Method and system for detecting force of cast-in-place bracket during the whole process based on image recognition

Abstract

The invention discloses a method and system for detecting the force of a cast-in-place full-process support based on image recognition. The method comprises: collecting an initial video of a cast-in-place construction area, performing segmentation of the initial video of the construction area, and obtaining an initial image after segmentation; then constructing a semantic segmentation network model, performing training on the semantic segmentation network model, calling the trained DeepLabv3 model, performing segmentation operation on the construction area represented by the initial image, and obtaining a target image; using the SAM2 model to track square holes, obtaining square hole masks, and calculating the actual area represented by each pixel in the target image according to the square hole area; finally, based on the actual area, calculating the real-time volume of the construction area and determining the force analysis result of the support in the construction area. It can calculate the area change during the pouring process in real time, thereby determining the force change of the support during the pouring process, reducing the manpower and material costs, while reducing the errors caused by cognitive factors, and improving construction safety.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge safety monitoring, and in particular to a method and system for detecting the force of a cast-in-place full-process support based on image recognition. Background Art

[0002] In the construction process of the construction industry, concrete pouring technology is one of the most important links in the entire bridge construction process. Concrete pouring has a very obvious force on the support and is an important basis for the successful construction of the bridge. According to existing research, the existing technology for safety assessment and detection during the bridge construction stage is still at the stage of traditional instrument detection and manual observation. This method is not only time-consuming and labor-intensive, but also easily affected by human factors, making it difficult to ensure the accuracy and reliability of the detection results. This has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for detecting the force of a whole-process cast-in-place support based on image recognition to address the deficiencies in the prior art. The method aims to monitor the bridge construction and concrete pouring process in real time based on computer image recognition technology, and calculate the area changes during the pouring process in real time, thereby determining the force changes of the support during the pouring process, reducing manpower and material costs, and at the same time reducing errors caused by cognitive factors, thereby improving construction safety.

[0004] An embodiment of the present application provides a method for detecting force of a cast-in-place bracket during the entire process based on image recognition, the method comprising:

[0005] Collect initial video of the cast-in-place construction area;

[0006] Execute segmentation of the initial video of the construction area to obtain an initial image after segmentation;

[0007] Constructing a semantic segmentation network model, executing training on the semantic segmentation network model, calling the trained DeepLabv3 model, performing a segmentation operation on the construction area represented by the initial image, and obtaining a target image;

[0008] The square hole is tracked by using the SAM2 model to obtain a square hole mask, and the actual area represented by each pixel in the target image is calculated according to the square hole area;

[0009] Based on the actual area, the real-time volume of the construction area is calculated and the force analysis results of the support in the construction area are determined.

[0010] Optionally, after performing segmentation of the initial video of the construction area and obtaining the segmented initial image, the method further includes:

[0011] Performing preprocessing on the initial image;

[0012] The preprocessing includes image scaling and normalization.

[0013] Optionally, the step of constructing a semantic segmentation network model and training the semantic segmentation network model includes:

[0014] The semantic segmentation network model is trained by setting the loss function and optimizer of the semantic segmentation network model through the cross entropy loss and the Adam optimizer, and the fourth layer of the semantic segmentation network model classifier is replaced by the preset convolution layer so that the number of output channels is adjusted to 2;

[0015] The loss function is expressed by the following formula:

[0016] ;

[0017] in, represents the cross entropy loss, represents the number of samples, represents the number of categories, Represents the one-hot encoding of the true label, Represents the probability of the output category c of the semantic segmentation network model.

[0018] Optionally, the DeepLabv3 model includes:

[0019] The encoder module, the hole convolution fusion multi-scale information module, the image-level feature module and the decoder module are connected by communication, wherein:

[0020] The encoder module is used to extract input features of the initial image frame;

[0021] The hole convolution fusion multi-scale information module is used to capture information of different scales of the initial image frame and perform multi-scale feature extraction;

[0022] The image-level feature module is used to fuse the context information of the multi-scale information module according to the encoder module and the dilated convolution to improve the image segmentation result of the construction area;

[0023] The decoder module is used to output a target image.

