A home collocation scheme generation method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application provide a home collocation scheme generation method and device, electronic equipment and storage medium, wherein the method comprises: constructing a feature extraction model; training the feature extraction model to obtain a trained feature extraction model; constructing a scoring model; training the scoring model to obtain a trained scoring model; determining a first home; generating the home collocation scheme using a bundle search algorithm according to the first home, wherein the search results of each step of the bundle search are filtered according to the trained feature extraction model and the trained scoring model to obtain a matching scheme in the search results of each step. By implementing the above embodiments, the search results of the materials are mathematized and abstracted, and subjective factors are separated, so that the generated furniture collocation scheme is more coordinated and objective.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of home furnishing, in particular to a home furnishing collocation scheme generation method and device, electronic equipment and computer readable storage medium. BACKGROUND

[0002] The existing home furnishing collocation theme color matching method first selects available furniture materials for each home furnishing pit according to the category and size of each home furnishing pit in the home furnishing material library, and then determines the filled furniture materials for each home furnishing pit according to the similarity of the theme colors. This method only relies on color to collocate home furnishing materials, which is prone to the situation of inconsistent and inharmonious style, and lacks an objective evaluation method for the harmony and aesthetics of home furnishing material collocation. SUMMARY

[0003] The purpose of the embodiments of the present application is to provide a home furnishing collocation scheme generation method, device, electronic equipment and storage medium, which can accurately and objectively generate a home furnishing collocation method.

[0004] The embodiments of the present application provide a home furnishing collocation scheme generation method, which comprises:

[0005] Constructing a feature extraction model;

[0006] Training the feature extraction model to obtain a trained feature extraction model;

[0007] Constructing a scoring model;

[0008] Training the scoring model to obtain a trained scoring model;

[0009] Determining a first home furnishing;

[0010] Generating the home furnishing collocation scheme by using a bundle search algorithm according to the first home furnishing, wherein the search results of each step of the bundle search are screened according to the trained feature extraction model and the trained scoring model to obtain a matching scheme in the search results of each step.

[0011] In the above implementation process, unlike the prior art, the feature extraction model and the scoring model are used to mathematize and abstract the search results of the materials in the home furnishing collocation scheme, and the subjective factors are abstracted, so that the harmony and objectivity of the generated furniture collocation scheme are higher.

[0012] Further, the step of training the feature extraction model comprises:

[0013] Obtaining a home furnishing sequence from the material library;

[0014] Obtaining a plurality of negative materials corresponding to the replacement sample in the home furnishing sequence in the material library;

[0015] generating a first training home sequence according to the home sequence replacement sample and the plurality of negative materials;

[0016] replacing the replacement sample in the home sequence with a mask to obtain a second training home sequence;

[0017] inputting the first training home sequence and the second training home sequence into the feature extraction model, so that the feature extraction model determines the replacement sample in the second training home sequence.

[0018] In the above implementation process, the negative material and the replacement sample correspond, the first training home sequence is generated by the negative material, the replacement sample in the home sequence is replaced with a mask to obtain a second training home sample, and the first training home sequence and the second training home sequence are input into the feature extraction model, so that the feature extraction model determines the replacement sample in the second training home sequence. Through continuous training, the feature extraction model obtains the same points and different points of the home materials, so as to generate different feature vectors according to different materials.

[0019] Further, the step of training the scoring model further comprises:

[0020] generating a third training home sequence according to the home sequence and the plurality of negative material sets;

[0021] inputting the second training home sequence and the third training home sequence into the feature extraction model, so that the feature extraction model learns the interaction relationship between different home materials in the material library.

[0022] In the above implementation process, the third training home sequence is generated based on the negative material, and the feature extraction model is trained according to the third training home sequence, so that the feature extraction model learns the interaction relationship between different home materials in the material library, thereby being able to generate different feature vectors according to different materials.

