A construction engineering intelligent monitoring method and system based on feature analysis

Abstract

The present invention discloses a method and system for intelligent monitoring of construction projects based on feature analysis. By setting a monitoring unit in a preset construction area to obtain construction video in real time, the video data is subjected to key frame extraction and PCA analysis processes to identify personnel areas and body movement features, form personnel information and movement data, and screen out target personnel and non-target personnel. During the analysis cycle, the non-target personnel movement feature data is used as the second movement feature data, and is vectorized together with the target personnel movement feature data and stored in the form of a queue. The gray correlation method is used to analyze the correlation between the two queue data, and the non-target personnel are quickly identified and recorded based on the correlation to form group identification information. Finally, based on the group information, personnel operation statistics, project progress analysis and monitoring resource dynamic planning are carried out, and the monitoring plan is effectively optimized, the efficiency of project statistical analysis is improved, and effective data support is provided for subsequent project progress analysis and construction plan optimization.

Description

Technical Field

[0001] The present invention relates to the field of construction engineering monitoring and analysis, and more specifically, to a construction engineering intelligent monitoring method and system based on feature analysis. Background Art

[0002] In the construction industry, as projects expand in scale and complexity, the need for monitoring and managing the construction process is becoming increasingly urgent. Traditional monitoring methods often rely on manual inspections, which are not only inefficient but also difficult to achieve real-time, comprehensive monitoring of the entire construction process. Furthermore, traditional construction project identification models struggle to quickly and accurately identify personnel and conduct activity statistics, limiting their application in construction progress management, personnel deployment, and resource optimization.

[0003] Therefore, there is an urgent need for an intelligent monitoring method and system for construction projects based on feature analysis. Summary of the Invention

[0004] The present invention overcomes the defects of the prior art and proposes a construction engineering intelligent monitoring method and system based on feature analysis.

[0005] A first aspect of the present invention provides a construction engineering intelligent monitoring method based on feature analysis, comprising:

[0006] Based on the construction project plan, monitoring units are set up in the preset construction area, and real-time video of the construction work is acquired through the monitoring units to obtain monitoring video data;

[0007] Extract key frames from the monitoring video data, and use the PCA analysis method to perform personnel area recognition and body movement feature extraction to form personnel recognition information and movement feature data;

[0008] Based on the personnel identification information, it is determined whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked;

[0009] In each analysis cycle, the motion feature data of non-target persons is obtained and marked as the second motion feature data. The target motion feature data and the second motion feature data are feature-vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance and serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group.

[0010] Based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed;

[0011] Based on group identification information, personnel work statistics and construction project progress analysis are carried out, and monitoring resources are dynamically planned to generate monitoring plans.

[0012] In this solution, based on the construction project plan, a monitoring unit is set up in a preset construction area, and real-time video of the construction work is acquired through the monitoring unit to obtain monitoring video data, specifically:

[0013] Analyze the construction workload of personnel in the preset construction area through the construction project plan, and set up monitoring units in the preset construction area;

[0014] The monitoring unit acquires real-time video of the construction operation to obtain monitoring video data.

[0015] In this solution, the monitoring video data is subjected to key frame extraction, and the personnel area recognition and body movement feature extraction are performed through the PCA analysis method to form personnel recognition information and movement feature data, specifically:

[0016] Extract key frames from monitoring video data to obtain an image set;

[0017] Perform denoising, image enhancement and standardization preprocessing on the image set;

[0018] Image frames are selected from the image set for analysis. Based on the PCA analysis method, feature extraction and personnel area recognition are performed on the image frames to obtain personnel identification information. Personnel areas are set based on the personnel identification information. Through multiple image frames, features of the personnel areas are integrated to form motion feature data.

[0019] In this solution, the personnel identification information is used to determine whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked, specifically:

[0020] Select a preset area and determine whether the number of corresponding personnel is greater than the preset number based on personnel identification information;

[0021] If so, the action feature data is imported into the recognition model for personnel operation recognition, the successfully recognized features are marked and the recognition result information is extracted to obtain the target person and target action feature data.

[0022] In this solution, in each analysis cycle, the motion feature data of non-target persons is obtained and marked as second motion feature data. The target motion feature data and the second motion feature data are feature vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance, and the sorted two sets of feature vectors are serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group, specifically:

[0023] Setting an analysis cycle, and for each analysis cycle, obtaining motion feature data of non-target persons and marking the data as second motion feature data;

[0024] Performing feature vectorization on the target motion feature data and the second motion feature data to form a first set of feature vectors and a second set of feature vectors respectively;

[0025] Using cosine distance, in the first group of eigenvectors, the sum of the distances between each eigenvector and the remaining eigenvectors is calculated, and the mean of the distance sum is calculated to obtain the mean distance. The overall similarity of each eigenvector is set by the mean distance, and the eigenvectors in the group are sorted by the overall similarity to achieve the ordering of the first group of eigenvectors.

