BIM-based construction component image evidence generation and state monitoring method and system

Abstract

The application relates to the field of building engineering informatization and computer vision technology, and discloses a construction component image evidence generation and state monitoring method and system based on BIM, which comprises the following steps: acquiring point clouds, panoramic images, viewpoints and BIM of a region where a target component is located, and establishing a unified coordinate mapping relationship; generating effective viewpoints of the target component based on the unified coordinate mapping relationship, and generating image evidence corresponding to the target component based on the effective viewpoints; inputting the image evidence into a pre-trained AI image classification model to obtain an initial appearance category and an initial confidence degree; aggregating the initial appearance category and the initial confidence degree to obtain an aggregated appearance category and an aggregated confidence degree; obtaining the construction state of the target component according to the aggregated appearance category and the aggregated confidence degree; and inputting the construction state into the BIM to complete the construction state monitoring of the target component. The application can realize component-level construction state monitoring based on BIM.

Description

Technical Field

[0001] This invention relates to the fields of building engineering information technology and computer vision technology, and in particular to a method and system for generating and monitoring the status of construction component image evidence based on BIM. Background Technology

[0002] Building Information Modeling (BIM) is a modeling software used in the fields of architecture, engineering, and civil engineering.

[0003] With the development of smart construction and digital twin technologies, BIM has expanded from its initial design and construction phases to applications such as construction process monitoring, schedule management, and status assessment. Construction component-level schedule monitoring is a crucial basis for construction plan control, quantity calculation, and phased payment verification. Conventional technical approaches include: geometric alignment analysis based on point cloud and BIM, image-based component detection and recognition, and schedule inference methods based on multi-source data fusion.

[0004] However, existing technologies still face several challenges. On the one hand, complex construction environments present problems such as large variations in lighting, severe component occlusion, and changing viewing angles, directly impacting the stable identification of components in images. Without effective constraints affecting viewpoints and component visibility, a large amount of invalid or misleading visual information can be introduced, reducing the reliability of automated monitoring results. On the other hand, existing methods focus on the overall analysis of images or point clouds, failing to fully utilize prior information such as spatial location contained in BIM, making it difficult to achieve truly meaningful component-level status monitoring and refined management. Furthermore, at the image recognition and analysis level, existing construction monitoring methods rely on artificial intelligence models to classify or identify construction components by appearance. However, the lack of effective integration with BIM often makes it difficult to map image-level recognition results to specific BIM component entities, limiting the direct application of results in construction management systems.

[0005] Existing technologies still have limitations in achieving reliable, component-level monitoring and refined management of construction conditions; therefore, existing technologies need further improvement. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a method for generating and monitoring the status of construction components based on BIM, in order to address the shortcomings of the existing technology and solve the problem that the existing technology is difficult to achieve reliable, component-level construction status monitoring and refined management.

[0007] The technical solution adopted by this invention to solve the technical problem is as follows: In a first aspect, the present invention provides a method for generating and monitoring the status of construction component image evidence based on BIM, including: Acquire point cloud, panoramic image, viewpoint and BIM of the area where the target component is located, and establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint and the BIM; Based on the unified coordinate mapping relationship, an effective viewpoint of the target component is generated, and based on the effective viewpoint, image evidence corresponding to the target component is generated. The image evidence is input into a pre-trained AI image classification model to obtain the initial appearance category and initial confidence level. The initial appearance category and the initial confidence level are aggregated to obtain the aggregated appearance category and aggregated confidence level; The construction status of the target component is obtained based on the aggregated appearance category and aggregated confidence level. The construction status is then input into the BIM to complete the monitoring of the construction status of the target component.

[0008] In one implementation, establishing a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM includes: The point cloud is registered with the BIM to obtain the spatial transformation relationship between the measured coordinate system and the BIM coordinate system; Based on the spatial transformation relationship, the point cloud and the viewpoint are transformed into the BIM coordinate system; wherein, the viewpoint is the camera viewpoint pose corresponding to the panoramic image; The unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM is obtained.

[0009] In one implementation, generating an effective viewpoint of the target component based on the unified coordinate mapping relationship, and generating image evidence corresponding to the target component based on the effective viewpoint, includes: Based on the unified coordinate mapping relationship, candidate viewpoints of the target component under BIM constraints are generated. The candidate viewpoints are filtered for visibility to obtain the visible viewpoints; The visible viewpoints are divided into a preset number of observation groups, and a preset number of visible viewpoints are selected from each observation group as valid viewpoints. Based on the effective viewpoint, the target component is projected onto the panoramic image to obtain a two-dimensional projection area; Based on the two-dimensional projection area, the panoramic image is cropped, a mask is generated, and background suppression is performed to obtain image evidence corresponding to the target component.

[0010] In one implementation, the aggregation of the initial appearance category and the initial confidence level to obtain an aggregated appearance category and an aggregated confidence level includes: Obtain the observation correspondence between the observation group, the effective viewpoint, and the image evidence; Based on the observation correspondence, calculate the aggregate score of all initial appearance categories in the same observation group; The largest initial appearance category obtained from the aggregation of each observation group is taken as the aggregated appearance category, and the aggregated confidence level corresponding to the aggregated appearance category is calculated based on the initial confidence level.

[0011] In one implementation, obtaining the construction status of the target component based on the aggregated appearance category and aggregated confidence level includes: Establish a state mapping relationship between appearance category and construction status, and obtain the observation correspondence between the observation group, the effective viewpoint, and the image evidence; Based on the observation correspondence, the aggregated appearance category and aggregated confidence level of each observation group corresponding to the target component are obtained; The observation group is filtered for validity based on a preset aggregation confidence threshold to obtain a valid observation group. The effective observation group is determined based on the state mapping relationship to obtain the construction status of the target component.

