Engineering construction progress intelligent monitoring method, system, equipment and medium

Abstract

The invention discloses an intelligent engineering construction progress monitoring method, system, equipment and medium, and particularly relates to the technical field of engineering monitoring, and the technical key points are as follows: carrying out the shielding detection of each engineering structure image in an engineering time sequence image set at different construction stages through a pre-constructed shielding detection algorithm, generating a shielding mask engineering image with mask holes so as to construct a mask engineering time sequence image set of different construction stages; based on hole position information in different shielding mask engineering images, effective position nodes in the shielding mask engineering images are judged; selecting a corresponding shielding complementing method based on a judgment result, complementing mask holes in combination with mask engineering time sequence image sets at different construction stages, and generating a complemented engineering structure image; and obtaining a current construction stage engineering model according to the complemented engineering structure image, and comparing the current construction stage engineering model with a pre-constructed BIM engineering progress model to obtain an engineering progress monitoring result.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering monitoring, and in particular to a method, system, equipment and medium for intelligent monitoring of engineering construction progress. Background Art

[0002] During construction, it is necessary to monitor the progress of the building to avoid delays due to construction problems, which could affect subsequent projects and revenue. During the construction process, aerial photography is used to capture building image information during the construction process. The construction progress in the captured image information is analyzed against the expected construction progress, and the difference between the two is analyzed. The current construction situation is then judged and an early warning is issued to avoid delays caused by the construction progress.

[0003] A Chinese patent application, publication number CN114708216A, discloses a method and system for intelligently identifying construction progress based on image recognition. This system uses a building construction information model established based on construction plan information and monitoring equipment information, combined with image information captured by multiple camera stations around the construction site during construction. The system first determines whether the building has completed foundation construction. If so, the system estimates the building's construction progress based on the identified building top outline, the distance from the center of the building top outline to the center of the foundation, and the designed height of each floor. This method uses images obtained from a fixed-height camera, rather than drones, to identify construction progress, thereby reducing the cost of the monitoring system while improving its reliability.

[0004] In the above-mentioned existing technology, the construction progress of the building is estimated by obtaining the distance from the center of the building's top outline to the center of the foundation and the designed height of each floor of the building; however, in the actual construction process, scaffolding, protective nets and other obstructions are built on the exterior of the building, and the above-mentioned technical solution does not take into account the impact of obstructions on building recognition, resulting in reduced recognition accuracy and excessive deviation in construction progress estimation.

[0005] Therefore, the present invention aims to provide a method, system, equipment and medium for intelligent monitoring of engineering construction progress to solve the above-mentioned related problems. Summary of the Invention

[0006] The technical problem to be solved by the present invention is that the existing technology does not take into account the impact of obstructions on building recognition, resulting in reduced recognition accuracy and excessive deviation in construction progress estimation. The purpose is to provide a method, system, equipment and medium for intelligent monitoring of engineering construction progress. By combining the occlusion detection algorithm obtained by training the improved U-net model with semantic segmentation technology, pixel-level recognition of obstructions and buildings in construction images is performed, and a mask image with hole marks is output to clearly mark the building area covered by the occlusion and the non-occluded area, thereby avoiding misjudging the obstruction as a building structure and reducing the error recognition rate; by selecting the corresponding occlusion completion method in the multimodal completion strategy based on the judgment result of the valid position node in the occlusion mask engineering image, the corresponding completion method can be selected for different occlusion situations, and the image data of different construction stages can be fully utilized to maximize the avoidance of completion defects caused by missing data in the occlusion area, thereby solving the problems that the existing technology does not take into account the impact of obstructions on building recognition, resulting in reduced recognition accuracy and excessive deviation in construction progress estimation.

[0007] The present invention is achieved through the following technical solutions:

[0008] A method for intelligently monitoring construction progress, comprising:

[0009] Collect engineering structure images at different construction stages according to the preset image acquisition path to construct a time-series image set of the engineering structure at different construction stages;

[0010] Using a pre-built occlusion detection algorithm, occlusion detection is performed on each engineering structure image in the engineering time series image set at different construction stages, generating occlusion mask engineering images with mask holes to construct a mask engineering time series image set at different construction stages;

[0011] Based on the hole position information in different occlusion mask engineering images, the valid position nodes in the occlusion mask engineering images are judged; based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time-series image sets of mask engineering at different construction stages to generate the completed engineering structure image;

[0012] The current construction phase engineering model is obtained based on the completed engineering structure image, and the current construction phase engineering model is compared with the pre-built BIM engineering progress model to obtain the engineering progress monitoring results.

