Building construction progress intelligent monitoring method and device and electronic equipment

Abstract

The invention discloses a building construction progress intelligent monitoring method and device and electronic equipment, and belongs to the technical field of intelligent monitoring, and the method comprises the steps: comparing a label of a building construction image with a building information model library, and obtaining a two-dimensional visual angle image data set; performing semantic segmentation on a bounding box of the building construction image to obtain a segmentation mask set; calculating the cosine similarity of the two-dimensional visual angle image data set and the segmentation mask set according to the feature vector cosine similarity, and determining the angle corresponding to the maximum value of the cosine similarity as the best matching angle; based on the optimal matching angle, a two-dimensional optimal matching view angle image and an optimal matching mask are determined, and the construction progress of the building construction image is determined; semantic segmentation is carried out on the bounding box, and parts related to the building are extracted in a targeted mode; and by calculating the cosine similarity between the two-dimensional visual angle image data set and the segmentation mask set, similarity matching is carried out on the building construction image and the building information model library, so that the monitoring accuracy is further ensured.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring technology, and in particular to a method, device and electronic equipment for intelligent monitoring of construction progress. Background Art

[0002] In the construction industry, accurate monitoring and prediction of construction progress is crucial to project management. Through real-time monitoring, safety hazards in the construction process, such as structural deformation and cracks, can be discovered in a timely manner, so that timely measures can be taken to repair and adjust to avoid accidents. This can not only protect the lives of construction workers, but also prevent construction delays and cost increases caused by construction accidents.

[0003] However, traditional progress monitoring methods usually rely on manual recording and supervision, which is not only time-consuming but also prone to data lag and subjective errors. In addition, the construction site environment is complex, and it is difficult to obtain accurate progress data in real time using manual methods.

[0004] Therefore, in the process of monitoring the progress of construction in the prior art, there are problems of low monitoring efficiency and low monitoring accuracy. Summary of the invention

[0005] In view of this, it is necessary to provide a method, device and electronic equipment for intelligent monitoring of construction progress to solve the problems of low monitoring efficiency and low monitoring accuracy in the process of monitoring construction progress in the prior art.

[0006] In order to solve the above problems, the present invention provides a method for intelligent monitoring of construction progress, comprising: Compare the labels of the building construction images with the building information model library to obtain a two-dimensional perspective image dataset; Perform semantic segmentation on the bounding box of the building construction image to obtain a segmentation mask set; The cosine similarity of the two-dimensional view image dataset and the segmentation mask set is calculated according to the cosine similarity of the feature vectors, and the angle corresponding to the maximum value of the cosine similarity is determined as the best matching angle; Based on the best matching angle, a two-dimensional best matching view image and a best matching mask are determined respectively, and the construction progress of the building construction image is determined.

[0007] In a possible implementation, before comparing the label of the building construction image with the building information model library, the method further includes: Identify unit projects in construction images through the target detection model, detect the building elements of the unit projects, and obtain labels; Identify the location of the label in the building construction image and get the bounding box.

[0008] In a possible implementation, the labels of the building construction images are compared with the building information model library to obtain a two-dimensional perspective image dataset, including: Based on the label-model mapping rule, the label is mapped and matched with the building unit model name of the building information model library to obtain a three-dimensional matching model of the label; The 3D matching model is sampled at all angles to obtain a 2D perspective image dataset.

[0009] In a possible implementation, semantic segmentation is performed on the bounding box of the building construction image to obtain a segmentation mask set, including: Divide the pixels in the bounding box into regions to obtain multiple target building unit regions; The target building unit area is semantically segmented according to the image segmentation model, the target building unit contour is extracted, and the segmentation mask set is obtained.

[0010] In a possible implementation, after semantic segmentation is performed on the target building unit area according to the image segmentation model, the target building unit contour is extracted to obtain a segmentation mask set, the following further includes: The segmentation mask area of ​​the segmentation mask set is divided by the bounding box area to obtain the area ratio; The segmentation masks whose area ratios are within a preset ratio threshold range are selected as target segmentation masks.

[0011] In a possible implementation, before calculating the cosine similarity of the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors and determining the angle corresponding to the maximum cosine similarity as the best matching angle, the method further includes: Generate full-angle sample images of each building unit based on the building information model library; The full-angle sample image is rotated three-dimensionally to obtain a two-dimensional perspective image dataset.