[0024] Optionally, before tracking the square hole using the SAM2 model to obtain the square hole mask and calculating the actual area represented by each pixel in the target image according to the square hole area, the method further includes:

[0025] Set a label for the center of the square hole and calculate the center point of the square hole label; wherein the calculation of the center point of the square hole label is performed by the following formula:

[0026] ;

[0027] ;

[0028] in, Indicates the center coordinate in the horizontal direction, Indicates the center coordinate in the vertical direction, is the number of points, are the coordinates of each point.

[0029] Optionally, the method of tracking the square hole using the SAM2 model to obtain a square hole mask and calculating the actual area represented by each pixel in the target image according to the square hole area includes:

[0030] The SAM2 model is used to track the square holes, generate square hole masks, and determine the actual area of ​​the square hole region represented by each pixel by pixel counting; wherein the actual area calculation formula of the square hole region is:

[0031] ;

[0032] in, Indicates the actual area. represents the number of pixels in the target area, Indicates the actual area represented by each pixel; the number of pixels in the target area satisfies:

[0033] ;

[0034] in, Indicates the mask height, Indicates the mask width, If it is 1, the pixel is counted; if it is 0, it is not counted.

[0035] Optionally, the calculating the real-time volume of the construction area and determining the force analysis result of the support in the construction area based on the actual area includes:

[0036] The real-time volume of the construction area is calculated as follows:

[0037] ;

[0038] in, Represents the real-time volume of the construction area, Indicates the pouring height of the construction area;

[0039] The force analysis results of the support in the construction area are determined by the following methods:

[0040] ;

[0041] in, represents the force caused by the volume of the construction area, Indicates the pouring density in the construction area. represents the acceleration due to gravity and m represents the weight.

[0042] Another embodiment of the present application provides a cast-in-place full-process support force detection system based on image recognition, the system comprising:

[0043] The acquisition module is used to collect the initial video of the cast-in-place construction area;

[0044] An execution module, used for executing segmentation of the initial video of the construction area to obtain an initial image after segmentation;

[0045] A construction module is used to construct a semantic segmentation network model, execute training on the semantic segmentation network model, call the trained DeepLabv3 model, perform a segmentation operation on the construction area represented by the initial image, and obtain a target image;

[0046] The first calculation module is used to track the square hole using the SAM2 model to obtain the square hole mask, and calculate the actual area represented by each pixel in the target image according to the square hole area;

[0047] The second calculation module is used to calculate the real-time volume of the construction area and determine the force analysis results of the support in the construction area based on the actual area.

[0048] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to implement the above-mentioned method when executed.

[0049] Another embodiment of the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to implement the above-mentioned method.

[0050] Compared with the prior art, this application first collects the initial video of the cast-in-place construction area, performs segmentation on the initial video of the construction area, and obtains the initial image after segmentation; then constructs a semantic segmentation network model, performs training on the semantic segmentation network model, calls the trained DeepLabv3 model, performs segmentation operations on the construction area represented by the initial image, and obtains the target image; uses the SAM2 model to track the square holes, obtains the square hole mask, and calculates the actual area represented by each pixel in the target image based on the square hole area; finally, based on the actual area, calculates the real-time volume of the construction area and determines the force analysis results of the support in the construction area. It uses computer image recognition technology to monitor the bridge construction and concrete pouring process in real time, calculates the area changes during the pouring process in real time, and thus determines the force changes of the support during the pouring process, reducing manpower and material costs, while reducing errors caused by cognitive factors and improving construction safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A hardware structure block diagram of a computer terminal for a method for detecting force of a cast-in-place bracket throughout the entire process based on image recognition provided by an embodiment of the present invention;

