Building surface number prediction method and device, equipment and storage medium

Abstract

The invention provides a building surface number prediction method and device, equipment and a storage medium. The method comprises the following steps: acquiring geometric feature information of a to-be-measured building; inputting the geometric feature information into a GMM (Gaussian Mixture Model) surface number prediction model to obtain a predicted surface number of the to-be-detected building; according to the method, the geometric features of the building are input into the GMM surface number prediction model, so that the distribution of the surface number of the building can be effectively modeled, and a relatively accurate prediction result is obtained. Through feature extraction and standardization processing, the influence of different building feature scale differences on model training is eliminated, and the robustness of the model is ensured. Moreover, the prediction result is optimized according to the building style category by combining a surface number optimization strategy, the prediction surface number can be adjusted according to different building types, the reasonability and practicability of the prediction result are improved, and errors possibly occurring in a single model are avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of building design, and in particular to a method, device, equipment and storage medium for predicting the number of building surfaces. Background Art

[0002] With the rapid development of the construction industry, the prediction of building faces has become an important part of building design, construction and evaluation. The face number of a building affects its spatial layout, exterior design, functional division, energy consumption and other aspects. Therefore, accurate building face number prediction is of great significance to project cost control, design optimization and resource planning. Traditional building face number prediction methods mostly rely on manual experience or simple mathematical models, and have problems such as low prediction accuracy and poor applicability. Especially when facing complex building structures, the prediction effect is often not ideal.

[0003] In recent years, with the rapid development of computer vision technology and machine learning algorithms, face number prediction based on building model data has gradually become a research hotspot. As a classic probabilistic model, Gaussian mixture model (GMM) is widely used in the prediction of building face number due to its good flexibility and adaptability. By modeling the distribution of building face number as a weighted sum of multiple Gaussian distributions, GMM can effectively process building data with different face number characteristics and achieve better prediction results. However, in practical applications, the complexity and diversity of building model data require refined processing of the model training process, such as data standardization, noise removal, feature extraction and enhancement, to improve the accuracy and robustness of the prediction.

[0004] In summary, the problems existing in the prior art need to be solved urgently. Summary of the invention

[0005] The present invention provides a method, device, equipment and storage medium for predicting the number of building surfaces, which are used to solve the defects in the prior art and realize accurate prediction of the number of building surfaces.

[0006] The present invention provides a method for predicting the number of building surfaces, comprising: Obtain geometric feature information of the building to be tested; Inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0007] According to a method for predicting the number of building faces provided by the present invention, after the step of inputting the geometric feature information into a GMM face number prediction model to obtain the predicted face number of the building to be tested, the method further comprises: The predicted number of faces is optimized according to a face number optimization strategy.

[0008] According to a method for predicting the number of building faces provided by the present invention, the step of optimizing the predicted number of faces according to the face number optimization strategy specifically includes: Determining the architectural style category of the building to be tested; According to the architectural style category, matching the corresponding face number optimization strategy; The predicted number of faces is optimized according to the face number optimization strategy.

[0009] According to a method for predicting the number of building faces provided by the present invention, the step of optimizing the predicted number of faces according to the face number optimization strategy is specifically as follows:

[0010] in, is the final predicted number of faces, is the predicted number of faces, The adjustment factor corresponding to the face number optimization strategy.

[0011] According to a method for predicting the number of building surfaces provided by the present invention, after the step of acquiring building model data, the method further includes: Preprocessing the building model data, wherein the preprocessing includes removing noise and outliers; Standardizing various features in the building model data so that the building model data are within the same scale range; The building model data is enhanced to obtain a training data set.

[0012] According to a method for predicting the number of building faces provided by the present invention, the step of inputting the building feature vector into the GMM face number prediction model and executing the EM algorithm specifically includes: Initialize the model parameters of each Gaussian distribution in the GMM face number prediction model, wherein the model parameters include a mean, a covariance matrix, and a mixing coefficient; Iteratively calculate the responsibility probability of each data point belonging to each Gaussian distribution, and update the mean, covariance matrix and mixing coefficient of the Gaussian distribution until the preset training conditions are met.

[0013] According to a method for predicting the number of building faces provided by the present invention, the GMM face number prediction model is:

[0014] in, is the probability distribution of the number of building surfaces, It is k The mixing coefficient of the Gaussian distribution satisfies =1 =1, The mean is , the covariance is Gaussian distribution, are the parameters of the model.