[0026] The second set of eigenvectors is calculated and analyzed by cosine distance, and the second set of eigenvectors is ordered;

[0027] The ordered first set of feature vectors and the second set of feature vectors are sequentially entered into the queue structure for data storage based on the feature vector sequence, forming two queue data.

[0028] In this solution, the grey correlation method is used to perform correlation analysis on the two queue data as the original analysis series. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed, specifically:

[0029] Based on the grey correlation method, the two cohort data were used as the original analysis series, one cohort was selected as the reference series, and the other cohort was selected as the comparison series;

[0030] Through the queue structure, the eigenvectors are output in sequence for difference calculation, and the absolute difference, maximum difference, and minimum difference between the reference series and the comparison series are calculated. The difference calculation process is based on the cosine distance calculation of the eigenvectors, and the corresponding correlation coefficient and correlation degree are obtained based on the absolute difference, maximum difference, and minimum difference.

[0031] Determine whether the correlation is greater than a preset value. If so, update the corresponding non-target person to a target person and identify the corresponding action feature data within the current analysis cycle;

[0032] In one analysis cycle, a correlation analysis is performed on all non-target persons, and the identification information of the target persons is recorded to obtain group identification information.

[0033] In this solution, personnel operation statistics and construction project progress analysis are performed based on group identification information, and monitoring resources are dynamically planned to generate a monitoring plan. Specifically:

[0034] Through group identification information, the workload and type of work of personnel are counted, and the project progress is analyzed in combination with the construction project plan. Further monitoring demand analysis and monitoring unit adjustment are carried out on the preset construction area to generate a new monitoring plan.

[0035] A second aspect of the present invention further provides a construction project intelligent monitoring system based on feature analysis, the system comprising: a memory and a processor, the memory comprising a construction project intelligent monitoring program based on feature analysis, the construction project intelligent monitoring program based on feature analysis, when executed by the processor, implementing the following steps:

[0036] Based on the construction project plan, monitoring units are set up in the preset construction area, and real-time video of the construction work is acquired through the monitoring units to obtain monitoring video data;

[0037] Extract key frames from the monitoring video data, and use the PCA analysis method to perform personnel area recognition and body movement feature extraction to form personnel recognition information and movement feature data;

[0038] Based on the personnel identification information, it is determined whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked;

[0039] In each analysis cycle, the motion feature data of non-target persons is obtained and marked as the second motion feature data. The target motion feature data and the second motion feature data are feature-vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance and serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group.

[0040] Based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed;

[0041] Based on group identification information, personnel work statistics and construction project progress analysis are carried out, and monitoring resources are dynamically planned to generate monitoring plans.

[0042] The third aspect of the present invention also provides a computer-readable storage medium, which includes a construction project intelligent monitoring program based on feature analysis. When the construction project intelligent monitoring program based on feature analysis is executed by a processor, the steps of the construction project intelligent monitoring method based on feature analysis as described in any one of the above items are implemented.

[0043] The present invention discloses a method and system for intelligent monitoring of construction projects based on feature analysis. By setting a monitoring unit in a preset construction area to obtain construction video in real time, the video data is subjected to key frame extraction and PCA analysis processes to identify personnel areas and body movement features, form personnel information and movement data, and screen out target personnel and non-target personnel. During the analysis cycle, the non-target personnel movement feature data is used as the second movement feature data, and is vectorized together with the target personnel movement feature data and stored in the form of a queue. The gray correlation method is used to analyze the correlation between the two queue data, and the non-target personnel are quickly identified and recorded based on the correlation to form group identification information. Finally, based on the group information, personnel operation statistics, project progress analysis and monitoring resource dynamic planning are carried out, and the monitoring plan is effectively optimized, the efficiency of project statistical analysis is improved, and effective data support is provided for subsequent project progress analysis and construction plan optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A flow chart of a construction engineering intelligent monitoring method based on feature analysis according to the present invention is shown;

[0045] Figure 2 A block diagram of a construction engineering intelligent monitoring system based on feature analysis of the present invention is shown. DETAILED DESCRIPTION

[0046] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited to the specific embodiments disclosed below.

[0048] Figure 1 The flowchart of the intelligent monitoring method for construction projects based on feature analysis of the present invention is shown.