[0013] Furthermore, a pre-built occlusion detection algorithm is used to perform occlusion detection on each engineering structure image in the engineering time series image set at different construction stages, generating occlusion mask engineering images with mask holes to construct a mask engineering time series image set at different construction stages. Specifically:

[0014] Using a pre-built occlusion detection algorithm, each engineering structure image is sequentially tested for occlusion, and multiple initial occlusion mask engineering images are obtained;

[0015] Perform morphological optimization and hole marker fusion processing on multiple initial occlusion mask engineering images to obtain multiple fused occlusion mask engineering images. Then, construct a set of mask engineering time series images at different construction stages through multiple fused occlusion mask engineering images.

[0016] The occlusion detection algorithm is obtained by training the improved U-net model. The improved U-net model includes an encoder and a decoder. A multi-scale feature extraction module is provided in the fifth layer of the encoder. The multi-scale feature extraction module includes a first convolution unit, a first hole convolution unit, a second hole convolution unit, a third hole convolution unit, a global average pooling unit and a second convolution unit. The input ends of the first convolution unit, the first hole convolution unit, the second hole convolution unit, the third hole convolution unit and the global average pooling unit are all connected to the output end of the fifth layer of the encoder, the output ends of the first convolution unit, the first hole convolution unit, the second hole convolution unit, the third hole convolution unit and the global average pooling unit are all connected to the input end of the second convolution unit, and the output end of the second convolution unit is connected to the decoder.

[0017] Furthermore, based on the hole position information in different occlusion mask engineering images, the valid position nodes in the occlusion mask engineering images are judged, specifically:

[0018] Obtain hole position information of a mask hole in one of the occlusion mask engineering images and a first mask value of the mask hole;

[0019] Matching a second mask value at the same position as the mask hole in the remaining occlusion mask engineering image of the current construction stage based on the hole position information;

[0020] Based on the first mask value and multiple second mask values, the valid position nodes in the occlusion mask engineering image are judged to obtain a judgment result; wherein the judgment result includes unoccluded positions in the same stage, unoccluded positions across stages, and fully occluded positions.

[0021] Furthermore, based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time-series image sets of the mask engineering at different construction stages to generate the completed engineering structure image; the occlusion completion methods include projection completion method, cross-time series completion method and model completion method, specifically:

[0022] If the result is that the position at the same stage is not blocked, the projection repair method is selected to combine the mask engineering time series image set of the current construction stage to fill the mask holes and generate the completed engineering structure image;

[0023] If the result of the judgment is that the cross-stage position is not blocked, the cross-time series completion method is selected to combine the mask engineering time series image set of other construction stages to complete the mask holes and generate the completed engineering structure image;

[0024] If the judgment result is that the position is fully occluded, the model completion method is selected to complete the mask holes and generate a completed engineering structure image.

[0025] Furthermore, the projection repair method is selected to complete the mask holes in combination with the mask engineering time series image set of the current construction stage to generate the completed engineering structure image, specifically:

[0026] Selecting, from the mask engineering time-series image set of the current construction phase, an occlusion mask engineering projection image whose second mask value is different from the first mask value of the occlusion mask engineering image to be completed;

[0027] Based on the hole position information of the occlusion mask engineering image to be completed, obtaining the position information of the corresponding completion reference position in the occlusion mask engineering projection image;

[0028] The RGB pixel value and depth value of the reference position are extracted using the position information, and the extracted RGB pixel value and depth value are mapped to the hole position of the occlusion mask engineering image to be completed to generate the completed engineering structure image.

[0029] Furthermore, the cross-temporal completion method is selected to combine the engineering time series image sets of other construction stages to complete the mask holes and generate the completed engineering structure image, specifically:

[0030] Constructing an alignment reference point in the occluded mask engineering image to be completed; based on the coordinate information of the alignment reference point, aligning each occluded mask engineering image in the mask engineering time-series image set of the remaining construction phases to obtain an aligned mask engineering time-series image set;

[0031] Selecting, from the aligned occlusion mask engineering images, an occlusion mask engineering cross-temporal image whose second mask value is different from the first mask value of the occlusion mask engineering image to be completed;

[0032] Based on the hole position information of the occlusion mask engineering image to be completed, the position information of the corresponding completion reference position in the cross-time sequence image of the occlusion mask engineering is obtained;

[0033] The RGB pixel value and depth value of the reference position are extracted using the position information, and the extracted RGB pixel value and depth value are mapped to the hole position of the occlusion mask engineering image to be completed to generate the completed engineering structure image.

[0034] Furthermore, the model completion method is selected to complete the mask holes and generate the completed engineering structure image, specifically:

[0035] Obtain a structural surface matching image within a preset range around the mask hole in the occluded mask engineering image to be completed; input the structural surface matching image into the trained 3D generative adversarial network to generate a structural generative image at the mask hole location;

[0036] The structure generation image is fused to the hole position of the occluded mask engineering image to be completed to generate the completed engineering structure image.

