Machine vision-based concrete vibrating pipe control method, device, equipment and storage medium

Abstract

The present application relates to the technical field of machine vision, and provides a concrete vibrating control method, device and equipment based on machine vision and a storage medium. Through real-time video acquisition and target detection, the vibrating personnel, vibrating rod and concrete storage surface are automatically identified, high-precision identification of the storage surface can be realized, manual intervention is reduced, and monitoring efficiency is improved. Through equivalent vibrating point fitting of the vibrating range, the vibrating state is comprehensively analyzed and judged, the working state and the idle state are accurately distinguished, and misjudgment is reduced. Through equivalent vibrating point extraction, merging and consideration of the vibrating influence radius, the vibrating time is accurately calculated, and the vibrating quality is ensured. In combination with the construction scheme, under-vibration and over-vibration early warning is carried out, the timeliness and scientificity of construction management are improved. Through visualization of the vibrating quality heat map, the vibrating effect is intuitively displayed, and support is provided for subsequent construction optimization and data analysis.

Description

Technical Field

[0001] This invention relates to the field of machine vision technology, and in particular to a method, apparatus, equipment and storage medium for controlling concrete vibration based on machine vision. Background Technology

[0002] Concrete pouring is typically carried out in sections and layers, and vibration is a crucial step in ensuring concrete density and structural strength. Traditional vibration operations rely entirely on the engineering experience of operators, leading to issues such as high subjectivity, arbitrariness, difficulty in quality traceability, and inability to consistently guarantee vibration quality. Furthermore, prolonged, high-intensity on-site supervision can cause fatigue among supervisors, further increasing the risk of missed inspections and misjudgments. In addition, traditional manual on-site supervision cannot accurately determine in real time whether the coverage of vibration points and the vibration time meet the requirements of the plan, easily resulting in under-vibration or over-vibration, affecting the durability and safety of the structure, and consequently increasing later maintenance costs.

[0003] With the rapid development of artificial intelligence (AI) technology, it has demonstrated powerful capabilities in image recognition, target detection, and behavior analysis, providing technical support for the intelligent transformation of various industries. In the field of hydropower engineering, especially in the process of concrete vibration, how to utilize AI technology to achieve automated and intelligent vibration control has become an urgent industry need. Currently, there is no mature solution capable of collaboratively identifying and analyzing vibration personnel, equipment, and the concrete surface. Furthermore, there is a lack of intelligent judgment and visual feedback regarding the location, time, and coverage of vibration points, making it difficult to meet the requirements of high-quality, high-efficiency, and low-risk modern construction management.

[0004] Currently, the results of controlling concrete vibration based on visual detection algorithms are all based on a single vibration point, without comprehensively considering the radius of action of the concrete vibrator and ignoring the vibration influence between adjacent vibration points. At the same time, during the construction process, the same vibration area is repeatedly vibrated multiple times. The existing control based on a single vibration point is difficult to achieve accurate superposition control, and the final time heat map has a large deviation.

[0005] Therefore, it is necessary to propose a machine vision-based method, device, equipment, and storage medium for concrete vibration control to solve or at least alleviate the above-mentioned defects. Summary of the Invention

[0006] The main objective of this invention is to provide a machine vision-based method, device, equipment, and storage medium for concrete vibration control, in order to solve the technical problem that existing vision-based vibration control methods suffer from inaccurate vibration quality control due to neglecting the radius of action of the vibrator and the inability to calculate the cumulative vibration time in overlapping areas.

[0007] To achieve the above objectives, the present invention provides a machine vision-based method for controlling concrete vibration, comprising the following steps:

[0008] S1, real-time acquisition of video stream of vibration operation on concrete surface;

[0009] S2, the video stream of the vibration operation is framed, and a pre-trained image recognition model is used to identify the area of ​​the vibrator operator, the area of ​​the vibrator rod, and the area of ​​the concrete surface in each frame.

