CN114462207B - Matching method, system, equipment and medium for home decoration template - Google Patents
Matching method, system, equipment and medium for home decoration template Download PDFInfo
- Publication number
- CN114462207B CN114462207B CN202210016951.2A CN202210016951A CN114462207B CN 114462207 B CN114462207 B CN 114462207B CN 202210016951 A CN202210016951 A CN 202210016951A CN 114462207 B CN114462207 B CN 114462207B
- Authority
- CN
- China
- Prior art keywords
- template
- ornament
- information
- templates
- furniture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a matching method, a system, equipment and a medium for a home decoration template, wherein the method comprises the following steps: acquiring historical furniture information and historical ornament information; determining the setting relationship between the furniture and the ornaments; generating a plurality of first ornament templates according to the setting relationship and the spatial relationship; integrating the first ornament template to generate a spatial distribution map of the integrated first ornament template; performing density clustering on the spatial distribution map to generate a layout template; screening the first ornament templates according to the layout templates to obtain a plurality of second ornament templates, and training a machine learning model through the second ornament templates and historical spatial information; inputting the target space information into the machine learning model after training is completed, and predicting to obtain a third ornament template; the scheme can greatly shorten the time and the cost for constructing the ornament template, enables the ornament arrangement to be quickly iterated along with the change of requirements, improves the design service quality and the design efficiency, and can be widely applied to the technical field of computer data processing.
Description
Technical Field
The invention relates to the technical field of computer data processing, in particular to a method, a system, equipment and a medium for matching a home decoration template.
Background
Whether an attractive design scheme can be quickly generated or not in a business scene of a home decoration design store often determines the key of whether an order is a deal or not; in addition to placing various furniture, the indoor design needs to make the soft installation design beautiful, so that the household is also needed to be arranged for placing various artworks such as artworks, textiles, collections, lamps, flower art, plants and the like. When the on-line demonstration and the transaction are carried out, a designer is difficult to place a large number of home ornaments in a short time, and the personal aesthetics and example accumulation of the designer are tested by selecting proper ornaments from a large ornament library to carry out layout and placement.
In the related art, since the kinds of the home decoration designs are various, a large amount of manpower and material resources are required to be consumed for manufacturing decoration combinations with various kinds of the decorations, and for most of the decorations placed around the furniture, such as a cloth doll and a toy decoration placed around a bed, it is difficult to automatically place the decorations at a reasonable angle and distance according to rules. In addition, because the types and the placing modes of the ornaments are frequently updated, the variable and huge market demands are difficult to meet only by limited designer resources.
Disclosure of Invention
In view of the above, in order to at least partially solve one of the above technical problems, embodiments of the present invention provide a method for matching a home decoration template, which is more convenient, practical and flexible; meanwhile, the technical scheme of the application also provides a system, equipment and a computer readable and writable storage medium which can correspondingly realize the method.
On one hand, the technical scheme of the application provides a matching method of a home decoration template, and the method comprises the following steps:
acquiring historical furniture information and historical ornament information in a design scheme;
determining the setting relationship between the furniture and the ornaments according to the furniture information and the ornament information;
generating a plurality of first ornament templates according to the setting relationship and the spatial relationship;
integrating the first ornament template to generate a spatial distribution diagram of the integrated first ornament template;
performing density clustering on the ornament template on the spatial distribution map, and generating a layout template according to a density clustering result;
screening the first ornament templates according to the layout templates to obtain a plurality of second ornament templates, and training a machine learning model through the second ornament templates and historical spatial information;
and inputting the target space information into the trained machine learning model, and predicting to obtain a third ornament template, wherein the third ornament template comprises the target ornament information.
In one possible embodiment of the present solution, the setting relationship includes a hanging relationship, a supporting relationship, and an abutting relationship; the first ornament template comprises a hanging ornament template, a supporting ornament template and an adjacent ornament template, and the step of generating a plurality of first ornament templates according to the setting relationship and the spatial relationship comprises the following steps:
acquiring the spatial relationship, and screening the historical furniture information and the historical ornament information according to the spatial relationship to determine a central object;
generating a suspended ornament template according to the suspension relation of the central object in a first preset range;
generating a supporting ornament template according to the supporting relation of the central object in a second preset range;
and generating a neighborhood ornament template according to the adjacency relation of the central object in a third preset range.
In a possible embodiment of the present disclosure, the step of integrating the first ornament template to generate the spatial distribution map of the integrated first ornament template includes:
establishing a coordinate system on the horizontal plane according to the horizontal plane of the central object;
projecting the suspended ornament template, the supported ornament template and the neighborhood ornament template into the coordinate system to obtain a two-dimensional distribution map;
the horizontal plane comprises a vertical plane and a horizontal plane; the two-dimensional distribution map comprises projected points of the ornaments and the furniture.
In a possible embodiment of the present application, the step of performing density clustering on the ornament template on the spatial distribution map and generating a layout template according to a density clustering result includes:
classifying the projection points in the two-dimensional distribution map according to a preset distance threshold value to obtain a cluster;
calculating to obtain a local density value according to the preset distance threshold value through a density clustering algorithm;
determining the boundary distance of the projection point according to the local density value;
and screening out noise points in the projection points according to the boundary distance and the local density value to obtain the layout template.