[0049] like Figure 1 As shown, the first aspect of the present invention provides a construction engineering intelligent monitoring method based on feature analysis, comprising:

[0050] S102, based on the construction project plan, setting up monitoring units in a preset construction area, and acquiring real-time video of the construction work through the monitoring units to obtain monitoring video data;

[0051] S104, extracting key frames from the monitoring video data, and performing personnel area recognition and body movement feature extraction using the PCA analysis method to generate personnel recognition information and movement feature data;

[0052] S106, based on the personnel identification information, determining whether the number of personnel is greater than a preset number; if so, performing preliminary personnel operation identification based on the motion feature data, marking the target personnel and target motion feature data;

[0053] S108: In each analysis cycle, motion feature data of non-target individuals is obtained and marked as second motion feature data. The target motion feature data and the second motion feature data are feature-vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance, and the sorted two sets of feature vectors are serialized and stored in a queue form to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group.

[0054] S110, based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than a preset value. If so, non-target personnel are marked as target personnel and recorded within the current analysis cycle. By analyzing all non-target personnel, group identification information is generated.

[0055] S112, based on the group identification information, personnel work statistics and construction project progress analysis are carried out, and monitoring resources are dynamically planned to generate a monitoring plan.

[0056] According to an embodiment of the present invention, based on the construction project plan, a monitoring unit is set in a preset construction area, and real-time video of the construction work is acquired through the monitoring unit to obtain monitoring video data, specifically:

[0057] Analyze the construction workload of personnel in the preset construction area through the construction project plan, and set up monitoring units in the preset construction area;

[0058] The monitoring unit acquires real-time video of the construction operation to obtain monitoring video data.

[0059] It should be noted that the construction project plan includes information such as construction plan, project quantity, and construction period.

[0060] According to an embodiment of the present invention, the monitoring video data is subjected to key frame extraction, and personnel area recognition and body movement feature extraction are performed through PCA analysis to form personnel recognition information and movement feature data, specifically:

[0061] Extract key frames from monitoring video data to obtain an image set;

[0062] Perform denoising, image enhancement and standardization preprocessing on the image set;

[0063] Image frames are selected from the image set for analysis. Based on the PCA analysis method, feature extraction and personnel area recognition are performed on the image frames to obtain personnel identification information. Personnel areas are set based on the personnel identification information. Through multiple image frames, features of the personnel areas are integrated to form motion feature data.

[0064] It should be noted that the personnel area generally includes multiple ones, which is determined by the number of people recognized by the image frame.

[0065] According to an embodiment of the present invention, the determination of whether the number of personnel is greater than a preset number based on personnel identification information is performed. If so, preliminary identification of personnel operations is performed through motion feature data, and target personnel and target motion feature data are marked, specifically:

[0066] Select a preset area and determine whether the number of corresponding personnel is greater than the preset number based on personnel identification information;

[0067] If so, the action feature data is imported into the recognition model for personnel operation recognition, the successfully recognized features are marked and the recognition result information is extracted to obtain the target person and target action feature data.

[0068] It should be noted that in larger construction projects, a pre-set area can be selected for construction crowd research and adjusted based on the scale of the construction project. The recognition model is a feature recognition model based on the CNN model, which can be used to identify and analyze personnel motion characteristics, facial features, and other features, and further implement job type analysis. It should be understood that the recognition model is only a preliminary recognition and analysis. When the number of workers is large, traditional recognition models have difficulty in achieving rapid recognition and job statistics. Non-target personnel corresponding to unsuccessfully identified personnel.

[0069] According to an embodiment of the present invention, in each analysis cycle, the motion feature data of non-target persons is obtained and marked as second motion feature data, the target motion feature data and the second motion feature data are feature vectorized to form two sets of feature vectors respectively, and the two sets of feature vectors are sorted based on feature importance, and the sorted two sets of feature vectors are serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group, specifically:

[0070] Setting an analysis cycle, and for each analysis cycle, obtaining motion feature data of non-target persons and marking the data as second motion feature data;

[0071] Performing feature vectorization on the target motion feature data and the second motion feature data to form a first set of feature vectors and a second set of feature vectors respectively;

[0072] Using cosine distance, in the first group of eigenvectors, the sum of the distances between each eigenvector and the remaining eigenvectors is calculated, and the mean of the distance sum is calculated to obtain the mean distance. The overall similarity of each eigenvector is set by the mean distance, and the eigenvectors in the group are sorted by the overall similarity to achieve the ordering of the first group of eigenvectors.

[0073] The second set of eigenvectors is calculated and analyzed by cosine distance, and the second set of eigenvectors is ordered;

[0074] The ordered first set of feature vectors and the second set of feature vectors are sequentially entered into the queue structure for data storage based on the feature vector sequence, forming two queue data.

[0075] It should be noted that the queue is a storage structure that helps improve the efficiency of subsequent association analysis. The mean distance is inversely proportional to the similarity. The ordering process of the first and second sets of eigenvectors is the same.