[0012] In one implementation, determining the effective observation group based on the state mapping relationship to obtain the construction state of the target component includes: The state determination conditions are obtained based on the state mapping relationship; wherein, the state determination conditions include: reliability verification conditions, pure state determination conditions, intermediate state determination conditions, and fallback determination conditions; The preset state determination conditions are sequentially determined for the effective observation group; When the effective observation group satisfies any one of the state determination conditions, the construction status of the target component corresponding to the state determination condition is output. When the effective observation group does not meet all the state determination conditions, the construction status of the target component is set to "pending verification" and then output.

[0013] In one implementation, inputting the construction status into the BIM to complete the construction status monitoring of the target component includes: Based on the construction status corresponding to the target component, obtain the viewpoint and the aggregate confidence level corresponding to the target component; In the BIM, a structured write-back is performed on the construction status of the target component, the viewpoint, and the aggregate confidence level; Based on the write-back status, the target component is redrawn and a construction report is generated to complete the construction status monitoring of the target component.

[0014] Secondly, the present invention provides a BIM-based system for generating and monitoring the image evidence of construction components, comprising: The data acquisition module is used to acquire point cloud, panoramic image, viewpoint and BIM of the area where the target component is located, and to establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint and the BIM. The image evidence generation module is used to generate an effective viewpoint of the target component based on the unified coordinate mapping relationship, and to generate image evidence corresponding to the target component based on the effective viewpoint. The appearance classification module is used to input the image evidence into a pre-trained AI image classification model to obtain the output initial appearance category and initial confidence level. The appearance aggregation module is used to aggregate the initial appearance category and the initial confidence level to obtain aggregated appearance category and aggregated confidence level; The status monitoring module is used to obtain the construction status of the target component based on the aggregated appearance category and aggregated confidence level, input the construction status into the BIM, and complete the construction status monitoring of the target component.

[0015] Thirdly, the present invention provides a terminal, comprising: a processor and a memory, wherein the memory stores a BIM-based construction component image evidence generation and status monitoring program, wherein when the processor executes the BIM-based construction component image evidence generation and status monitoring program, the BIM-based construction component image evidence generation and status monitoring program is used to implement the operation of the BIM-based construction component image evidence generation and status monitoring method as described in the first aspect.

[0016] Fourthly, the present invention also provides a computer-readable storage medium storing a BIM-based construction component image evidence generation and status monitoring program, wherein the BIM-based construction component image evidence generation and status monitoring program, when executed by a processor, is used to implement the operation of the BIM-based construction component image evidence generation and status monitoring method as described in the first aspect.

[0017] The present invention, by employing the above technical solution, has the following effects: The system acquires point cloud, panoramic image, viewpoint, and BIM data of the area where the target component is located, establishes a unified coordinate mapping relationship between the point cloud, panoramic image, viewpoint, and BIM, and generates effective viewpoints of the target component based on the unified coordinate mapping relationship. This enables the selection of effective observation viewpoints with sufficient information under complex construction site conditions, reduces the impact of obstruction and unfavorable perspectives on the recognition results, and thus improves the stability and consistency of component status monitoring. Based on the effective viewpoint, image evidence corresponding to the target component is generated, which can obtain component-level image evidence that corresponds one-to-one with the target component instance, so that the judgment focuses on the component area, reduces the interference of background and irrelevant objects, and thus improves the accuracy, robustness and traceability of appearance category recognition. The image evidence is input into a pre-trained AI image classification model to obtain the initial appearance category and initial confidence score. The initial appearance category and initial confidence score are aggregated to obtain the aggregated appearance category and aggregated confidence score. The construction status of the target component is obtained based on the aggregated appearance category and aggregated confidence score. The construction status is input into the BIM to complete the monitoring of the construction status of the target component. The component status results can be associated with the BIM model and used for visualization and progress statistical analysis, forming a queryable and traceable engineering management closed loop, improving the implementation and management efficiency of on-site applications. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the method for generating and monitoring the status of construction component images based on BIM in this invention.

[0020] Figure 2 This is a schematic diagram of the structure 21 of the BIM-based construction component image evidence generation and status monitoring system in one implementation of the present invention.

[0021] Figure 3 This is a schematic diagram of the measured data and design data obtained at the construction site in one implementation of the present invention.

[0022] Figure 4 This is a schematic diagram illustrating the data registration of point cloud, panoramic image, and BIM in one implementation of the present invention.

[0023] Figure 5 This is a schematic diagram of the viewpoint filtering process in one implementation of the present invention.

[0024] Figure 6 This is a schematic diagram of the process of component evidence retrieval and projection clipping to generate a mask in one implementation of the present invention.

[0025] Figure 7 This is a schematic diagram of cross-viewpoint aggregation decision-making for observation groups in one implementation of the present invention.

[0026] Figure 8 This is a flowchart of a method for construction status mapping and validity screening in one implementation of the present invention.

[0027] Figure 9 This is a functional schematic diagram of the terminal in one implementation of the present invention.

[0028] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0030] Exemplary methods With the development of smart construction and digital twin technologies, BIM has expanded from its initial design and construction phases to applications such as construction process monitoring, schedule management, and status assessment. Construction component-level schedule monitoring is a crucial basis for construction plan control, quantity calculation, and phased payment verification. Conventional technical approaches include: geometric alignment analysis based on point cloud and BIM, image-based component detection and recognition, and schedule inference methods based on multi-source data fusion.

[0031] However, existing technologies still face several challenges. On the one hand, complex construction environments present problems such as large variations in lighting, severe component occlusion, and changing viewing angles, directly impacting the stable identification of components in images. Without effective constraints affecting viewpoints and component visibility, a large amount of invalid or misleading visual information can be introduced, reducing the reliability of automated monitoring results. On the other hand, existing methods focus on the overall analysis of images or point clouds, failing to fully utilize prior information such as spatial location contained in BIM, making it difficult to achieve truly meaningful component-level status monitoring and refined management. Furthermore, at the image recognition and analysis level, existing construction monitoring methods rely on artificial intelligence models to classify or identify construction components by appearance. However, the lack of effective integration with BIM often makes it difficult to map image-level recognition results to specific BIM component entities, limiting the direct application of results in construction management systems.