In a feasible embodiment of the scheme of the application, the step of screening the first ornament templates according to the layout templates to obtain a plurality of second ornament templates, and training a machine learning model through the second ornament templates and the historical spatial information includes:
performing feature coding on the central object in the layout template to obtain a first feature code;
performing feature coding on the non-central object in the layout template to obtain a second feature code;
generating a first relation code according to a position relation between the central object and the non-central object, wherein the position relation relates to the distance and the orientation between the central object and the non-central object;
and training through the first feature codes, the second feature codes and the first relation codes to obtain the machine learning model.
In a possible embodiment of the present application, before the step of inputting the target space information into the trained machine learning model and predicting to obtain the third ornament template, the matching method includes:
determining core furniture in the target space information;
sequencing the first ornament templates according to the clustering density to obtain a template sequence;
traversing the template sequence according to preset furniture intervals, and determining the target position of the core furniture.
In a possible embodiment of the present application, the step of inputting the target space information into the trained machine learning model to predict and obtain the third ornament template includes:
inputting the target position to a machine learning model after training is completed, and outputting to obtain a target environment code of the target position;
and determining the third ornament template according to the cosine similarity between the target environment code and the historical environment code, wherein the historical environment code is generated according to the furniture feature code, the ornament feature code and the second relation code in the first ornament template.
On the other hand, this application technical scheme still provides a matching system of house ornamentation ornament template, and the system includes:
the data acquisition unit is used for acquiring historical furniture information and historical ornament information in the design scheme;
the data processing unit is used for determining the setting relationship between the furniture and the ornaments according to the furniture information and the ornament information; generating a plurality of first ornament templates according to the setting relationship and the spatial relationship; integrating the first ornament template to generate a spatial distribution diagram of the integrated first ornament template; performing density clustering on the ornament template on the spatial distribution map, and generating a layout template according to a density clustering result;
the model training unit is used for screening the first ornament templates according to the layout templates to obtain a plurality of second ornament templates and training a machine learning model through the second ornament templates and historical spatial information;
and the template matching unit is used for inputting the target space information into the trained machine learning model and predicting to obtain a third ornament template, wherein the third ornament template comprises the target ornament information.
On the other hand, the technical scheme of the invention also provides a matching device of the home decoration template, which comprises the following components:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is caused to execute a method for matching a home decoration template as described above.
In another aspect, the present invention further provides a storage medium, in which a processor-executable program is stored, and the processor-executable program is used to execute the aforementioned method for matching a home decoration template when executed by a processor.
Advantages and benefits of the present invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention:
according to the technical scheme, firstly, according to information of historical furniture and ornaments, and according to the furnishing modes and spatial position relations of the home furnishings and the ornaments in the space, a corresponding ornament template is generated; and then manufacturing a furniture ornament layout template for training through density clustering, then training a machine learning model to determine a new scene placing position of the layout template of the target ornament according to the layout template, and matching the ornament template according to the scene code and the position information.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the description below are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flow chart illustrating steps of a method for matching a home decoration template according to an embodiment of the present invention;
FIG. 2 is a schematic view of an embodiment of an accessory template of the present invention;
FIG. 3 is a schematic representation of density clustering of furniture surface templates in an embodiment of the present invention;
FIG. 4 is a diagram illustrating an exemplary self-supervision model according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to the embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
The technical scheme of the application provides a matching method of the home decoration template based on data mining, and aims to automatically extract decoration templates and decoration layout templates aiming at different furniture from a high-quality design template, find the decoration template which is most matched with a historical environment through furniture environment codes, automatically place soft decorations, automatically construct the decoration template by using data and place decorations by using the template, so that the automatic decoration placing effect can be iterated quickly, the work of artificially designing the template is reduced, and the design efficiency of a designer is improved.
In one aspect, as shown in fig. 1, an embodiment of the present application provides a method for matching a home decoration template, where the method includes steps S100-S700:
s100, obtaining historical furniture information and historical ornament information in a design scheme;
in the embodiment, firstly, a high-quality historical whole house design scheme is collected as a data source; for example, historical home information and historical ornament information can be obtained from excellent indoor design schemes, and corresponding furniture information and ornament information are extracted from the indoor design schemes through necessary target identification processing; the historical furniture information may include, but is not limited to, furniture items such as beds, sofas, wardrobes, tables and chairs; the historical ornament information can include, but is not limited to, artware, textiles, collectibles, lamps, flower art, plants and other decorative articles; in some embodiments, a home database of services and designs and an accessories database may be constructed.