[0037] The present invention further provides an intelligent monitoring system for construction progress, which is used in any one of the above-mentioned intelligent monitoring methods for construction progress, and the system includes:

[0038] The engineering time-series image set construction module is used to collect engineering structure images at different construction stages according to a preset image acquisition path to construct engineering time-series image sets at different construction stages;

[0039] The mask engineering time-series image set construction module is used to use a pre-built occlusion detection algorithm to perform occlusion detection on each engineering structure image in the engineering time-series image set at different construction stages, generate occlusion mask engineering images with mask holes, and thus construct the mask engineering time-series image set at different construction stages;

[0040] The mask hole completion module is used to determine the valid position nodes in the occluded mask engineering images based on the hole position information in the images. Based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time-series image sets of mask engineering at different construction stages to generate the completed engineering structure image.

[0041] The project progress monitoring module is used to obtain the current construction phase project model based on the completed project structure image, and compare the current construction phase project model with the pre-built BIM project progress model to obtain the project progress monitoring results.

[0042] The present invention also provides a computer device, comprising a system memory and a processor, wherein the system memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0043] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any one of the methods described above are implemented.

[0044] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0045] 1. In the present invention, the occlusion detection algorithm obtained by training the improved U-net model is combined with semantic segmentation technology to perform pixel-level recognition of occlusions and buildings in construction images, output a mask image with hole marks, clearly mark the building area covered by the occlusion and the non-occlusion area, avoid misjudging the occlusion as a building structure, and reduce the error recognition rate; by selecting the corresponding occlusion completion method in the multimodal completion strategy according to the judgment result of the valid position node in the occlusion mask engineering image, it is possible to select the corresponding completion method for different occlusion situations, and make full use of image data in different construction stages to maximize the avoidance of completion defects caused by missing data in the occlusion area, thereby solving the problems that the existing technology does not take into account the impact of occlusions on building recognition, resulting in reduced recognition accuracy and excessive deviation in construction progress estimation.

[0046] 2. In the present invention, by improving the multi-scale feature extraction module in the U-net model, the first, second and third dilated convolution units can be utilized to effectively capture the multi-scale information in the engineering structure image, and to enhance the occlusion detection algorithm's ability to segment the occluded areas of the engineering structure image. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the examples. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can also derive other relevant drawings based on these drawings without inventive effort. In the drawings:

[0048] Figure 1 This is a method flow chart of an intelligent monitoring method for engineering construction progress in this embodiment;

[0049] Figure 2 Schematic diagram of the model structure of the improved U-net model in this embodiment;

[0050] Figure 3 Schematic diagram of the module structure of the multi-scale feature extraction module in this embodiment;

[0051] Figure 4 Schematic diagram of system modules of an intelligent monitoring system for construction progress in this embodiment;

[0052] Figure 5 This is a structural diagram of a computer device in this embodiment. DETAILED DESCRIPTION

[0053] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0054] In this disclosure, unless otherwise specified, the use of terms such as "first" and "second" to describe various elements is not intended to limit the positional relationship, temporal relationship, or importance relationship of these elements. Such terms are only used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of the element, while in some cases, based on the context of the description, they may also refer to different instances.

[0055] The terms used in the descriptions of various examples in this disclosure are for the purpose of describing specific examples only and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element may be one or more. In addition, the term "and / or" used in this disclosure encompasses any one and all possible combinations of the listed items.

[0056] Example 1

[0057] See also Figure 1 , Figure 1 A method flow chart of an intelligent monitoring method for construction progress is shown, wherein the method includes:

[0058] S1: Collect engineering structure images at different construction stages according to the preset image acquisition path to construct a time-series image set of the engineering structure at different construction stages;

[0059] It should be noted that, in this embodiment, the drone is used to capture images of the construction project building according to a preset image capture path. According to the preset image capture path, engineering structure images of the construction project building in different directions can be captured, including but not limited to top view images, front view images, rear view images, left view images and right view images. The specific image capture path depends on the actual situation and is not subject to excessive restrictions here. At the same time, in this embodiment, multiple images can be captured at each angle to ensure that at least one image is available through image redundancy. At the same time, different construction stages are specifically divided according to a certain time interval. In this embodiment, the time interval is 1 day, that is, the engineering structure image of the previous construction stage is the engineering structure image captured on the previous day, the engineering structure image of the current construction stage is the engineering structure image captured on the same day, and the engineering structure image of the next construction stage is the engineering structure image captured on the next day. In other embodiments, the time interval can also be set to 2 days or 3 days, which is not subject to excessive restrictions here. At the same time, the engineering time series image set is an image set formed by arranging and combining the captured engineering structure images in sequence according to the acquisition timestamps.

[0060] At the same time, it should be noted that in this embodiment, after the engineering structure image is collected, the image needs to be preprocessed such as denoising, illumination normalization and distortion correction; at the same time, the collected engineering structure image is time-series aligned through the acquisition timestamp. This technical solution is a conventional technical means in this field and will not be elaborated here.