[0032] Existing technologies still have limitations in achieving reliable, component-level monitoring and refined management of construction conditions; therefore, existing technologies need further improvement.

[0033] To address the above technical problems, this invention provides a BIM-based method for generating image evidence and monitoring the status of construction components. The method includes: acquiring point cloud data, panoramic images, viewpoints, and BIM data of the area where the target component is located; establishing a unified coordinate mapping relationship between the point cloud data, panoramic images, viewpoints, and BIM data; generating effective viewpoints of the target component based on the unified coordinate mapping relationship; generating image evidence corresponding to the target component based on the effective viewpoints; inputting the image evidence into a pre-trained AI image classification model to obtain an initial appearance category and an initial confidence score; aggregating the initial appearance category and the initial confidence score to obtain an aggregated appearance category and an aggregated confidence score; obtaining the construction status of the target component based on the aggregated appearance category and the aggregated confidence score; inputting the construction status into the BIM data to complete the monitoring of the construction status of the target component. This embodiment enables BIM-based component-level construction status monitoring, reducing manual inspection workload and improving the stability and deployability of component status identification.

[0034] like Figure 1 As shown, this embodiment of the invention provides a method for generating and monitoring the status of construction component image evidence based on BIM, including the following steps: Step S100: Obtain the point cloud, panoramic image, viewpoint, and BIM of the area where the target component is located, and establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM.

[0035] Specifically, in one implementation of this embodiment, step S100 includes the following steps: Step S101: Obtain the point cloud, panoramic image, viewpoint, and BIM of the area where the target component is located.

[0036] In this embodiment, point cloud, panoramic image, viewpoint, and BIM data of the area where the target component is located are acquired. Specifically, measured data and design data of the construction site containing the target component are acquired. Figure 3 The diagram shown is a schematic of the measured data and design data obtained at the construction site in this embodiment. The measured data includes point clouds collected by the mobile mapping system and panoramic image sequences acquired simultaneously, and camera trajectory / viewpoint pose information can be obtained at the same time. The data involved is Building Information Model (BIM), which contains at least the unique identifier of the target component and its three-dimensional geometric information.

[0037] Step S102: Register the point cloud with the BIM to obtain the spatial transformation relationship between the measured coordinate system and the BIM coordinate system.

[0038] It should be noted that the acquired measured data and BIM data are usually in different coordinate systems, such as the measured coordinate system. BIM coordinate system Therefore, it is necessary to establish the same spatial benchmark to achieve cross-data source association.

[0039] In this embodiment, the point cloud is aligned / registered with the BIM to obtain the spatial transformation relationship between the measured coordinate system and the BIM coordinate system.

[0040] Step S103: Based on the spatial transformation relationship, the point cloud and the viewpoint are transformed into the BIM coordinate system; wherein, the viewpoint is the camera viewpoint pose corresponding to the panoramic image.

[0041] In this embodiment, the viewpoint is defined as the camera viewpoint pose corresponding to the panoramic image. Based on the spatial transformation relationship obtained in step S102, the point cloud and the viewpoint are transformed into the BIM coordinate system to achieve data alignment. For example... Figure 4 The diagram shown is a data registration diagram of point cloud, panoramic image and BIM in this embodiment. By converting the point cloud and the viewpoint to the BIM coordinate system, it is used to represent the fusion effect of point cloud and panoramic image after registration in the BIM coordinate system, forming fused data of "point cloud + panoramic image + BIM", realizing the superimposed display and index association in the same coordinate system (i.e. BIM coordinate system).

[0042] Step S104: Obtain the unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM.

[0043] Furthermore, in this embodiment, the viewpoint is the camera viewpoint pose corresponding to the panoramic image, that is, there is a correspondence between the viewpoint and the panoramic image. Based on this, on the basis of step S103, a unified coordinate mapping relationship of "BIM component - point cloud - viewpoint - panoramic image" is established, which provides a basis for subsequent component visibility analysis, image evidence extraction and construction status determination.

[0044] like Figure 1 As shown, this embodiment of the invention provides a method for generating and monitoring the status of construction component image evidence based on BIM, including the following steps: Step S200: Generate an effective viewpoint of the target component based on the unified coordinate mapping relationship, and generate image evidence corresponding to the target component based on the effective viewpoint.

[0045] Specifically, in one implementation of this embodiment, step S200 includes the following steps: Step S201: Based on the unified coordinate mapping relationship, generate candidate viewpoints of the target component under BIM constraints.

[0046] In this embodiment, each panoramic image is associated with the pose of a known viewpoint in the BIM coordinate system based on the same coordinate mapping relationship, thereby enabling the retrieval of stable visual evidence for constructing state recognition under BIM constraints. For example... Figure 5 The diagram shown is a schematic of the viewpoint filtering process in this embodiment. First, candidate viewpoints are generated for the target component in the BIM.

[0047] In this embodiment, the method for generating candidate viewpoints includes: setting a buffer radius based on the outer contour of the component. Remove distance greater than The distant viewpoint is used to avoid insufficient resolution and excessively small observation angle; the observable range of the component, i.e., the field-of-observation angle (FOA), is calculated for the remaining viewpoints. In the horizontal projection plane, the angle between the viewpoint as the vertex and the line connecting the two endpoints of the component boundary as the target is defined as FOA.

[0048] Step S202: Visibility filtering is performed on the candidate viewpoints to obtain visible viewpoints.

[0049] In this embodiment, visibility filtering is performed on candidate viewpoints to obtain visible viewpoints, specifically including: Considering the calculation of the observable angle due to actual construction occlusion, the occlusion angle in the direction of this viewpoint is estimated using the aligned point cloud. (For example, obtained from the occlusion sector formed by occlusion objects in the point cloud along the line of sight), thus obtaining the visible observation angle. ,have: .