[0023] Further, the step of training the feature extraction model further comprises:

[0024] using a cross-entropy loss function as an objective function;

[0025] determining that the feature extraction model learns to determine the replacement sample in the second training home sequence when the cross-entropy loss function converges.

[0026] Further, the step of training the scoring model comprises:

[0027] replacing the replacement sample in the home sequence with the plurality of negative materials to obtain a plurality of fourth training home sequences;

[0028] obtaining feature vectors of the home sequence and the plurality of fourth training home sequences according to the trained feature extraction model;

[0029] inputting the feature vector of the home sequence and the feature vectors of the plurality of fourth training home sequences into the scoring model to obtain a probability that each fourth training home sequence belongs to the home sequence;

[0030] setting a cross-entropy loss function as an objective function;

[0031] judging whether the cross-entropy loss function converges according to the probability that each fourth training home sequence belongs to the home sequence;

[0032] if yes, stopping training.

[0033] In the implementation process, the plurality of fourth training home sequences are generated based on negative materials, and the fourth training home sequences are input into the scoring model, so that the scoring model can learn the matching degree of the home in different home sequences.

[0034] Further, the step of screening the search result of each step of the bundle search according to the trained feature extraction model and the trained scoring model to obtain a matching scheme in the search result of each step, comprises:

[0035] obtaining feature vectors corresponding to a plurality of candidate schemes corresponding to the search result according to the trained feature extraction model;

[0036] inputting the feature vectors into the trained scoring model to obtain a scoring result;

[0037] determining the matching scheme according to the scoring result.

[0038] In the implementation process, the matching degree of the home sequence in the scoring result is measured by the trained feature extraction model and the scoring model, so that the matching scheme can be obtained in the search result of each step.

[0039] Further, the step of determining the first home comprises:

[0040] determining the first home according to a heuristic search or a color matching method.

[0041] In a second aspect, the embodiments of the present application provide a home collocation scheme generation device, comprising:

[0042] a construction module configured to construct a feature extraction model;

[0043] a training module configured to train the feature extraction model to obtain a trained feature extraction model;

[0044] The construction module is further configured to construct a scoring model.

[0045] The training module is further configured to train the scoring model to obtain a trained scoring model.

[0046] A determination module configured to determine a first home;

[0047] A generation module configured to generate the home furnishing scheme according to the first home by using a beam search algorithm, wherein a search result of each step of the beam search is filtered according to the trained feature extraction model and the trained scoring model to obtain a matching scheme in the search result of each step.

[0048] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the steps of the method according to any one of the first aspect when executing the computer program.

[0049] In a fourth aspect, a computer readable storage medium is provided, and the computer readable storage medium stores instructions, and the instructions, when running on a computer, cause the computer to execute the method according to any one of the first aspect.

[0050] Other features and advantages of the present application will be illustrated in the following description, or can be known or determined by no doubt from the description, or can be known by implementing the above-mentioned technologies disclosed in the present application.

[0051] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are specifically described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation to the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0053] Figure 1 The flowchart of the home furnishing method provided by the embodiments of the present application is shown in the figure;

[0054] Figure 2 The flowchart of the training feature extraction model provided by the embodiments of the present application is shown in the figure;

[0055] Figure 3A schematic diagram of the pre-training sample provided for the embodiment of the present application;

[0056] Figure 4 A schematic diagram of the process of training the feature extraction model provided for the embodiment of the present application;

[0057] Figure 5 A schematic diagram of the pre-training sample provided for the embodiment of the present application;

[0058] Figure 6 A schematic diagram of the process of training the scoring model provided for the embodiment of the present application;

[0059] Figure 7 Generating a home furnishing collocation scheme according to the first home furnishing by using the beam search algorithm provided for the embodiment of the present application;

[0060] Figure 8 A schematic diagram of the beam search algorithm provided for the embodiment of the present application;

[0061] Figure 9 A schematic diagram of the structure of the home furnishing collocation device provided for the embodiment of the present application;

[0062] Figure 10 A schematic diagram of the structure of the electronic device provided for the embodiment of the present application. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.