[0061] S2: Using a pre-built occlusion detection algorithm, perform occlusion detection on each engineering structure image in the engineering time series image set at different construction stages, generate occlusion mask engineering images with mask holes, and construct a mask engineering time series image set at different construction stages;

[0062] Specifically, in this embodiment, occlusion detection is performed on each engineering structure image in turn using a pre-built occlusion detection algorithm to obtain multiple initial occlusion mask engineering images; morphological optimization and hole marker fusion processing are performed on the multiple initial occlusion mask engineering images to obtain multiple fused occlusion mask engineering images, and a set of mask engineering time series images of different construction stages is constructed using the multiple fused occlusion mask engineering images;

[0063] It should be noted that in this embodiment, different locations in the obtained initial occlusion mask engineering image have corresponding category labels (0 for background, 1 for building, and 2 for occlusion); morphological optimization includes opening and closing operations. The opening operation is used to eliminate small noise in the occlusion mask; the closing operation is used to fill holes in the building mask. Hole marking uses image contour detection methods to detect holes larger than 500 pixels. The occlusion mask (category label 2) is superimposed and fused with the building hole marker to generate a composite mask to obtain multiple fused occlusion mask engineering images.

[0064] The occlusion detection algorithm is obtained by training the improved U-net model. The training process is as follows: a large number of original image datasets and annotated data are obtained, wherein the original image dataset includes a large number of engineering structure images, and the annotated data refers to the pixel-level semantic segmentation mask (single channel or multi-channel) corresponding to each image, and the annotated categories are such as background, building, and occlusion; then the large number of original image datasets and annotated data are input into the improved U-net model for training, thereby obtaining the occlusion detection algorithm. The training scheme adopts conventional technical means in this field and will not be elaborated on here.

[0065] See also Figure 2 , Figure 2 The improved U-net model includes an encoder and a decoder, wherein both the encoder and the decoder adopt a five-layer structure, and a multi-scale feature extraction module is set in the fifth layer of the encoder. Figure 3 , Figure 3 A schematic diagram of the module structure of the multi-scale feature extraction module is shown. The multi-scale feature extraction module includes a first convolution unit (convolution kernel is 1×1, step size is 1, number of channels is 256), a first dilated convolution unit, a second dilated convolution unit, a third dilated convolution unit, a global average pooling unit, and a second convolution unit (convolution kernel is 1×1, step size is 1, number of channels is 256). The input ends of the first convolution unit, the first dilated convolution unit, the second dilated convolution unit, the third dilated convolution unit, and the global average pooling unit are all connected to the output end of the fifth layer of the encoder, and the output ends of the first convolution unit, the first dilated convolution unit, the second dilated convolution unit, the third dilated convolution unit, and the global average pooling unit are all connected to the input end of the second convolution unit, and the output end of the second convolution unit is connected to the decoder.

[0066] It should be noted that, in this embodiment, the first layer of the encoder includes two sequentially connected convolution units (channel number 64) and a maximum pooling unit; the second layer of the encoder includes two sequentially connected convolution units (channel number 128) and a maximum pooling unit; the third layer of the encoder includes three sequentially connected convolution units (channel number 256) and a maximum pooling unit; the fourth layer of the encoder includes three sequentially connected convolution units (channel number 512) and a maximum pooling unit; the fifth layer of the encoder also includes three sequentially connected convolution units (channel number 512) and a maximum pooling unit; wherein, each convolution unit is followed by a ReLU function, and the convolution kernel of each convolution unit is 3×3 with a step size of 1; the pooling sum of each maximum pooling unit is 2×2 with a step size of 2; through the layer-by-layer convolution and pooling of the encoder, the low-level texture to high-level semantic features of the image are gradually extracted;

[0067] The fifth layer of the decoder includes upsampling units and convolution units (channel number 256) connected in sequence; the fourth layer of the decoder includes upsampling units and convolution units (channel number 128) connected in sequence; the third layer of the decoder includes upsampling units and convolution units (channel number 64) connected in sequence; the second layer of the decoder includes upsampling units and convolution units (channel number 32) connected in sequence; the first layer of the decoder includes convolution units (channel number 3) connected in sequence and a softmax activation function; wherein, the upsampling factor of each upsampling unit is 2, the convolution kernels of the convolution units between the fifth layer and the second layer of the decoder are all 3×3 with a stride of 1, and the convolution kernels of the convolution units of the first layer of the decoder are 1×1 with a stride of 1; the resolution is gradually restored through upsampling + skip connection of the decoder, multi-scale features are fused, and the segmentation result is finally output.

[0068] At the same time, the inputs of the first convolution unit, the first dilated convolution unit, the second dilated convolution unit, the third dilated convolution unit, and the global average pooling unit are all connected to the output of the maximum pooling unit in the fifth layer of the encoder. The output of the second convolution unit is connected to the upsampling unit in the fifth layer of the decoder and embedded at the end of the encoder through the ASPP module for multi-scale context information fusion, enhancing the model's perception of defects of different sizes.