[0015] The present invention also provides a device for predicting the number of building surfaces, comprising: An information acquisition module is used to obtain geometric feature information of the building to be tested; A face number prediction module, used for inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0016] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for predicting the number of building surfaces as described above is implemented.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for predicting the number of building surfaces as described in any one of the above is implemented.

[0018] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for predicting the number of building surfaces.

[0019] The method, device, equipment and storage medium for predicting the number of building faces provided by the present invention obtain the geometric feature information of the building to be tested; input the geometric feature information into the GMM face number prediction model to obtain the predicted number of faces of the building to be tested; the present invention can effectively model the distribution of the number of building faces and obtain more accurate prediction results by inputting the building geometric features into the GMM face number prediction model. Through feature extraction and standardization processing, the influence of the scale differences of different building features on model training is eliminated, ensuring the robustness of the model. In addition, by combining the face number optimization strategy, the prediction results are optimized according to the architectural style category, and the predicted number of faces can be adjusted according to different building types, thereby improving the rationality and practicality of the prediction results and avoiding the errors that may occur in a single model. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 It is a schematic flow chart of the method for predicting the number of building surfaces provided by the present invention; Figure 2 It is a structural schematic diagram of a device for predicting the number of building surfaces provided by the present invention; Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] In order to solve the problems in the prior art, the present invention proposes a method for predicting the number of building surfaces to achieve the prediction of the number of building surfaces. The following is a description of the method for predicting the number of building surfaces. Figure 1 As shown, including but not limited to the following steps: Step 110: Obtain geometric feature information of the building to be tested.

[0024] In this step, we first need to obtain the geometric feature information of the building to be tested. This feature information can be obtained in many ways, such as: Building floor plan: Extract the building's floor plan information through architectural design drawings or CAD drawings, including the area, shape and height of each floor.

[0025] Building facade drawing: Obtain the building's appearance features, such as the building's height, number of floors, boundary length, etc., through the building's facade drawing or 3D modeling data.

[0026] LiDAR data: LiDAR data is used to obtain accurate 3D data of buildings, which can obtain the geometric size and shape characteristics of the building from different perspectives.

[0027] Image recognition technology: Computer vision technology is used to extract the geometric outline and detailed features of buildings from satellite images or pictures taken by drones.

[0028] Once the data is obtained, data preprocessing can be performed to ensure the accuracy and completeness of the data. Preprocessing includes data cleaning (noise removal, outlier processing, etc.) and data standardization (converting the data to a uniform scale).

[0029] Step 120: Input the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested.

[0030] In this step, the geometric feature information of the building to be tested is used as input and sent to the GMM face number prediction model for processing. The specific process is as follows: The geometric feature information after data preprocessing includes the building's spatial features (such as area, floor height, window-to-wall ratio, etc.) and appearance features (such as floor distribution and structural type, etc.). This information will be converted into feature vectors and input into the trained GMM face count prediction model.

[0031] The model is based on the Gaussian mixture model (GMM), which models the probability distribution of the number of building faces as a weighted sum of multiple Gaussian distributions. The goal of the model is to find the best Gaussian distribution parameters (mean, covariance matrix, mixing coefficient) by maximizing the likelihood function of the data, and then predict the number of faces.

[0032] The GMM model can be used to calculate the predicted number of faces of the building to be tested. Specifically, the model will output the probability of the number of faces corresponding to each Gaussian distribution, and finally combine the weight coefficient to obtain the most likely number of faces of the building.

[0033] The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0034] In the training phase, we first need to collect a set of building model data with a known number of faces. This data can be obtained in the following ways: Real building dataset: collects information such as the number of faces and geometric features of completed buildings. This data can come from architectural design companies, urban planning departments, etc.

[0035] Simulated building data: Building data generated by computer-aided design (CAD) systems or 3D modeling tools.

[0036] After obtaining the building model data, it is necessary to extract its features. The specific process is as follows: Geometric feature extraction: including building height, area, number of floors, window ratio, wall shape, etc. These geometric features are extracted based on the floor plan, elevation or 3D model of the building.

[0037] Feature vector construction: The extracted geometric features are converted into feature vectors to form a feature representation of each building. The dimension of the feature vector depends on the number and type of geometric features selected.

[0038] Through feature extraction, building data can be converted into a standardized numerical format, which is convenient for subsequent model training and analysis.