[0076] The value obtained by calculating the mean of the sum of the distances is the degree of deviation of a particular feature vector from the group. This value can effectively reflect the degree of similarity of a particular feature vector to the overall feature set within the group and can be used as the overall similarity for analysis. The smaller the similarity, the greater the distance from the overall feature set, and therefore the greater the feature's discriminability. After ordering, it is possible to screen for feature vectors with discriminative significance and improve the efficiency of subsequent association recognition.

[0077] According to an embodiment of the present invention, the grey correlation method is used to perform correlation analysis on two queue data as original analysis series. The difference between the queue data is calculated by applying the cosine distance of the eigenvector to obtain the correlation degree. It is determined whether the correlation degree is greater than a preset value. If so, non-target persons are marked as target persons and recorded within the current analysis cycle. By analyzing all non-target persons, group identification information is formed, specifically:

[0078] Based on the grey correlation method, the two cohort data were used as the original analysis series, one cohort was selected as the reference series, and the other cohort was selected as the comparison series;

[0079] Through the queue structure, the eigenvectors are output in sequence for difference calculation, and the absolute difference, maximum difference, and minimum difference between the reference series and the comparison series are calculated. The difference calculation process is based on the cosine distance calculation of the eigenvectors, and the corresponding correlation coefficient and correlation degree are obtained based on the absolute difference, maximum difference, and minimum difference.

[0080] Determine whether the correlation is greater than a preset value. If so, update the corresponding non-target person to a target person and identify the corresponding action feature data within the current analysis cycle;

[0081] In one analysis cycle, a correlation analysis is performed on all non-target persons, and the identification information of the target persons is recorded to obtain group identification information.

[0082] It should be noted that, generally speaking, the number series corresponding to the target action feature data is used as a reference sequence.

[0083] It is worth mentioning here that in the construction process of multiple personnel or large-scale, traditional recognition models are difficult to achieve rapid recognition and operation statistics, and there are cases where the analysis error is large. In addition, the traditional monitoring and analysis process is often based on the identification and statistics of individual personnel one by one. The process is repetitive and inefficient, and the recognition target features are not effectively used for feature mining and analysis.

[0084] Based on this, the present invention uses a monitoring unit to perform preliminary identification of personnel in the construction project area, extracts the relevant action features of the target personnel, and on the basis of preliminary identification, by distinguishing the action feature data of the target personnel and collecting them, constructs a reference sequence for the target features, and constructs a comparison sequence for the non-target personnel features, and uses the gray correlation method to perform correlation analysis. By calculating the correlation degree, the similarity and correlation degree of the action features of the non-target personnel and the target personnel can be quickly analyzed, the correlation characteristics in the feature data can be mined, and the non-target personnel can be effectively identified and counted. In the sequence comparison process, a queue structure is introduced for sequence analysis, which effectively improves the efficiency of the comparison process, and the queue can effectively limit the subsequent comparison process and comparison order, reduce data processing and analysis anomalies, thereby improving processing efficiency and reducing data analysis anomalies. Through the present invention, it is possible to realize the rapid identification of operating personnel and the feature correlation analysis between the crowd, realize the rapid and accurate personnel identification and operation statistics of large-scale engineering operations, realize intelligent engineering monitoring, and provide effective data support for subsequent engineering progress analysis and construction plan optimization.

[0085] According to an embodiment of the present invention, the personnel work statistics and construction project progress analysis based on group identification information, and the dynamic planning of monitoring resources and generation of monitoring plans are specifically as follows:

[0086] Through group identification information, the workload and type of work of personnel are counted, and the project progress is analyzed in combination with the construction project plan. Further monitoring demand analysis and monitoring unit adjustment are carried out on the preset construction area to generate a new monitoring plan.

[0087] It should be noted that the monitoring solution includes position control of the monitoring unit, setting of the number of devices, etc.

[0088] According to an embodiment of the present invention, the further embodiment includes:

[0089] During the second analysis cycle, group identification information from the first analysis cycle was obtained;

[0090] Based on group identification information, the ratio of non-target personnel to target personnel is calculated;

[0091] If the ratio is greater than the preset ratio and the number of target persons is greater than the second preset value, the target person action feature data of the first analysis cycle is obtained and marked as the target feature data set;

[0092] Cluster the target feature data set to form multiple feature sets. Based on the DCA feature fusion algorithm, set the weight parameters, generate feature maps and fuse the feature data in each feature set to form multiple feature fusion data.

[0093] The multiple feature fusion data are used as identification and comparison features for the second analysis cycle and are imported into a system database for storage.