[0069] At the same time, the maximum pooling unit of the first layer of the encoder and the upsampling unit of the second layer of the decoder are jump-connected, the maximum pooling unit of the second layer of the encoder and the upsampling unit of the third layer of the decoder are jump-connected, the maximum pooling unit of the third layer of the encoder and the upsampling unit of the fourth layer of the decoder are jump-connected; the maximum pooling unit of the fourth layer of the encoder and the upsampling unit of the fifth layer of the decoder are jump-connected.

[0070] At the same time, it should be noted that the expansion rate of the first dilated convolution unit is 2, the convolution kernel is 3×3, and the step size is 1; the expansion rate of the second dilated convolution unit is 4, the convolution kernel is 3×3, and the step size is 1; the expansion rate of the third dilated convolution unit is 6, the convolution kernel is 3×3, and the step size is 1; the pooling kernel of the global average pooling unit is 16×16.

[0071] S3: Based on the hole position information in different occlusion mask engineering images, the valid position nodes in the occlusion mask engineering images are judged; based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time-series image sets of mask engineering at different construction stages to generate the completed engineering structure image;

[0072] Specifically, in this embodiment, the hole position information of the mask hole in one of the occlusion mask engineering images and the first mask value of the mask hole are first obtained; based on the hole position information, the second mask value at the same position as the mask hole is matched in the remaining occlusion mask engineering images of the current construction stage; based on the first mask value and multiple second mask values, the valid position nodes in the occlusion mask engineering image are judged to obtain a judgment result; wherein the judgment result includes unoccluded positions in the same stage, unoccluded positions across stages, and fully occluded positions; the occlusion completion methods include projection completion method, cross-time series completion method, and model completion method;

[0073] It should be noted that, in this embodiment, it is first necessary to judge the occlusion position of each occlusion mask engineering image to determine whether the occlusion mask engineering images of the same stage or across stages can complement each other; first, by comparing whether the mask values of the same position of the two occlusion mask engineering images of the same stage are the same, if the first mask value of the mask hole position of the occlusion mask engineering image to be complemented is 2, if the second mask value of the same position of the compared occlusion mask engineering image is 0 or 1, it means that the position is not occluded, and the output is that the position of the same stage is not occluded; if the second mask value of the same position of the compared occlusion mask engineering image is 2, then the same method is used to compare Whether the mask value of the same position in the occlusion mask engineering image across stages is the same, if the second mask value of the same position in the occlusion mask engineering image across stages is 0 or 1, it means that the position is not occluded, and the output is the position not occluded across stages; if the second mask value of the same position in the occlusion mask engineering image across stages is 2, the output is the position fully occluded; therefore, unoccluded position in the same stage means that the building structure at the mask hole position is not completely occluded in the same construction stage; unoccluded position across stages means that the building structure at the mask hole position is not completely occluded in other construction stages; fully occluded position means that the building structure at the mask hole position is completely occluded in all construction stages.

[0074] If the judgment result is that the position in the same stage is not blocked, the projection repair method is selected to complete the mask holes in combination with the mask engineering time-series image set of the current construction stage to generate a completed engineering structure image. Specifically, in the mask engineering time-series image set of the current construction stage, a projection image of the occluded mask engineering whose second mask value is different from the first mask value of the occluded mask engineering image to be completed is selected; based on the hole position information of the occluded mask engineering image to be completed, the position information of the corresponding completion reference position in the occluded mask engineering projection image is obtained; the position information is used to extract the RGB pixel value and depth value of the completion reference position, and the extracted RGB pixel value and depth value are mapped to the hole position of the occluded mask engineering image to be completed to generate the completed engineering structure image.

[0075] It should be noted that, in this embodiment, the hole area of the occlusion mask engineering image to be completed is first located through the depth map to obtain the hole position information (3D vertex coordinates) of the hole position; then the hole position is projected into the corresponding occlusion mask engineering projection image through the camera parameters of other occlusion mask engineering projection images, and the position information of the corresponding completion reference position is calculated; then the RGB pixel value at the completion reference position is directly read according to the position information of the completion reference position, and the extracted RGB pixels are mapped to the occlusion area of the current perspective to restore its color and surface texture (such as wall material, painting details); then the depth value at the completion reference position is read from the depth map of the occlusion mask engineering projection image, or calculated by multi-view stereo matching (MVS), and the 3D geometric shape of the occlusion area (such as wall convexity and crack depth) is restored in combination with the depth value, thereby completing the hole position and obtaining the completed engineering structure image.