[0052] Figure 2 A schematic flow chart of a method for detecting force on a cast-in-place bracket during the entire process based on image recognition provided by an embodiment of the present invention;

[0053] Figure 3 A schematic diagram of the DeepLabv3 model structure provided in an embodiment of the present invention;

[0054] Figure 4 A schematic diagram of construction area segmentation provided by an embodiment of the present invention;

[0055] Figure 5 A schematic diagram of square hole tracking using the SAM2 model provided in an embodiment of the present invention;

[0056] Figure 6 A schematic diagram of the structure of a SAM2 model provided in an embodiment of the present invention;

[0057] Figure 7 A schematic diagram of an area calculation result provided by an embodiment of the present invention;

[0058] Figure 8 A schematic diagram of a support structure based on a Midas Civil finite element model provided in an embodiment of the present invention;

[0059] Fig. 9 A structural schematic diagram of a cast-in-place full-process support force detection system based on image recognition provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, but should not be construed as limiting the present invention.

[0061] The embodiment of the present invention first provides a method for detecting the force of a cast-in-place whole-process bracket based on image recognition. The method can be applied to electronic devices, such as computer terminals, specifically ordinary computers, tablets, etc.

[0062] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal for a method for detecting the force of a cast-in-place bracket during the entire process based on image recognition provided by an embodiment of the present invention. Figure 1 As shown, the computer device includes a processor, a memory, and a network interface connected via a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0063] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any one of the cast-in-place full-process support force detection methods based on image recognition.

[0064] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0065] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any one of the cast-in-place full-process support force detection methods based on image recognition.

[0066] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0067] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be 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. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0068] See also Figure 2 , Figure 2 A schematic flow chart of a method for detecting force of a cast-in-place bracket during the entire process based on image recognition provided by an embodiment of the present invention includes:

[0069] S201: Collect initial video of the cast-in-place construction area.

[0070] Specifically, before the bridge concrete pouring process, a high-definition camera can be placed at a suitable position to enable it to observe the overall bridge pouring process. The purpose is to be able to capture all positions of the concrete pouring process, and select a camera that can clearly observe the position of concrete pouring on the upper part of the bracket to collect the initial video of the cast-in-place construction area.

[0071] S202: Execute segmentation of the initial video of the construction area to obtain an initial image after segmentation.

[0072] Specifically, the collected initial video of the cast-in-place construction area can be segmented into frames using FFmpeg to obtain the segmented initial image, for example, 5 minutes as one frame, and the time can be selected at will without restriction.

[0073] It should be noted that FFmpeg is an open source multimedia framework that can process audio and video files of all formats and allow users to extract individual image frames from videos. The on-site concrete pouring video is collected and divided into frames to achieve the purpose of real-time monitoring. In addition, by segmenting the video, it can also meet the needs of various large, medium and small bridges for the instant grasp of the stress conditions of the bracket.

[0074] In an optional implementation, after performing segmentation of the initial video of the construction area to obtain the segmented initial image, the method further includes:

[0075] Preprocessing of the initial image is performed; the preprocessing includes image scaling and normalization.

[0076] For example, image preprocessing can first scale the initial image to 0.5, and the scaled image size will be 50% of the original initial image size. Image scaling maintains the aspect ratio of the image, that is, the width and height are scaled at the same ratio, which is intended to improve computational efficiency, reduce memory usage, and enhance the flexibility of subsequent models at different input sizes. Secondly, normalization is to scale the pixel values ​​of the input initial image to [0,1] to ensure that the input initial image data remains consistent during model training, which helps to improve the model convergence speed and performance, allowing the model to learn faster and make more accurate predictions.

[0077] S203: construct a semantic segmentation network model, execute training on the semantic segmentation network model, call the trained DeepLabv3 model, perform a segmentation operation on the construction area represented by the initial image, and obtain a target image.