[0012] In a possible implementation, the cosine similarity of the two-dimensional view image data set and the segmentation mask set is calculated according to the cosine similarity of the feature vectors, and the angle corresponding to the maximum value of the cosine similarity is determined as the most matching angle, including: A first high-dimensional mapping feature matrix of the segmentation mask set and a second high-dimensional mapping feature matrix of the two-dimensional view image data set are respectively extracted through a convolutional neural network model; The cosine similarity of the first high-dimensional mapping feature matrix and the second high-dimensional mapping feature matrix is ​​determined according to the cosine similarity calculation formula, and the angle corresponding to the maximum value of the cosine similarity is determined as the most matching angle.

[0013] In a possible implementation, based on the best matching angle, a two-dimensional best matching view image and a best matching mask are determined respectively, and a construction progress of the building construction image is determined, including: Determine the two-dimensional perspective image data corresponding to the best matching angle in the two-dimensional perspective image data set as the two-dimensional best matching perspective image; Determine the best matching mask of the two-dimensional best matching view image in the segmentation mask set by a feature point matching algorithm; The area comparison between the best matching mask and the two-dimensional best matching view image is performed to determine the construction progress of the building construction image.

[0014] In order to solve the above problems, the present invention also provides a construction progress intelligent monitoring device, comprising: A two-dimensional perspective image data set acquisition module is used to compare the labels of the building construction images with the building information model library to obtain a two-dimensional perspective image data set; A segmentation mask set acquisition module is used to perform semantic segmentation on the bounding box of the building construction image to obtain a segmentation mask set; A best matching angle determination module is used to calculate the cosine similarity of the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determine the angle corresponding to the maximum value of the cosine similarity as the best matching angle; The construction progress intelligent monitoring module is used to determine the two-dimensional best-matching view image and the best-matching mask based on the best-matching angle, and determine the construction progress of the building construction image.

[0015] In order to solve the above problem, the present invention further provides an electronic device, including a memory and a processor, wherein: Memory, used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the intelligent monitoring method for construction progress as described above.

[0016] The beneficial effects of adopting the above embodiments are as follows: the present invention provides a method, device and electronic equipment for intelligent monitoring of construction progress, which realizes intelligent monitoring of construction progress by converting the problem of monitoring construction progress into a data comparison problem of construction images, thereby greatly improving the monitoring efficiency; by semantically segmenting the boundary box of the construction image, the part related to the building in the image can be extracted in a targeted manner; by calculating the cosine similarity of the two-dimensional perspective image data set and the segmentation mask set through the cosine similarity of the feature vector, the current construction image is matched with the data in the building information model library for similarity, thereby determining the most matching angle, further ensuring the accuracy of image recognition, and then ensuring the accuracy of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a flow chart of an embodiment of a method for intelligent monitoring of construction progress provided by the present invention; Figure 2A schematic diagram of a flow chart of an embodiment of obtaining a segmentation mask set provided by the present invention; Figure 3 A schematic diagram of a flow chart of an embodiment of determining the best matching angle provided by the present invention; Figure 4 A schematic diagram of a flow chart of an embodiment of determining the construction progress of a building construction image provided by the present invention; Figure 5 A schematic diagram of a flow chart of another embodiment of the intelligent monitoring method for determining construction progress provided by the present invention; Figure 6 A structural block diagram of an embodiment of an intelligent monitoring device for construction progress provided by the present invention; Figure 7 This is a structural block diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0018] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0019] Monitoring data can provide a basis for optimizing construction plans. By comparing the actual construction progress with the planned progress, analyzing the reasons and taking corresponding measures, the construction plan can be optimized and construction efficiency can be improved. For example, the construction sequence, construction method or construction resources can be adjusted according to the monitoring data to ensure that the construction progress is carried out as planned.

[0020] The existing technology mainly includes two methods: one is the progress monitoring method based on manual records, and the other is the construction status analysis based on simple image recognition. However, the traditional method is difficult to handle the complex environment and multi-angle changes of the construction site, while the image recognition method is difficult to accurately segment the building units and automatically generate the progress. Therefore, in the process of monitoring the progress of construction in the existing technology, there are problems of low monitoring efficiency and low monitoring accuracy.

[0021] In order to solve the above problems, the present invention provides a method, device and electronic equipment for intelligent monitoring of construction progress, which are described in detail below.

