information may include a total budget amount that the user can invest in interior construction or budget limits per item (e.g., material cost limit, furniture cost limit, construction cost limit, etc.). The processor (230) may compare the calculated estimated construction cost with the user's budget information to determine whether the estimated construction cost exceeds the budget information.

[0227] If the calculated estimated construction cost exceeds the budget information, the processor (230) can calculate the contribution to the cost overrun based on the unit price information of each material and furniture item included in the generated list of material and furniture quantities. Specifically, the processor (230) can calculate the total cost per item by multiplying the unit price and quantity of each item, and calculate the contribution to the cost overrun based on the ratio that the total cost per item accounts for in the total estimated construction cost. The processor (230) can prioritize selecting items with a higher calculated contribution as replacement items. For example, if the unit price of floor tiles is 50,000 won per sheet and the required quantity is 100 sheets, the total cost of the item is 5,000,000 won, and if the unit price of a sofa is 2,000,000 won and the quantity is 1, the total cost of the item is 2,000,000 won; therefore, the processor (230) can prioritize selecting floor tiles, which have a higher contribution to the cost overrun, as replacement items over sofas. Through this, replacements can be systematically carried out starting with the items that offer the greatest overall cost savings.

[0228] Next, the processor (230) can search for alternative candidate materials and candidate furniture belonging to the same category in the materials and furniture database of the external server (100) via the communication unit (210) for each selected replacement target item. Here, the same category refers to materials or furniture corresponding to the same use and type as the replacement target item. For example, if the replacement target item is floor tiles, the processor (230) can search for candidate materials belonging to the floor tile category in the materials database of the external server (100), and if the replacement target item is a 3-seater fabric sofa, it can search for candidate furniture belonging to the 3-seater sofa category in the furniture database. When searching, the processor (230) prioritizes searching for candidates with a lower unit price than the currently applied materials or furniture, but can also search for candidates that can be included within the total budget through combination with other items, even if the unit price is the same or slightly higher.

[0229] Next, the processor (230) can use an AI model (similarity calculation model) to calculate color similarity, pattern similarity, and finishing material similarity with reference design feature information extracted for each of the searched candidate materials and candidate furniture. Color similarity, pattern similarity, and finishing material similarity can each be calculated as a value between 0 and 1, and the closer the value is to 1, the higher the similarity with reference design feature information.

[0230] Specifically, color similarity is a numerical value representing the degree of visual similarity between the color of a candidate material or candidate furniture and the color palette of the corresponding area included in the reference design feature information. The processor (230) can calculate the color distance between the color value of the candidate material or candidate furniture and the color value included in the color palette of the reference design feature information based on the CIEDE2000 color difference in the CIE Lab color space or the Euclidean distance in the HSV color space, and calculate the color similarity by normalizing the calculated color distance. Through this, it is possible to quantitatively evaluate whether the replacement material or replacement furniture maintains a color consistent with the color tone of the reference design image (e.g., warm tone series, neutral tone series, cool tone series, etc.).

[0231] Pattern similarity is a numerical value indicating the degree of visual similarity between the surface pattern of a candidate material or candidate furniture and the pattern information of the corresponding area included in the reference design feature information. The processor (230) can calculate pattern similarity by comprehensively analyzing whether the pattern type of the candidate material (e.g., plain, stripe, check, herringbone, marble pattern, wood grain, etc.) belongs to the same or similar category as the pattern type of the reference design feature information, the degree of similarity in the scale (size) and direction of the pattern, and the degree of similarity in the density and repetition cycle of the pattern. For example, if the pattern of the wall area in the reference design image is a herringbone pattern, the processor (230) can calculate a high pattern similarity for the herringbone pattern tile among the candidate materials and a low pattern similarity for the plain (solid) tile.

[0232] The similarity of the finishing material is a numerical value indicating the degree of similarity in material characteristics between the material of the candidate material or candidate furniture and the finishing material information of the corresponding area included in the reference design feature information. The processor (230) can calculate the similarity of the finishing material by comprehensively analyzing whether the material category of the candidate material (e.g., natural stone, artificial stone, solid wood, plywood, tile, metal, fabric, leather, etc.) belongs to the same or similar series as the finishing material category of the reference design feature information, the degree of similarity in surface texture (glossiness, roughness) characteristics, and the degree of similarity in material grade (high-end, mid-range, standard, etc.). For example, if the floor finishing material in the reference design image is natural marble, the processor (230) can calculate a high similarity of the finishing material (e.g., 0.85) for artificial marble, a medium level similarity of the finishing material (e.g., 0.70) for marble pattern polished tiles, and a low similarity of the finishing material (e.g., 0.30) for economy vinyl tiles.

[0233] The processor (230) can select candidate materials and candidate furniture as replacement candidates, wherein the calculated color similarity, pattern similarity, and finishing material similarity are each greater than or equal to a preset threshold value. Here, the preset threshold value can be set individually for each type of similarity, for example, the threshold value for color similarity can be set to 0.7, the threshold value for pattern similarity to 0.6, and the threshold value for finishing material similarity to 0.65. The threshold value can be set directly by the user or set as a default value of the system. By excluding candidate materials and candidate furniture that are below the threshold value from the replacement candidate group, the processor (230) can ensure at least minimal similarity with the visual characteristics of the reference design image. Subsequently, the processor (230) can select combinations that can be constructed within the budget information from the selected replacement candidate group. The processor (230) can recalculate the total estimated construction cost by applying the unit prices of candidate materials and candidate furniture included in the replacement candidate group to each replacement target item, and can select a combination in which the recalculated estimated construction cost falls within the budget information.

[0234] Meanwhile, according to one embodiment, the processor (230) may determine the priority of proposals for alternative candidates based on a user preference learning model when proposing a plurality of candidate materials and candidate furniture included in the alternative candidate group to the user. The user preference learning model may be a machine learning model constructed by utilizing the history of material and furniture selection, history of alternative candidate approval, history of finishing material changes, and history of feedback on design results from design projects performed by the user in the past as training data. Specifically, the processor (230) may accumulate data in the memory (220) regarding the characteristics of finishing materials and furniture that the user finally approved in past design projects (brand preference, material preference, color family preference, price range preference, pattern type preference, etc.). Additionally, the processor (230) may accumulate data regarding items approved and items rejected by the user after receiving alternative candidate proposals. The processor (230) may input the accumulated data into the user preference learning model to learn the user's potential preference patterns. The processor (230) can predict the user acceptance probability for each candidate material and candidate furniture included in the replacement candidate group using a user preference learning model. The user acceptance probability is a numerical value between 0 and 1 indicating the likelihood that the candidate material or candidate furniture will be approved by the user. The processor (230) can calculate a comprehensive recommendation score by weighting the user acceptance probability to the calculated color similarity, pattern similarity, and finishing material similarity, and can propose replacement candidates to the user by ranking them in order of highest comprehensive recommendation score. Through this, even among multiple replacement candidates with the same level of similarity, a candidate that matches the user's taste is proposed first, thereby improving user satisfaction in the process of selecting replacement materials and replacement furniture.Furthermore, if the training data of the user preference learning model falls short of a preset minimum number of data (e.g., three or more project histories), the processor (230) can estimate the probability of user acceptance by referencing preference data of similar user groups in a collaborative filtering manner based on the user's demographic characteristics (e.g., age group, residential area) or initial design selection patterns. Through this, personalized alternative candidate suggestions may be possible even for new users.

[0235] The processor (230) can receive approval input by providing the selected combination to the user along with comparison information with existing material and furniture items. Specifically, for each replaced item, the processor (230) can display the product name, specifications, and unit price of the existing material or furniture and the product name, specifications, and unit price of the replacement candidate material or furniture in parallel, and output the color similarity, pattern similarity, and finishing material similarity scores, cost savings per item, and total cost savings of each replacement item on the display of the user terminal. In addition, the processor (230) can provide a 3D rendering image in the form of a before / after that allows visual comparison of the existing design result and the design result with the replacement candidate applied.

[0236] When the processor (230) receives approval input from the user, it can regenerate a three-dimensional interior design result by reflecting the combination selected based on the approval input. Specifically, the processor (230) can regenerate a three-dimensional interior design result reflecting the replacement candidate by reapplying a finishing texture corresponding to the replaced material to the corresponding area of ​​the space model and repositioning a three-dimensional furniture object corresponding to the replaced furniture within the space model. Subsequently, the processor (230) can recalculate the list of material and furniture quantities and the estimated construction cost based on the regenerated three-dimensional interior design result. In this case, subsequent steps S520 through S540 can be performed based on the regenerated three-dimensional interior design result.

[0237] According to one embodiment, the processor (230) may apply different replacement strategies depending on the extent to which the estimated construction cost exceeds the budget information. If the estimated construction cost slightly exceeds the budget (e.g., within 10%), the processor (230) may select only the top few items that contribute significantly to the cost overrun as replacement targets to meet the budget with minimal changes. If the estimated construction cost significantly exceeds the budget (e.g., more than 30%), the processor (230) may select all material and furniture items as replacement targets to perform comprehensive replacement. Meanwhile, the processor (230) may re-verify the overall harmony of the design results reflecting the replacement materials and replacement furniture using an AI model. If the harmony is below a preset standard, the processor (230) may identify replacement items that lower the harmony and re-select replacement candidates by setting a higher standard value for color similarity, pattern similarity, or finishing material similarity for those items. Through this, a balance between budget reduction and design harmony can be maintained.

[0238] Next, the processor (230) can generate construction drawing data based on the finishing material information and furniture information of each design area included in the three-dimensional interior design result determined in step S400 (S520). Construction drawing data refers to technical drawing information required to actually construct the design result.

[0239] Specifically, the processor (230) can automatically generate a two-dimensional construction drawing from three-dimensional model data of a three-dimensional interior design result. The construction drawing data may include at least one of a floor plan, a wall elevation, a reflected ceiling plan, and a detailed section.

[0240] A floor plan is a drawing that represents a horizontal cross-section based on the floor surface of a design result, and may include the wall layout of each room, the location of openings (doors, windows), the type and scope of construction of floor finishing materials, the location and layout pattern of furniture placement, and major dimension lines. A processor (230) can automatically generate a floor plan by projecting the wall data and furniture placement data of a three-dimensional space model into two dimensions. At this time, the processor (230) may express the scope of construction of each finishing material by hatching or color coding, and may add the type and specifications of the finishing material as drawing annotations.

[0241] An elevation view is a frontal projection of each wall surface and may include the type and scope of wall finishing materials, boundary lines between finishing materials, locations of outlets and switches, locations of wall lighting, and locations and dimensions of shelves or cabinets attached to the wall surface. A processor (230) can automatically generate an elevation view by projecting each wall surface data of a three-dimensional space model in the frontal view and can represent the construction pattern of the finishing material texture applied to the wall surface (e.g., tile joint pattern, wallpaper seam location, etc.) in the drawing.

[0242] A ceiling plan is a drawing of a ceiling surface viewed from the floor direction and may include the type and scope of construction of ceiling finishing materials, the location and type of lighting fixtures, the location of air conditioner outlets, the range and height difference of a drop ceiling, and the type and location of moldings. A processor (230) can automatically generate a ceiling plan by mirroring the ceiling surface data of a three-dimensional space model.

[0243] A detailed cross-sectional drawing is a drawing that shows the construction method of the finishing material, the interlayer configuration (e.g., floor finishing material → adhesive → leveling → waterproofing layer → structure), the construction structure of the wall finishing material (e.g., gypsum board → putty → paint, or tile → adhesive mortar → waterproofing layer → structure), and the construction structure of the ceiling finishing material (e.g., finishing material → lightweight steel frame → hanger bolt → structure). The processor (230) can automatically generate a detailed cross-sectional drawing by searching for standard construction structure data corresponding to each type of finishing material from a construction standard database stored in memory (220) or an external server (100).

[0244] According to one embodiment, the processor (230) can convert and store construction drawing data in a CAD (Computer-Aided Design) format (e.g., DWG, DXF, etc.) or a BIM (Building Information Modeling) format (e.g., IFC, Revit, etc.). Construction drawing data in CAD format can be directly used as a work drawing at the construction site, and construction drawing data in BIM format can be used for construction management and maintenance by including metadata such as the material, quantity, and construction method of each building element.

[0245] Meanwhile, the processor (230) can automatically verify constructability when generating construction drawing data. The processor (230) verifies whether the construction conditions of each finishing material (e.g., minimum construction thickness, substrate conditions, curing period, etc.) match the actual space conditions, and if there is a problem with constructability, it can notify the user of the relevant matters. For example, if a natural stone finishing material with a thickness exceeding 50 mm is applied to an area where floor heating pipes are installed, the processor (230) can detect the possibility of reduced heating efficiency, warn the user, and suggest an alternative finishing material.

[0246] Next, the processor (230) can generate priority and process sequence information for each construction process based on the construction drawing data generated in step S520 (S530).

[0247] Specifically, the processor (230) can analyze each construction process included in the construction drawing data and automatically determine the construction priority based on the dependency relationship between the processes.

[0248] The processor (230) can classify the construction process into multiple construction stages. The construction stages may be classified, for example, into a first stage (demolition and dismantling process), a second stage (equipment process: electrical wiring, plumbing, heating and cooling equipment), a third stage (foundation finishing process: waterproofing, floor leveling, substructure work), a fourth stage (floor finishing process), a fifth stage (wall finishing process), a sixth stage (ceiling finishing process), a seventh stage (furniture delivery and installation process), and an eighth stage (finishing process: lighting installation, decoration placement, cleaning). The processor (230) can analyze the dependency relationships between each construction stage to identify sequential dependency relationships in which a subsequent stage can only be initiated after a preceding stage is completed. For example, floor tile installation is possible only after the floor waterproofing process is completed, wallpaper or paint installation is possible only after the wall substructure work is completed, and ceiling finishing material installation is possible only after the ceiling lightweight steel frame installation is completed.

[0249] According to one embodiment, the processor (230) can adjust the construction priority by taking into account the curing time of each construction process. The curing time refers to the waiting time required for the finishing material to acquire sufficient strength after installation. For example, a curing time of about 24 to 48 hours may be required after floor leveling (self-leveling mortar) installation, a curing time of about 24 hours may be required after waterproofing layer installation, and a curing time of about 24 hours may be required from tile installation to grouting. The processor (230) can prioritize the process with a long curing time and optimize the construction order so that other areas or other types of construction processes can be performed concurrently during the curing period. Through this, the total construction time can be minimized.

[0250] Additionally, the processor (230) can identify construction processes that can be performed concurrently. If construction processes in different rooms can be performed concurrently without mutual interference, the processor (230) can arrange those processes in parallel within the same construction phase. For example, if floor tiling in the living room and wall painting in the bedroom can be performed simultaneously without mutual interference, the processor (230) can classify them as processes that can be performed concurrently within the same construction period, thereby shortening the overall construction schedule.

[0251] Furthermore, the processor (230) can adjust the construction priority by considering the lead time of the materials. The processor (230) can calculate the estimated lead time for each material by referring to supplier information stored in a material unit price database or an external server (100). In the case of a construction process that includes materials with long lead times (e.g., imported materials, custom-made furniture, etc.), the processor (230) can adjust the construction schedule by advancing the ordering time of the materials to before the start of construction and prioritizing other processes during the lead time waiting period. This prevents construction delays caused by the non-delivery of materials.

[0252] Meanwhile, the processor (230) can calculate a critical path based on the dependency relationships and required durations between construction processes. A critical path refers to a sequence of processes that, if delayed in the overall construction schedule, directly affects the overall construction completion schedule. The processor (230) identifies and separately marks the construction processes corresponding to the critical path and may include precautions to prevent delays in the corresponding processes (e.g., pre-ordering materials, securing skilled personnel, strict adherence to curing periods, etc.).

[0253] The processor (230) can generate construction schedule data in the form of a Gantt chart based on the aforementioned construction priority, inter-process dependency, parallel possible processes, and critical path information, and include this data in the priority and process sequence information for each construction process. The Gantt chart may include the start date, end date, duration, precedence dependency, parallel process indication, curing waiting period, and critical path indication for each construction process. Through this, the construction company can systematically manage the process according to the construction priority and predict the total construction period to establish an efficient plan for manpower and material input.

[0254] Meanwhile, according to one embodiment, the processor (230) may adjust the construction schedule by additionally referring to weather condition data expected during the construction period when generating priority and process sequence information for each construction process. The processor (230) may receive weather forecast data, such as daily average temperature, humidity, and probability of precipitation during the scheduled construction period, from an external weather data server through the communication unit (210). Specifically, the processor (230) may automatically adjust the schedule of construction processes sensitive to weather conditions based on the received weather forecast data. Construction processes sensitive to weather conditions may include, for example, a paint coating process (appropriate construction temperature: 5°C, appropriate humidity: 85% or less), a waterproofing layer construction process (appropriate construction temperature: 5°C or higher, cannot be constructed during precipitation), a tile construction process (appropriate construction temperature: 5°C), and a leveling process (appropriate construction temperature: 10°C or higher). If, based on weather forecast data, the temperature on a specific date falls outside the appropriate construction temperature range for the process or the humidity exceeds the appropriate range, the processor (230) can automatically reschedule the construction date of the process to a date with appropriate weather conditions. In this case, the processor (230) can sequentially readjust the schedule of subsequent processes following the rescheduling to maintain consistency in the overall construction schedule. Additionally, the processor (230) can reflect variations in the curing period according to the temperature. For example, the curing period of leveling mortar may be approximately 24 hours at a temperature of 20°C, but may increase to approximately 48 hours at a temperature of 10°C. The processor (230) can dynamically calculate the curing period based on the expected temperature of the construction date based on weather forecast data and reflect the calculated curing period in the construction schedule to prevent construction defects caused by insufficient curing. Furthermore, according to one embodiment, the processor (230) can automatically reschedule the construction schedule based on process status data received in real time during construction.A construction company or site manager can input the actual start date, actual completion date, and progress rate of each construction process through the input interface of the electronic device (200). The processor (230) can compare the input process status data with the originally planned construction schedule to calculate the number of days of delay in the event of a delay in a specific process and determine whether the delayed process corresponds to a critical path. If the delayed process corresponds to a critical path, the processor (230) can automatically readjust the schedule of subsequent processes by the number of days of delay and perform optimized rescheduling to minimize the increase in the total construction period through the rearrangement of parallel processes. The processor (230) can provide the rescheduled construction schedule to the user in the form of an updated Gantt chart and visually display changes compared to the original plan to improve the efficiency of construction management.

[0255] Next, the processor (230) can generate a construction detail drawing including a list of material and furniture quantities calculated in step S510, an estimated construction cost, construction drawing data generated in step S520, and priority and process sequence information for each construction process generated in step S530, and store it in memory (220) (S540).

[0256] Specifically, the construction detail drawing is document data that combines construction drawing data (plan layout, elevation view, ceiling view, detailed cross-section, etc.), a material quantity list (type, specification, quantity, unit price, subtotal by material), a furniture quantity list (type, style, specification, quantity, unit price, subtotal by furniture), an estimated cost list (material cost, furniture cost, construction cost, contingency reserve, total), information on priority and sequence of construction processes, and precautions for construction. The processor (230) can generate the construction detail drawing in a format such as PDF, CAD, or BIM, store it in memory (220), output it to a user terminal, or transmit it to a construction company via an external server (100).

[0257] According to one embodiment, the processor (230) may request final confirmation from a user or customer before generating a construction detail drawing. The processor (230) may provide a confirmation interface to the user terminal that combines a 3D rendering image of the design result, a preview of the construction drawing data, details of material and furniture quantities, and details of estimated costs. When the user transmits a final approval input through the confirmation interface, the processor (230) may confirm the design result and construction-related data at that time as the final construction detail drawing and store it in memory (220). If the user requests a modification, the processor (230) may update the design result or construction drawing data based on the modification request and re-perform steps S510 to S540 to generate an updated construction detail drawing.

[0258] Additionally, the processor (230) may include purchasing information for materials and furniture in the construction detail drawing. The processor (230) may include purchasing-related information in the construction detail drawing, such as a list of suppliers where each material and furniture can be purchased, delivery lead times, and minimum order quantities, by referring to supplier information stored in the material unit price database and the furniture unit price database.

[0259] Through this, the processor (230) can automatically generate a comprehensive construction detail drawing based on a three-dimensional interior design result, including a list of materials and furniture quantities required for actual construction, estimated construction costs, construction drawing data, priority and process sequence information for each construction process. In particular, by automatically selecting alternative materials and alternative furniture that can reduce costs while maintaining consistency in color schemes, material grades, and style categories based on the user's budget information, it can provide a construction detail drawing that maintains design quality to the maximum extent even under budget constraints.

[0260] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using a general-purpose computer or a special-purpose computer, such as a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on the operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. In addition, other processing configurations, such as parallel processors, are also possible.

[0261] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0262] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0263] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based thereon. For example, appropriate results may be achieved even if the described techniques are performed in a different order than described and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from the described method, or are replaced or substituted by other components or equivalents.

[0264] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

Claims

Claim 1 An electronic device that generates interior design results and construction detail drawings of a new space based on reference design images using AI (Artificial Intelligence), comprising: a communication unit that transmits and receives data to and from an external server; and a memory that stores at least one instruction. and includes a processor that executes at least one of the above instructions, wherein the processor receives spatial information for the new space and generates a spatial model of the new space, obtains at least one reference design image by receiving it directly from a user or by searching from the external server through the communication unit, analyzes the obtained reference design image with an AI model to extract reference design feature information, applies the extracted reference design feature information to the generated spatial model to generate a three-dimensional interior design result for the new space, and generates a construction detail drawing based on the three-dimensional interior design result, wherein in generating the spatial model, the processor receives spatial information including at least one of floor plan data, actual measurement data, and spatial photograph images of the new space, calculates spatial structure data including wall locations, opening locations, ceiling height, and floor area of ​​the new space based on the input spatial information, generates a three-dimensional spatial model based on the calculated spatial structure data, determines the generated three-dimensional spatial model as the spatial model for the new space, and the processor [receives] the reference design feature information In the extraction process, the acquired reference design image is divided into multiple design areas including a wall area, a floor area, a ceiling area, and a furniture area using the AI ​​model; for each of the divided wall area, floor area, and ceiling area, area-specific feature data including color palette information, pattern information, and finishing material information is extracted; and for the divided furniture area, the type and style of the furnitureExtracting furniture feature data including size and placement patterns, integrating the extracted area-specific feature data and the extracted furniture feature data to generate the reference design feature information, and in the process of the processor generating the 3D interior design result, deriving spatial constraints including the size, shape, structural location, and opening location of the new space from the generated spatial model, matching and applying color palette information, pattern information, and finishing material information corresponding to each of the wall area, floor area, and ceiling area of ​​the spatial model based on the extracted area-specific feature data, determining the type, size, and placement location of furniture and equipment suitable for the new space based on the extracted furniture feature data and the spatial constraints using the AI ​​model, and integrating the applied area-specific finishing material, color, and pattern information and the determined furniture and equipment placement information to generate the 3D interior design result, and in the process of the processor generating the construction detail drawing, calculating a list of materials and furniture quantities required for construction and estimated construction costs based on the generated 3D interior design result, and based on the generated 3D interior design result, a floor plan layout, A construction drawing data is generated that includes at least one of a ceiling plan, an elevation plan, and a detailed cross-section plan; a construction process is classified into a plurality of construction stages based on the generated construction drawing data; a sequential dependency relationship is identified by analyzing the precedence dependency relationship between the plurality of construction stages such that a subsequent construction stage can be initiated only after a preceding construction stage is completed; a construction stage with a relatively longer curing period is prioritized based on the curing period of the finishing material for each of the plurality of construction stages, and a construction order is adjusted so that other construction stages that do not violate the sequential dependency relationship are carried out in parallel during the curing period.Among construction processes performed in different rooms, construction processes that can be performed