[0010] S3, determine whether the vibrator is in an effective working state based on the recognition result, and when the vibrator is in an effective working state, extract the equivalent vibration point in the current frame image and record the number of vibrations.

[0011] S4, repeat steps S2~S3 until the current layer of concrete vibration construction is completed, and obtain the first set; wherein, the first set includes multiple equivalent vibration points, and each equivalent vibration point includes the equivalent vibration point pixel coordinates and the cumulative number of vibrations;

[0012] S5, Spatial merging of equivalent vibration points in the first set is performed to generate a second set;

[0013] S6, based on the frame-sampling frequency of the vibration operation video stream, convert the number of vibrations at each equivalent vibration point in the second set into the actual vibration time, and generate the third set;

[0014] S7. Based on the preset standard vibration time range and the actual vibration time in the third set, determine whether each equivalent vibration point has under-vibration or over-vibration; if so, generate a warning message.

[0015] S8 generates a vibration quality heat map based on the actual vibration time of each equivalent vibration point for visualization.

[0016] Preferably, the image recognition model in step S2 is an instance segmentation model; the instance segmentation model is obtained through the following steps:

[0017] S21, Collect images of on-site vibration operation and construct a training dataset;

[0018] S22, Use the annotation tool to annotate each image in the training dataset; wherein, the areas of the vibrator operator, the vibrator rod, and the concrete surface are annotated to generate corresponding instance segmentation mask labels;

[0019] S23, the labeled training dataset is divided into training set, validation set and test set according to a preset ratio;

[0020] S24, the instance segmentation neural network is trained using the training set, the parameters are tuned using the validation set, and the performance of the instance segmentation model is evaluated using the test set until the model converges or the performance on the validation set reaches a preset index, thus obtaining the trained instance segmentation model.

[0021] Preferably, step S3 includes the following steps:

[0022] S31, calculate the intersection ratio of the vibrating personnel area and the vibrating rod area in the identification result. When the intersection ratio is greater than a set threshold, the vibrating rod detection is determined to be correct; when the intersection ratio is less than or equal to the set threshold, the vibrating rod detection result is filtered.

[0023] S32 performs a geometric intersection calculation between the vibrator area and the concrete surface area. If there is an intersection between the vibrator area and the concrete surface area, the current operation state is marked as a valid operation state; if there is no intersection between the vibrator area and the concrete surface area, the current operation state is marked as an idle hovering state.

[0024] S33, when the current operation state is vibration operation state, extract the contour points of the vibrating rod area, and select the point at the bottom of the vibrating rod as the equivalent vibration point according to the contour points, and record the pixel coordinates of the equivalent vibration point and the number of vibrations in one operation.

[0025] Preferably, step S5 includes the following steps:

[0026] S51, traverse the first set. If the pixel distance between any two equivalent vibration points is less than the first preset threshold, merge the two equivalent vibration points to obtain the merged first set. The merged pixel coordinates are the average pixel coordinates of the merged equivalent vibration points, and the merged vibration count is the sum of the vibration counts of the merged equivalent vibration points.

[0027] S52, obtain the theoretical influence radius of the vibrating rod, and determine the second preset threshold based on the theoretical influence radius. Then, traverse the first set merged in step S51 again. If the pixel distance between any two equivalent vibration points is less than the second preset threshold, the vibration count of the two equivalent vibration points is accumulated, and the pixel coordinates of each equivalent vibration point remain unchanged, thereby obtaining the second set.

[0028] Preferably, step S6 includes the following steps:

[0029] Using formula The number of vibrations at each equivalent vibration point in the second set is converted into the actual vibration time. Generate a third set; where, This refers to the cumulative number of vibrations at the equivalent vibration points in the second set. To extract the number of original video frames that are spaced between two adjacent frames. This refers to the number of frames per second in the video stream of the vibration operation.