[0050] Visibility threshold filtering performs filtering based on a preset visibility threshold, retaining candidate viewpoints that meet the visibility threshold. Examples include: ; in, This indicates that the visible range of the component is insufficient (the field of view is too narrow or severely obstructed). This indicates that the field of view is too wide and the observation is not focused. Optionally, for planar components (such as walls), the filtered viewpoint can be categorized by... Sort from largest to smallest, and select on both sides of the component's normal direction. The largest front Multiple viewpoints are used to improve coverage and robustness. It should be understood that the preset visibility threshold and corresponding conditions are only examples, and single or multiple index combinations based on indicators such as projected area, occlusion ratio, line-of-sight angle, and image sharpness can also be used for screening.

[0051] Step S203: Divide the visible viewpoints into a preset number of observation groups, and select a preset number of visible viewpoints from each observation group as valid viewpoints.

[0052] In addition, to facilitate the subsequent fusion of multi-viewpoint evidence of the same component and to take into account the differences between different observation directions or different visible surfaces, the visible viewpoints can be divided into a preset number of observation groups, and a preset number of visible viewpoints can be selected from each observation group as effective viewpoints.

[0053] Specifically, visible viewpoints are divided into one or more sets according to preset observation conditions. Each set corresponds to a class of similar observation directions / observation surfaces / viewpoint sectors, i.e., an observation group. This observation group is denoted as... ( ).

[0054] In this embodiment, the strategy for dividing observation groups can be determined based on component type, component geometric orientation, and the distribution of accessible viewpoints on site. For example, it can be divided by the two sides of the component's normal direction, by viewpoint azimuth sector, by component key surfaces, or by layering based on the degree of occlusion. For scenarios requiring only unidirectional observation or only able to acquire a single-sided viewpoint, a more suitable strategy can be adopted. And all valid viewpoints are grouped into the same observation group.

[0055] In this embodiment, to reduce subsequent computational load and improve evidence quality, several representative visible viewpoints are selected as effective viewpoints within each observation group to participate in subsequent identification and fusion, for example, according to... Select the top or bottom viewpoint quality indicators after sorting them from high to low. One perspective.

[0056] For example, for planar components, the viewpoints on both sides of the component's normal direction can be assigned to different observation groups, and within each group, the front viewpoints can be selected separately. One viewpoint is selected to improve coverage and robustness; for other different types of components, representative visible viewpoints are selected based on the characteristics of the component.

[0057] Step S204: Project the target component onto the panoramic image based on the effective viewpoint to obtain a two-dimensional projection area.

[0058] In this embodiment, after selecting representative visible viewpoints from the observation group as valid viewpoints, component evidence retrieval and projection clipping are performed, such as... Figure 6 The diagram shown illustrates the process of component evidence retrieval and projection clipping to generate a mask in this embodiment.

[0059] First, the target component is projected onto the panoramic image based on the selected effective viewpoints to obtain a two-dimensional projection area. By reading the panoramic image corresponding to each effective viewpoint, the camera intrinsic parameters corresponding to the effective viewpoint are obtained. Based on the camera intrinsic parameters and the pose of the viewpoint, the geometric boundary or projection contour of the target component is projected onto the panoramic image coordinate system to obtain the two-dimensional projection area of ​​the target component.

[0060] Step S205: Based on the two-dimensional projection area, the panoramic image is cropped, a mask is generated, and background suppression is performed to obtain image evidence corresponding to the target component.

[0061] In this embodiment, as Figure 6 As shown, the panoramic image is cropped, a mask is generated, and background suppression is performed based on the two-dimensional projection area obtained in step S204 to form component-level image evidence.

[0062] In this embodiment, component-level image evidence is preferably bound to component ID and viewpoint ID, and optionally, its corresponding observation group identifier can be recorded. Viewpoint quality indicators (e.g.) (and metadata such as timestamps for tracing, for subsequent status analysis and identification.)

[0063] like Figure 1 As shown, this embodiment of the invention provides a method for generating and monitoring the status of construction component image evidence based on BIM, including the following steps: Step S300: Input the image evidence into the pre-trained AI image classification model to obtain the output initial appearance category and initial confidence level.

[0064] In this embodiment, a pre-trained AI image classification model is used to identify and classify the input component-level image evidence, obtaining the initial appearance category and initial confidence level corresponding to the image evidence. In this embodiment, the AI ​​image classification module can be implemented using convolutional neural networks, visual Transformers, retrieval classifiers, or other equivalent model structures; its output can be discrete category labels, category probability distributions, or confidence scores, and the appearance category system can be predefined according to the engineering scenario, such as appearance subcategories like "not started / wall template erection / wall plastering in progress / wall plastering completed," or more fine-grained appearance categories.

[0065] In one implementation of this embodiment, the pre-trained AI image classification model includes: a preprocessing module, a feature extraction module, a classifier module, and an output management module, which are used for input preprocessing, feature extraction, classification, and output management, respectively.

[0066] In this embodiment, the component-level image evidence formed by the same component under various viewpoints is input into the AI ​​image classification module, and the initial appearance category and initial confidence level corresponding to each viewpoint are output. The initial confidence level can also be expressed as the initial probability.

[0067] Specifically, in one implementation of this embodiment, step S300 includes the following steps: Step S301: The image evidence is preprocessed by the input preprocessing module.

[0068] In this embodiment, the component-level image evidence is preprocessed, including cropping, normalization, and necessary enhancements, to adapt to the model input.

[0069] Step S302: Extract image features of the preprocessed image evidence through the feature extraction module.

[0070] In this embodiment, high-dimensional features of preprocessed component-level image evidence are extracted using convolutional neural networks, visual Transformers, or other equivalent structures.

[0071] Step S303: The image features are classified by the classifier module to obtain the initial appearance category and initial confidence level corresponding to the image evidence.