[0078] Specifically, the step of constructing a semantic segmentation network model and training the semantic segmentation network model may include:

[0079] The semantic segmentation network model is trained by setting the loss function and optimizer of the semantic segmentation network model through the cross entropy loss and the Adam optimizer, and the fourth layer of the semantic segmentation network model classifier is replaced by the preset convolution layer so that the number of output channels is adjusted to 2;

[0080] The loss function is expressed by the following formula:

[0081] ;

[0082] in, represents the cross entropy loss, represents the number of samples, represents the number of categories, Represents the one-hot encoding of the true label, Represents the probability of the output category c of the semantic segmentation network model.

[0083] The Adam optimizer is an adaptive learning rate optimization algorithm that combines the advantages of AdaGrad and RMSProp. It can effectively accelerate the training process of deep learning. During the model training process, the learning rate can be set to 0.0001.

[0084] Among them, the DeepLabv3 model is a deep learning model for semantic image segmentation, which may include: a communication-connected encoder module, a dilated convolution fusion multi-scale information module, an image-level feature module and a decoder module, wherein the encoder module is used to extract the input features of the initial image frame; the dilated convolution fusion multi-scale information module is used to capture information of different scales of the initial image frame and perform multi-scale feature extraction; the image-level feature module is used to improve the image segmentation results of the construction area based on the contextual information of the encoder module and the dilated convolution fusion multi-scale information module; the decoder module is used to output the target image.

[0085] It should be noted that the DeepLabv3 model is a semantic segmentation architecture used to identify and distinguish different objects and scene elements in images at the pixel level. It is improved on the basis of DeepLabv2 to better handle the problem of multi-scale object segmentation.

[0086] See also Figure 3 , Figure 3 A schematic diagram of the DeepLabv3 model structure provided for an embodiment of the present invention, wherein the DeepLabv3 model mainly includes an encoder module, a dilated convolution fusion multi-scale information module, an image-level feature module, and a decoder module. The number of channels output by the last layer of the classifier of the DeepLabv3 model is the number of categories. In the original implementation, the number of output channels of this layer is usually a predetermined category. After replacement, the number of output channels is set to 2, corresponding to the background and the target, respectively, so that the model can be used for binary classification tasks.

[0087] See also Figure 4 , Figure 4 A schematic diagram of construction area segmentation is provided for an embodiment of the present invention. In the segmentation operation of the construction area represented by the initial image, construction area segmentation refers to calling the trained DeepLabv3 model to segment the input image, and separating the construction area and background in the image data through the DeepLabv3 model.

[0088] S204: Track the square hole using the SAM2 model to obtain a square hole mask, and calculate the actual area represented by each pixel in the target image according to the area of ​​the square hole.

[0089] Before tracking the square hole using the SAM2 model to obtain the square hole mask and calculating the actual area represented by each pixel in the target image according to the square hole area, the method further includes:

[0090] Set a label for the center of the square hole and calculate the center point of the square hole label; wherein, to set the label for the center of the square hole, first sort the video frames and set the starting frame, and then use annotations to annotate the data. Annotations are pixel-level or even finer-grained labels in images or videos, which can effectively indicate the location, shape and category of specific objects. Calculate the center point of the square hole label, and the calculation of the center point of the square hole label is performed by the following formula:

[0091] ;

[0092] ;

[0093] in, Indicates the center coordinate in the horizontal direction, Indicates the center coordinate in the vertical direction, is the number of points, are the coordinates of each point.

[0094] The method of tracking the square hole using the SAM2 model to obtain the square hole mask and calculating the actual area represented by each pixel in the target image according to the square hole area includes:

[0095] The SAM2 model is used to track the square holes, generate square hole masks, and determine the actual area of ​​the square hole region represented by each pixel by pixel counting; wherein the actual area calculation formula of the square hole region is:

[0096] ;

[0097] in, Indicates the actual area. represents the number of pixels in the target area, Indicates the actual area represented by each pixel; the number of pixels in the target area satisfies:

[0098] ;

[0099] in, Indicates the mask height, Indicates the mask width, If it is 1, the pixel is counted; if it is 0, it is not counted.