[0030] Preferably, step S7 includes the following steps:

[0031] Compare the time vibration time of each equivalent vibration point in the third set with the preset standard vibration time range;

[0032] If the actual vibration time at the equivalent vibration point is less than the minimum value of the preset standard vibration time range, the equivalent vibration point is determined to be an under-vibration point, and an under-vibration warning message is generated and pushed to the on-site management personnel in real time.

[0033] If the actual vibration time at the equivalent vibration point exceeds the maximum value of the preset standard vibration time range, the equivalent vibration point is determined to be an over-vibration point, an over-vibration warning message is generated, and it is pushed to the site management personnel in real time.

[0034] Preferably, step S8 includes the following steps:

[0035] Using each equivalent vibration point in the third set as an interpolation node and the actual vibration time of each equivalent vibration point as an interpolation attribute, a continuous vibration time distribution surface is generated in the concrete placement area through a spatial interpolation algorithm.

[0036] Based on the vibration time distribution surface, a color gradient is used to render and generate a vibration quality heat map;

[0037] The vibration quality thermal map is overlaid and fused with the pre-stored silo surface bottom map;

[0038] The combined and merged results are sent to the backend management system for storage and display.

[0039] The present invention also provides a machine vision-based concrete vibration control device for performing the machine vision-based concrete vibration control method described above, comprising:

[0040] The image recognition unit is used to acquire the video stream of the vibration operation of the concrete surface in real time, extract frames from the vibration operation video stream, and use a pre-trained image recognition model to identify the area of ​​the vibrator operator, the area of ​​the vibrator rod, and the area of ​​the concrete surface in each frame.

[0041] The feature extraction unit is used to extract frames from the vibration operation video stream and use a pre-trained image recognition model to identify the vibration personnel area, vibration rod area and concrete surface area in each frame image.

[0042] The loop execution unit is used to repeatedly execute steps S2 to S3 until the current layer of concrete vibration construction is completed, and a first set is obtained; wherein, the first set includes multiple equivalent vibration points, and each equivalent vibration point includes the equivalent vibration point pixel coordinates and the cumulative number of vibrations;

[0043] The merging processing unit is used to perform spatial merging processing on the equivalent vibration points in the first set to generate a second set.

[0044] The conversion calculation unit is used to convert the number of vibrations at each equivalent vibration point in the second set into the actual vibration time based on the frame extraction frequency of the vibration operation video stream, and generate a third set.

[0045] The real-time control unit is used to determine whether each equivalent vibration point is under-vibrated or over-vibrated based on the preset standard vibration time range and the actual vibration time in the third set; if so, it generates a warning message; and generates a vibration quality heat map based on the actual vibration time of each equivalent vibration point for visualization.

[0046] The present invention also provides a machine vision-based concrete vibration control device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the machine vision-based concrete vibration control method described above.

[0047] The present invention also provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the machine vision-based concrete vibration control method described above.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] This invention provides a machine vision-based method, device, equipment, and storage medium for concrete vibration control. Through real-time video acquisition and target detection, it automatically identifies vibration personnel, vibrators, and the concrete surface, achieving high-precision surface identification, reducing manual intervention, and improving monitoring efficiency. By fitting equivalent vibration points to the vibration range, it comprehensively analyzes and judges the vibration state, accurately distinguishing between working and idle states, reducing misjudgments. Through equivalent vibration point extraction and merging, and considering the radius of vibration influence, it accurately calculates vibration time, ensuring vibration quality. Combined with the construction plan, it provides early warnings for under-vibration and over-vibration, improving the timeliness and scientific nature of construction management. Visualizing the vibration quality heatmap provides a clear view of the vibration effect, supporting subsequent construction optimization and data analysis. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0051] Figure 1 This is a schematic flowchart of one embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of the vibration mass thermal curve in one embodiment of the present invention.

[0053] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0054] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0056] In this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. If the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0057] Please refer to Figures 1 to 2 The present invention provides a machine vision-based method for controlling concrete vibration, comprising the following steps:

[0058] S1, real-time acquisition of video streams of the vibration operation on the concrete pouring surface; for example, at a concrete pouring site of a certain project, a portable intelligent sensing terminal is set up to acquire video of the vibration operation surface in real time and transmit it to the back-end server via network signal. The video shooting terminal can use existing equipment, which will not be elaborated here.