[0094] It should be noted that if the ratio is greater than the preset ratio and the number of target personnel is greater than the second preset value, it means that the number of personnel working in the current construction project is large and the recognition efficiency is low, which reduces the efficiency and accuracy of operation analysis and statistics. Therefore, the present invention collects features of the target personnel in the previous analysis cycle and fuses the corresponding features based on feature fusion to extract the principal components and important features. In the second cycle, the fused features are used to identify the characteristics of the crowd operation, effectively improving the efficiency of crowd identification and realizing efficient operation statistics and monitoring analysis in multi-person and large-scale construction operations. Clustering can be based on algorithm models such as kmeans and DBSCAN.

[0095] According to an embodiment of the present invention, the further embodiment includes:

[0096] During an analysis cycle, building materials are identified and counted through monitoring units to form material change information;

[0097] Obtain personnel operation statistics through group identification information;

[0098] Based on the personnel work statistics and material change information, a comprehensive assessment of the construction progress is conducted, and the target construction progress is compared to evaluate the work efficiency per unit time.

[0099] In N analysis cycles, N operation efficiency evaluation values ​​are calculated, and the linear changes of the N operation efficiency evaluation values ​​are evaluated by a linear regression algorithm, and a linear change curve is drawn;

[0100] The linear change curve is used as the monitoring demand change curve, and linear prediction is performed to obtain a prediction curve. The monitoring demand of future construction projects is analyzed through the prediction curve, and the monitoring plan based on the preset building area is dynamically adjusted based on the prediction curve.

[0101] It should be noted that the material change information includes material quantity change, location change, etc. The unit time here is set as one analysis cycle.

[0102] Figure 2 A block diagram of a construction engineering intelligent monitoring system based on feature analysis of the present invention is shown.

[0103] A second aspect of the present invention further provides a construction engineering intelligent monitoring system 2 based on feature analysis, the system comprising: a memory 21 and a processor 22, wherein the memory 21 includes a construction engineering intelligent monitoring program based on feature analysis, and when the construction engineering intelligent monitoring program based on feature analysis is executed by the processor 22, the following steps are implemented:

[0104] Based on the construction project plan, monitoring units are set up in the preset construction area, and real-time video of the construction work is acquired through the monitoring units to obtain monitoring video data;

[0105] Extract key frames from the monitoring video data, and use the PCA analysis method to perform personnel area recognition and body movement feature extraction to form personnel recognition information and movement feature data;

[0106] Based on the personnel identification information, it is determined whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked;

[0107] In each analysis cycle, the motion feature data of non-target persons is obtained and marked as the second motion feature data. The target motion feature data and the second motion feature data are feature-vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance and serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group.

[0108] Based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed;

[0109] Based on group identification information, personnel work statistics and construction project progress analysis are carried out, and monitoring resources are dynamically planned to generate monitoring plans.

[0110] According to an embodiment of the present invention, based on the construction project plan, a monitoring unit is set in a preset construction area, and real-time video of the construction work is acquired through the monitoring unit to obtain monitoring video data, specifically:

[0111] Analyze the construction workload of personnel in the preset construction area through the construction project plan, and set up monitoring units in the preset construction area;

[0112] The monitoring unit acquires real-time video of the construction operation to obtain monitoring video data.

[0113] It should be noted that the construction project plan includes information such as construction plan, project quantity, and construction period.

[0114] According to an embodiment of the present invention, the monitoring video data is subjected to key frame extraction, and personnel area recognition and body movement feature extraction are performed through PCA analysis to form personnel recognition information and movement feature data, specifically:

[0115] Extract key frames from monitoring video data to obtain an image set;

[0116] Perform denoising, image enhancement and standardization preprocessing on the image set;

[0117] Image frames are selected from the image set for analysis. Based on the PCA analysis method, feature extraction and personnel area recognition are performed on the image frames to obtain personnel identification information. Personnel areas are set based on the personnel identification information. Through multiple image frames, features of the personnel areas are integrated to form motion feature data.

[0118] It should be noted that the personnel area generally includes multiple ones, which is determined by the number of people recognized by the image frame.

[0119] According to an embodiment of the present invention, the determination of whether the number of personnel is greater than a preset number based on personnel identification information is performed. If so, preliminary identification of personnel operations is performed through motion feature data, and target personnel and target motion feature data are marked, specifically:

[0120] Select a preset area and determine whether the number of corresponding personnel is greater than the preset number based on personnel identification information;

[0121] If so, the action feature data is imported into the recognition model for personnel operation recognition, the successfully recognized features are marked and the recognition result information is extracted to obtain the target person and target action feature data.

[0122] It should be noted that in larger construction projects, a pre-set area can be selected for construction crowd research and adjusted based on the scale of the construction project. The recognition model is a feature recognition model based on the CNN model, which can be used to identify and analyze personnel motion characteristics, facial features, and other features, and further implement job type analysis. It should be understood that the recognition model is only a preliminary recognition and analysis. When the number of workers is large, traditional recognition models have difficulty in achieving rapid recognition and job statistics. Non-target personnel corresponding to unsuccessfully identified personnel.