[0100] The SAM2 model is an image and video segmentation model. During the casting process of a cast-in-place continuous box girder bridge, a hole is usually reserved at the top. By tracking the other hole, the memory encoder and memory attention mechanism of the SAM2 model are used to identify the area increment. The reason for tracking the other hole is that for some extra-large continuous box bridges, the entire appearance cannot be directly captured during the image acquisition process. The camera is rotated to capture the uncaptured area. Therefore, it is proposed to track the other hole to track the total area of ​​concrete pouring in real time, which is convenient, fast and accurate.

[0101] For example, see Figure 5 , Figure 5 A schematic diagram of square hole tracking using the SAM2 model provided in an embodiment of the present invention, firstly, all square holes are tracked using the SAM2 model to generate a square hole mask, and the actual area represented by each pixel is calculated by pixel counting. The square hole area calculation formula is: actual area = number of pixels in the target area × actual area represented by each pixel.

[0102] See also Figure 6 , Figure 6 A schematic diagram of the structure of a SAM2 model provided by an embodiment of the present invention, the SAM2 model includes an image encoder, a memory module, a segmentation encoder, a memory encoder, and a memory storage device. The center point obtained above is passed into the SAM2 model as a prompt. Due to the long length of the bridge, one photo cannot fully observe the global appearance. According to the characteristics of the memory encoder and memory attention mechanism of the SAM2 model, the module uses a cross-attention mechanism to fuse the image features, prediction results, and prompt information of the historical frames and provide them to the current frame as a reference, thereby achieving association and consistency between multiple frames. SAM2 creates a memory based on the current prediction through the memory encoder, and the memory library retains information about the past predictions of the video target object. The memory attention mechanism conditions the current frame features and adjusts them according to the features of the past frames to generate an embedding, which is then passed to the mask decoder to generate a mask prediction for the frame, and the subsequent frames repeat this operation. Therefore, the SAM2 model can track and identify the area changes of the target object throughout the image sequence, thereby realizing the recognition of the area increment.

[0103] See also Figure 7 , Figure 7A schematic diagram of an area calculation result provided by an embodiment of the present invention. The calculation of the construction area area (shown in red) first defines the timestamp of each tag, which is conducive to ensuring the consistency of image recognition and conveniently recording the construction area area data. Secondly, traverse the image file, read the construction area mask, record the initial frame area increment as 0, and calculate the construction area increment for each frame except the first frame. Finally, according to the pixel counting method, calculate the number of pixels in the newly added area, and finally obtain the actual total area of ​​each frame. The actual total area of ​​the current frame = the total area of ​​the previous frame + the actual area represented by each pixel x the pixel increment.

[0104] S205: Based on the actual area, calculate the real-time volume of the construction area and determine the force analysis result of the support in the construction area.

[0105] Specifically, the calculating the real-time volume of the construction area and determining the force analysis result of the support in the construction area based on the actual area includes:

[0106] The real-time volume of the construction area is calculated as follows:

[0107] ;

[0108] in, Represents the real-time volume of the construction area, Indicates the pouring height of the construction area.

[0109] Bracket stress refers to the analysis and calculation of various forces that the bracket bears in engineering structure and mechanical design, which is directly related to the safety and stability of the structure. The stability of the bracket structure directly affects the performance and safety of the entire system. The bracket structure should have high strength, high stability and durability to withstand the weight of equipment and components as well as external loads, prevent displacement or tilt caused by vibration, wind load or other external forces, and ensure construction safety and subsequent smooth opening of the bridge. MIDAS / Civil software provides strong support in the establishment and analysis of the bracket finite element model, which not only improves the accuracy and safety of the design, but also helps to optimize the construction plan, reduce costs and improve economic benefits.