[0059] S2, the video stream of the vibration operation is framed, and a pre-trained image recognition model is used to identify the area of ​​the vibrator operator, the area of ​​the vibrator rod, and the area of ​​the concrete surface in each frame.

[0060] S3. Based on the recognition result, determine whether the vibrator is in an effective working state. When the vibrator is in an effective working state, extract the equivalent vibration point in the current frame image and record the number of vibrations. The association verification can be performed by calculating the intersection ratio between the vibrator and the personnel, and the effective working state can be determined by calculating the intersection area between the vibrator and the outline of the concrete surface. Then, extract the point at the bottom of the vibrator as the equivalent vibration point and record the number of vibrations. The following embodiments will describe this in detail.

[0061] S4, repeat steps S2~S3 until the current layer of concrete vibration construction is completed, and obtain the first set; wherein, the first set includes multiple equivalent vibration points, and each equivalent vibration point includes the equivalent vibration point pixel coordinates and the cumulative number of vibrations;

[0062] S5, Spatial merging of equivalent vibration points in the first set is performed to generate a second set;

[0063] S6, based on the frame-sampling frequency of the vibration operation video stream, convert the number of vibrations at each equivalent vibration point in the second set into the actual vibration time, and generate the third set;

[0064] Further, step S6 includes the following steps:

[0065] Using formula The number of vibrations at each equivalent vibration point in the second set is converted into the actual vibration time. Generate a third set; where, This refers to the cumulative number of vibrations at the equivalent vibration points in the second set. To extract the number of original video frames that are spaced between two adjacent frames. This refers to the number of frames per second in the vibration operation video stream. Using a mathematical conversion model based on the fundamental principle of frame-sampling detection, the equivalent time of the continuous vibration operation process is accurately calculated and quantified, providing a reliable data foundation for subsequent quality assessment.

[0066] S7. Based on the preset standard vibration time range and the actual vibration time in the third set, determine whether each equivalent vibration point has under-vibration or over-vibration; if so, generate a warning message.

[0067] S8 generates a vibration quality heat map based on the actual vibration time of each equivalent vibration point for visualization.

[0068] In this application, real-time video acquisition and target detection automatically identify vibration personnel, vibrators, and the concrete surface, achieving high-precision surface identification, reducing manual intervention, and improving monitoring efficiency. By fitting equivalent vibration points that integrate the vibration range, the vibration status is comprehensively analyzed and judged, accurately distinguishing between working and idle states, reducing misjudgments. By extracting and merging equivalent vibration points and considering the radius of vibration influence, the vibration time is accurately calculated to ensure vibration quality. Under-vibration and over-vibration warnings are provided in conjunction with the construction plan, improving the timeliness and scientific nature of construction management. The vibration quality heatmap visualization intuitively displays the vibration effect, providing support for subsequent construction optimization and data analysis.

[0069] In a preferred embodiment, the image recognition model in step S2 is an instance segmentation model; the instance segmentation model is obtained through the following steps:

[0070] S21, Collect images of on-site vibration operation and construct a training dataset;

[0071] S22, each image in the training dataset is labeled using a labeling tool; wherein, the vibrator area, vibrator rod area, and concrete surface area are labeled to generate corresponding instance segmentation mask labels; the labeling tool can be the open-source labeling tool LabelMe, and the generated instance segmentation mask labels label the precise contour information of the vibrator area, vibrator rod area, and concrete surface area, so that the finally trained instance segmentation model can not only identify where there are labels, but also identify what the corresponding shape is, thereby improving the final judgment accuracy.

[0072] S23, the labeled training dataset is divided into training set, validation set and test set according to a preset ratio;

[0073] S24, the instance segmentation neural network is trained using the training set, the parameters are tuned using the validation set, and the performance of the instance segmentation model is evaluated using the test set until the model converges or the performance on the validation set reaches a preset index, thus obtaining the trained instance segmentation model.