[0076] If the judgment result is that the cross-stage position is not blocked, the cross-temporal completion method is selected to combine the mask engineering time-series image sets of other construction stages to complete the mask holes and generate the completed engineering structure image, specifically: construct an alignment reference point in the occluded mask engineering image to be completed; based on the coordinate information of the alignment reference point, align each occluded mask engineering image in the mask engineering time-series image set of the remaining construction stages to obtain the aligned mask engineering time-series image set; in the aligned occluded mask engineering image, select the occluded mask engineering cross-temporal image whose second mask value is different from the first mask value of the occluded mask engineering image to be completed; based on the hole position information of the occluded mask engineering image to be completed, obtain the position information of the corresponding completion reference position in the occluded mask engineering cross-temporal image; use the position information to extract the RGB pixel value and depth value of the completion reference position, and map the extracted RGB pixel value and depth value to the hole position of the occluded mask engineering image to be completed to generate the completed engineering structure image.

[0077] It should be noted that, in this embodiment, first, an alignment reference point is constructed in the occlusion mask engineering image to be completed. The alignment reference point can be a foundation corner point or a core column, and the coordinate information of the alignment reference point is obtained; then, the transformation matrix is calculated based on the feature matching algorithm. Among them, R represents the rotation matrix, r ij Indicates the position of the new coordinate system in the original coordinate system after rotation, i=1,2,3,j=1,2,3; t represents the translation vector, t x , t y , t z Respectively represent the translation amount in different directions; each occluded mask engineering image in the mask engineering time series image set of the remaining construction stages is aligned through the transformation matrix; the completion step after alignment is consistent with the above judgment result that the position in the same stage is not occluded.

[0078] If the judgment result is that the position is fully occluded, the model completion method is selected to complete the mask holes and generate the completed engineering structure image. Specifically, the structure surface matching image within the preset range around the mask hole in the occluded mask engineering image to be completed is obtained; the structure surface matching image is input into the trained 3D generative adversarial network to generate a structure generation image of the mask hole position; the structure generation image is fused to the hole position of the occluded mask engineering image to be completed to generate the completed engineering structure image.

[0079] It should be noted that in this embodiment, the hole boundary is first extracted, and then the target repair area is defined. The structural surface matching image within a radius of 0.5m is intercepted with the hole position as the center; the structural surface matching image is input into the generator of the SD-GAN generative adversarial network and passes through the encoder and decoder in sequence. The encoder extracts multi-scale features, and the decoder gradually generates a complete structural image, and finally obtains the structural generation image of the mask hole position; the obtained structural generation image of the mask hole position is then input into the discriminator for discrimination, and the true and false probability is output; the structural generation image of the mask hole position is then optimized based on the output result until the loss function converges to obtain the final structural generation image; the structural generation image is fused to the hole position of the occluded mask engineering image to be completed to generate the completed engineering structure image.

[0080] S4: Obtain the current construction phase engineering model based on the completed engineering structure image, and compare the current construction phase engineering model with the pre-built BIM engineering progress model to obtain the engineering progress monitoring result.

[0081] It should be noted that, in this embodiment, the BIM project progress model is constructed in advance through construction drawings and the estimated project period. This technology is a conventional technical means in this field and will not be elaborated on here. At the same time, the current construction stage project model is compared with the pre-built BIM project progress model, which also adopts conventional technical means in this field and will not be elaborated on here.

[0082] Specifically, in this embodiment, the occlusion detection algorithm obtained by training the improved U-net model is combined with semantic segmentation technology to perform pixel-level recognition of occlusions and buildings in the construction image, output a mask image with hole marks, clearly mark the building area covered by the occlusion and the non-occlusion area, avoid misjudging the occlusion as a building structure, and reduce the error recognition rate; by selecting the corresponding occlusion completion method in the multimodal completion strategy according to the judgment result of the valid position node in the occlusion mask engineering image, it is possible to select the corresponding completion method for different occlusion situations, and make full use of image data in different construction stages to maximize the avoidance of completion defects caused by missing data in the occlusion area, thereby solving the problem that the existing technology does not take into account the impact of occlusions on building recognition, resulting in reduced recognition accuracy and excessive deviation in construction progress estimation.

[0083] Example 2

[0084] See also Figure 4 The present invention further provides an intelligent monitoring system for construction progress, which is used in any one of the above-mentioned intelligent monitoring methods for construction progress, and the system includes:

[0085] The engineering time-series image set construction module 100 is used to collect engineering structure images at different construction stages according to a preset image acquisition path to construct engineering time-series image sets at different construction stages;

[0086] The mask engineering time-series image set construction module 200 is used to perform occlusion detection on each engineering structure image in the engineering time-series image set of different construction stages using a pre-built occlusion detection algorithm, and generate occlusion mask engineering images with mask holes to construct the mask engineering time-series image set of different construction stages;

[0087] The mask hole completion module 300 is used to determine valid position nodes in the occlusion mask engineering image based on the hole position information in different occlusion mask engineering images; select a corresponding occlusion completion method based on the judgment result, and complete the mask holes in combination with the time-series image set of the mask engineering at different construction stages to generate a completed engineering structure image;

[0088] The project progress monitoring module 400 is used to obtain the current construction phase project model based on the completed project structure image, and compare the current construction phase project model with the pre-built BIM project progress model to obtain the project progress monitoring result.