[0123] According to an embodiment of the present invention, in each analysis cycle, the motion feature data of non-target persons is obtained and marked as second motion feature data, the target motion feature data and the second motion feature data are feature vectorized to form two sets of feature vectors respectively, and the two sets of feature vectors are sorted based on feature importance, and the sorted two sets of feature vectors are serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group, specifically:

[0124] Setting an analysis cycle, and for each analysis cycle, obtaining motion feature data of non-target persons and marking the data as second motion feature data;

[0125] Performing feature vectorization on the target motion feature data and the second motion feature data to form a first set of feature vectors and a second set of feature vectors respectively;

[0126] Using cosine distance, in the first group of eigenvectors, the sum of the distances between each eigenvector and the remaining eigenvectors is calculated, and the mean of the distance sum is calculated to obtain the mean distance. The overall similarity of each eigenvector is set by the mean distance, and the eigenvectors in the group are sorted by the overall similarity to achieve the ordering of the first group of eigenvectors.

[0127] The second set of eigenvectors is calculated and analyzed by cosine distance, and the second set of eigenvectors is ordered;

[0128] The ordered first set of feature vectors and the second set of feature vectors are sequentially entered into the queue structure for data storage based on the feature vector sequence, forming two queue data.

[0129] It should be noted that the queue is a storage structure that helps improve the efficiency of subsequent association analysis. The mean distance is inversely proportional to the similarity. The ordering process of the first and second sets of eigenvectors is the same.

[0130] The value obtained by calculating the mean of the sum of the distances is the degree of deviation of a particular feature vector from the group. This value can effectively reflect the degree of similarity of a particular feature vector to the overall feature set within the group and can be used as the overall similarity for analysis. The smaller the similarity, the greater the distance from the overall feature set, and therefore the greater the feature's discriminability. After ordering, it is possible to screen for feature vectors with discriminative significance and improve the efficiency of subsequent association recognition.

[0131] According to an embodiment of the present invention, the grey correlation method is used to perform correlation analysis on two queue data as original analysis series. The difference between the queue data is calculated by applying the cosine distance of the eigenvector to obtain the correlation degree. It is determined whether the correlation degree is greater than a preset value. If so, non-target persons are marked as target persons and recorded within the current analysis cycle. By analyzing all non-target persons, group identification information is formed, specifically:

[0132] Based on the grey correlation method, the two cohort data were used as the original analysis series, one cohort was selected as the reference series, and the other cohort was selected as the comparison series;

[0133] Through the queue structure, the eigenvectors are output in sequence for difference calculation, and the absolute difference, maximum difference, and minimum difference between the reference series and the comparison series are calculated. The difference calculation process is based on the cosine distance calculation of the eigenvectors, and the corresponding correlation coefficient and correlation degree are obtained based on the absolute difference, maximum difference, and minimum difference.

[0134] Determine whether the correlation is greater than a preset value. If so, update the corresponding non-target person to a target person and identify the corresponding action feature data within the current analysis cycle;

[0135] In one analysis cycle, a correlation analysis is performed on all non-target persons, and the identification information of the target persons is recorded to obtain group identification information.

[0136] It should be noted that, generally speaking, the number series corresponding to the target action feature data is used as a reference sequence.

[0137] It is worth mentioning here that in the construction process of multiple personnel or large-scale, traditional recognition models are difficult to achieve rapid recognition and operation statistics, and there are cases where the analysis error is large. In addition, the traditional monitoring and analysis process is often based on the identification and statistics of individual personnel one by one. The process is repetitive and inefficient, and the recognition target features are not effectively used for feature mining and analysis.

[0138] Based on this, the present invention uses a monitoring unit to perform preliminary identification of personnel in the construction project area, extracts the relevant action features of the target personnel, and on the basis of preliminary identification, by distinguishing the action feature data of the target personnel and collecting them, constructs a reference sequence for the target features, and constructs a comparison sequence for the non-target personnel features, and uses the gray correlation method to perform correlation analysis. By calculating the correlation degree, the similarity and correlation degree of the action features of the non-target personnel and the target personnel can be quickly analyzed, the correlation characteristics in the feature data can be mined, and the non-target personnel can be effectively identified and counted. In the sequence comparison process, a queue structure is introduced for sequence analysis, which effectively improves the efficiency of the comparison process, and the queue can effectively limit the subsequent comparison process and comparison order, reduce data processing and analysis anomalies, thereby improving processing efficiency and reducing data analysis anomalies. Through the present invention, it is possible to realize the rapid identification of operating personnel and the feature correlation analysis between the crowd, realize the rapid and accurate personnel identification and operation statistics of large-scale engineering operations, realize intelligent engineering monitoring, and provide effective data support for subsequent engineering progress analysis and construction plan optimization.