[0072] In this embodiment, a classifier module is used to classify image features and output the probability distribution or confidence level of each category. The category system can be predefined according to the engineering scenario, such as subcategories like "not started," "wall formwork construction," "wall plastering in progress," and "wall plastering completed."

[0073] Step S304: The output management module outputs the initial appearance category and initial confidence level corresponding to the image evidence.

[0074] In this embodiment, the output management module outputs the initial appearance category and initial confidence level corresponding to the image evidence, and records the component ID and observation surface information. The initial appearance category and initial confidence level are output for multi-viewpoint aggregation and BIM write-back.

[0075] like Figure 1 As shown, this embodiment of the invention provides a method for generating and monitoring the status of construction component image evidence based on BIM, including the following steps: Step S400: Aggregate the initial appearance category and the initial confidence level to obtain aggregated appearance category and aggregated confidence level.

[0076] It should be noted that before aggregating the initial appearance categories and initial confidence scores, the aggregation conditions for the initial appearance categories and initial confidence scores need to be determined. Specifically, by dividing the observation groups, the aggregation of the initial appearance categories and initial confidence scores within the same observation group is achieved.

[0077] In this embodiment, based on the observation group recorded in step S203 ( To accommodate the implementation of "each observation group selects several representative viewpoints", the set of viewpoints within the group participating in the aggregation is defined as follows: ,have: .

[0078] In one implementation of this embodiment, Top-k selection within a group can also be used. For viewpoint quality indicators (e.g.) ) Selected One viewpoint; if no filtering is performed, then take Among them, the symbols This is used to ensure that subsequent aggregation expressions remain consistent across different implementations.

[0079] In this embodiment, the initial appearance category and initial confidence score output in step S300 are classified according to the divided observation groups, or the pre-trained AI image classification model is directly adjusted to output the initial appearance category and initial confidence score according to the divided observation group categories; for any observation group Internal viewpoint The AI ​​image classification module outputs that it belongs to the first... Probability or confidence level of appearance category ,in For appearance category number or category name, For viewpoint numbering, The observation group is labeled.

[0080] Specifically, in one implementation of this embodiment, step S400 includes the following steps: Step S401: Obtain the observation correspondence between the observation group, the effective viewpoint, and the image evidence.

[0081] In this embodiment, the observation correspondence between the observation group, the effective viewpoint, and the image evidence is obtained. The observation group is obtained by dividing the visible viewpoints according to preset observation conditions, while the effective viewpoint is a preset number of representative visible viewpoints selected from the observation group. The image evidence is generated based on the effective viewpoint, thereby obtaining the observation correspondence of "observation group - effective viewpoint - image evidence".

[0082] Step S402: Based on the observation correspondence, calculate the aggregate score of all initial appearance categories in the same observation group.

[0083] In this embodiment, based on the observation correspondence, the aggregated appearance category and aggregated confidence level of each observation group corresponding to the target component are obtained.

[0084] like Figure 7 The diagram shown illustrates cross-viewpoint aggregation decision-making for the observation group in this embodiment. Specifically, to reduce single-viewpoint misjudgments and enhance robustness in occluded scenarios, the same component is aggregated within the same observation group. The results from multiple valid viewpoints within the data are aggregated to construct a category scoring function and make a decision. Preferably, a logarithmic accumulation evidence fusion method is used for any observation group. ,have: ; in, For the observation group Category The aggregate score; For the first The weights of each viewpoint are used to reflect the amount of information at that viewpoint, and are preferably determined by the visible observation angle. Sure( Larger viewpoints take higher weights, which can be denoted as In an unweighted implementation, the preferred approach is... In this case, aggregation is equivalent to logarithmic summation of the probabilities of multiple viewpoints.

[0085] Step S403: Take the largest initial appearance category obtained from the aggregation in each observation group as the aggregated appearance category, and calculate the aggregated confidence level corresponding to the aggregated appearance category based on the initial confidence level.

[0086] In this embodiment, the category with the highest aggregation score in each observation group is taken as the appearance category determination result of that observation group: ; Through the above process, aggregated appearance category results for the same component across one or more observation groups are obtained. And calculate the aggregate confidence level corresponding to the aggregate appearance category based on the initial confidence level. .

[0087] In this embodiment, the aggregate confidence score is obtained by fusing the initial confidence scores of each valid viewpoint within the same observation group for the aggregate appearance category. The fusing calculation includes weighted averaging or normalization processing.

[0088] like Figure 1As shown, this embodiment of the invention provides a method for generating and monitoring the status of construction component image evidence based on BIM, including the following steps: Step S500: Obtain the construction status of the target component based on the aggregated appearance category and aggregated confidence level, input the construction status into the BIM, and complete the construction status monitoring of the target component.

[0089] like Figure 8 The diagram shown is a flowchart of the construction state mapping and validity screening method in this embodiment. First, the aggregated appearance category and aggregated confidence level of each observation group of the target component are input; valid observation groups are screened based on the aggregated confidence level threshold; then, statistics are compiled within the valid observation groups. , , , Subsequently, the effective observation group is determined based on the state mapping relationship to obtain the construction status of the target component.

[0090] Specifically, in one implementation of this embodiment, step S500 includes the following steps: Step S501: Establish a state mapping relationship between appearance category and construction status, and obtain the observation correspondence between the observation group, the effective viewpoint, and the image evidence.

[0091] In this embodiment, a state mapping relationship between appearance categories and construction states is first established. The construction state set in this embodiment includes: a completion state set. During construction and unconstructed collection Appearance categories that cannot be mapped to the above set are denoted as exception categories.

[0092] In this embodiment, the observation correspondence between the observation group, the effective viewpoint, and the image evidence is obtained in the same way as in step S401.

[0093] Step S502: Based on the observation correspondence, obtain the aggregated appearance category and aggregated confidence level of each observation group corresponding to the target component.