[0089] It should be noted that the modules in the system of Example 2 correspond to the steps in the method of Example 1. The steps in the method of Example 1 have been described in detail in Example 1. The contents of the modules in the system will not be described in detail in this Example 2.

[0090] Example 3

[0091] See also Figure 5 As shown, this embodiment further provides a computer device, including a system memory 1005 and a processor 1001, wherein the system memory 1005 stores a computer program, and the processor 1001 implements the steps of any of the above methods when executing the computer program.

[0092] It should be noted that the processor 1001 is configured to execute the steps of the above method embodiments according to the instructions in the program code. Alternatively, the processor 1001 implements the functions of the modules / units in the above system / device embodiments when executing the computer program.

[0093] Specifically, in this embodiment, the computer program may be divided into one or more modules / units, one or more modules / units being stored in the system memory 1005 and executed by the processor 1001 to complete the present application. One or more modules / units may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0094] The terminal device may be a computing device such as a desktop computer, laptop, PDA, or cloud server. The terminal device may include, but is not limited to, a processor 1001 and a system memory 1005. Those skilled in the art will appreciate that this does not limit the terminal device and may include more or fewer components than shown, or a combination of certain components, or different components. For example, the terminal device may also include an input / output device 1003, a network access device 1002, a bus 1006, and the like.

[0095] The processor 1001 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0096] The system memory 1005 can be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. The system memory 1005 can also be the storage device 1004 of the terminal device, such as a plug-in hard disk, a SmartMedia Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the terminal device. Furthermore, the system memory 1005 can also include both the internal storage unit of the terminal device and the storage device 1004. The system memory 1005 is used to store computer programs and other programs and data required by the terminal device. The system memory 1005 can also be used to temporarily store data that has been output or is about to be output.

[0097] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0098] Example 4

[0099] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of any one of the above methods are implemented.

[0100] Among them, the computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, system or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk. Random Access Memory (RAM), Read-Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), a register, a hard disk, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above, or any other form of computer-readable storage medium known in the art.

[0101] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and the storage medium can be located in an application-specific integrated circuit (ASIC). In an embodiment of the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device.

[0102] Example 5

[0103] This embodiment further provides a computer program product comprising instructions. When the instructions are executed by a computer device cluster, the computer device cluster executes the method described in Embodiment 1.

[0104] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for intelligent monitoring of construction progress, characterized in that: include: Collect engineering structure images at different construction stages according to the preset image acquisition path to construct a time-series image set of the engineering structure at different construction stages; Using a pre-built occlusion detection algorithm, occlusion detection is performed on each engineering structure image in the engineering time series image set at different construction stages, generating occlusion mask engineering images with mask holes to construct a mask engineering time series image set at different construction stages; Based on the hole position information in different occlusion mask engineering images, the valid position nodes in the occlusion mask engineering images are judged; Based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time-series image sets of mask projects at different construction stages to generate the completed engineering structure image; The current construction phase engineering model is obtained based on the completed engineering structure image, and the current construction phase engineering model is compared with the pre-built BIM engineering progress model to obtain the engineering progress monitoring results.

2. The method for intelligent monitoring of construction progress according to claim 1, characterized in that: Using the pre-built occlusion detection algorithm, occlusion detection is performed on each engineering structure image in the engineering time series image set at different construction stages, generating occlusion mask engineering images with mask holes to construct the mask engineering time series image set at different construction stages. Specifically: Using a pre-built occlusion detection algorithm, each engineering structure image is sequentially tested for occlusion, and multiple initial occlusion mask engineering images are obtained; Perform morphological optimization and hole marker fusion processing on multiple initial occlusion mask engineering images to obtain multiple fused occlusion mask engineering images. Then, construct a set of mask engineering time series images at different construction stages through multiple fused occlusion mask engineering images. The occlusion detection algorithm is obtained by training the improved U-net model. The improved U-net model includes an encoder and a decoder. A multi-scale feature extraction module is provided in the fifth layer of the encoder. The multi-scale feature extraction module includes a first convolution unit, a first hole convolution unit, a second hole convolution unit, a third hole convolution unit, a global average pooling unit and a second convolution unit. The input ends of the first convolution unit, the first hole convolution unit, the second hole convolution unit, the third hole convolution unit and the global average pooling unit are all connected to the output end of the fifth layer of the encoder, the output ends of the first convolution unit, the first hole convolution unit, the second hole convolution unit, the third hole convolution unit and the global average pooling unit are all connected to the input end of the second convolution unit, and the output end of the second convolution unit is connected to the decoder.