[0139] According to an embodiment of the present invention, the personnel work statistics and construction project progress analysis based on group identification information, and the dynamic planning of monitoring resources and generation of monitoring plans are specifically as follows:

[0140] Through group identification information, the workload and type of work of personnel are counted, and the project progress is analyzed in combination with the construction project plan. Further monitoring demand analysis and monitoring unit adjustment are carried out on the preset construction area to generate a new monitoring plan.

[0141] It should be noted that the monitoring solution includes position control of the monitoring unit, setting of the number of devices, etc.

[0142] The third aspect of the present invention also provides a computer-readable storage medium, which includes a construction project intelligent monitoring program based on feature analysis. When the construction project intelligent monitoring program based on feature analysis is executed by a processor, the steps of the construction project intelligent monitoring method based on feature analysis as described in any one of the above items are implemented.

[0143] The present invention discloses a method and system for intelligent monitoring of construction projects based on feature analysis. By setting a monitoring unit in a preset construction area to obtain construction video in real time, the video data is subjected to key frame extraction and PCA analysis processes to identify personnel areas and body movement features, form personnel information and movement data, and screen out target personnel and non-target personnel. During the analysis cycle, the non-target personnel movement feature data is used as the second movement feature data, and is vectorized together with the target personnel movement feature data and stored in the form of a queue. The gray correlation method is used to analyze the correlation between the two queue data, and the non-target personnel are quickly identified and recorded based on the correlation to form group identification information. Finally, based on the group information, personnel operation statistics, project progress analysis and monitoring resource dynamic planning are carried out, and the monitoring plan is effectively optimized, the efficiency of project statistical analysis is improved, and effective data support is provided for subsequent project progress analysis and construction plan optimization.

[0144] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0145] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0146] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0147] Those skilled in the art will appreciate that all or part of the steps of the above-mentioned method embodiments may be implemented by hardware associated with program instructions, and the aforementioned program may be stored in a computer-readable storage medium. When the program is executed, the program executes the steps of the above-mentioned method embodiments. The aforementioned storage medium includes various media that can store program codes, such as mobile storage devices, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0148] Alternatively, if the integrated units described above are implemented as software modules and sold or used as standalone products, they can also be stored on a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product, stored on a storage medium, includes instructions for enabling a computer device (such as a personal computer, server, or network device) to execute all or part of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as removable storage devices, ROM, RAM, magnetic disks, or optical disks.

[0149] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A construction engineering intelligent monitoring method based on feature analysis, characterized in that: include: Based on the construction project plan, monitoring units are set up in the preset construction area, and real-time video of the construction work is acquired through the monitoring units to obtain monitoring video data; Extract key frames from the monitoring video data, and use the PCA analysis method to perform personnel area recognition and body movement feature extraction to form personnel recognition information and movement feature data; Based on the personnel identification information, it is determined whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked; In each analysis cycle, the motion feature data of non-target persons is obtained and marked as the second motion feature data. The target motion feature data and the second motion feature data are feature-vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance and serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group. Based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed; Based on group identification information, personnel work statistics and construction project progress analysis are carried out, and monitoring resources are dynamically planned to generate monitoring plans.

2. The construction engineering intelligent monitoring method based on feature analysis according to claim 1 is characterized in that: Based on the construction project plan, a monitoring unit is set in a preset construction area, and real-time video of the construction work is acquired through the monitoring unit to obtain monitoring video data, specifically: Analyze the construction workload of personnel in the preset construction area through the construction project plan, and set up monitoring units in the preset construction area; The monitoring unit acquires real-time video of the construction operation to obtain monitoring video data.

3. The construction engineering intelligent monitoring method based on feature analysis according to claim 1 is characterized in that: The monitoring video data is subjected to key frame extraction, and personnel area recognition and body movement feature extraction are performed through PCA analysis to form personnel recognition information and movement feature data, specifically: Extract key frames from monitoring video data to obtain an image set; Perform denoising, image enhancement and standardization preprocessing on the image set; Image frames are selected from the image set for analysis. Based on the PCA analysis method, feature extraction and personnel area recognition are performed on the image frames to obtain personnel identification information. Personnel areas are set based on the personnel identification information. Through multiple image frames, features of the personnel areas are integrated to form motion feature data.

4. The construction engineering intelligent monitoring method based on feature analysis according to claim 3 is characterized in that: The method is based on the personnel identification information to determine whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked, specifically: Select a preset area and determine whether the number of corresponding personnel is greater than the preset number based on personnel identification information; If so, the action feature data is imported into the recognition model for personnel operation recognition, the successfully recognized features are marked and the recognition result information is extracted to obtain the target person and target action feature data.