S200, determining the setting relationship of the furniture and the ornaments according to the furniture information and the ornament information;
specifically, in the embodiment, the category C, the position XYZ, the bounding box size WDH and the direction IJK of each object in the design scheme are identified according to the historical furniture information and the historical ornament information extracted in step S100, where the object covers article objects such as furniture and ornaments, and for an ornament object, for example, the embodiment may determine whether the ornament object leans against a corresponding wall according to the relative position distance between the ornament object and the surrounding wall in the horizontal direction, and establish the hanging relationship between the wall and the ornament; or judging whether the ornament is placed on certain furniture according to the position and the height of the ornament, and recording the supporting relation between the furniture and the ornament.
S300, generating a plurality of first ornament templates according to the setting relationship and the spatial relationship;
wherein, the first ornament template can comprise a plurality of elements such as furniture and/or ornaments; in the embodiment, according to setting a corresponding distance threshold, taking the specific position of the furniture or the ornament in the target space determined in step S100 as an area center according to the distance threshold as a radius, marking the formed enclosed area as a bounding box of the template, and screening the articles in the target space according to the formed bounding box and the set relationship among the furniture or the ornament articles determined in step S200; the furniture and the ornaments falling into the range of the enclosure box are used as elements in the ornament template.
To more clearly describe the set relationship between the furniture and the video, in some alternative embodiments, the set relationship may include, but is not limited to, a hanging relationship, a supporting relationship, and an abutting relationship; therefore, the first ornament template in step S300 includes a suspended ornament template, a supported ornament template, and an adjacent ornament template; further, in the matching method, the step S300 generates a plurality of first ornament templates according to the setting relationship and the spatial relationship, which may include steps S310 to S340:
s310, obtaining a spatial relationship, and screening the historical furniture information and the historical ornament information according to the spatial relationship to determine a central object;
in particular, in the embodiment, according to the default display principle of indoor design, any ornament should be placed with indoor core furniture as a reference; in an embodiment, identification processing of the furniture may be added in the process of extracting the furniture information, for example, the furniture with a large volume or a large occupied space is actively used as a central object in the space, the top surfaces of the bed, the table, the sofa, and the cabinet are selected as candidate central objects for placing the accessories, the size of the bounding box is set, and the related accessories having a suspension relationship, a support relationship, and an adjacent relationship with the central object are screened to obtain the first accessory template.
And S320, generating a suspended ornament template according to the suspension relation of the central object in the first preset range.
Exemplarily, in the embodiment, as shown in fig. 2, for the candidate furniture object in each data source, a rectangular coordinate system is established with the candidate furniture object, a wall directly opposite to the x-axis and y-axis directions is searched, the projection width of the furniture on the wall is calculated, decorations on the wall within the distance threshold search range are added, and the projection width and the decorations are packed into a group of wall decoration templates, that is, a suspended decoration template is obtained, as shown in fig. 2, three drawing boards directly opposite to the wall are used for a dining table, a total bounding box of the three drawing boards is calculated as a bounding box of the template, and the suspended decoration template is constructed according to decorations contained in the bounding box.
S330, generating a supporting ornament template according to the supporting relation of the central object in the second preset range.
Illustratively, in an embodiment, as shown in fig. 2, the embodiment detects whether there is an ornament in the area of the top surface of the candidate center object, and if yes, packs all ornaments on the top surface into a group of supporting ornament templates; for example, in fig. 2, the total enclosure of all the ornaments on the dining table, such as all dishes, candles, chopsticks and spoons, is calculated as the enclosure of the template, and the supported ornament template is constructed according to the ornaments contained in the enclosure.
And S340, generating a neighborhood ornament template according to the adjacency relation of the center object in the third preset range.
Illustratively, in the embodiment, as shown in fig. 2, the ground around the candidate center object is searched for whether ornaments are placed on the ground with the empirical radius threshold corresponding to different kinds of furniture, for example, 0.8m of bed; for the ornaments with the minimum distance of the surrounding frame within the threshold value, dividing the plurality of retrieved ornaments into a plurality of groups of neighborhood ornament templates; such as the pot culture in fig. 2. It should be noted that, in the embodiment, the aggregated neighborhood ornament template may be obtained by calculating the overall bounding box of each ornament group.
S400, integrating the first ornament template to generate a spatial distribution map of the integrated first ornament template;
specifically, in the embodiment, the plurality of first jewelry modules generated in step S300 may be projected into a spatial plane with the position point of the central object as the origin of coordinates after being collected, so as to obtain the spatial distribution map of the corresponding jewelry template.
In some optional embodiments, the process of integrating the first ornament template in step S400 and generating the spatial distribution map of the integrated first ornament template in the method may include steps S410 and S420:
s410, establishing a coordinate system on a horizontal plane according to the horizontal plane of the central object;
in particular embodiments, horizontal includes vertical planes as well as horizontal planes.
S420, projecting the suspended ornament template, the supported ornament template and the neighborhood ornament template into a coordinate system to obtain a two-dimensional distribution map;
in particular, in an embodiment, the two-dimensional distribution map comprises a plurality of projected points of the ornaments and the furniture; illustratively, in an embodiment, for each template, the coordinates of its planar center point and the co-planar center point coordinates of the corresponding candidate center object are calculated, being a vertical plane for the wall adornment template and a horizontal plane for the neighborhood adornment templates and the support adornment templates. Calculating the coordinate value (x ', y') of the central point projection point of each template total bounding box to be used as a two-dimensional distribution map; as shown in FIG. 3, a table is taken as a candidate center object, the center of the bounding box of the table is taken as the origin of coordinates, a supporting ornament template of the table is sampled, and a two-dimensional graph is drawn by taking the center of the template as a sample point.