[0064] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second", etc. are only used for distinguishing description, and cannot be understood as indicating or implying relative importance.

[0065] Embodiment 1

[0066] Referring to Figure 1 The embodiment of the present application provides a home furnishing collocation method, comprising:

[0067] S1: constructing a feature extraction model;

[0068] S2: training the feature extraction model to obtain a trained feature extraction model;

[0069] S3: constructing a scoring model;

[0070] S4: training the scoring model to obtain a trained scoring model;

[0071] S5: determining a first home furnishing;

[0072] S6: According to the first home, a home collocation scheme is generated by using a beam search algorithm, wherein the search results of each step of the beam search are screened according to the trained feature extraction model and the trained scoring model, and a matching scheme in the search results of each step is obtained.

[0073] In the above implementation process, unlike the prior art, the feature extraction model and the scoring model are used to mathematize and abstract the search results of the home collocation scheme and the materials, and to abstract subjective factors, so that the generated furniture collocation scheme is more coordinated and objective.

[0074] Referring to Figure 2 , considering that the feature extraction model and the scoring model affect the aesthetics and coordination of the furniture collocation scheme generation. The first embodiment of the present application provides a task of training the feature extraction model, which includes:

[0075] S21: Obtain a home sequence from a material library;

[0076] S22: Obtain a plurality of negative materials corresponding to a replacement sample in the home sequence from the material library;

[0077] S23: Generate a first training home sequence according to the replacement sample in the home sequence and the plurality of negative materials;

[0078] S24: Replace the replacement sample in the home sequence with a mask to obtain a second training home sequence;

[0079] S25: Input the first training home sequence and the second training home sequence into the feature extraction model, so that the feature extraction model determines the replacement sample in the second training home sequence.

[0080] Further, the cross-entropy loss function is used as the training task objective function;

[0081] When the cross-entropy loss function converges, it is determined that the feature extraction model learns to determine the replacement sample in the second training home sequence.

[0082] Exemplarily, referring to Figure 3from the designer sample room, a home sequence [home 1, home 2, home 3, home 4, home 5] is extracted, and multiple candidate homes of the same category as home 5 are selected from the home material library as negative materials, which together with home 5 form a candidate set [home 5, candidate 1, candidate 2, candidate 3]. The last home in the home sequence, that is, home 5, is replaced by a mask, and the home sequence becomes [home 1, home 2, home 3, home 4, mask]. The sequence is sent into the model together with the candidate set sequence, and the model is allowed to learn which candidate home in the candidate set is the home replaced by the mask. In this task, home 5 is the positive sample, and candidate 1-3 are negative materials. The cross-entropy loss function is used as the task optimization objective, and the task ends when the loss function converges.

[0083] Further, referring to Figure 4 The embodiments of the present application provide a second task of training a feature extraction model, comprising:

[0084] S26: generating a third training home sequence according to the home sequence and the plurality of negative material sets;

[0085] S27: inputting the second training home sequence and the third training home sequence into the feature extraction model, so that the feature extraction model learns the interaction relationship between different home materials in the material library.

[0086] Exemplarily, referring to Figure 5 [START, home 1, home 2, home 3, home 4] and the candidate set [home 5, candidate 1, candidate 2, candidate 3] are sent into the model, and the model learns the interaction relationship between different home materials, wherein START represents the entire furniture sequence. The matching score is calculated by calculating the feature of START and the feature of the home in the candidate set, and the matching score can be calculated by calculating the norm distance.

[0087] Further, the difference between Score(Start, home 5) and Max(Score(Start, candidate 1), Score(Start, candidate 2), Score(Start, candidate 3)) is maximized by the SGD algorithm, and the training ends when the difference converges.