[0094] In this embodiment, based on the observation correspondence, the aggregated appearance category and aggregated confidence level of each observation group corresponding to the target component are obtained.

[0095] Step S503: The observation group is screened for validity based on a preset aggregation confidence threshold to obtain a valid observation group.

[0096] Specifically, for the target component, its aggregate confidence threshold in the observation group dimension is read, and the observation group is filtered for validity based on the aggregate confidence threshold, and those that meet the aggregate confidence threshold are included. The set of observations is denoted as the set of valid observations, and the total number of valid observations is denoted as . .

[0097] In addition, statistics were compiled separately within the valid observation group: This is mapped to the number of observation groups in the completed state; This is mapped to the number of observation groups during construction; This is mapped to the number of observation groups that have not yet been constructed; This is mapped to the number of observation groups for anomaly categories.

[0098] Step S504: Based on the state mapping relationship, determine the effective observation group to obtain the construction status of the target component.

[0099] In this embodiment, the effective observation group is determined based on the state mapping relationship to obtain the construction status of the target component, including the following steps: Step S504a: Obtain state determination conditions based on the state mapping relationship; wherein, the state determination conditions include: reliability verification conditions, pure state determination conditions, intermediate state determination conditions, and fallback determination conditions.

[0100] In this embodiment, state determination conditions are obtained based on the state mapping relationship. The state determination conditions include reliability verification conditions, pure state determination conditions, intermediate state determination conditions, and fallback determination conditions.

[0101] In this embodiment, the construction state is defined as follows: Other thresholds and flag values ​​are as follows: The confidence threshold for the observation group aggregation; Minimum number of valid observation groups threshold; The threshold for the proportion of abnormal evidence; The threshold for determining the state of completion can be expressed as an absolute number or a percentage. Key observation coverage indicators, among which This indicates that the key directions / key surfaces are effectively covered by observations. This indicates that key observational evidence is missing.

[0102] In this implementation example, reliability verification conditions are defined based on the aforementioned thresholds and standard quality. These conditions are used to verify the reliability of valid evidence. If a valid observation set meets any of the following conditions, the valid observation set is determined to be "unresolvable," i.e.: ; ; ; When the effective observation set meets any of the above conditions, output This indicates that the quality / coverage of evidence from the effective observation group is insufficient.

[0103] The pure state determination criteria, used to determine whether a valid observation group reaches a consistent conclusion without contradictory evidence, without triggering a reliability check, include: like Then determine ; like Then determine .

[0104] The intermediate state determination condition, used to determine whether the valid observation set satisfies the intermediate state condition when the pure state determination is not hit, is as follows: like Then determine ; like and Then determine .

[0105] The catch-all condition is used when none of the above conditions are met, indicating that the cross-observation group evidence is contradictory or insufficient to form a unique conclusion, and outputs the result. And record the observation groups and their aggregate confidence information corresponding to opposing evidence for manual verification.

[0106] Step S504b: Sequentially determine the preset state determination conditions for the effective observation group.

[0107] In this embodiment, the sequential determination of the preset state determination conditions for the effective observation group is as follows: Figure 8 As shown, starting with reliability verification, each judgment condition is checked item by item; when a certain condition is met, the corresponding construction status is output. The judgment ends; if the condition is not met, the judgment continues to the next condition.

[0108] Step S504c: When the effective observation group satisfies any one of the state determination conditions, the construction state of the target component corresponding to the state determination condition is output.

[0109] In this embodiment, as Figure 8 As shown, when the effective observation group satisfies any one of the state determination conditions, the construction state of the target component corresponding to the state determination condition is output. For example, when the effective observation group satisfies the pure state determination condition, and When, output .

[0110] In step S504d, when the effective observation group does not meet all the state determination conditions, the construction status of the target component is set to "to be verified" and then output.

[0111] In this embodiment, if the effective observation group fails to meet all state determination conditions, the construction status of the target component is set to "pending verification" before being output, i.e., output. And record the observation groups and their aggregate confidence information corresponding to opposing evidence for manual verification.

[0112] Step S505: Based on the construction status corresponding to the target component, obtain the viewpoint and the aggregate confidence level corresponding to the target component.

[0113] In this embodiment, a globally unique identifier is used as a data anchor point to perform structured write-back on the acquired final construction status and its associated metadata. That is, based on the construction status corresponding to the target component, the viewpoint and the aggregate confidence level corresponding to the target component are obtained for structured write-back.

[0114] The metadata includes the determination timestamp, the aggregate confidence level after multi-source evidence aggregation, and the original evidence index (including viewpoint ID sequence and corresponding unstructured image storage path) supporting the determination conclusion.

[0115] Step S506: Perform structured write-back on the construction status, viewpoint, and aggregate confidence level of the target component in the BIM.

[0116] In this embodiment, by establishing database triggers or API interfaces, the dynamic updating of the BIM model property set is realized, ensuring a high degree of synchronization between the digital twin model and the physical entity's progress.

[0117] In addition, to achieve intuitive monitoring of construction progress, a state-color topology mapping mechanism was established. By calling the rendering interface of the BIM engine, the geometry of the components is redrawn based on the written-back state attributes, as follows: Completed: Mapped to Green, indicating that the component forms a consistent conclusion within the effective observation set, satisfying the requirement for closure of the completed state evidence; Partially Complete: Mapped in yellow, indicating that the component has shown characteristics of being completed but has not yet reached the completion threshold, and there is no evidence of construction in progress; Under construction: Mapped to orange, indicating that there is evidence of construction on the component, which is in the process of progress, alternation, or partial implementation stage; Not started: Mapped to red, indicating that the component has not yet triggered installation or the process has not started, and the observational evidence consistently points to no construction; Pending verification: Mapped to gray, indicating that there is conflicting evidence across observation groups or that is insufficient to form a unique conclusion, requiring manual verification; Undecidable: Mapped to dark gray, indicating insufficient valid evidence, missing key observation coverage, or an excessively high proportion of anomalous evidence, resulting in failure to pass reliability verification.