S500, performing density clustering on the ornament templates on the spatial distribution map, and generating a layout template according to a density clustering result;
specifically, in the embodiment, the cluster analysis may be performed on the projection points in the spatial distribution map obtained in step S400 to remove isolated points or noise points in the spatial distribution map, and after the outliers such as the isolated points or the noise points are removed, the result of the subsequent matching processing is more accurate.
In some alternative embodiments, the method performs density clustering on the ornament template in the spatial distribution map in step S400, and the step of generating the layout template according to the result of the density clustering may include steps S410-S440:
s410, classifying the projection points in the two-dimensional distribution map according to a preset distance threshold value to obtain a cluster;
specifically, in the embodiment, a density clustering algorithm is adopted, a plurality of clustering clusters are automatically calculated according to a preset truncation distance threshold dc of the embodiment, and then the templates are classified according to the result; the center of each cluster is a candidate layout template of the furniture under the condition of a wall, a top surface or a neighborhood, wherein the layout template not only comprises the center position of the cluster, but also can sequence the priority of the cluster according to the point density in the cluster.
S420, calculating to obtain a local density value according to a preset distance threshold value through a density clustering algorithm;
specifically, in the process of clustering by the density clustering algorithm of the embodiment, the local density value ρ of each projection point is calculated according to the truncation distance threshold dc:
wherein, for each projection point i, the distance d between other points j and the point i is obtained ij If d is ij <dc, the result is 1, otherwise, the result is 0. And summing to obtain the number of all other points of the point i within the range of the distance threshold dc as the local density value.
S430, determining the boundary distance of the projection point according to the local density value;
specifically, in the embodiment, after the local density value ρ of each projection point is determined, the boundary distance δ of each projection point is calculated for all the points, and for the point with the maximum local density value, the boundary distance is the straight-line distance between the maximum local density value ρ and all the points; for the remaining points, the boundary distance δ is a straight line distance from the closest point that is greater than the local density value ρ of this point, and a larger boundary value evidences a greater distance from the other clusters.
S440, screening noise points in the projection points according to the boundary distance and the local density value to obtain a layout template;
specifically, in the embodiment, after the boundary distance is determined, screening out a plurality of points as the starting points of the clusters according to the preset threshold of the local density value ρ and the threshold of the boundary distance δ; for other points, it follows the nearest point that is larger than its local density value ρ when clustering is performed; if the distance between some points in the cluster and all other points in the same cluster is greater than the truncation distance threshold dc, the points are taken as noise points to be excluded; searching all ornament templates to check whether the center points of the ornament templates belong to noise points, and screening the ornament templates if the center points belong to the noise points; thereby obtaining a range for each cluster, and calculating the center position of the range. After determining the range and position of each cluster and removing noise points in the distribution diagram, training materials of the layout template in the embodiment are obtained.
S600, screening the first ornament templates according to the layout templates to obtain a plurality of second ornament templates, and training a machine learning model through the second ornament templates and historical spatial information;
specifically, in the embodiment, after the layout template is generated in step S500, the first ornament template obtained in the foregoing step is screened according to each cluster in the layout template and the corresponding relationship between the projection point in each cluster and the ornament to obtain a second ornament template, and then a machine learning model is used to automatically judge that a specific template needs to be placed at the corresponding position in the target scene; the machine learning model in the embodiment may adopt an auto-supervision model, for example, an encoding expression of a scene trained by adopting a method of a Recursive automatic encoding model (Recursive automatic encoders). It should be noted that the training data of the recursive automatic coding model in the embodiment may include the second adornment template and the history space information.
In some alternative embodiments, the step S600 of the method may be to filter the first ornament template according to the layout template to obtain a plurality of second ornament templates, and the training of the machine learning model by the second ornament templates and the historical spatial information may include steps S610 to S640:
s610, performing feature coding on a central object in the layout template to obtain a first feature code;
s620, performing feature coding on the non-central object in the layout template to obtain a second feature code;
s630, generating a first relation code according to the position relation between the central object and the non-central object;
s640, training through the first feature codes, the second feature codes and the first relation codes to obtain a machine learning model;
in particular, in embodiments, the location relates to the distance and orientation between a central object and a non-central object. Due to the complex scene environment, the self-supervision model is used for carrying out feature coding on the furniture of the candidate central object and the peripheral object. The method comprises the steps that a code character string obtained by feature coding of a center object is marked as a first feature code; recording a code character string obtained by performing feature coding on other furniture around the central object as a second feature code; the embodiment sets a radius threshold according to the furniture of each candidate center object, acquires furniture within the radius range, wherein each furniture object has the category C and the dimension WDH as the furniture characteristics, and codes the relative distances dx, dy and dz between the center furniture and the surrounding furniture and the orientation angle theta as the relation between the two furniture.