[0088] Further, referring to Figure 6 The step of training the scoring model comprises:

[0089] S41: replacing the replacement sample in the home sequence with the plurality of negative materials respectively to obtain a plurality of fourth training home sequences;

[0090] S42: obtaining the feature vectors of the home sequence and the plurality of fourth training home sequences according to the trained feature extraction model;

[0091] S43: input the feature vector of the home sequence and the feature vectors of the plurality of fourth training home sequences into the scoring model to obtain a probability that each fourth training home sequence belongs to the home sequence;

[0092] S44: set the cross-entropy loss function as an objective function;

[0093] S45: determine whether the cross-entropy loss function converges according to the probability that each fourth training home sequence belongs to the home sequence, and if yes, perform S213;

[0094] S46: stop training.

[0095] In a possible implementation, S7 comprises: determining the first home by using a heuristic search according to the home position, size and category.

[0096] In the above implementation process, the search range can be expanded according to the heuristic search algorithm, and the first home most suitable for the input home position, size and category can be obtained.

[0097] In a possible implementation, the first home can be obtained based on a color matching method.

[0098] In the above implementation process, the first home matching the color input by the user can be obtained, so that the generated home matching scheme is more in line with the user's requirements.

[0099] Referring to Figure 7 In a possible implementation, S6 comprises:

[0100] S61: obtain the feature vectors corresponding to the plurality of candidate schemes in the search result according to the trained feature extraction model;

[0101] S62: input the feature vectors into the trained scoring model to obtain a scoring result;

[0102] S63: determine the matching scheme according to the scoring result.

[0103] Exemplarily, assuming that we need to generate 2 matching schemes, then the width of the bundle search is 2, such as Figure 8As shown, the first furniture in the home pit is determined by heuristic search or color matching method, that is, home 1. The second furniture pit has 4 candidate materials, so the scoring model needs to score 4 combinations, [home 1, home 2-1], [home 1, home 2-2], [home 1, home 2-3], [home 1, home 2-4], and finally select home 2-1 and home 2-4 with the highest score as the candidate materials for further branching.(4) The third home pit also has 4 candidate materials, so the scoring model needs to score 2*4 combinations, and finally select [home 1, home 2-1, home 3-1] and [home 1, home 2-4, home 3-6] as the combination scheme. Each subsequent home pit is combined and scored to determine the combination material for further branching until the last pit candidate material is determined, where each home pit needs to score (search width*candidate number) combinations. Finally, 2 combination schemes are generated.

[0104] Embodiment 2

[0105] Referring to Figure 9 The embodiment of the present application provides a home combination scheme generation device, comprising:

[0106] The construction module 1 is configured to construct a feature extraction model.

[0107] The training module 2 is configured to train the feature extraction model to obtain a trained feature extraction model.

[0108] The construction module 1 is further configured to construct a scoring model.

[0109] The training module 2 is further configured to train the scoring model to obtain a trained scoring model.

[0110] The determination module 3 is configured to determine a first home.

[0111] The generation module 4 is configured to generate a home combination scheme according to the first home by using a beam search algorithm, wherein the trained feature extraction model and the trained scoring model are used to screen the search result of each step of the beam search to obtain a matching scheme in the search result of each step.

[0112] In a possible implementation, the training module is further configured to obtain a home sequence from a material library.

[0113] The plurality of negative materials corresponding to the replacement sample in the home sequence are obtained in the material library; a first training home sequence is generated according to the replacement sample in the home sequence and the plurality of negative materials; the replacement sample in the home sequence is replaced by using a mask to obtain a second training home sequence; and the first training home sequence and the second training home sequence are input into the feature extraction model, so that the feature extraction model determines the replacement sample in the second training home sequence.

[0114] In a possible implementation, the training module is further configured to generate a third training home sequence according to the home sequence and the plurality of negative material sets; and input the second training home sequence and the third training home sequence into the feature extraction model, so that the feature extraction model learns the interaction relationship between different home materials in the material library.