[0039] The building feature vector is input into the GMM face number prediction model and trained using the expectation maximization (EM) algorithm. The specific process is as follows: Initialization: Initialize the model parameters, including the mean (μ), covariance matrix (Σ), and mixing coefficient (π) of each Gaussian distribution. These initial parameters can be obtained by random selection, K-means clustering, or other initialization methods.

[0040] Expectation step (E step): In each iteration, the probability of each data point belonging to each Gaussian distribution is calculated. This probability indicates the degree to which each data point belongs to a Gaussian distribution.

[0041] Maximization step (M step): Update the mean, covariance matrix and mixing coefficient of the Gaussian distribution according to the responsibility probability. The update formula is calculated based on the maximum log-likelihood function to ensure that each update improves the model's fit.

[0042] Iteration: Repeat the E and M steps until the preset training conditions are met. For example, when the change in the log-likelihood function is less than the set threshold, the training process terminates.

[0043] After multiple rounds of EM algorithm iterations, the model parameters will gradually converge, and eventually a GMM face number prediction model that can effectively predict the number of building faces will be obtained.

[0044] The method for predicting the number of building faces provided by the present invention uses the GMM model to model the building geometric feature information, and trains the model through the EM algorithm, so as to accurately predict the number of building faces to be measured. The method combines technical means such as feature extraction, standardization, noise removal and data enhancement of building data, which can effectively improve the accuracy and stability of prediction, and is particularly suitable for processing complex and nonlinear building face prediction problems.

[0045] As a further optional embodiment, after the step of inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested, the method further includes: The predicted number of faces is optimized according to a face number optimization strategy.

[0046] As a further optional embodiment, the step of optimizing the predicted number of faces according to the face number optimization strategy specifically includes: Determining the architectural style category of the building to be tested; According to the architectural style category, matching the corresponding face number optimization strategy; The predicted number of faces is optimized according to the face number optimization strategy.

[0047] In this step, firstly, it is necessary to apply the face number optimization strategy to correct or adjust the predicted face number based on the prediction results and the architectural style category. Specifically, for the building to be tested, its architectural style category is determined. The determination of the architectural style category can be achieved in the following ways: Based on building appearance features: By analyzing the appearance features of the building facade, such as shape, decoration, and building materials, computer vision technology or deep learning models (such as convolutional neural networks) are used to automatically identify the style type of the building, such as modern style, classical style, minimalist style, etc.

[0048] Based on building function type: The style is classified according to the building function type (such as residential, commercial office, industrial plant, etc.). When designing buildings of each functional type, the distribution of the number of faces and the architectural form often follow certain rules.

[0049] Based on architectural history or cultural background: Identify the style type of the building through historical architectural databases or cultural background information, especially suitable for buildings with strong historical and cultural characteristics.

[0050] Once the style category of the building is determined, different facet optimization strategies can be selected based on the characteristics of different architectural styles. Facet optimization strategies mainly adjust the predicted facets based on the understanding of different architectural styles. Common optimization strategies include: Style feature matching: For some typical architectural styles (such as simple modern style or traditional European style), the distribution rules of the number of faces are determined through historical data analysis, and the predicted number of faces is corrected based on these rules. For example, simple modern style buildings are generally high in height and large in area, and the predicted number of faces may be too low, so it is necessary to increase the number of faces through optimization.

[0051] Functional requirement matching: The number of faces is adjusted according to the functional requirements of the building. For example, the number of faces required for commercial office buildings and residential buildings is different. Office buildings may have a larger usable area per floor and may have relatively fewer faces, while residential buildings may have relatively more faces. The number of faces is adjusted according to the predicted architectural style and functional type, combined with the functional requirements.

[0052] Regional or geographical considerations: The geographical location of the building, urban planning, and local building standards will also affect the number of building sides required. In some geographic areas, the number of building sides may be affected by local building codes or environmental factors. For example, in high-density urban centers, the number of building sides is usually higher; while in low-density suburban areas, the number of building sides may be lower.

[0053] As a further optional embodiment, the step of optimizing the predicted number of faces according to the face number optimization strategy is specifically as follows:

[0054] in, is the final predicted number of faces, is the predicted number of faces, The adjustment factor corresponding to the face number optimization strategy.