3. The method for intelligent monitoring of construction progress according to claim 1, characterized in that: Based on the hole position information in different occlusion mask engineering images, the valid position nodes in the occlusion mask engineering images are judged, specifically: Obtain hole position information of a mask hole in one of the occlusion mask engineering images and a first mask value of the mask hole; Matching a second mask value at the same position as the mask hole in the remaining occlusion mask engineering image of the current construction stage based on the hole position information; Based on the first mask value and multiple second mask values, the valid position nodes in the occlusion mask engineering image are judged to obtain a judgment result; wherein the judgment result includes unoccluded positions in the same stage, unoccluded positions across stages, and fully occluded positions.

4. The method for intelligent monitoring of construction progress according to claim 3, characterized in that: Based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time series image sets of the mask engineering at different construction stages to generate the completed engineering structure image; the occlusion completion methods include projection completion method, cross-time series completion method and model completion method, specifically: If the result is that the position at the same stage is not blocked, the projection repair method is selected to combine the mask engineering time series image set of the current construction stage to fill the mask holes and generate the completed engineering structure image; If the result of the judgment is that the cross-stage position is not blocked, the cross-time series completion method is selected to combine the mask engineering time series image set of other construction stages to complete the mask holes and generate the completed engineering structure image; If the judgment result is that the position is fully occluded, the model completion method is selected to complete the mask holes and generate a completed engineering structure image.

5. The method for intelligent monitoring of construction progress according to claim 4, characterized in that: The projection repair method is selected to combine the mask engineering time series image set of the current construction stage to fill the mask holes and generate the completed engineering structure image. Specifically: Selecting, from the mask engineering time-series image set of the current construction phase, an occlusion mask engineering projection image whose second mask value is different from the first mask value of the occlusion mask engineering image to be completed; Based on the hole position information of the occlusion mask engineering image to be completed, obtaining the position information of the corresponding completion reference position in the occlusion mask engineering projection image; The RGB pixel value and depth value of the reference position are extracted using the position information, and the extracted RGB pixel value and depth value are mapped to the hole position of the occlusion mask engineering image to be completed to generate the completed engineering structure image.

6. The method for intelligent monitoring of construction progress according to claim 4, characterized in that: The cross-temporal completion method is selected to combine with the engineering time series image set of other construction stages to complete the mask holes and generate the completed engineering structure image. Specifically: Constructing an alignment reference point in the occluded mask engineering image to be completed; based on the coordinate information of the alignment reference point, aligning each occluded mask engineering image in the mask engineering time-series image set of the remaining construction phases to obtain an aligned mask engineering time-series image set; Selecting, from the aligned occlusion mask engineering images, an occlusion mask engineering cross-temporal image whose second mask value is different from the first mask value of the occlusion mask engineering image to be completed; Based on the hole position information of the occlusion mask engineering image to be completed, the position information of the corresponding completion reference position in the cross-time sequence image of the occlusion mask engineering is obtained; The RGB pixel value and depth value of the reference position are extracted using the position information, and the extracted RGB pixel value and depth value are mapped to the hole position of the occlusion mask engineering image to be completed to generate the completed engineering structure image.

7. The method for intelligent monitoring of construction progress according to claim 4, characterized in that: Select the model completion method to complete the mask holes and generate the completed engineering structure image, specifically: Obtain a structural surface matching image within a preset range around the mask hole in the occluded mask engineering image to be completed; input the structural surface matching image into the trained 3D generative adversarial network to generate a structural generative image at the mask hole location; The structure generation image is fused to the hole position of the occluded mask engineering image to be completed to generate the completed engineering structure image.

8. An intelligent monitoring system for construction progress, characterized in that: The system is used in an intelligent monitoring method for construction progress according to any one of claims 1 to 7, and the system comprises: The engineering time-series image set construction module is used to collect engineering structure images at different construction stages according to a preset image acquisition path to construct engineering time-series image sets at different construction stages; The mask engineering time-series image set construction module is used to use a pre-built occlusion detection algorithm to perform occlusion detection on each engineering structure image in the engineering time-series image set at different construction stages, generate occlusion mask engineering images with mask holes, and thus construct the mask engineering time-series image set at different construction stages; The mask hole completion module is used to determine the valid position nodes in the occluded mask engineering images based on the hole position information in the images. Based on the judgment results, the corresponding occlusion completion method is selected, and the mask holes are completed by combining the time-series image sets of mask engineering at different construction stages to generate the completed engineering structure image. The project progress monitoring module is used to obtain the current construction phase project model based on the completed project structure image, and compare the current construction phase project model with the pre-built BIM project progress model to obtain the project progress monitoring results.

9. A computer device comprising a system memory and a processor, wherein the system memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images