[0115] In a possible implementation, the training module is further configured to use a cross-entropy loss function as a target function; and determine that the feature extraction model learns to determine the replacement sample in the second training home sequence when the cross-entropy loss function converges.

[0116] In a possible implementation, the training module is further configured to replace the plurality of negative materials in the home sequence to obtain a plurality of fourth training home sequences; obtain feature vectors of the home sequence and the plurality of fourth training home sequences according to the trained feature extraction model; input the feature vectors of the home sequence and the plurality of training home sequences into the scoring model to obtain a probability that each fourth training home sequence belongs to the home sequence; set the cross-entropy loss function as a target function; determine whether the cross-entropy loss function converges according to the probability that each fourth training home sequence belongs to the home sequence; and if yes, stop training.

[0117] In a possible implementation, the generation module is further configured to obtain feature vectors corresponding to a plurality of candidate schemes corresponding to the search result according to the trained feature extraction model; input the feature vectors into the trained scoring model to obtain a scoring result; and determine the matching scheme according to the scoring result.

[0118] In a possible implementation, the determination module is further configured to determine the first home according to a heuristic search or a color matching method.

[0119] The application further provides an electronic device, please see Figure 10 , Figure 10 A structural block diagram of an electronic device provided by the embodiment of the application. The electronic device can include a processor 101, a communication interface 102, a memory 103, and at least one communication bus 104. Wherein, the communication bus 104 is used to realize the direct connection communication of these components. Wherein, the communication interface 102 of the electronic device in the embodiment of the application is used for signaling or data communication with other node devices. The processor 101 can be an integrated circuit chip with signal processing capability.

[0120] The processor 101 can be a general processor, including a central processing unit (CPU), a network processor (NP), etc. It can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. The disclosed methods, steps and logic block diagrams in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor 101 can also be any conventional processor.

[0121] The memory 103 can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read only memory (PROM), an erasable programmable read only memory (EPROM), an electrically erasable programmable read only memory (EEPROM), etc. The memory 103 stores computer readable instructions, which are executed by the processor 101, so that the electronic device can perform various steps involved in the embodiments.

[0122] Optionally, the electronic device can also include a storage controller, an input output unit.

[0123] The memory 103, the storage controller, the processor 101, the peripheral interface, the input output unit are directly or indirectly electrically connected to each other to realize the transmission or interaction of data. For example, these elements can be electrically connected to each other through one or more communication buses 104. The processor 101 is used to execute the executable modules stored in the memory 103, such as software function modules or computer programs included in the electronic device.

[0124] The input output unit is used to provide the user with creating tasks and creating optional time periods or preset execution times for the tasks to realize the interaction between the user and the server. The input output unit can be, but is not limited to, a mouse and a keyboard, etc.

[0125] It can be understood that Figure 10 The structure shown is only schematic, and the electronic device can include more or less components than those shown in the figure, or have the same components as those shown in the figure. Figure 10 The structure shown is only schematic, and the electronic device can include more or less components than those shown in the figure, or have the same components as those shown in the figure. Figure 10Different configurations are shown. Figure 10 The components shown in the various configurations can be implemented in hardware, software, or a combination thereof.

[0126] The embodiments of the present application also provide a computer readable storage medium, and the computer readable storage medium stores instructions. When the instructions are run on a computer, the computer program is executed by a processor to implement the method of the method embodiments. To avoid repetition, details are not described here.