[0055] In this step, based on the combination of the predicted building surface number and the style optimization factor, the coefficient is adjusted to obtain the final predicted surface number that better meets the actual needs. The specific optimization process can be expressed as the following mathematical formula:

[0056] in, is the final predicted number of faces, is the preliminary predicted number of faces calculated by the GMM face number prediction model, that is, the unoptimized prediction result. The adjustment factor corresponding to the face number optimization strategy is set based on the building's style, functional type, geographical location and other relevant characteristics.

[0057] Firstly, the GMM face prediction model (Gaussian mixture model) is used to analyze and predict the geometric feature information of the building to be tested, and the preliminary face prediction results are obtained.

[0058] Next, the adjustment factor corresponding to the face number optimization strategy is determined based on factors such as the style category, function type, and geographical location of the building to be tested. The adjustment factor is calculated based on the following parameters: Architectural style categories: such as modern style, classical style, simple style, etc. Buildings of different styles usually have different number of facets distributed.

[0059] Building Function Type: Buildings with different functional types such as residential, commercial, industrial, etc. may have different number of sides. For example, residential buildings usually have more sides, while office buildings have fewer sides.

[0060] Geographic and climatic factors: The climate and geography of the region where a building is located may affect the number of sides it is designed for. For example, a building in a tropical region may have more vents, increasing the number of sides of the building.

[0061] Finally, the above formula is used to multiply the initial predicted number of sides by the adjustment factor to obtain the optimized number of building sides.

[0062] As a further optional embodiment, after the step of acquiring the building model data, the method further includes: Preprocessing the building model data, wherein the preprocessing includes removing noise and outliers; Standardizing various features in the building model data so that the building model data are within the same scale range; The building model data is enhanced to obtain a training data set.

[0063] In this step, a series of preprocessing operations are performed on the acquired building model data to ensure the quality and consistency of the data, thereby improving the effect of subsequent model training. The specific preprocessing steps are as follows: In the building model data, there may be noise and outliers introduced due to measurement errors, data collection problems or other reasons. To ensure the accuracy and reliability of the data, these outliers need to be detected and removed. Commonly used denoising and outlier processing methods include: Outlier detection based on statistical methods: such as box plot method and Z-Score method, to identify data points that are significantly deviated from the normal range.

[0064] Model-based outlier detection: Use clustering algorithms (such as K-means or DBSCAN) to identify outliers that are significantly different from other data points.

[0065] Interpolation method: For missing data, linear interpolation or polynomial interpolation methods are used to fill in the missing data to avoid data loss affecting model training.

[0066] Since building model data may contain features of different dimensions and scales, in order to avoid the deviation of the model training process caused by the scale of some features being too large or too small, all features need to be standardized. The purpose of standardization is to convert all feature values ​​into a common scale range. Commonly used standardization methods include: Z-score standardization: By subtracting the mean from the data and dividing it by the standard deviation, the data of each feature is made to follow a standard normal distribution, with a mean of 0 and a variance of 1.

[0067] Minimum-maximum normalization: By scaling the value of each feature, the data range is mapped to between 0 and 1, or any specified interval.

[0068] The standardized data enables the model to better learn the relationship between features and improve the convergence speed and stability of model training.

[0069] In order to avoid overfitting and increase the diversity of the training data set, the building model data can be enhanced. By transforming, rotating, mirroring, and other operations on the original data, new training samples can be generated to expand the training set and improve the generalization ability of the model. Common data enhancement methods include: Geometric transformation: Perform operations such as translation, rotation, and scaling on the building model to generate samples of different positions, angles, and sizes.

[0070] Adding noise: Add a small amount of random noise to the data to simulate changes in the actual environment and increase the robustness of the model.

[0071] Feature transformation: Enhance the expressiveness of data by performing linear or nonlinear transformations on eigenvalues ​​(such as logarithmic transformation, square root transformation, etc.).

[0072] Synthetic sample generation: Using algorithms to generate synthetic samples, such as the SMOTE (Synthetic Minority Over-sampling Technique) method, can generate new samples in the feature space and enhance the diversity of samples.

[0073] As a further optional embodiment, the step of inputting the building feature vector into the GMM face number prediction model and executing the EM algorithm specifically includes: Initialize the model parameters of each Gaussian distribution in the GMM face number prediction model, wherein the model parameters include a mean, a covariance matrix, and a mixing coefficient; Iteratively calculate the responsibility probability of each data point belonging to each Gaussian distribution, and update the mean, covariance matrix and mixing coefficient of the Gaussian distribution until the preset training conditions are met.