[0110] See also Figure 8 , Figure 8 A schematic diagram of a support structure based on a Midas Civil finite element model is provided in an embodiment of the present invention. According to the construction plan, the properties of materials and cross-sections are defined, the boundary conditions between structures are defined, and a Midas Civil finite element model is established. The force analysis result of the support in the construction area is determined by loading the model with deadweight, construction personnel load, pouring and vibrating concrete load and wind load, as well as the force brought by the volume of the construction area in step six, that is, by the following method:

[0111] ;

[0112] in, represents the force caused by the volume of the construction area, Indicates the pouring density in the construction area. represents the acceleration due to gravity and m represents the weight.

[0113] It can be seen that this application can monitor the changes of every minute and every second in real time through image recognition of the concrete pouring process, and the construction is safe and reliable with simple operation. For some bridges with long mileage and relatively harsh construction environment, image recognition can effectively ensure construction safety. Compared with traditional support force analysis, support force analysis based on image recognition reduces manpower and material costs, and at the same time reduces errors caused by human factors; the support force condition can be analyzed in real time with a wide coverage; at the same time, this method is simple to operate, safe and reliable.

[0114] Compared with the prior art, this application first collects the initial video of the cast-in-place construction area, performs segmentation on the initial video of the construction area, and obtains the initial image after segmentation; then constructs a semantic segmentation network model, performs training on the semantic segmentation network model, calls the trained DeepLabv3 model, performs segmentation operations on the construction area represented by the initial image, and obtains the target image; uses the SAM2 model to track the square holes, obtains the square hole mask, and calculates the actual area represented by each pixel in the target image based on the square hole area; finally, based on the actual area, calculates the real-time volume of the construction area and determines the force analysis results of the support in the construction area. It uses computer image recognition technology to monitor the bridge construction and concrete pouring process in real time, calculates the area changes during the pouring process in real time, and thus determines the force changes of the support during the pouring process, reducing manpower and material costs, while reducing errors caused by cognitive factors and improving construction safety.

[0115] See also Fig. 9 , Fig. 9 A structural schematic diagram of a cast-in-place full-process support force detection system based on image recognition provided by an embodiment of the present invention includes:

[0116] The acquisition module 901 is used to acquire the initial video of the cast-in-place construction area;

[0117] An execution module 902 is used to segment the initial video of the construction area to obtain an initial image after segmentation;

[0118] A construction module 903 is used to construct a semantic segmentation network model, execute training on the semantic segmentation network model, call the trained DeepLabv3 model, perform a segmentation operation on the construction area represented by the initial image, and obtain a target image;

[0119] The first calculation module 904 is used to track the square hole using the SAM2 model to obtain the square hole mask, and calculate the actual area represented by each pixel in the target image according to the square hole area;

[0120] The second calculation module 905 is used to calculate the real-time volume of the construction area and determine the force analysis result of the support in the construction area based on the actual area.

[0121] Compared with the prior art, this application first collects the initial video of the cast-in-place construction area, performs segmentation on the initial video of the construction area, and obtains the initial image after segmentation; then constructs a semantic segmentation network model, performs training on the semantic segmentation network model, calls the trained DeepLabv3 model, performs segmentation operations on the construction area represented by the initial image, and obtains the target image; uses the SAM2 model to track the square holes, obtains the square hole mask, and calculates the actual area represented by each pixel in the target image based on the square hole area; finally, based on the actual area, calculates the real-time volume of the construction area and determines the force analysis results of the support in the construction area. It uses computer image recognition technology to monitor the bridge construction and concrete pouring process in real time, calculates the area changes during the pouring process in real time, and thus determines the force changes of the support during the pouring process, reducing manpower and material costs, while reducing errors caused by cognitive factors and improving construction safety.

[0122] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to implement the steps in the above method embodiment when running.