Then, the embodiment enters a training process of a Recursive automatic coding model, as shown in fig. 4, in the training process, firstly, furniture features are converted into high-dimensional furniture codes through box embedding, and then, from left to right, two furniture codes and one relation code are combined each time, and a higher-dimensional scene code is generated step by step through a Recursive Encoder, for example:
y 1 ＝f RecursvieEncoder (x′ furniture 1 ,x′ furniture2 ,x relation 1 )
y 2 ＝f RecursvieEncoder (y 1 ,x′ furniture ,x relation 2 )
in the output process of the model, decoding the code by using the Decoder model with different weights and symmetrical structure, comparing the decoded furniture characteristics and relationship characteristics with the original input, and taking the error as the model weight suitable for the prediction loss training. After the structure is trained, for each scene, the Encoder can generate the corresponding scene code.
S700, inputting the target space information into the trained machine learning model, and predicting to obtain a third ornament template, wherein the third ornament template comprises target ornament information;
specifically, in the embodiment, before matching and placing the accessory templates, the candidate placing position needs to be determined in the embodiment, the embodiment may first determine core furniture in a target environment, determine the placing position of the core furniture according to the position information of placing furniture in the previous excellent design scheme, then output an environment code of the placing position of the core furniture according to a trained self-supervision model, and perform matching screening on the generated suspended accessory template, the supported accessory template, and the neighboring accessory template according to the output environment code to obtain an accessory template matched with the placing position of the core furniture.
In some alternative embodiments, before step S700 of inputting the target space information into the trained machine learning model and predicting a third ornament template, the method may further include steps S650 to S670:
s650, determining core furniture in the target space information;
s660, sequencing the first ornament templates according to the clustering density to obtain a template sequence;
s670, traversing the template sequence according to a preset furniture interval, and determining the target position of the core furniture;
specifically, in the embodiment, for each indoor core furniture which is placed as a candidate, the furniture wall ornament layout template, the supporting ornament layout template and the neighborhood ornament layout template can be respectively applied to the embodiment. In the process of applying the template, the embodiment firstly substitutes the candidate position with high clustering density into the smallest ornament template area in the cluster by taking the candidate position as the center, and observes whether a certain distance is left between the position and the surrounding furniture or other candidate positions, and if so, takes the position as the position point needing to be placed. And circulating the process until all candidate nodes in the ornament layout template are screened once, thereby finding out a reasonable ornament template placing position point.
In some alternative embodiments, step S700 of the method inputs the target space information into the trained machine learning model, and predicts a third ornament template, which may include steps S710-S720:
s710, inputting the target position into the trained machine learning model, and outputting to obtain a target environment code of the target position;
s720, determining a third ornament template according to the cosine similarity of the target environment code and the historical environment code, wherein the historical environment code is generated according to the furniture feature code, the ornament feature code and the second relation code in the first ornament template;
specifically, in the embodiment, the core furniture is used as the center, the trained self-supervision model is used for generating the environment code of the position, the environment code is combined with the spatial relative distances dx, dy and dz between the candidate placing position point and the position point of the furniture, the ornament templates corresponding to the history codes which are most similar to the current codes in the cluster at the same position are screened according to cosine similarity, and the ornament templates are placed on the candidate position point. And circulating the steps until all the position points are placed on the ornament template with the most similar scene in the history.
In a second aspect, the present application further provides a matching system for a home decoration template, where the system includes:
the data acquisition unit is used for acquiring historical furniture information and historical ornament information in the design scheme;
the data processing unit is used for determining the setting relationship of the furniture and the ornaments according to the furniture information and the ornament information; generating a plurality of first ornament templates according to the setting relationship and the spatial relationship; integrating the first ornament template to generate a spatial distribution map of the integrated first ornament template; performing density clustering on the ornament template on the spatial distribution map, and generating a layout template according to a density clustering result;
the model training unit is used for screening the first ornament templates according to the layout templates to obtain a plurality of second ornament templates and training the machine learning model through the second ornament templates and the historical spatial information;
and the template matching unit is used for inputting the target space information into the trained machine learning model and predicting to obtain a third ornament template, and the third ornament template comprises the target ornament information.
In a third aspect, the present disclosure further provides a matching device for a home decoration template, which includes at least one processor; at least one memory for storing at least one program; when the at least one program is executed by the at least one processor, the at least one processor may be caused to execute a method of matching a home accessories template as in the first aspect.
The embodiment of the invention also provides a program stored in the storage medium, and the program is executed by the processor to realize the matching method of the home decoration template.