[0074] In a preferred embodiment, step S3 includes the following steps:

[0075] S31, calculate the intersection ratio of the vibrator personnel area and the vibrator rod area in the identification results. When the intersection ratio is greater than a set threshold, the vibrator rod detection is determined to be correct; when the intersection ratio is less than or equal to the set threshold, the vibrator rod detection result is filtered. The set threshold can be set based on experience during the current layer surface construction, or determined through trial operation in the early stages of deployment of this application scheme. As a preferred example, the set threshold is 0.15.

[0076] S32 performs a geometric intersection calculation between the vibrator area and the concrete surface area. If there is an intersection between the vibrator area and the concrete surface area, the current operation state is marked as a valid operation state; if there is no intersection between the vibrator area and the concrete surface area, the current operation state is marked as an idle hovering state.

[0077] As a better example, a closed surface can be constructed by the outer contour of the vibrator area and the outer contour of the concrete surface area, and the intersection area can be solved. If the area of ​​the intersection area is greater than 0.3 of the area of ​​the vibrator area, the current state is marked as the vibration operation state; otherwise, it is marked as the idle hovering state. This step can accurately determine whether the vibration point is in an effective operation state and avoid including invalid time in the operation time.

[0078] S33, when the current operation state is vibration operation state, extract the contour points of the vibrating rod area, and select the point at the bottom of the vibrating rod as the equivalent vibration point according to the contour points, and record the pixel coordinates of the equivalent vibration point and the number of vibrations in one operation.

[0079] Specifically, the point at the bottom of the vibrator can be selected as the equivalent vibration point along the length of the vibrator area. The number of vibrations and the pixel coordinates of this point are recorded in the format (1,x,y), where x and y are pixel coordinates, thereby achieving precise point determination of the vibration point.

[0080] In a preferred embodiment, step S5 includes the following steps:

[0081] S51, traverse the first set. If the pixel distance between any two equivalent vibration points is less than the first preset threshold, merge the two equivalent vibration points to obtain the merged first set. The merged pixel coordinates are the average pixel coordinates of the merged equivalent vibration points, and the merged vibration count is the sum of the vibration counts of the merged equivalent vibration points.

[0082] Specifically, in the first merging process, a small initial preset threshold (e.g., 5-15 pixels) is set. The first set is traversed, and equivalent vibration points with a pixel distance less than the first preset threshold are merged. The coordinates of the merged new point are the average coordinates of the merged points, and the number of vibrations is the sum of their cumulative values. This merging process filters out redundant and duplicate points spatially generated due to interference factors (such as equipment vibration or detection errors), improving data accuracy and cleanliness. It effectively solves the problem in existing vision-based detection methods that do not consider detection errors, resulting in a large amount of noise in the original data, affecting the accuracy of subsequent analysis.

[0083] S52, obtain the theoretical influence radius of the vibrating rod, and determine the second preset threshold based on the theoretical influence radius. Then, traverse the first set merged in step S51 again. If the pixel distance between any two equivalent vibration points is less than the second preset threshold, the vibration count of the two equivalent vibration points is accumulated, and the pixel coordinates of each equivalent vibration point remain unchanged, thereby obtaining the second set.

[0084] Through a second merging process, considering the theoretical influence radius of the vibrator, and setting a larger second preset threshold (e.g., 40-60 pixels), the merged first set is traversed again. For equivalent vibration points with a distance less than the second preset threshold, only their vibration counts are accumulated, while their individual pixel coordinates remain unchanged. It is worth noting that the effects of vibration within the influence range are additive. Without changing the actual position of the work point, the total vibration energy received within the area of ​​influence between two equivalent vibration points is accurately quantified by accumulating the vibration counts. This ensures that the final generated actual vibration time accurately reflects the actual vibration conditions in the area, improving the accuracy of the assessment. This effectively solves the problem of existing technologies that rely on a single vibration point for control, completely ignoring the influence between adjacent vibration points, ultimately achieving precise control over the quality of concrete vibration.