[0118] Step S507: Redraw the target component according to the write-back status and generate a construction report to complete the construction status monitoring of the target component.

[0119] In this embodiment, based on the structured status data after write-back, the system logic layer executes a multi-dimensional calculation algorithm to automatically extract spatial topology information (such as floors and zones) and management attributes (such as construction teams and work sections) from the BIM model. Through aggregate calculation, the system periodically generates multi-granularity construction progress analysis reports, supporting automatic reconciliation of the Percentage of Completion (POC), thereby forming an automated closed-loop management process from on-site perception and status determination to management decision-making.

[0120] For example, in a certain project, the system statistics on the construction status of all “wall plastering” type components located on the 3rd floor, 2nd section, with construction team A and the second section of the process flow on March 6, 2026, calculated that the completion rate of this type of component was 75%.

[0121] This embodiment achieves the following technical effects through the above technical solution: (1) Obtain point cloud, panoramic image, viewpoint and BIM of the area where the target component is located, establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint and the BIM, generate effective viewpoints of the target component based on the unified coordinate mapping relationship, and select effective observation viewpoints with sufficient information under complex construction site conditions, reduce the impact of occlusion and unfavorable viewpoints on the recognition results, thereby improving the stability and consistency of component status monitoring.

[0122] (2) Based on the effective viewpoint, the image evidence corresponding to the target component is generated, which can obtain component-level image evidence that corresponds one-to-one with the target component instance, so that the judgment is focused on the component area, reducing the interference of background and irrelevant objects, thereby improving the accuracy, robustness and traceability of appearance category recognition.

[0123] (3) Perform visibility filtering on the candidate viewpoints to obtain visible viewpoints; The visible viewpoints are divided into a preset number of observation groups, and a preset number of visible viewpoints are selected from each observation group as valid viewpoints. The observation correspondence between the observation group, the valid viewpoints, and the image evidence is obtained. Based on the observation correspondence, the aggregate score of all initial appearance categories in the same observation group is calculated. The initial appearance category with the largest aggregate score in each observation group is taken as the aggregate appearance category, and the aggregate confidence score corresponding to the aggregate appearance category is calculated based on the initial confidence score. This method can fuse the multi-viewpoint recognition results of the same component under the same observation plane, reduce the influence of single-viewpoint misjudgment and occasional noise, output a more robust observation plane-level final appearance category, and support reliable component-level state inference.

[0124] (4) Perform structured write-back on the construction status, viewpoint and aggregate confidence of the target component in the BIM; redraw the target component according to the write-back status and generate construction reports to complete the construction status monitoring of the target component; the component status results can be associated with the BIM model and used for visualization and progress statistical analysis to form a queryable and traceable engineering management closed loop, thereby improving the implementation and management efficiency of on-site applications.

[0125] Exemplary device Based on the above embodiments, the present invention also provides a BIM-based system for generating and monitoring the image evidence of construction components, such as... Figure 2 As shown, the BIM-based construction component image evidence generation and status monitoring system includes: The data acquisition module 21 is used to acquire point cloud, panoramic image, viewpoint and BIM of the area where the target component is located, and to establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint and the BIM. Image evidence generation module 22 is used to generate an effective viewpoint of the target component based on the unified coordinate mapping relationship, and to generate image evidence corresponding to the target component based on the effective viewpoint; The appearance classification module 23 is used to input the image evidence into a pre-trained AI image classification model to obtain the output initial appearance category and initial confidence level. Appearance aggregation module 24 is used to aggregate the initial appearance category and the initial confidence level to obtain aggregated appearance category and aggregated confidence level; The status monitoring module 25 is used to obtain the construction status of the target component based on the aggregated appearance category and aggregated confidence level, input the construction status into the BIM, and complete the construction status monitoring of the target component.

[0126] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 9 As shown.

[0127] The terminal includes: a processor, a memory, an interface, a display screen, and a communication module connected via a system bus; wherein, the processor of the terminal provides computing and control capabilities; the memory of the terminal includes a computer-readable storage medium and internal memory; the computer-readable storage medium stores an operating system and computer programs; the internal memory provides an environment for the operation of the operating system and computer programs in the computer-readable storage medium; the interface is used to connect to external devices; the display screen is used to display relevant information; and the communication module is used to communicate with a cloud server or other devices.

[0128] When executed by the processor, this computer program is used to implement the operation of a BIM-based method for generating image evidence of construction components and monitoring their condition.

[0129] It will be understood by those skilled in the art that Figure 9 The schematic diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the terminal to which the present invention is applied. A specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0130] In one embodiment, a terminal is provided, comprising: a processor and a memory, the memory storing a BIM-based construction component image evidence generation and status monitoring program, which, when executed by the processor, is used to implement the operation of the above-described BIM-based construction component image evidence generation and status monitoring method.

[0131] In one embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a BIM-based construction component image evidence generation and status monitoring program, which, when executed by a processor, is used to implement the operation of the above-described BIM-based construction component image evidence generation and status monitoring method.

[0132] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, database, or other media used in the embodiments provided by this invention can include both non-volatile and volatile memory.

[0133] In summary, this invention provides a method and system for generating image evidence and monitoring the status of construction components based on BIM, comprising: acquiring point cloud, panoramic image, viewpoint, and BIM of the area where the target component is located; establishing a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM; generating effective viewpoints of the target component based on the unified coordinate mapping relationship; generating image evidence corresponding to the target component based on the effective viewpoints; inputting the image evidence into a pre-trained AI image classification model to obtain an initial appearance category and an initial confidence score; aggregating the initial appearance category and the initial confidence score to obtain an aggregated appearance category and an aggregated confidence score; obtaining the construction status of the target component based on the aggregated appearance category and the aggregated confidence score; inputting the construction status into the BIM to complete the construction status monitoring of the target component. This invention can realize component-level construction status monitoring based on BIM, reduce the workload of manual inspection, and improve the stability and project deployability of component status identification.