From the above specific implementation process, it can be concluded that the technical solution provided by the present invention has the following advantages or advantages compared to the prior art:
1. according to the technical scheme, the work of automatically arranging ornaments can be finished without manufacturing a large number of ornament templates and compiling arrangement logics corresponding to the ornament templates by a designer; the scheme can realize automatic extraction of the design template and layout information from the elegant design scheme, greatly shorten the time and cost for constructing the ornament template, ensure that the ornament arrangement method can be iterated rapidly along with the change of requirements, and improve the design service quality and the design efficiency;
2. according to the technical scheme, the recommended object is the template in a template mode, the template comprises a plurality of ornaments and positions of the ornaments, and the method can be used for screening bad ornament placing results in some application scenes;
3. the method provided by the technical scheme of the application is not used for recommendation, and is characterized in that a position point range which is most likely to appear in the template is screened out by a position-specific density clustering method, so that candidate templates which are not frequently used in position deployment are excluded from a candidate library; the method for searching for similar recommended targets in the scheme does not relate to clustering, but adopts the size of a vector dot product for comparison; for a large batch of data templates, the matching speed is higher, the efficiency is higher, and various requirements of practical application conditions can be better met.
In alternative embodiments, the functions/acts noted in the block diagrams may occur out of the order noted in the operational illustrations. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality/acts involved. Furthermore, the embodiments presented and described in the flow charts of the present invention are provided by way of example in order to provide a more thorough understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed, and in which sub-operations described as part of larger operations are performed independently.
Furthermore, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and/or features may be integrated in a single physical device and/or software module, or one or more of the functions and/or features may be implemented in a separate physical device or software module. It will also be appreciated that a detailed discussion of the actual implementation of each module is not necessary for an understanding of the present invention. Rather, the actual implementation of the various functional modules in the apparatus disclosed herein will be understood within the ordinary skill of an engineer given the nature, function, and interrelationships of the modules. Accordingly, those skilled in the art can, using ordinary skill, practice the invention as set forth in the claims without undue experimentation. It is also to be understood that the specific concepts disclosed are merely illustrative of and not intended to limit the scope of the invention, which is defined by the appended claims and their full scope of equivalents.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions.
In the description of the specification, reference to the description of "one embodiment," "some embodiments," "an example," "a specific example," or "some examples" or the like means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A matching method of a home decoration template is characterized by comprising the following steps:
acquiring historical furniture information and historical ornament information in a design scheme;
determining the setting relationship between the furniture and the ornaments according to the furniture information and the ornament information;
generating a plurality of first ornament templates according to the setting relationship and the spatial relationship, acquiring the spatial relationship, and screening the historical furniture information and the historical ornament information according to the spatial relationship to determine a central object;
establishing a coordinate system on the horizontal plane according to the horizontal plane of the central object, integrating the first ornament template into the coordinate system, and generating a spatial distribution map of the integrated first ornament template;
classifying the projection points in the spatial distribution map according to a preset distance threshold to obtain cluster clusters, screening out noise points in the projection points according to local density values of the cluster clusters, and generating a layout template;
screening the first ornament template according to the corresponding relation between the projection points in the cluster of the layout template and the ornaments to obtain a plurality of second ornament templates, and training a machine learning model through the second ornament templates and historical spatial information;
and inputting the target space information into the trained machine learning model, and predicting to obtain a third ornament template, wherein the third ornament template comprises target ornament information.
2. The method of matching a home decoration template of claim 1 wherein the set relationship comprises a hanging relationship, a supporting relationship and an abutting relationship; the first ornament template comprises a suspended ornament template, a supported ornament template and a neighborhood ornament template, and the step of generating a plurality of first ornament templates according to the setting relationship and the spatial relationship comprises the following steps:
generating a suspended ornament template according to the suspension relation of the central object in a first preset range;
generating a supporting ornament template according to the supporting relation of the central object in a second preset range;
and generating a neighborhood ornament template according to the adjacency relation of the central object in a third preset range.
3. The method for matching a home decoration template as claimed in claim 2, wherein the step of integrating the first decoration template to generate the spatial distribution map of the integrated first decoration template comprises:
projecting the suspended ornament template, the supported ornament template and the neighborhood ornament template into the coordinate system to obtain a two-dimensional distribution map;
the horizontal plane comprises a vertical plane and a horizontal plane; the two-dimensional distribution map comprises projected points of the ornaments and the furniture.
4. The method for matching a home decoration template according to claim 3, wherein the step of performing density clustering on the decoration template on the spatial distribution map and generating a layout template according to a density clustering result comprises:
classifying the projection points in the two-dimensional distribution map according to a preset distance threshold value to obtain a cluster;
calculating to obtain a local density value according to the preset distance threshold value through a density clustering algorithm;
determining the boundary distance of the projection point according to the local density value;
and screening out noise points in the projection points according to the boundary distance and the local density value to obtain the layout template.
5. The method for matching home decoration templates of claim 2, wherein the step of screening the first decoration templates according to the layout templates to obtain a plurality of second decoration templates, and training a machine learning model according to the second decoration templates and historical spatial information comprises:
performing feature coding on the central object in the layout template to obtain a first feature code;
performing feature coding on the non-central object in the layout template to obtain a second feature code;
generating a first relation code according to a position relation between the central object and the non-central object, wherein the position relation relates to the distance and the orientation between the central object and the non-central object;
and training through the first feature codes, the second feature codes and the first relation codes to obtain the machine learning model.