concurrently without mutual interference are arranged in parallel within the same construction period; the delivery date of each material included in the calculated list of material and furniture quantities is calculated; for construction processes containing materials with a calculated delivery date longer than a preset standard period, the ordering time of the corresponding materials is advanced to before the construction start time, and the construction order is adjusted so that other construction processes that do not violate the sequential dependency relationship are performed first during the waiting period for the delivery of the corresponding materials; a critical path, which is a chain of construction processes that affects the overall construction completion schedule if a delay occurs, is calculated based on the dependency relationship and required period between the construction processes; priority and process order information for each construction process is generated to shorten the overall construction period based on the identified sequential dependency relationship, the construction processes arranged in parallel, and the calculated critical path; and a construction detailed drawing is generated including the calculated list of material and furniture quantities, the calculated estimated construction cost, the generated construction drawing data, and the generated priority and process order information for each construction process. The processor, in generating the reference design feature information In this, a plurality of reference design images are obtained by receiving input directly from a user or by searching from the external server through the communication unit, and each of the obtained plurality of reference design images is divided into the plurality of design areas including the wall area, the floor area, the ceiling area, and the furniture area using the AI ​​model, and for each of the divided wall area, floor area, and ceiling area, area-specific feature data including at least the color palette information, the pattern information, and the finishing material information is extracted, and for the divided furniture area, furniture feature data including the type, style, size, and arrangement pattern of the furniture is extracted.The area-specific feature data and furniture feature data extracted from each of the plurality of reference design images are sorted for comparison by reference design image for each of the wall area, floor area, ceiling area, and furniture area, and provided to the user; input is received from the user to select a reference design image from the plurality of reference design images to be applied to each of the wall area, floor area, ceiling area, and furniture area; based on the selection, feature data of the reference design images corresponding to each area are combined to generate new integrated reference design feature information; an AI model is used to determine whether the color scheme, finishing material grade range, and furniture style category match among the area-specific feature data included in the generated integrated reference design feature information; if, as a result of the determination, there exists an area where the color scheme, finishing material grade range, or furniture style category does not match, the AI ​​model is used to automatically correct the feature data of that area to match the color scheme, finishing material grade range, and furniture style category of the adjacent area; the integrated reference design feature information is updated by reflecting the automatically corrected feature data; and the updated integrated reference design feature information is provided to the user for approval. When the processor receives input and generates the construction detail drawing, it receives budget information from the user, compares the calculated estimated construction cost with the received budget information, and if the calculated estimated construction cost exceeds the budget information, it selects items with a high contribution to the cost overrun as priority replacement items based on the unit price information of each material and furniture item included in the calculated list of material and furniture quantities.For each of the above-mentioned selected replacement target items, the processor searches for replaceable candidate materials and candidate furniture belonging to the same category in the materials and furniture database of the external server via the communication unit; calculates the color similarity, pattern similarity, and finishing material similarity with the extracted reference design feature information for each of the searched candidate materials and candidate furniture using the AI ​​model; selects candidate materials and candidate furniture whose calculated color similarity, pattern similarity, and finishing material similarity are each greater than or equal to a preset threshold value as a replacement candidate group; selects combinations that can be constructed within the budget information from the selected replacement candidate group; provides the selected combinations along with comparison information with existing material and furniture items to the user to receive approval input; regenerates the 3D interior design result by reflecting the selected combinations based on the received approval input; regenerates the construction drawing data, the list of material and furniture quantities, and the estimated construction cost based on the regenerated 3D interior design result to regenerate the construction detail drawing; and in selecting the replacement candidate group, the processor uses a user preference learning model to replace An electronic device that calculates user acceptance probabilities for each candidate material and candidate furniture included in a candidate group, wherein the user preference learning model is a machine learning model constructed using the user's past history of material and furniture selection, history of approval of alternative candidates, history of changes to finishing materials, and history of feedback on design deliverables as training data, and adjusts the priority of the alternative candidate group by reflecting the calculated user acceptance probabilities. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 delete

Citation Information

Patent Citations

  • Ai-based interior layout design system

    KR102725985B1

  • Designing Branding Space to Improve User Experience Based on Artificial Intelligence Model and Interior Construction Using it

    KR102853887B1

  • Customized interior concept design system

    KR102921594B1