[0127] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are only schematic. For example, the flowcharts and block diagrams in the drawings show the possible implementation architectures, functions and operations of the devices, methods and computer program products according to the embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different orders from those described in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and the combination of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system for executing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0128] In addition, each functional module in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0129] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the method. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0130] The above merely provides an example of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0131] The above merely provides an example of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0132] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one from another entity or action without necessarily requiring or implying any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

Claims

1. A home collocation scheme generation method, characterized by, The method comprises the following steps: constructing a feature extraction model; training the feature extraction model to obtain a trained feature extraction model; constructing a scoring model; training the scoring model to obtain a trained scoring model; determining a first home; generating a home matching scheme by using a beam search algorithm according to the first home, wherein the search results of each step of the beam search are screened according to the trained feature extraction model and the trained scoring model, and a matching scheme in the search results of each step is obtained; wherein the step of training the feature extraction model comprises: obtaining a home sequence from a material library; obtaining a plurality of negative materials corresponding to a replacement sample in the home sequence from the material library; generating a first training home sequence according to the home sequence replacement sample and the plurality of negative materials; replacing the replacement sample in the home sequence with a mask to obtain a second training home sequence; inputting the first training home sequence and the second training home sequence into the feature extraction model, so that the feature extraction model determines the replacement sample in the second training home sequence. 2.The home collocation scheme generation method of claim 1, wherein, The step of training the feature extraction model further comprises: generating a third training home sequence according to the home sequence and the plurality of negative material sets; inputting the second training home sequence and the third training home sequence into the feature extraction model, so that the feature extraction model learns the interaction relationship between different home materials in the material library. 3.The home collocation scheme generation method of claim 1, wherein, The step of training the feature extraction model further comprises: using a cross-entropy loss function as an objective function; when the cross-entropy loss function converges, it is judged that the feature extraction model has learned to determine the replacement sample in the second training home sequence. 4.The home collocation scheme generation method of claim 1, wherein, The step of training the scoring model comprises: replacing the replacement sample in the home sequence with the plurality of negative materials respectively to obtain a plurality of fourth training home sequences; obtaining feature vectors of the home sequence and the plurality of fourth training home sequences according to the trained feature extraction model; inputting the feature vectors of the home sequence and the plurality of fourth training home sequences into the scoring model to obtain the probability that each fourth training home sequence belongs to the home sequence; setting a cross-entropy loss function as an objective function; determining whether the cross-entropy loss function converges according to the probability that each fourth training home sequence belongs to the home sequence; if yes, stop training. 5.The home collocation scheme generation method of claim 1, wherein, The step of screening the search results of each step of the beam search according to the trained feature extraction model and the trained scoring model to obtain a matching scheme in the search results of each step comprises: obtaining feature vectors corresponding to a plurality of candidate schemes corresponding to the search results according to the trained feature extraction model; inputting the feature vectors into the trained scoring model to obtain a scoring result; determining the matching scheme according to the scoring result. 6.The home collocation scheme generation method of claim 5, wherein, The step of determining a first home comprises: determining the first home according to a heuristic search or a color matching method. 7.A home collocation scheme generation device, characterized by comprising: The method comprises the following steps: constructing a feature extraction model; The training module is configured to train the feature extraction model to obtain a trained feature extraction model. The construction module is further configured to construct a scoring model. The training module is further configured to train the scoring model to obtain a trained scoring model. The determination module is configured to determine a first home. The generation module is configured to generate the home collocation scheme according to the first home by using a beam search algorithm, wherein the trained feature extraction model and the trained scoring model are used to screen the search result of each step of the beam search to obtain a matching scheme in the search result of each step. The training module is further configured to obtain a home sequence from a material library. The plurality of negative materials corresponding to the replacement sample in the home sequence are obtained from the material library, the first training home sequence is generated according to the replacement sample in the home sequence and the plurality of negative materials, the replacement sample in the home sequence is replaced by using a mask to obtain the second training home sequence, and the first training home sequence and the second training home sequence are input into the feature extraction model to enable the feature extraction model to determine the replacement sample in the second training home sequence.

8. An electronic device, comprising: The computer readable storage medium stores instructions, and when the instructions are run on the computer, the computer executes the steps of the method. The computer readable storage medium stores instructions, and when the instructions are run on the computer, the computer executes the steps of the method.

9. A computer-readable storage medium, characterized in that, ​

Images