6. The method for matching a home decoration template of claim 1, wherein before the step of inputting the target space information into the trained machine learning model and predicting a third decoration template, the method comprises:
determining core furniture in the target space information;
sequencing the first ornament templates according to the clustering density to obtain a template sequence;
traversing the template sequence according to preset furniture intervals, and determining the target position of the core furniture.
7. The method for matching a home decoration template of claim 6, wherein the step of inputting the target space information to the trained machine learning model and predicting a third decoration template comprises:
inputting the target position to the machine learning model after training is completed, and outputting to obtain a target environment code of the target position;
and determining the third ornament template according to the cosine similarity of the target environment code and the historical environment code, wherein the historical environment code is generated according to the furniture feature code, the ornament feature code and the second relation code in the first ornament template.
8. A matching system for a home decoration template, comprising:
the data acquisition unit is used for acquiring historical furniture information and historical ornament information in the design scheme;
the data processing unit is used for generating a plurality of first ornament templates according to a set relationship and a spatial relationship, acquiring the spatial relationship, and screening the historical furniture information and the historical ornament information according to the spatial relationship to determine a central object; establishing a coordinate system on the horizontal plane according to the horizontal plane of the central object, integrating the first ornament template into the coordinate system, and generating a spatial distribution map of the integrated first ornament template; classifying the projection points in the spatial distribution map according to a preset distance threshold to obtain cluster clusters, screening out noise points in the projection points according to local density values of the cluster clusters, and generating a layout template;
the model training unit is used for screening the first ornament template according to the corresponding relation between the projection points in the cluster of the layout template and the ornaments to obtain a plurality of second ornament templates, and training a machine learning model through the second ornament templates and historical spatial information;
and the template matching unit is used for inputting the target space information into the trained machine learning model and predicting to obtain a third ornament template, wherein the third ornament template comprises the target ornament information.
9. A matching device of a home decoration template is characterized by comprising:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to perform a method of matching a home decoration template as claimed in any one of claims 1 to 7.
10. A storage medium in which a processor-executable program is stored, wherein the processor-executable program is configured to execute a method of matching a home decoration template according to any one of claims 1 to 7 when executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210016951.2A CN114462207B (en) | 2022-01-07 | 2022-01-07 | Matching method, system, equipment and medium for home decoration template |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210016951.2A CN114462207B (en) | 2022-01-07 | 2022-01-07 | Matching method, system, equipment and medium for home decoration template |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114462207A CN114462207A (en) | 2022-05-10 |
| CN114462207B true CN114462207B (en) | 2023-03-14 |
Family
ID=81408779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210016951.2A Active CN114462207B (en) | 2022-01-07 | 2022-01-07 | Matching method, system, equipment and medium for home decoration template |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114462207B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115795627B (en) * | 2022-12-28 | 2023-09-26 | 广州极点三维信息科技有限公司 | Furniture feature construction method, system, device and medium |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737409A (en) * | 2012-06-28 | 2012-10-17 | 北京中科广视科技有限公司 | Method for generating three-dimensional virtual interior design plan |
| CN105023201A (en) * | 2015-07-24 | 2015-11-04 | 中建三局第一建设工程有限责任公司 | Assembled building detailed design and construction method based on BIM (Building Information Modeling) and large data |
| CN105976434A (en) * | 2016-05-04 | 2016-09-28 | 杭州群核信息技术有限公司 | Interior decoration intelligent design method |
| CN106202352A (en) * | 2016-07-05 | 2016-12-07 | 华南理工大学 | The method that indoor furniture style based on Bayesian network designs with colour match |
| CN106484940A (en) * | 2016-07-09 | 2017-03-08 | 陈志静 | A kind of home decoration Intelligentized design method and system |
| CN107767465A (en) * | 2017-10-25 | 2018-03-06 | 武汉华酷科技有限公司 | A kind of Decoration Design scheme intelligence construction method |
| CN108984904A (en) * | 2018-07-17 | 2018-12-11 | 北京理工大学 | A kind of Home Fashion & Design Shanghai method based on deep neural network |
| US10628546B1 (en) * | 2018-06-29 | 2020-04-21 | Cadence Design Systems, Inc. | Method and system for automatically extracting layout design patterns for custom layout design reuse through interactive recommendations |