[0022] like Figure 1 As shown, Figure 1 This is a flow chart of an embodiment of a method for intelligently monitoring construction progress provided by the present invention. The method for intelligently monitoring construction progress includes: S101: Compare the label of the building construction image with the building information model library to obtain a two-dimensional perspective image dataset; S102: performing semantic segmentation on the bounding box of the building construction image to obtain a segmentation mask set; S103: Calculating the cosine similarity of the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determining the angle corresponding to the maximum value of the cosine similarity as the most matching angle; S104: Based on the best matching angle, determine the two-dimensional best matching view image and the best matching mask respectively, and determine the construction progress of the building construction image.

[0023] In this embodiment, by converting the problem of monitoring the progress of construction into a problem of comparing data of construction images, intelligent monitoring of the progress of construction is achieved, which greatly improves the monitoring efficiency; by semantically segmenting the bounding box of the construction image, the part of the image related to the building can be extracted in a targeted manner; by calculating the cosine similarity of the two-dimensional perspective image data set and the segmentation mask set through the cosine similarity of the feature vector, the current construction image is matched with the data in the building information model library for similarity, thereby determining the most matching angle, further ensuring the accuracy of image recognition, and then ensuring the accuracy of monitoring.

[0024] It should be noted that the building information model library is an existing database that can be directly obtained by consulting building-related materials and will not be elaborated here.

[0025] As a preferred embodiment, in S101, before comparing the label of the building construction image with the building information model library, it is necessary to perform preliminary processing on the building construction image to obtain the label and boundary box of the building construction image.

[0026] Specifically, the unit project of the construction image is identified through the target detection model, and the construction elements of the unit project are detected to obtain a label; the position of the label in the construction image is identified to obtain a bounding box.

[0027] It should be noted that the object detection model takes an image as input and then outputs the bounding box coordinates of the detected objects and the labels identifying these objects. An image may contain multiple objects, each with its own bounding box and label. The basic principle of the object detection algorithm is to use a deep neural network to extract features from image or video data, and apply classifiers and bounding box regressors on the extracted features to identify and locate objects.

[0028] In one embodiment, the building construction image The target detection model is used to obtain the labels of the main building elements in the unit project. The building elements in each picture are given multiple labels through the target detection model. , these labels are unit projects identified by the model, such as raw material mill plant, comprehensive office building, etc. Bounding box The position of each label in the image is identified as the input for the subsequent SAM model semantic segmentation.

[0029] For a given building construction image , the target detection model output is as follows:

[0030] in, are multiple labels obtained by the target detection model for the building elements in each picture. are the multiple bounding boxes of the building elements in each picture obtained by the target detection model, A sequence representing the architectural elements in each image, Represents the input image The one-to-one combination of the label and bounding box output by the target detection model.

[0031] By identifying labels, building construction images can be associated with standardized model libraries, providing an information basis for subsequent matching and comparison.

[0032] It should be noted that the SAM model, namely Segment Anything Model, is a large image segmentation model proposed by Meta. The SAM model is a hint model, and its architecture mainly consists of three parts: image encoder, hint encoder and mask decoder.

[0033] Furthermore, in order to compare the labels of construction images with the building information model library and obtain a two-dimensional perspective image dataset, firstly, based on the label-model mapping rule, the labels are mapped and matched with the building unit model names of the building information model library to obtain a three-dimensional matching model of the labels; then, the three-dimensional matching model is sampled at all angles to obtain a two-dimensional perspective image dataset.

[0034] In one embodiment, the detected tags Match the model names of the corresponding building units in the building information model library, and associate the detected building element labels with the standard 3D models in the building information model library through the label-model mapping rules to ensure that standard engineering sample images are used for subsequent analysis.

[0035] Given a label , search for matching models in the model library , the formula is as follows:

[0036] in, Represents the building information model library with tags The corresponding three-dimensional building information model.

[0037] By matching tags, we can ensure that the selected building information model is consistent with the current construction unit, thus achieving accurate comparative analysis.

[0038] It should be noted that Building Information Modeling (BIM) is a digital building information modeling technology. The BIM model uses the relevant information data of the construction project as the basis of the model to establish the building model, and simulates the real information of the building through digital information simulation. It integrates non-geometric information such as geometric information, functional requirements, performance attributes, etc. of the construction project to form a complete building information database. This database can provide decision support for various stages such as design, construction, and operation throughout the life cycle of the construction project.