[0085] Further, step S7 includes the following steps:

[0086] Compare the time vibration time of each equivalent vibration point in the third set with the preset standard vibration time range;

[0087] If the actual vibration time at the equivalent vibration point is less than the minimum value of the preset standard vibration time range, the equivalent vibration point is determined to be an under-vibration point, and an under-vibration warning message is generated and pushed to the on-site management personnel in real time.

[0088] If the actual vibration time at the equivalent vibration point exceeds the maximum value of the preset standard vibration time range, the equivalent vibration point is determined to be an over-vibration point, an over-vibration warning message is generated, and it is pushed to the site management personnel in real time.

[0089] This embodiment automates the comparison between the actual vibration time at each equivalent vibration point and a preset standard vibration time range, thereby shifting quality judgment from relying on manual experience to basing it on objective data and eliminating human subjectivity and arbitrariness. The preset standard vibration time range can be determined in advance by those skilled in the art based on the actual construction plan.

[0090] Further, step S8 includes the following steps:

[0091] Using each equivalent vibration point in the third set as an interpolation node and the actual vibration time of each equivalent vibration point as an interpolation attribute, a continuous vibration time distribution surface is generated within the concrete placement area using a spatial interpolation algorithm. The calculation is performed within the concrete placement area using each equivalent vibration point in the third set as an interpolation node, its actual vibration time as an interpolation attribute, and a spatial interpolation algorithm (such as the inverse distance weighting method, Kriging method, or radial basis function method).

[0092] Based on the vibration time distribution surface, a color gradient is used for rendering to generate a vibration quality heatmap; as shown... Figure 2 The heatmap shown uses a gradient from blue to red, with redder areas representing longer vibration times and lighter areas (like blue) representing shorter times. This improves data interpretation efficiency and allows for instantaneous assessment of complex construction quality. Vibration time is a core quality control factor for ensuring concrete compaction; it cannot be too long or too short. Therefore, the heatmap's color-coding allows for a direct visual identification of under- or over-vibrated areas. Furthermore, the uniformity of color distribution on the heatmap helps workers assess the uniformity of concrete vibration.

[0093] The vibration quality thermal map is overlaid and fused with a pre-stored silo surface base map; wherein, the pre-stored silo surface base map can be the first frame image of the silo surface stored as the base map before starting the algorithm of this application by adjusting the zoom, magnification, etc. of the portable smart terminal.

[0094] The combined and merged results are sent to the backend management system for storage and display.

[0095] The present invention also provides a machine vision-based concrete vibration control device for performing the machine vision-based concrete vibration control method described above, comprising:

[0096] The image recognition unit is used to acquire the video stream of the vibration operation of the concrete surface in real time, extract frames from the vibration operation video stream, and use a pre-trained image recognition model to identify the area of ​​the vibrator operator, the area of ​​the vibrator rod, and the area of ​​the concrete surface in each frame.

[0097] The feature extraction unit is used to extract frames from the vibration operation video stream and use a pre-trained image recognition model to identify the vibration personnel area, vibration rod area and concrete surface area in each frame image.

[0098] The loop execution unit is used to repeatedly execute steps S2 to S3 until the current layer of concrete vibration construction is completed, and a first set is obtained; wherein, the first set includes multiple equivalent vibration points, and each equivalent vibration point includes the equivalent vibration point pixel coordinates and the cumulative number of vibrations;

[0099] The merging processing unit is used to perform spatial merging processing on the equivalent vibration points in the first set to generate a second set.

[0100] The conversion calculation unit is used to convert the number of vibrations at each equivalent vibration point in the second set into the actual vibration time based on the frame extraction frequency of the vibration operation video stream, and generate a third set.