[0123] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for performing the following steps:

[0124] S201: Collecting the initial video of the cast-in-place construction area;

[0125] S202: Segmenting the initial video of the construction area to obtain an initial image after segmentation;

[0126] S203: constructing a semantic segmentation network model, executing training on the semantic segmentation network model, calling the trained DeepLabv3 model, performing a segmentation operation on the construction area represented by the initial image, and obtaining a target image;

[0127] S204: Track the square hole using the SAM2 model to obtain a square hole mask, and calculate the actual area represented by each pixel in the target image according to the square hole area;

[0128] S205: Based on the actual area, calculate the real-time volume of the construction area and determine the force analysis result of the support in the construction area.

[0129] Specifically, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0130] Compared with the prior art, this application first collects the initial video of the cast-in-place construction area, performs segmentation on the initial video of the construction area, and obtains the initial image after segmentation; then constructs a semantic segmentation network model, performs training on the semantic segmentation network model, calls the trained DeepLabv3 model, performs segmentation operations on the construction area represented by the initial image, and obtains the target image; uses the SAM2 model to track the square holes, obtains the square hole mask, and calculates the actual area represented by each pixel in the target image based on the square hole area; finally, based on the actual area, calculates the real-time volume of the construction area and determines the force analysis results of the support in the construction area. It uses computer image recognition technology to monitor the bridge construction and concrete pouring process in real time, calculates the area changes during the pouring process in real time, and thus determines the force changes of the support during the pouring process, reducing manpower and material costs, while reducing errors caused by cognitive factors and improving construction safety.

[0131] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in the above method embodiment.

[0132] Specifically, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0133] Specifically, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0134] S201: Collecting the initial video of the cast-in-place construction area;

[0135] S202: Segmenting the initial video of the construction area to obtain an initial image after segmentation;

[0136] S203: constructing a semantic segmentation network model, executing training on the semantic segmentation network model, calling the trained DeepLabv3 model, performing a segmentation operation on the construction area represented by the initial image, and obtaining a target image;

[0137] S204: Track the square hole using the SAM2 model to obtain a square hole mask, and calculate the actual area represented by each pixel in the target image according to the square hole area;

[0138] S205: Based on the actual area, calculate the real-time volume of the construction area and determine the force analysis result of the support in the construction area.

[0139] Compared with the prior art, this application first collects the initial video of the cast-in-place construction area, performs segmentation on the initial video of the construction area, and obtains the initial image after segmentation; then constructs a semantic segmentation network model, performs training on the semantic segmentation network model, calls the trained DeepLabv3 model, performs segmentation operations on the construction area represented by the initial image, and obtains the target image; uses the SAM2 model to track the square holes, obtains the square hole mask, and calculates the actual area represented by each pixel in the target image based on the square hole area; finally, based on the actual area, calculates the real-time volume of the construction area and determines the force analysis results of the support in the construction area. It uses computer image recognition technology to monitor the bridge construction and concrete pouring process in real time, calculates the area changes during the pouring process in real time, and thus determines the force changes of the support during the pouring process, reducing manpower and material costs, while reducing errors caused by cognitive factors and improving construction safety.

[0140] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0141] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0142] In the several embodiments provided by the present invention, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the above units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0143] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0144] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0145] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the above-mentioned methods of each embodiment of the present invention. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0146] The embodiments of the present invention are described in detail above. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for detecting the force of a cast-in-place bracket during the entire process based on image recognition, the method comprising: Collect initial video of the cast-in-place construction area; Execute segmentation of the initial video of the construction area to obtain an initial image after segmentation; Constructing a semantic segmentation network model, executing training on the semantic segmentation network model, calling the trained DeepLabv3 model, performing a segmentation operation on the construction area represented by the initial image, and obtaining a target image; wherein the DeepLabv3 model represents the trained semantic segmentation network model; The square hole is tracked by using the SAM2 model to obtain a square hole mask, and the actual area represented by each pixel in the target image is calculated according to the square hole area; Based on the actual area, the real-time volume of the construction area is calculated and the force analysis results of the support in the construction area are determined.