[0134] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for generating and monitoring the status of construction component image evidence based on BIM, characterized in that, include: Acquire point cloud, panoramic image, viewpoint and BIM of the area where the target component is located, and establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint and the BIM; Based on the unified coordinate mapping relationship, an effective viewpoint of the target component is generated, and based on the effective viewpoint, image evidence corresponding to the target component is generated. The image evidence is input into a pre-trained AI image classification model to obtain the initial appearance category and initial confidence level. The initial appearance category and the initial confidence level are aggregated to obtain the aggregated appearance category and aggregated confidence level; The construction status of the target component is obtained based on the aggregated appearance category and aggregated confidence level. The construction status is then input into the BIM to complete the monitoring of the construction status of the target component.

2. The method for generating and monitoring the status of construction component image evidence based on BIM according to claim 1, characterized in that, Establishing a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM includes: The point cloud is registered with the BIM to obtain the spatial transformation relationship between the measured coordinate system and the BIM coordinate system; Based on the spatial transformation relationship, the point cloud and the viewpoint are transformed into the BIM coordinate system; wherein, the viewpoint is the camera viewpoint pose corresponding to the panoramic image; The unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint, and the BIM is obtained.

3. The method for generating and monitoring the status of construction component image evidence based on BIM according to claim 1, characterized in that, The process of generating an effective viewpoint for the target component based on the unified coordinate mapping relationship, and generating image evidence corresponding to the target component based on the effective viewpoint, includes: Based on the unified coordinate mapping relationship, candidate viewpoints of the target component under BIM constraints are generated. The candidate viewpoints are filtered for visibility to obtain the visible viewpoints; The visible viewpoints are divided into a preset number of observation groups, and a preset number of visible viewpoints are selected from each observation group as valid viewpoints. Based on the effective viewpoint, the target component is projected onto the panoramic image to obtain a two-dimensional projection area; Based on the two-dimensional projection area, the panoramic image is cropped, a mask is generated, and background suppression is performed to obtain image evidence corresponding to the target component.

4. The method for generating and monitoring the status of construction component image evidence based on BIM according to claim 3, characterized in that, The aggregation of the initial appearance category and the initial confidence level to obtain the aggregated appearance category and aggregated confidence level includes: Obtain the observation correspondence between the observation group, the effective viewpoint, and the image evidence; Based on the observation correspondence, calculate the aggregate score of all initial appearance categories in the same observation group; The largest initial appearance category obtained from the aggregation of each observation group is taken as the aggregated appearance category, and the aggregated confidence level corresponding to the aggregated appearance category is calculated based on the initial confidence level.

5. The method for generating and monitoring the status of construction component image evidence based on BIM according to claim 3, characterized in that, The step of obtaining the construction status of the target component based on the aggregated appearance category and aggregated confidence level includes: Establish a state mapping relationship between appearance category and construction status, and obtain the observation correspondence between the observation group, the effective viewpoint, and the image evidence; Based on the observation correspondence, the aggregated appearance category and aggregated confidence level of each observation group corresponding to the target component are obtained; The observation group is filtered for validity based on a preset aggregation confidence threshold to obtain a valid observation group. The effective observation group is determined based on the state mapping relationship to obtain the construction status of the target component.

6. The method for generating and monitoring the status of construction component image evidence based on BIM according to claim 5, characterized in that, The step of determining the effective observation group based on the state mapping relationship to obtain the construction status of the target component includes: The state determination conditions are obtained based on the state mapping relationship; wherein, the state determination conditions include: reliability verification conditions, pure state determination conditions, intermediate state determination conditions, and fallback determination conditions; The preset state determination conditions are sequentially determined for the effective observation group; When the effective observation group satisfies any one of the state determination conditions, the construction status of the target component corresponding to the state determination condition is output. When the effective observation group does not meet all the state determination conditions, the construction status of the target component is set to "pending verification" and then output.

7. The method for generating and monitoring the status of construction component image evidence based on BIM according to claim 1, characterized in that, The step of inputting the construction status into the BIM to complete the construction status monitoring of the target component includes: Based on the construction status corresponding to the target component, obtain the viewpoint and the aggregate confidence level corresponding to the target component; In the BIM, a structured write-back is performed on the construction status of the target component, the viewpoint, and the aggregate confidence level; Based on the write-back status, the target component is redrawn and a construction report is generated to complete the construction status monitoring of the target component.

8. A BIM-based system for generating and monitoring the status of construction component image evidence, used to implement the BIM-based method for generating and monitoring the status of construction component image evidence as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to acquire point cloud, panoramic image, viewpoint and BIM of the area where the target component is located, and to establish a unified coordinate mapping relationship between the point cloud, the panoramic image, the viewpoint and the BIM. The image evidence generation module is used to generate an effective viewpoint of the target component based on the unified coordinate mapping relationship, and to generate image evidence corresponding to the target component based on the effective viewpoint. The appearance classification module is used to input the image evidence into a pre-trained AI image classification model to obtain the output initial appearance category and initial confidence level. The appearance aggregation module is used to aggregate the initial appearance category and the initial confidence level to obtain aggregated appearance category and aggregated confidence level; The status monitoring module is used to obtain the construction status of the target component based on the aggregated appearance category and aggregated confidence level, input the construction status into the BIM, and complete the construction status monitoring of the target component.

9. A terminal, characterized in that, include: The processor and memory, wherein the memory stores a BIM-based construction component image evidence generation and status monitoring program, which, when executed by the processor, is used to implement the operation of the BIM-based construction component image evidence generation and status monitoring method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a BIM-based construction component image evidence generation and status monitoring program, which, when executed by a processor, is used to implement the operation of the BIM-based construction component image evidence generation and status monitoring method as described in any one of claims 1-7.

Images