[0101] The real-time control unit is used to determine whether each equivalent vibration point is under-vibrated or over-vibrated based on the preset standard vibration time range and the actual vibration time in the third set; if so, it generates a warning message; and generates a vibration quality heat map based on the actual vibration time of each equivalent vibration point for visualization.

[0102] The present invention also provides a machine vision-based concrete vibration control device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the machine vision-based concrete vibration control method described above.

[0103] The present invention also provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the machine vision-based concrete vibration control method described above.

[0104] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may 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. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0105] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A machine vision-based method for controlling a concrete vibrating pipe, characterized in that, The method comprises the following steps: S1, collecting a concrete surface vibration operation video stream in real time; S2, frame extraction is performed on the vibration operation video stream, and a pre-trained image recognition model is used to identify a vibration personnel region, a vibration rod region and a concrete surface region in each frame of image; S3, whether the vibration rod is in an effective operation state is determined according to the identification result, and when the vibration rod is in the effective operation state, an equivalent vibration point in the current frame of image is extracted, and a vibration frequency is recorded; S4, steps S2-S3 are repeatedly executed until the concrete vibration construction of the current layer of surface is completed, and a first set is obtained; wherein the first set comprises a plurality of equivalent vibration points, and each equivalent vibration point comprises equivalent vibration point pixel coordinates and cumulative vibration frequency; S5, spatial merging processing is performed on the equivalent vibration points in the first set, and a second set is generated; S6, according to the frame extraction frequency of the vibration operation video stream, the vibration frequency of each equivalent vibration point in the second set is converted into actual vibration time, and a third set is generated; S7, according to a preset standard vibration time range and the actual vibration time in the third set, whether each equivalent vibration point is under-vibrated or over-vibrated is determined; if yes, an alarm information is generated; S8, a vibration quality heat map is generated according to the actual vibration time of each equivalent vibration point for visual display. 2.The machine vision-based concrete vibrating pipe control method according to claim 1, wherein, The image recognition model in step S2 is an instance segmentation model; the instance segmentation model is obtained by the following steps: S21, collecting images of on-site vibration operation to construct a training data set; S22, using a labeling tool to label each image in the training data set; wherein the vibration personnel region, the vibration rod region and the concrete surface region are labeled respectively to generate corresponding instance segmentation mask labels; S23, dividing the labeled training data set into a training set, a validation set and a test set according to a preset proportion; S24, using the training set to train an instance segmentation neural network, using the validation set to optimize parameters, using the test set to evaluate the performance of the instance segmentation model, until the model converges or the performance on the validation set reaches a preset index, and obtaining the trained instance segmentation model. 3.The machine vision-based concrete vibrating pipe control method according to claim 1, wherein, The step S3 comprises the following steps: S31, calculating the intersection-over-union of the vibration personnel region and the vibration rod region in the identification result, when the intersection-over-union is greater than a set threshold, determining that the vibration rod detection is correct; when the intersection-over-union is less than or equal to the set threshold, filtering the vibration rod detection result; S32, performing a geometric intersection operation between the vibration rod region and the concrete surface region, when there is an intersection between the vibration rod region and the concrete surface region, marking the current operation state as an effective operation state; when there is no intersection between the vibration rod region and the concrete surface region, marking the current operation state as an idle hovering state; S33, when the current operation state is a vibration operation state, extracting contour points of the vibration rod region, and selecting a lowermost point of the vibration rod as an equivalent vibration point according to the contour points, and recording pixel coordinates and vibration frequency of the equivalent vibration point.