2. The method according to claim 1, characterized in that After performing segmentation of the initial video of the construction area and obtaining the segmented initial image, the method further includes: Preprocessing of the initial image is performed; the preprocessing includes image scaling and normalization.

3. The method according to claim 2, characterized in that The step of constructing a semantic segmentation network model and training the semantic segmentation network model includes: The semantic segmentation network model is trained by setting the loss function and optimizer of the semantic segmentation network model through the cross entropy loss and Adam optimizer, and the fourth layer of the semantic segmentation network model classifier is replaced by the preset convolution layer so that the number of output channels is adjusted to 2; The loss function is expressed by the following formula: ； in, represents the cross entropy loss, represents the number of samples, represents the number of categories, Represents the one-hot encoding of the true label, Represents the probability of the output category c of the semantic segmentation network model.

4. The method according to claim 3, characterized in that The DeepLabv3 model includes: An encoder module, a dilated convolution fusion multi-scale information module, an image-level feature module, and a decoder module that are communicatively connected, wherein the encoder module is used to extract input features of the initial image frame; The hole convolution fusion multi-scale information module is used to capture information of different scales of the initial image frame and perform multi-scale feature extraction; The image-level feature module is used to fuse the context information of the multi-scale information module according to the encoder module and the dilated convolution to improve the image segmentation result of the construction area; The decoder module is used to output a target image.

5. The method according to claim 4, characterized in that Before tracking the square hole using the SAM2 model to obtain the square hole mask and calculating the actual area represented by each pixel in the target image according to the square hole area, the method further includes: Set a label for the center of the square hole and calculate the center point of the square hole label; wherein the calculation of the center point of the square hole label is performed by the following formula: ； ； in, Indicates the center coordinate in the horizontal direction, Indicates the center coordinate in the vertical direction, is the number of points, are the coordinates of each point.

6. The method according to claim 5, characterized in that The method of tracking the square hole using the SAM2 model to obtain the square hole mask and calculating the actual area represented by each pixel in the target image according to the square hole area includes: The SAM2 model is used to track the square holes, generate square hole masks, and determine the actual area of ​​the square hole region represented by each pixel by pixel counting; wherein the actual area calculation formula of the square hole region is: ； in, Indicates the actual area. represents the number of pixels in the target area, Indicates the actual area represented by each pixel; the number of pixels in the target area satisfies: ； in, Indicates the mask height, Indicates the mask width, If it is 1, the pixel is counted; if it is 0, it is not counted.

7. The method according to any one of claims 1 to 6, characterized in that: The method of calculating the real-time volume of the construction area and determining the force analysis result of the support in the construction area based on the actual area includes: The real-time volume of the construction area is calculated as follows: ； in, Represents the real-time volume of the construction area, Indicates the pouring height of the construction area; The force analysis results of the support in the construction area are determined by the following methods: ； in, represents the force caused by the volume of the construction area, Indicates the pouring density in the construction area. represents the acceleration due to gravity and m represents the weight.

8. A cast-in-place full-process support force detection system based on image recognition, the system comprising: The acquisition module is used to collect the initial video of the cast-in-place construction area; An execution module, used for executing segmentation of the initial video of the construction area to obtain an initial image after segmentation; A construction module is used to construct a semantic segmentation network model, execute training of the semantic segmentation network model, call the trained DeepLabv3 model, perform a segmentation operation on the construction area represented by the initial image, and obtain a target image; wherein the DeepLabv3 model represents the trained semantic segmentation network model; The first calculation module is used to track the square hole using the SAM2 model to obtain the square hole mask, and calculate the actual area represented by each pixel in the target image according to the square hole area; The second calculation module is used to calculate the real-time volume of the construction area and determine the force analysis results of the support in the construction area based on the actual area.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to implement the method according to any one of claims 1 to 7 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the computer program to implement the method according to any one of claims 1 to 7.

Images