[0074] Through the above preprocessing operations, the obtained building model data will be cleaner, more standardized and more diverse, thus forming a training data set for the subsequent GMM face number prediction model training. This training data set can effectively improve the performance of the model and reduce the deviation caused by data quality issues.

[0075] This embodiment not only improves the quality of the data by preprocessing and enhancing the building model data, but also enhances the robustness and generalization ability of the model, which contributes to the accuracy and stability of the final prediction results.

[0076] As a further optional embodiment, the GMM face number prediction model is:

[0077] in, is the probability distribution of the number of building surfaces, It is k The mixing coefficient of the Gaussian distribution satisfies =1 =1, The mean is , the covariance is Gaussian distribution, are the parameters of the model.

[0078] By using GMM (Gaussian mixture model) to predict the number of building faces, multimodal data distribution can be effectively processed. For building model data with multiple modes or distributions, different parts of the data can be fit through multiple Gaussian distributions to improve the prediction ability of the model.

[0079] The prediction device for the number of building surfaces provided by the present invention is described below. Figure 2 As shown, the device for predicting the number of building surfaces described below and the method for predicting the number of building surfaces described above can refer to each other.

[0080] A device for predicting the number of building surfaces, comprising: The information acquisition module 210 is used to acquire geometric feature information of the building to be tested; A face number prediction module 220 is used to input the geometric feature information into a GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0081] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor 310, a communications interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communications interface 320 and the memory 330 communicate with each other via the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the method for predicting the number of building surfaces, and the method includes: Obtain geometric feature information of the building to be tested; Inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0082] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, 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 a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0083] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the method for predicting the number of building surfaces provided by the above methods, the method comprising: Obtain geometric feature information of the building to be tested; Inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0084] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the method for predicting the number of building surfaces provided by the above methods, the method comprising: Obtain geometric feature information of the building to be tested; Inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

[0085] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0086] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the number of building surfaces, characterized in that: include: Obtain geometric feature information of the building to be tested; Inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

2. The method for predicting the number of building surfaces according to claim 1, characterized in that: After the step of inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested, the method further comprises: The predicted number of faces is optimized according to a face number optimization strategy.

3. The method for predicting the number of building surfaces according to claim 2, characterized in that: The step of optimizing the predicted number of faces according to the face number optimization strategy specifically includes: Determining the architectural style category of the building to be tested; According to the architectural style category, matching the corresponding face number optimization strategy; The predicted number of faces is optimized according to the face number optimization strategy.

4. The method for predicting the number of building surfaces according to claim 3, characterized in that: The step of optimizing the predicted number of faces according to the face number optimization strategy is specifically as follows: in, is the final predicted number of faces, is the predicted number of faces, The adjustment factor corresponding to the face number optimization strategy.

5. The method for predicting the number of building surfaces according to claim 1, characterized in that: After the step of acquiring the building model data, the method further includes: Preprocessing the building model data, wherein the preprocessing includes removing noise and outliers; Standardizing various features in the building model data so that the building model data are within the same scale range; The building model data is enhanced to obtain a training data set.

6. The method for predicting the number of building surfaces according to claim 1, characterized in that: The step of inputting the building feature vector into the GMM face number prediction model and executing the EM algorithm specifically includes: Initialize the model parameters of each Gaussian distribution in the GMM face number prediction model, wherein the model parameters include a mean, a covariance matrix, and a mixing coefficient; Iteratively calculate the responsibility probability of each data point belonging to each Gaussian distribution, and update the mean, covariance matrix and mixing coefficient of the Gaussian distribution until the preset training conditions are met.

7. The method for predicting the number of building surfaces according to claim 1, characterized in that: The GMM face number prediction model is: in, is the probability distribution of the number of building surfaces, It is k The mixing coefficient of the Gaussian distribution satisfies =1 =1, The mean is , the covariance is Gaussian distribution, are the parameters of the model.

8. A device for predicting the number of building surfaces, characterized in that: include: An information acquisition module is used to obtain geometric feature information of the building to be tested; A face number prediction module, used for inputting the geometric feature information into the GMM face number prediction model to obtain the predicted face number of the building to be tested; The GMM face number prediction model is trained by the following steps: Obtain building model data; Extracting features from the building model data to obtain a building feature vector; The building feature vector is input into the GMM face number prediction model, and the EM algorithm is executed until the preset training conditions are met.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for predicting the number of building surfaces as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting the number of building surfaces as described in any one of claims 1 to 7 is implemented.

Images