5. The construction engineering intelligent monitoring method based on feature analysis according to claim 4 is characterized in that: In each analysis cycle, the motion feature data of non-target persons is obtained and marked as second motion feature data. The target motion feature data and the second motion feature data are feature vectorized to form two sets of feature vectors respectively. The two sets of feature vectors are sorted based on feature importance, and the sorted two sets of feature vectors are serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group, specifically: Setting an analysis cycle, and for each analysis cycle, obtaining motion feature data of non-target persons and marking the data as second motion feature data; Performing feature vectorization on the target motion feature data and the second motion feature data to form a first set of feature vectors and a second set of feature vectors respectively; Using cosine distance, in the first group of eigenvectors, the sum of the distances between each eigenvector and the remaining eigenvectors is calculated, and the mean of the distance sum is calculated to obtain the mean distance. The overall similarity of each eigenvector is set by the mean distance, and the eigenvectors in the group are sorted by the overall similarity to achieve the ordering of the first group of eigenvectors. The second set of eigenvectors is calculated and analyzed by cosine distance, and the second set of eigenvectors is ordered; The ordered first set of feature vectors and the second set of feature vectors are sequentially entered into the queue structure for data storage based on the feature vector sequence, forming two queue data.

6. The construction engineering intelligent monitoring method based on feature analysis according to claim 5 is characterized in that: Based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvector to obtain the correlation degree. It is judged whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed, specifically: Based on the grey correlation method, the two cohort data were used as the original analysis series, one cohort was selected as the reference series, and the other cohort was selected as the comparison series; Through the queue structure, the eigenvectors are output in sequence for difference calculation, and the absolute difference, maximum difference, and minimum difference between the reference series and the comparison series are calculated. The difference calculation process is based on the cosine distance calculation of the eigenvectors, and the corresponding correlation coefficient and correlation degree are obtained based on the absolute difference, maximum difference, and minimum difference. Determine whether the correlation is greater than a preset value. If so, update the corresponding non-target person to a target person and identify the corresponding action feature data within the current analysis cycle; In one analysis cycle, a correlation analysis is performed on all non-target persons, and the identification information of the target persons is recorded to obtain group identification information.

7. The construction engineering intelligent monitoring method based on feature analysis according to claim 6 is characterized in that: The above mentioned method is to conduct personnel operation statistics and construction project progress analysis based on group identification information, dynamically plan monitoring resources, and generate monitoring plans, specifically: Through group identification information, the workload and type of work of personnel are counted, and the project progress is analyzed in combination with the construction project plan. Further monitoring demand analysis and monitoring unit adjustment are carried out on the preset construction area to generate a new monitoring plan.

8. An intelligent monitoring system for construction projects based on feature analysis, characterized in that: The system includes: a memory and a processor. The memory includes a construction engineering intelligent monitoring program based on feature analysis. When the construction engineering intelligent monitoring program based on feature analysis is executed by the processor, the following steps are implemented: Based on the construction project plan, monitoring units are set up in the preset construction area, and real-time video of the construction work is acquired through the monitoring units to obtain monitoring video data; Extract key frames from the monitoring video data, and use the PCA analysis method to perform personnel area recognition and body movement feature extraction to form personnel recognition information and movement feature data; Based on the personnel identification information, it is determined whether the number of personnel is greater than the preset number. If so, the personnel operation is preliminarily identified through the action feature data, and the target personnel and target action feature data are marked; In each analysis cycle, the motion feature data of non-target persons is obtained and marked as the second motion feature data. The target motion feature data and the second motion feature data are feature-vectorized to form two sets of feature vectors. The two sets of feature vectors are sorted based on feature importance and serialized and stored in the form of queues to obtain two queue data. The feature importance is determined by the similarity between the feature vectors in the group. Based on the grey correlation method, the two queue data are used as the original analysis series for correlation analysis. The difference between the queue data is calculated by applying the cosine distance of the eigenvectors to obtain the correlation degree. It is determined whether the correlation degree is greater than the preset value. If so, non-target personnel are marked as target personnel and recorded in the current analysis cycle. By analyzing all non-target personnel, group identification information is formed; Based on group identification information, personnel work statistics and construction project progress analysis are carried out, and monitoring resources are dynamically planned to generate monitoring plans.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a construction engineering intelligent monitoring program based on feature analysis. When the construction engineering intelligent monitoring program based on feature analysis is executed by a processor, the steps of the construction engineering intelligent monitoring method based on feature analysis as described in any one of claims 1 to 7 are implemented.

Images