[0039] As a preferred embodiment, in S102, in order to perform semantic segmentation on the bounding box of the building construction image, a segmentation mask set is obtained, such as Figure 2 As shown, Figure 2 A schematic diagram of a flow chart of an embodiment of obtaining a segmentation mask set provided by the present invention includes: S201: Divide the pixels in the boundary box into regions to obtain multiple target building unit regions; S202: Perform semantic segmentation on the target building unit area according to the image segmentation model, extract the target building unit contour, and obtain a segmentation mask set.

[0040] In this embodiment, by first dividing the bounding box into regions, the building units within the bounding box are refined and identified, and multiple target building unit regions are obtained; by performing zero-sample semantic segmentation on the target building unit regions, the target building unit contours are extracted, making the building unit contour segmentation map more accurate, that is, the reliability of the segmentation mask set is higher.

[0041] In a specific embodiment, the bounding box is set as a rectangular box area, and the rectangular box area obtained by target detection is semantically segmented using a SAM model. The SAM model further divides the area based on the pixel distribution within the bounding box to extract the specific outline of the target building unit.

[0042] For each detection area , perform semantic segmentation through the SAM model and output a multi-scale segmentation mask set :

[0043] Among them, SAM stands for SAM model, Represents from the image The bounding box area extracted from , Representatives will take pictures Each bounding box The set of segmentation masks obtained after input into the SAM model.

[0044] As a preferred embodiment, in S202, after obtaining the segmentation mask set, in order to improve the reliability of the segmentation mask, it is also necessary to take the quotient of the segmentation mask area of ​​the segmentation mask set and the bounding box area to obtain the area ratio; then, the segmentation mask whose area ratio is within the preset ratio threshold range is screened out as the target segmentation mask.

[0045] In a specific embodiment, the area ratio of the segmented area is calculated, and a threshold is applied to ensure that the area of ​​the segmented result meets the preset ratio requirement. Area ratio calculation: Calculate the ratio of the segmented area area to the detection area area :

[0046] in, Representatives will take pictures Each bounding box The area of ​​the segmentation mask set obtained after input into the SAM model, Representative pictures Each bounding box The area of Represents the area and bounding box of the segmentation mask set The ratio of the area of Represents the ratio threshold set to control screening.

[0047] Set a larger threshold , the default value is (0.8, 1), to control the filtering conditions. If the area ratio meets If the following conditions are met, the segmentation result is retained. If not, the segmented area is considered to not meet the expected ratio requirements and needs to be discarded or further processed. Indicates that from the image The bounding box area extracted from .

[0048] After semantic segmentation, the area ratio of the segmentation mask is calculated To filter out the segmented areas that meet the size requirements. This ensures that the segmented building units will not be too small or too large to affect the subsequent image matching process.

[0049] In another specific embodiment, the preset ratio threshold range can also be adjusted, preferably set to (0.5, 0.8). Generally, the threshold The larger the value is, the higher the overlap between the area of ​​the expected segmentation mask and the area of ​​the bounding box is. However, there may also be problems with unclear image recognition, and a smaller ratio threshold range needs to be set. Therefore, the ratio threshold range can be adjusted according to actual needs and is not limited here.

[0050] As a preferred embodiment, in S103, before calculating the cosine similarity of the two-dimensional view image dataset and the segmentation mask set according to the cosine similarity of the feature vectors and determining the angle corresponding to the maximum cosine similarity as the most matching angle, it is necessary to obtain the two-dimensional view image dataset.

[0051] Specifically, firstly, a full-angle sample image of each building unit is generated based on a building information model library; then, the full-angle sample image is three-dimensionally rotated to obtain a two-dimensional perspective image dataset.

[0052] In this embodiment, a full-angle sample image is generated for each building unit by using the BIM model library to match the on-site image. A full-angle image dataset is generated by three-dimensional rotation, and the standard model of each unit project is Perform full-angle sampling to ensure the generation of a two-dimensional image dataset Full angle, even segmentation, and accurate positioning ensure that each building unit has complete viewing angle coverage, making it easier to compare on-site images with models.

[0053]

[0054] in, represents the horizontal rotation angle (i.e. azimuth, usually rotated in the horizontal plane), Indicates the vertical rotation angle (i.e. pitch angle, usually rotated in the vertical direction).

[0055] Representative label The corresponding two-dimensional perspective image dataset, Label The standard three-dimensional model of the corresponding unit project.

[0056] It should be noted that the building unit can be adjusted according to actual needs. It can be a larger building, such as a house, a playground, a lawn, etc., or a smaller and more trivial building, such as a flower bed, a corridor, a stone bench, etc., without limitation here.

[0057] Furthermore, in order to determine the best matching angle, Figure 3 As shown, Figure 3 A schematic diagram of a process for determining the best matching angle according to an embodiment of the present invention includes: S301: extracting a first high-dimensional mapping feature matrix of a segmentation mask set and a second high-dimensional mapping feature matrix of a two-dimensional view image data set respectively through a convolutional neural network model; S302: Determine the cosine similarity of the first high-dimensional mapping feature matrix and the second high-dimensional mapping feature matrix according to a cosine similarity calculation formula, and determine the angle corresponding to the maximum value of the cosine similarity as the most matching angle.

[0058] In one embodiment, after segmentation, the mask and sample image sets Perform CNN feature extraction and cosine similarity calculation to extract the CNN feature vectors of the segmented area and the sample image, and select the closest standard perspective image through cosine similarity. This perspective represents the "same-angle simulation image" that is closest to the on-site image. The formula is as follows:

[0059] represent Model, Representative will label Corresponding 3D model by The horizontal rotation angle and The vertical rotation angle is used to obtain the two-dimensional graphics. Represents the mask after segmentation pass The first high-dimensional mapping feature matrix obtained by the model, Represents a 2D view image dataset pass The second highest dimensional mapping feature matrix obtained by the model.

[0060] Cosine similarity The calculation formula is as follows:

[0061] represent The cosine similarity of is the best matching angle, It is to find the norm.

[0062] In this embodiment, by performing cosine similarity calculation on the first high-dimensional mapping feature matrix and the second high-dimensional mapping feature matrix, the standard angle view that is most similar to the building construction image can be determined.

[0063] As a preferred embodiment, in S104, in order to determine the two-dimensional best matching view image and the best matching mask based on the best matching angle, and determine the construction progress of the building construction image, such as Figure 4 As shown, Figure 4 A schematic diagram of a flow chart of an embodiment of determining the construction progress of a building construction image provided by the present invention includes: S401: Determine that the two-dimensional perspective image data corresponding to the most matching angle in the two-dimensional perspective image data set is the two-dimensional most matching perspective image; S402: Determine the best matching mask of the two-dimensional best matching view image in the segmentation mask set by using a feature point matching algorithm; S403: performing area comparison on the best matching mask and the two-dimensional best matching view angle image to determine the construction progress of the building construction image.

[0064] In a specific embodiment, a feature point matching algorithm is used to identify the features of the completed area and estimate the construction progress. The SIFT algorithm is used to match the feature points to calculate the mask. , the formula is as follows:

[0065] in, Representative will label Corresponding 3D model The graphics obtained at the best matching angle, Representatives will take pictures Each bounding box The segmentation mask set obtained after input into the SAM model, The mask obtained after the feature point matching algorithm.

[0066] The calculation value of the completion rate of the construction period is as follows:

[0067] in, Represents area, It means that the graphics are passed through the SAM model to obtain the mask with the largest area. Representative will label Corresponding 3D model The image obtained at the best matching angle.

[0068] In this embodiment, a mask is generated by matching feature points. , marking the completed building unit area. The overlap between the position of the mask and the actual completed area represents the completed area of ​​the unit project. By comparing the mask map with the model, the ratio of the completed part to the entire construction unit can be calculated, thereby obtaining an estimate of the construction progress. By associating the completion of the building unit with the spatial progress, a quantitative analysis of the current construction period status is formed.

[0069] In summary, this application proposes an intelligent prediction method for the construction unit project duration by combining deep learning, SAM large model, BIM model and image feature analysis. Figure 5 As shown, Figure 5 A flow chart of another embodiment of the intelligent monitoring method for determining the progress of construction provided by the present invention, first, the unit projects in the construction site image are marked by the target detection model, and the SAM semantic segmentation model is used to perform refined cutting. Then, a multi-angle model image data set for each unit project is constructed through the BIM three-dimensional model library, and the cosine similarity is used to analyze the matching degree between the segmented image and the model image, and the most similar angle image is selected. Finally, based on the feature point matching algorithm, a mask of the completed part is generated to estimate the progress of the construction period. This method can provide more accurate construction period prediction in a complex construction environment, which not only significantly improves the accuracy of monitoring, but also reduces the errors caused by human factors, and provides an efficient and reliable intelligent solution for the progress management of engineering projects.

[0070] Through the above method, the problem of monitoring the progress of construction is transformed into a data comparison problem of construction images, which realizes intelligent monitoring of construction progress and greatly improves the monitoring efficiency; by semantically segmenting the bounding box of the construction image, the part related to the building in the image can be extracted in a targeted manner; by calculating the cosine similarity of the two-dimensional view image data set and the segmentation mask set through the cosine similarity of the feature vector, it is possible to match the current construction image with the data in the building information model library for similarity, thereby determining the most matching angle, further ensuring the accuracy of image recognition, and then ensuring the accuracy of monitoring.

[0071] In order to solve the above problems, the present invention also provides a construction progress intelligent monitoring device, such as Figure 6 As shown, Figure 6 This is a structural block diagram of an embodiment of an intelligent monitoring device for construction progress provided by the present invention. The intelligent monitoring device for construction progress 600 includes: A two-dimensional perspective image data set acquisition module 601 is used to compare the label of the building construction image with the building information model library to obtain a two-dimensional perspective image data set; The segmentation mask set acquisition module 602 is used to perform semantic segmentation on the boundary box of the building construction image to obtain a segmentation mask set; The best matching angle determination module 603 is used to calculate the cosine similarity between the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determine the angle corresponding to the maximum value of the cosine similarity as the best matching angle; The construction progress intelligent monitoring module 604 is used to determine the two-dimensional best-matching view image and the best-matching mask based on the best-matching angle, and determine the construction progress of the building construction image.

[0072] like Figure 7As shown, the present invention also provides an electronic device 700. The electronic device 700 includes a processor 701, a memory 702 and a display 703. Figure 7 Only some components of the electronic device 700 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0073] In some embodiments, the processor 701 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 702, such as the intelligent monitoring method for construction progress in the present invention.

[0074] In some embodiments, the processor 701 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, the processor 701 may be local or remote. In some embodiments, the processor 701 may be implemented in a cloud platform. In one embodiment, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-cloud, etc., or any combination thereof.

[0075] In some embodiments, the memory 702 may be an internal storage unit of the electronic device 700, such as a hard disk or memory of the electronic device 700. In other embodiments, the memory 702 may also be an external storage device of the electronic device 700, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the electronic device 700.

[0076] Furthermore, the memory 702 may include both an internal storage unit of the electronic device 700 and an external storage device. The memory 702 is used to store application software installed in the electronic device 700 and various data.

[0077] In some embodiments, the display 703 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 703 is used to display information of the electronic device 700 and to display a visual user interface. The components 701-703 of the electronic device 700 communicate with each other via a system bus.

[0078] In one embodiment, when the processor 701 executes the building construction progress intelligent monitoring program in the memory 702, the following steps may be implemented: Compare the building construction images with the building information model library to obtain a two-dimensional perspective image dataset; Perform semantic segmentation on the bounding box of the building construction image to obtain a segmentation mask set; The cosine similarity of the two-dimensional view image dataset and the segmentation mask set is calculated according to the cosine similarity of the feature vectors, and the angle corresponding to the maximum value of the cosine similarity is determined as the best matching angle; Based on the best matching angle, a two-dimensional best matching view image and a best matching mask are determined respectively, and the construction progress of the building construction image is determined.

[0079] It should be understood that: when the processor 701 executes the building construction progress intelligent monitoring program in the memory 702, in addition to the above functions, other functions can also be realized. For details, please refer to the description of the corresponding method embodiment above.

[0080] Furthermore, the embodiment of the present invention does not specifically limit the type of the electronic device 700 mentioned, and the electronic device 700 may be a portable electronic device such as a mobile phone, a tablet computer, a personal digital assistant (PDA), a wearable device, a laptop computer, etc. Exemplary embodiments of portable electronic devices include but are not limited to portable electronic devices equipped with IOS, Android, Microsoft or other operating systems. The above-mentioned portable electronic device may also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device 700 may not be a portable electronic device, but a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0081] Correspondingly, an embodiment of the present invention also provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the program or instructions are executed by a processor, it can implement the steps or functions of the intelligent monitoring method of construction progress provided by the above-mentioned method embodiments.

[0082] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0083] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for intelligent monitoring of construction progress, characterized in that: include: Compare the labels of the building construction images with the building information model library to obtain a two-dimensional perspective image dataset; Performing semantic segmentation on the bounding box of the building construction image to obtain a segmentation mask set; Calculating the cosine similarity of the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determining the angle corresponding to the maximum value of the cosine similarity as the most matching angle; Based on the best matching angle, a two-dimensional best matching viewing angle image and a best matching mask are determined respectively, and a construction progress of the building construction image is determined.

2. The intelligent monitoring method for construction progress according to claim 1 is characterized in that: In comparing the labeling of building construction images with the Building Information Model library, previously also included: Identifying the unit project of the construction image through the target detection model, and detecting the construction elements of the unit project to obtain the label; The position of the label in the building construction image is identified to obtain the bounding box.

3. The intelligent monitoring method for construction progress according to claim 1 is characterized in that: The step of comparing the labels of the building construction images with the building information model library to obtain a two-dimensional perspective image dataset includes: Based on the label-model mapping rule, the label is mapped and matched with the building unit model name of the building information model library to obtain a three-dimensional matching model of the label; The three-dimensional matching model is sampled at all angles to obtain a two-dimensional viewing angle image data set.

4. The intelligent monitoring method for construction progress according to claim 1 is characterized in that: The semantic segmentation of the bounding box of the building construction image to obtain a segmentation mask set includes: Dividing the pixels within the boundary box into regions to obtain a plurality of target building unit regions; The target building unit area is semantically segmented according to the image segmentation model, the target building unit contour is extracted, and a segmentation mask set is obtained.

5. The method for intelligent monitoring of construction progress according to claim 4, characterized in that: After semantic segmentation is performed on the target building unit area according to the image segmentation model, the target building unit contour is extracted to obtain a segmentation mask set, the method further includes: Taking the quotient of the segmentation mask area of ​​the segmentation mask set and the bounding box area respectively to obtain an area ratio; The segmentation masks whose area ratios are within a preset ratio threshold range are selected as target segmentation masks.

6. The intelligent monitoring method for construction progress according to claim 1 is characterized in that: Before calculating the cosine similarity of the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determining the angle corresponding to the maximum value of the cosine similarity as the most matching angle, the method further includes: Generate a full-angle sample image of each building unit based on the building information model library; The full-angle template image is three-dimensionally rotated to obtain the two-dimensional viewing angle image data set.

7. The intelligent monitoring method for construction progress according to claim 1 is characterized in that: The calculating the cosine similarity between the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determining the angle corresponding to the maximum value of the cosine similarity as the most matching angle, includes: Extracting a first high-dimensional mapping feature matrix of the segmentation mask set and a second high-dimensional mapping feature matrix of the two-dimensional perspective image data set respectively through a convolutional neural network model; The cosine similarity between the first high-dimensional mapping feature matrix and the second high-dimensional mapping feature matrix is ​​determined according to a cosine similarity calculation formula, and the angle corresponding to the maximum value of the cosine similarity is determined as the most matching angle.

8. The intelligent monitoring method for construction progress according to claim 1 is characterized in that: The determining of the two-dimensional best-matching viewing angle image and the best-matching mask based on the best-matching angle, and determining the construction progress of the building construction image, includes: Determine the two-dimensional perspective image data corresponding to the best matching angle in the two-dimensional perspective image data set as the two-dimensional best matching perspective image; Determine the best matching mask of the two-dimensional best matching view image in the segmentation mask set by a feature point matching algorithm; An area comparison is performed between the best-matching mask and the two-dimensional best-matching viewing angle image to determine a construction progress of the building construction image.

9. An intelligent monitoring device for construction progress, characterized in that: include: A two-dimensional perspective image data set acquisition module is used to compare the labels of the building construction images with the building information model library to obtain a two-dimensional perspective image data set; A segmentation mask set acquisition module is used to perform semantic segmentation on the boundary box of the building construction image to obtain a segmentation mask set; A best matching angle determination module, used to calculate the cosine similarity between the two-dimensional view image data set and the segmentation mask set according to the cosine similarity of the feature vectors, and determine the angle corresponding to the maximum value of the cosine similarity as the best matching angle; The construction progress intelligent monitoring module is used to determine the two-dimensional best-matching viewing angle image and the best-matching mask based on the best-matching angle, and determine the construction progress of the building construction image.

10. An electronic device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store programs; The processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps in the intelligent monitoring method for construction progress as described in any one of claims 1 to 8 above.

Images