4. The machine vision-based concrete tremie control method of claim 3, wherein, The step S5 comprises the following steps: S51, traversing the first set, if the pixel distance of any two equivalent vibrating points is less than a first preset threshold, merging the two equivalent vibrating points to obtain a merged first set; wherein the merged pixel coordinates are the average of the pixel coordinates of the merged equivalent vibrating points, and the merged vibrating frequency is the sum of the vibrating frequencies of the merged equivalent vibrating points; S52, obtaining a theoretical influence radius of the vibrating rod, determining a second preset threshold according to the theoretical influence radius, and traversing the first set merged in step S51 again, if the pixel distance of any two equivalent vibrating points is less than the second preset threshold, accumulating the vibrating frequencies of the two equivalent vibrating points, and the pixel coordinates of each equivalent vibrating point remain unchanged, thereby obtaining the second set. 5.The machine vision-based concrete vibrating pipe control method according to claim 4, wherein, The step S6 includes the following steps: The formula is as follows: The number of vibrations of each equivalent vibrating point in the second set is converted into actual vibrating time to generate a third set; wherein, is the cumulative number of vibrations of the equivalent vibrating point in the second set, is the number of original video frames spaced between the adjacent two frames of images, is the number of frames per second of the vibrating operation video stream.

6. The machine vision-based concrete tremie control method of claim 1, wherein, The step S7 includes the following steps: Comparing the time vibrating time of each equivalent vibrating point in the third set with a preset standard vibrating time range; If the actual vibrating time of the equivalent vibrating point is less than the minimum value of the preset standard vibrating time range, the equivalent vibrating point is determined as an under-vibrating point, an under-vibration warning information is generated, and is pushed to the on-site management personnel in real time; If the actual vibrating time of the equivalent vibrating point is greater than the maximum value of the preset standard vibrating time range, the equivalent vibrating point is determined as an over-vibrating point, an over-vibration warning information is generated, and is pushed to the on-site management personnel in real time.

7. The machine vision-based concrete tremie control method of claim 1, wherein, The step S8 includes the following steps: Taking each equivalent vibrating point in the third set as an interpolation node and taking the actual vibrating time of each equivalent vibrating point as an interpolation attribute, a continuous vibrating time distribution surface is generated in the concrete storage surface area by a spatial interpolation algorithm; According to the vibrating time distribution surface, a color gradient is used for rendering to generate a vibrating quality heat map; The vibrating quality heat map is superimposed and fused with a pre-stored storage surface base map; The superimposed and fused result is sent to a background management system for storage and display. 8.A machine vision-based concrete vibrating pipe control device, configured to perform the machine vision-based concrete vibrating pipe control method according to any one of claims 1-7, characterized in that, It includes: An image recognition unit is configured to collect a vibrating operation video stream of a concrete storage surface in real time, frame the vibrating operation video stream, and identify a vibrating personnel area, a vibrating rod area, and a concrete storage surface area in each frame of image by using a pre-trained image recognition model; A feature extraction unit is configured to frame the vibrating operation video stream and identify a vibrating personnel area, a vibrating rod area, and a concrete storage surface area in each frame of image by using a pre-trained image recognition model; A loop execution unit is configured to loop execute steps S2-S3 until the concrete vibrating construction of the current layer storage surface is completed to obtain a first set; wherein the first set includes a plurality of equivalent vibrating points, and each equivalent vibrating point includes an equivalent vibrating point pixel coordinate and a cumulative vibrating frequency; A merging processing unit is configured to perform spatial merging processing on the equivalent vibrating points in the first set to generate a second set; A conversion calculation unit is configured to convert the vibrating frequency of each equivalent vibrating point in the second set into an actual vibrating time according to the frame frequency of the vibrating operation video stream to generate a third set. A real-time control unit is configured to determine whether each equivalent vibrating point is under-vibrated or over-vibrated according to a preset standard vibrating time range and the actual vibrating time in the third set; if so, generate an alert message; generate a vibrating quality heat map according to the actual vibrating time of each equivalent vibrating point for visual display. 9.A machine vision-based concrete vibrating pipe control device, characterized in that, A computer program product comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the machine vision-based concrete vibrating control method according to any one of claims 1 to 7 when executing the computer program.

10. A storage medium storing a computer program, characterized by The computer program product is configured to implement the steps of the machine vision-based concrete vibrating control method according to any one of claims 1 to 7 when executed by the processor.

Images