| CN111460552A (en) * | 2020-03-27 | 2020-07-28 | 杭州群核信息技术有限公司 | Automatic design method between sample plates in toilet |
| CN113222686A (en) * | 2021-04-08 | 2021-08-06 | 复旦大学 | Decoration design scheme recommendation method |
| CN113487038A (en) * | 2021-06-25 | 2021-10-08 | 青岛海尔科技有限公司 | Scene determination method and device, storage medium and electronic device |
| CN113705111A (en) * | 2021-09-22 | 2021-11-26 | 百安居信息技术(上海)有限公司 | Fitment furniture automatic layout method and system based on deep learning |
| CN113886911A (en) * | 2021-09-16 | 2022-01-04 | 杭州群核信息技术有限公司 | Household design scheme generation method and device and computer readable storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006076910A1 (en) * | 2005-01-19 | 2006-07-27 | Guetig David | Method for simulating the appearance of furnishing objects on the destination point thereof |
| CN113743009B (en) * | 2021-08-31 | 2022-07-01 | 广州极点三维信息科技有限公司 | Cabinet type intelligent design method, device, equipment and medium based on representation learning |
-
2022
- 2022-01-07 CN CN202210016951.2A patent/CN114462207B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102737409A (en) * | 2012-06-28 | 2012-10-17 | 北京中科广视科技有限公司 | Method for generating three-dimensional virtual interior design plan |
| CN105023201A (en) * | 2015-07-24 | 2015-11-04 | 中建三局第一建设工程有限责任公司 | Assembled building detailed design and construction method based on BIM (Building Information Modeling) and large data |
| CN105976434A (en) * | 2016-05-04 | 2016-09-28 | 杭州群核信息技术有限公司 | Interior decoration intelligent design method |
| CN106202352A (en) * | 2016-07-05 | 2016-12-07 | 华南理工大学 | The method that indoor furniture style based on Bayesian network designs with colour match |
| CN106484940A (en) * | 2016-07-09 | 2017-03-08 | 陈志静 | A kind of home decoration Intelligentized design method and system |
| CN107767465A (en) * | 2017-10-25 | 2018-03-06 | 武汉华酷科技有限公司 | A kind of Decoration Design scheme intelligence construction method |
| US10628546B1 (en) * | 2018-06-29 | 2020-04-21 | Cadence Design Systems, Inc. | Method and system for automatically extracting layout design patterns for custom layout design reuse through interactive recommendations |
| CN108984904A (en) * | 2018-07-17 | 2018-12-11 | 北京理工大学 | A kind of Home Fashion & Design Shanghai method based on deep neural network |
| CN111460552A (en) * | 2020-03-27 | 2020-07-28 | 杭州群核信息技术有限公司 | Automatic design method between sample plates in toilet |
| CN113222686A (en) * | 2021-04-08 | 2021-08-06 | 复旦大学 | Decoration design scheme recommendation method |
| CN113487038A (en) * | 2021-06-25 | 2021-10-08 | 青岛海尔科技有限公司 | Scene determination method and device, storage medium and electronic device |
| CN113886911A (en) * | 2021-09-16 | 2022-01-04 | 杭州群核信息技术有限公司 | Household design scheme generation method and device and computer readable storage medium |
| CN113705111A (en) * | 2021-09-22 | 2021-11-26 | 百安居信息技术(上海)有限公司 | Fitment furniture automatic layout method and system based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| java、装饰设计模式与递归算法_-;qinqinnibaobaoni的博客;《java、装饰设计模式与递归算法_-CSDN博客》;20130224;全文 * |
| 建筑装饰信息化设计与绿色度动态评价研究_;徐刚;《建筑装饰信息化设计与绿色度动态评价研究》;20181231;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114462207A (en) | 2022-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chang et al. | Shapenet: An information-rich 3d model repository | |
| CN109871604B (en) | Indoor function division method based on depth countermeasure network model | |
| Phan et al. | Color orchestra: Ordering color palettes for interpolation and prediction | |
| CN101930627A (en) | Three-dimensional dwelling size modeling method based on two-dimensional dwelling size diagram | |
| CN109446185B (en) | Collaborative filtering missing data processing method based on user clustering | |
| CN109408954B (en) | Indoor design method and device applied to electronic commerce | |
| Kán et al. | Automated interior design using a genetic algorithm | |
| CN114462207B (en) | Matching method, system, equipment and medium for home decoration template | |
| CN110197225A (en) | House type spatial match method and system based on deep learning | |
| CN103198522B (en) | Three-dimensional scene models generation method | |
| CN110555258A (en) | Automatic layout method based on probability and rule base | |
| CN111046478B (en) | Custom hard-fitting design method and design device | |
| CN110598017A (en) | Self-learning-based commodity detail page generation method | |
| CN109509057A (en) | A kind of indoor design method and device | |
| CN111199577A (en) | Virtual house decoration method | |
| CN111091618A (en) | Combined indoor layout method and system based on house type data driving | |
| JP2019168976A (en) | Three-dimensional model generation device | |
| CN115115846A (en) | Automatic generation method and device of house type layout, computer equipment and storage medium | |
| Park et al. | Analysis of pairings of colors and materials of furnishings in interior design with a data-driven framework | |
| Li et al. | 3d compat: Composition of materials on parts of 3d things | |
| CN110162645A (en) | Image search method, device and electronic equipment based on index | |
| CN112818229A (en) | Ornament recommendation method, system, device and medium based on home space | |
| CN111460552A (en) | Automatic design method between sample plates in toilet | |
| CN113946900B (en) | Method for quickly recommending similar house types based on house type profiles and distribution characteristics | |
| Zhang et al. | User guided 3D scene enrichment. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |