An integrated intelligent monitoring system and method for a smart construction site

Abstract

The application discloses an integrated intelligent monitoring system and method for a smart construction site, and relates to the technical field of construction site monitoring. The method comprises arranging monitoring cameras based on the construction scene of the construction site, collecting images of the construction site based on the arranged monitoring cameras, and obtaining collected image data; performing image recognition on the collected image data based on image recognition technology, obtaining article data, judging the structural risk of the article data, marking articles with risks to obtain first article data; determining the position information of the article data based on the collected image data, assessing the state risk of the articles according to the position information, and marking articles with risks as second article data; determining the danger of the articles according to the first article data and the second article data, marking articles with dangers as dangerous articles; and when it is determined that there are dangerous articles, marking the dangerous articles for alarm. The application has the effect of improving the accuracy of safety monitoring of the construction site.

Description

Technical Field

[0001] This application relates to the field of construction site monitoring technology, and in particular to an integrated intelligent monitoring system and method for smart construction sites. Background Technology

[0002] With the development of society and economy and the implementation of a large number of infrastructure projects, ensuring the safety of construction sites has become an urgent problem to be solved.

[0003] In existing technologies, monitoring equipment, such as cameras, is installed at construction sites to dynamically monitor personnel. When an accident is detected, an alarm signal can be issued in a timely manner to minimize the harm caused by the accident and thus ensure the safety of workers.

[0004] However, the aforementioned technologies can only address risks when they occur, and cannot provide early warnings to further reduce the harm caused by risks, thus requiring improvement. Summary of the Invention

[0005] To improve the accuracy of construction site safety monitoring, this application provides an integrated intelligent monitoring system and method for smart construction sites.

[0006] Firstly, this application provides an integrated intelligent monitoring method for smart construction sites, employing the following technical solution:

[0007] An integrated intelligent monitoring method for smart construction sites includes:

[0008] Based on the construction scene of the construction site, surveillance cameras are deployed, and images of the construction site are collected based on the deployed surveillance cameras to obtain the collected image data.

[0009] Based on image recognition technology, image recognition is performed on the collected image data to obtain item data. The structural risk of the item data is then assessed, and items with risk are marked to obtain the first item data.

[0010] The location information of the items is determined based on the collected image data, and the status risk of the items is assessed based on the location information. Items with risks are marked as second item data.

[0011] Based on the data of the first item and the data of the second item, determine the hazard of the item and mark the item with hazard as a dangerous item;

[0012] When a hazardous substance is detected, an alarm sign is set up for the hazardous substance.

[0013] Preferably, the construction scenario is determined based on the construction characteristics of the construction site; the construction scenario includes planar construction and three-dimensional construction.

[0014] If the construction scenario is a three-dimensional construction, then each construction layer is divided into a planar construction with corresponding spatial relationships to obtain the construction area;

[0015] Obtain the monitoring range of the surveillance cameras and use the monitoring range as the spacing between the surveillance cameras. Arrange the surveillance cameras in the construction area according to the spacing between the surveillance cameras.

[0016] The surveillance camera captures images of the monitored area, resulting in captured image data.

[0017] Preferably, arbitrary image acquisition of the construction area is performed based on surveillance cameras to obtain image data to be analyzed;

[0018] Based on image recognition technology, object identification is performed on the image data to be analyzed to obtain the data of the object to be judged.

[0019] The data of the item to be judged is compared with the data of the actual item to determine the recognition accuracy. The recognition accuracy is then compared with the preset accuracy value to determine whether the recognition accuracy meets the standard.

[0020] If the recognition accuracy does not meet the standard, the monitoring range of the surveillance camera will be adjusted until the image recognition technology can recognize images with an accuracy greater than the preset accuracy value.

[0021] Preferably, image recognition technology is used to perform image recognition on the collected image data to determine all items contained within the monitored area and obtain item data;

[0022] Based on the acquired image data, the location and shape / size of each item are determined to obtain the first location information and shape information;

[0023] Based on the first location information, the items are filtered to obtain items whose location is within the built-in space range, and the corresponding items are marked as the first item item;

[0024] The effective area of ​​the corresponding item is determined based on the first item;

[0025] Obtain the material properties of each item in the first item category, and substitute the material properties and the area of ​​effect into the built-in first evaluation formula to obtain the structural risk value of the item. Then match the structural risk value with the built-in first evaluation index, and mark the items with structural risk values ​​greater than the built-in first evaluation index as the first item data.

[0026] Preferably, a three-dimensional space is constructed based on the monitoring area of ​​the surveillance camera and the perspective relationship of the image;

[0027] Match the three-dimensional space with the area monitored by the surveillance camera to identify items that extend beyond the surface of the three-dimensional space within the monitored area and mark them as items to be identified.

[0028] Image recognition technology is used to identify the items to be identified and obtain item data.

[0029] Preferably, the location data is obtained by marking the location of the items based on the item data;

[0030] The location data of the same item in multiple images captured by the same surveillance camera are compared to determine whether the item's location has changed.

[0031] If it is determined that the position of an item has changed, the reason for the change in position should be determined to determine whether there was any human intervention.

[0032] If it is determined to be human intervention, the location of the item will be re-marked and reassessed;

[0033] If it is determined that there was no human intervention, the relationship between the displacement distance and time of the item is determined based on the collected image data of the item, the state change index is obtained, and the risk of the item is determined based on the state change index.

[0034] Preferably, the time period corresponding to the change of the item's location is obtained, and facial recognition is performed on the collected image data of the corresponding time period to determine whether there is a human face in the collected image data.

[0035] If it is determined that there is a human image in the acquired image data, the position movement data of the human image is determined based on the acquired image data, and the position movement data of the human image is matched with the position change data when the position of the object changes, to determine whether the position movement data of the human image coincides with the position change data when the position of the object changes.

[0036] If the data on the movement of the human figure coincides with the data on the position of the object before the change, the change in the object's position is determined to be due to human intervention; otherwise, it is determined to be due to non-human intervention.

[0037] Preferably, based on the construction layer, a hierarchical relationship is established for the construction scene, and the hierarchical data of each surveillance camera is determined according to the hierarchical relationship;

[0038] Obtain the hierarchical data and location information of the current monitoring area. Based on the hierarchical data, location information, and construction scenario, acquire image data located in the same area with a hierarchical data smaller than the current monitoring range. Statistically sort the acquired image data and the acquired image data corresponding to the current monitoring range according to the hierarchical relationship to obtain the group to be analyzed.

[0039] If the current monitoring range is located at the first level of data, then based on the built-in hazard assessment model, a hazard assessment is performed on the first item data and the second item data to determine whether the item is dangerous.

[0040] If the current monitoring range is not the first layer of data, the location status of the first item data is determined, and the first item data is divided into the first fixed item whose location does not change and the first moving item whose location changes according to the location status of each item.

[0041] Based on the built-in hazard assessment model, the first fixed item is judged to determine whether there is a hazard judgment standard for the item in the hazard assessment model. If there is a hazard judgment standard for the item, the hazard assessment is performed on the item based on the hazard assessment model to determine whether the item is dangerous.

[0042] If the hazard assessment model does not have a hazard judgment standard for the item, it outputs an auxiliary signal and receives the artificial hazard marking result after artificial hazard marking. The artificial hazard marking result is used as the hazard assessment result for the item, and the hazard assessment model is trained on the corresponding item based on the artificial hazard marking result to obtain the hazard judgment standard for the corresponding item.

[0043] Based on the group to be analyzed, and by performing facial recognition on the group to be analyzed, it is determined whether there is a human trajectory in the monitoring range at the level below the current monitoring range. If it is determined that there is a human trajectory, the risk of the corresponding item falling is determined based on the state change index of each item in the first moving item and second item data, and the first indicator data is obtained.

[0044] A hazard assessment is performed based on the built-in hazard assessment model to obtain second indicator data, and the hazard status of the item is determined based on the first and second indicator data.

[0045] Secondly, this application provides an integrated intelligent monitoring system for smart construction sites, which adopts the following technical solution:

[0046] An integrated intelligent monitoring system for smart construction sites includes: a data acquisition and planning module, a first risk assessment module, a second risk assessment module, a hazard assessment module, and an alarm module;

[0047] The data acquisition and planning module deploys surveillance cameras based on the construction scene of the construction site, and acquires images of the construction site based on the deployed surveillance cameras to obtain the acquired image data.

[0048] The first risk assessment module, based on image recognition technology, performs image recognition on the collected image data to obtain item data, and performs structural risk assessment on the item data, marking items with risks to obtain the first item data;

[0049] The second risk assessment module determines the location information of the item data based on the collected image data, and performs a status risk assessment on the item based on the location information, marking the risky items as the second item data;

[0050] The hazard assessment module determines the hazard level of an item based on the data of the first item and the data of the second item, and marks items with hazards as hazardous materials.

[0051] The alarm module marks the presence of hazardous materials when it detects their presence.

[0052] In summary, this application includes at least one of the following beneficial technical effects:

[0053] By deploying surveillance cameras on the construction site, comprehensive monitoring of the construction area can be achieved, thereby improving the accuracy of safety assessments. The captured images are used for object identification to make an initial hazard assessment of items within the monitored area, and potentially hazardous items are marked for monitoring. Then, by judging and evaluating the location information of the items, their stability is determined, further confirming their hazard level within the monitored area. This two-stage assessment provides a more comprehensive picture, leading to more accurate subsequent hazard assessments and ultimately improving the accuracy of safety monitoring at the construction site.

[0054] By using existing image recognition technology to perform image recognition on images captured by surveillance cameras, and comparing the recognition results with actual item data, the compatibility between the adopted image recognition technology and the current acquisition range of the surveillance camera is determined. This ensures that items within the adjusted monitoring range can be accurately and effectively identified, thereby improving the accuracy of security monitoring of the area.

[0055] By marking the location of the same object over a long period, it is possible to determine whether the object's location is prone to change. The stability of the object's state is determined based on the difficulty of the location change, and the stability value of the object's state is determined based on the relationship between the displacement of the object's location change and time. When it is determined that the object's location has changed, human intervention analysis is performed to investigate the cause of the location change. Facial recognition is performed on the time period when the location change occurs, and the overlap between the facial recognition movement trajectory and the location trajectory of the object when the location change occurs is judged to improve the accuracy of the judgment of the object's state stability, thereby improving the accuracy of the construction site safety assessment. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating the steps of the integrated intelligent monitoring method for smart construction sites in this embodiment.

[0057] Figure 2 This is a block diagram of the integrated intelligent monitoring system for smart construction sites in this embodiment.

[0058] Attached diagram labels: 1. Data acquisition and planning module; 2. First risk assessment module; 3. Second risk assessment module; 4. Hazard assessment module; 5. Alarm module. Detailed Implementation

[0059] The following is in conjunction with the appendix Figures 1-2 This application will be described in further detail.

[0060] This application discloses an integrated intelligent monitoring system and method for smart construction sites.

[0061] Example: Figure 1 As shown, the present invention provides an integrated intelligent monitoring method for smart construction sites, comprising:

[0062] S1, Based on the construction scene of the construction site, the surveillance cameras are deployed, and the images of the construction site are collected based on the deployed surveillance cameras to obtain the collected image data.

[0063] S2, based on image recognition technology, performs image recognition on the collected image data to obtain item data, and performs structural risk assessment on the item data, marking items with risks to obtain first item data; by identifying items and assessing the hazards of the identified items, the potential hazards that the items themselves can cause are determined.

[0064] S3 determines the location information of the item data based on the acquired image data, and performs a status risk assessment on the item based on the location information, marking the risky item as the second item data; the risk assessment is used to determine the status change of the item, and the more easily the status changes, the higher the risk.

[0065] S4. Based on the data of the first item and the data of the second item, determine the hazard of the item and mark the item with hazard as a dangerous item;

[0066] S5, when a hazard is detected, an alarm sign is set for the hazard.

[0067] In this embodiment, by deploying surveillance cameras on the construction site, comprehensive monitoring of the construction area can be achieved, thereby improving the accuracy of safety assessments. The collected images are used for object identification to make an initial hazard assessment of the objects within the monitored area, and potentially hazardous objects are marked for monitoring. Then, the location information of the objects is assessed to determine their stability, further confirming their hazard level within the monitored area. This two-step assessment provides a more comprehensive picture, leading to more accurate subsequent hazard assessments and ultimately improving the accuracy of safety monitoring at the construction site.

[0068] For example, by comprehensively monitoring the construction area of ​​a construction site using surveillance cameras, the safety assessment results are more comprehensive and accurate. When conducting a site safety assessment on the area monitored by the surveillance cameras, existing image recognition technology is used to identify objects within the monitored area and to assess the structural risk of these objects. This initial assessment identifies items such as bricks, steel bars, gravel, and ropes. The structural risk assessment determines which items in the area could directly cause harm. For instance, if bricks and steel bars can directly cause harm, they are identified as posing a risk and marked as the first type of object. Since more stable objects have a lower probability of posing a danger, and conversely, less stable objects pose more risks, the assessment process continues after determining which items pose a risk. Once an object can directly cause harm, its location information needs to be assessed to determine its stability. For example, a rope lacks stability and is therefore marked as a second item. By assessing the inherent hazard of an item and then its state, the accuracy of the hazard assessment is comprehensively improved. However, the very nature of an item and its changeable state do not guarantee its inherent danger. Further assessment is required, considering the hazard's action pattern when the item becomes dangerous, to determine if the current environment has the potential to trigger the hazard. Similarly, a hazard is defined as having an inaccessible target. For example, a mudslide is considered dangerous if there are villages or residences below it; otherwise, it is considered non-dangerous. Therefore, reassessing the item based on its target to determine whether the hazard would affect people or construction work leads to more accurate hazard assessments and improves the accuracy of site safety monitoring.

[0069] In step S1, surveillance cameras are deployed based on the construction scene at the construction site, and images of the construction site are acquired using these deployed surveillance cameras to obtain the acquired image data. This includes the following steps:

[0070] S11. Determine the construction scenario based on the construction characteristics of the construction site; the construction scenario includes planar construction and three-dimensional construction.

[0071] S12, If the construction scenario is a three-dimensional construction, then each construction layer is divided into a planar construction with corresponding spatial relationships to obtain the construction area;

[0072] S13, obtain the monitoring range of the surveillance cameras, and use the monitoring range as the spacing between the surveillance cameras. Arrange the surveillance cameras in the construction area according to the spacing between the surveillance cameras.

[0073] S131, Based on the monitoring camera, arbitrary images of the construction area are acquired to obtain image data to be analyzed;

[0074] S132, Based on image recognition technology, perform item recognition on the image data to be analyzed to obtain the item data to be judged;

[0075] S133, compare the data of the item to be judged with the actual item data to determine the recognition accuracy, and compare the recognition accuracy with the preset accuracy value to determine whether the recognition accuracy meets the standard;

[0076] S134 If the recognition accuracy does not meet the standard, the monitoring range of the surveillance camera will be adjusted until the image recognition technology recognizes the collected images with an accuracy greater than the preset accuracy value.

[0077] S14: Based on the surveillance camera, images are acquired within the monitoring range to obtain the acquired image data.

[0078] In this embodiment, due to the special nature of the construction area, multiple surveillance cameras are required to monitor the area. The installation location of the surveillance cameras is determined by utilizing their monitoring range. Since the monitoring range of the cameras has virtually unlimited depth, the monitoring range itself needs to be defined. Image recognition technology is used to perform image recognition on the images captured by the surveillance cameras, and the results are compared with the actual item data. This determines the compatibility between the adopted image recognition technology and the current acquisition range of the surveillance cameras, ensuring that items within the adjusted monitoring range can be accurately and effectively identified, thereby improving the accuracy of security monitoring of the area.

[0079] For example, since construction sites may involve both planar and three-dimensional construction, the sources of danger in three-dimensional construction include not only the ground but also the risk of objects falling from above. Therefore, it is necessary to deploy surveillance cameras according to the characteristics of the construction site to ensure comprehensive monitoring of the site.

[0080] If the monitoring range of the surveillance cameras is 10 meters, then the distribution of the surveillance cameras is determined according to the structural characteristics of the construction area, so that the distance between the surveillance cameras is less than or equal to 10 meters, and the construction area is arranged so that the surveillance cameras can provide comprehensive monitoring of the construction area.

[0081] Since surveillance cameras are limited in width but unlimited in depth, it is necessary to limit the depth of the surveillance cameras. By using the cameras to capture images of the construction area and using existing image recognition technology to identify objects in the images, the identified objects are compared with the actual objects in the area to determine the accuracy of the image recognition technology in identifying objects. This accuracy includes whether objects are identified and whether their categories are identified. By adjusting the monitoring range, the existing image recognition technology can achieve a good adaptation effect with the captured images, thereby improving the accuracy of identification.

[0082] In step S2, based on image recognition technology, image recognition is performed on the collected image data to obtain item data. The structural risk of the item data is then assessed, and items deemed risky are marked to obtain the first item data. This includes the following steps:

[0083] S21, Based on image recognition technology, perform image recognition on the collected image data to determine all items contained within the monitored area and obtain item data;

[0084] S22, based on the acquired image data, determine the position and shape / size of each item to obtain the first position information and shape information;

[0085] S23, Based on the first location information, the items are filtered to obtain items whose locations are within the built-in space range, and the corresponding items are marked as the first item item;

[0086] S24, determine the effective area of ​​the corresponding item based on the first item; wherein, the effective area is the contact area when the surface formed by the horizontal line, the vertical line and the 45-degree inclined line respectively contact the item within the built-in space, without changing the position of the item.

[0087] S25: Obtain the material properties of each item in the first item category, and substitute the material properties and the effective area into the built-in first evaluation formula to obtain the structural risk value of the item. Match this structural risk value with the built-in first evaluation index, and mark items with structural risk values ​​greater than the built-in first evaluation index as first item data. The material properties of the item can include its hardness, sharpness, structural properties, etc. The first evaluation formula can be: z = ax + by, where z is the evaluated structural risk value, a is the material property coefficient, x is the value of each material property of the item, b is the effective area coefficient, and y is the effective area of ​​the item.

[0088] Among them, the characteristics of an object determine its hardness and overall cohesion. For example, bricks and gravel of the same volume can be considered to have the same hardness, but in terms of overall cohesion, bricks have greater cohesion than gravel, thus making bricks more dangerous than gravel.

[0089] In this embodiment, image recognition is performed on the collected image data to clarify the categories of items within the monitoring area. At the same time, the location of each item is determined to filter the items, thereby reducing the number of items to be evaluated and improving the evaluation efficiency. Furthermore, by utilizing the relationship between pressure and the area of ​​action, as well as the relationship between action and reaction forces, a built-in first evaluation formula is constructed. The relevant data of the corresponding items are substituted into this formula, making the final evaluation result more accurate and effective, thereby improving the accuracy of the construction site safety assessment.

[0090] For example, firstly, the image is identified using image recognition technology to determine which items are contained in the image. Then, the shape, size, and location information of the corresponding items are determined based on the image. Assuming the space within the monitoring range is 3*3*3 cubic meters, and the average height of a person is 1.8 meters, and assuming the safety height is set at 2 meters, then only the structural risk assessment of items within the space below 2 meters in height is required, thereby reducing the amount of assessment data.

[0091] When determining the contact area of ​​an object, assuming the object is a cube and its position remains unchanged, a collision between a person and the object is most likely an oblique collision. To ensure comprehensiveness of collision scenarios, a complete analysis of the object's collision probability is conducted. This is achieved by establishing detection surfaces within the monitored space. For example, if the cube is placed upright on the ground, the contact area between the horizontal surface and the cube is the top surface; the contact area between the vertical surface and the cube are the two sides; and the contact area between the 45-degree inclined surface and the cube are the edges or vertices. Based on the relationship between pressure and the contact area, it can be determined that the smaller the contact area, the greater the impact force and the higher the risk of harm.

[0092] Furthermore, based on the relationship between action and reaction forces, when the same force acts on objects of different materials, the softer the material, the more easily its external shape changes under the force. For example, given two cubes, one made of metal and the other of foam, the metal cube will cause greater damage than the foam cube in the same collision. Therefore, based on these characteristics, a first assessment formula is constructed to obtain more accurate results when assessing the risk of an object's structure.

[0093] In step S21, image recognition technology is used to perform image recognition on the collected image data to determine all items contained within the monitored area, thereby obtaining item data. This includes the following steps:

[0094] S211 constructs a three-dimensional space based on the monitoring area and image perspective relationship of the surveillance camera;

[0095] S212, Match the three-dimensional space with the area monitored by the surveillance camera, identify items that extend beyond the surface of the three-dimensional space within the monitored area, and mark them as items to be identified;

[0096] S213, Based on image recognition technology, identify the item to be identified and obtain item data.

[0097] In this embodiment, by judging the spatial perspective relationship of the images captured by the surveillance camera, the spatial range represented by the image data is determined. By establishing a corresponding three-dimensional spatial model and matching the model with the actual captured image data, the additives present in the captured image data are identified, and the additives are identified, thereby improving the accuracy of item identification.

[0098] For example, since the images captured by the surveillance camera are planar images, it is necessary to determine the specific recognition range and identify what are the additions in that space when recognizing planar images, so that image recognition technology can accurately identify the additions and improve accuracy.

[0099] Because the space is three-dimensional, when establishing corresponding spatial relationships based on the collected image data, the spatial boundary of the monitored area coincides with the spatial boundary of the constructed three-dimensional space. When there are other objects within the monitored area, the objects will protrude from the plane. Therefore, when matching with the constructed three-dimensional space, objects protruding into the three-dimensional space are marked as objects to be identified, thereby reducing the difficulty of object identification and improving the accuracy of object identification.

[0100] In step S3, the location information of the item data is determined based on the acquired image data, and the status risk assessment of the item is performed based on the location information. Items with high risk are marked as second item data. This includes the following steps:

[0101] S31, mark the location of the items based on the item data to obtain the location data;

[0102] S32 compares the position data of the same item in multiple images captured by the same surveillance camera to determine whether the position of the item has changed;

[0103] S33, If it is determined that the position of the item has changed, the reason for the change in the position of the item shall be determined to determine whether there is any human intervention;

[0104] S331, obtain the time period corresponding to when the position of the item changes, and perform facial recognition on the collected image data of the corresponding time period to determine whether there is a human face in the corresponding collected image data;

[0105] S332, if it is determined that there is a human image in the acquired image data, the position movement data of the human image is determined according to the acquired image data, and the position movement data of the human image is matched with the position change data when the position of the object changes, to determine whether the position movement data of the human image coincides with the position change data when the position of the object changes.

[0106] S333, if the position movement data of the human figure coincides with the position data of the item before the position changed, the position change of the item is determined to be due to human intervention; otherwise, it is determined to be due to non-human intervention.

[0107] S34, if it is determined to be human intervention, the location of the item shall be re-marked and reassessed;

[0108] S35, if it is determined that it is not due to human intervention, the relationship between the displacement distance and time of the item is determined based on the collected image data of the item, the state change index is obtained, and the risk of the item is determined based on the state change index.

[0109] In this embodiment, by marking the location of the same item for a long period of time, it is determined whether the item's location is prone to change. The stability of the item's state is determined based on the difficulty of the location change, and the stability value of the item's state is determined based on the relationship between the displacement of the item's location change and time. When it is determined that the item's location has changed, the cause of the location change is analyzed by human intervention. By performing facial recognition on the time period when the location change occurs, and comparing the movement trajectory of the facial recognition with the location trajectory when the item's location changes, the accuracy of the determination of the item's state stability is improved, thereby improving the accuracy of the construction site safety assessment.

[0110] For example, after assessing the structural risk of an item by examining its material properties and area of ​​effect, the assessment can be further improved by examining changes in the item's state within the monitoring range. Since the less stable an item is, the greater its positional changes, and the more hazardous variables there are in the environment, assessing changes in the item's state improves the accuracy of the assessment.

[0111] For example, if item A exists within the monitoring range, and during the first data collection, item A is at location a, but during the second data collection, item A is at location b, then it is determined that the item's position has changed. Since a change in the item's state indicates instability and a greater risk of danger, the item needs to be marked. However, marking an item only confirms a change in position, not the cause. For instance, if item A is a tool, the change in its position might be due to human intervention. Therefore, it is necessary to determine the cause of the change in item A to ascertain whether it was due to human intervention.

[0112] By performing facial recognition on the time period during which the location of item A changes, it can be determined whether any person passed through the area during that time period. When it is determined that a person has arrived in the area, it is necessary to correlate the person with the location change of item A to determine whether the location change of item A is directly related to the person. By comparing the person's movement trajectory with the location data of the item before and after the location change, if the location change of item A is caused by human intervention, then the person's movement trajectory should coincide with the location of the location change of item A, thus determining the cause of the location change of item A.

[0113] If the change in the position of item A is determined to be due to human intervention, the stability of item A's state cannot be determined, and therefore the stability of item A's state needs to be reassessed. If the change in the position of item A is determined not to be due to human intervention, the stability of item A's state can be determined, and then the variability of item A can be determined by analyzing the relationship between the change in the item's position and time.

[0114] The more times item A changes position in a shorter period of time, the more unstable the state of item A is, and the more dangerous item A is.

[0115] In step S4, based on the first item data and the second item data, the hazard level of the item is determined, and the hazardous item is marked as a dangerous object. This includes the following steps:

[0116] S41, based on the construction layer, establishes a hierarchical relationship for the construction scene and determines the layer data for each surveillance camera according to the hierarchical relationship; the hierarchical relationship is based on the construction layer as the evaluation standard, marking the construction layer located on the ground as the first layer, and marking other construction layers accordingly based on changes in height. For example, the second layer, the third layer, etc.

[0117] S42, obtain the hierarchical data and location information of the current monitoring area, and based on the hierarchical data, location information and construction scenario, obtain the collected image data located in the same area and whose hierarchical data is smaller than the current monitoring range, and statistically sort the collected image data and the collected image data corresponding to the current monitoring range according to the hierarchical relationship to obtain the group to be analyzed;

[0118] S43, if the current monitoring range is located at the first level, then based on the built-in hazard assessment model, perform a hazard assessment on the first item data and the second item data to determine whether the item is dangerous;

[0119] S44, if the layer data of the current monitoring range is not the first layer, then the position status of the first item data is judged, and the first item data is divided into the first fixed item whose position does not change and the first moving item whose position changes according to the position status of each item.

[0120] S45, Based on the built-in hazard assessment model, the first fixed item is judged to determine whether there is a hazard judgment standard for the item in the hazard assessment model. If there is a hazard judgment standard for the item, the hazard assessment is performed on the item based on the hazard assessment model to determine whether the item is dangerous.

[0121] S46, if the hazard assessment model does not have a hazard judgment standard for the item, an auxiliary signal is output, and the result of the artificial hazard marking is received. The artificial hazard marking result is used as the hazard assessment result for the item, and the hazard assessment model is trained on the corresponding item based on the artificial hazard marking result to obtain the hazard judgment standard for the corresponding item. Before applying the hazard assessment model, the construction scene is collected once, and the collected image data is artificially hazard-marked once to obtain the artificial hazard marking result. The machine learning algorithm is trained on the artificial hazard marking result to obtain the hazard assessment model.

[0122] S47, based on the group to be analyzed, and performing facial recognition on the group to be analyzed, determine whether there is a human trajectory in the monitoring range at the level below the current monitoring range. If it is determined that there is a human trajectory, determine the risk of the corresponding item falling based on the state change index of each item in the first moving item and second item data, and obtain the first indicator data.

[0123] S48, performs a hazard assessment based on the built-in hazard assessment model, obtains second indicator data, and determines whether the item is hazardous based on the first and second indicator data.

[0124] In this embodiment, by establishing a hierarchical relationship for the construction scene, the hierarchical data and location information corresponding to the current monitoring area are clarified. Then, by adopting corresponding evaluation methods based on different hierarchical data, the evaluation results are made more accurate. By conducting hazard assessment model evaluations on items without the risk of falling, and conducting hazard assessment model evaluations and fall risk assessments on items with the risk of falling, a comprehensive judgment can be made on the dangers posed by items in the current monitoring area, thereby improving the accuracy of construction site safety monitoring.

[0125] For example, when constructing a building, after the foundation is completed, a platform needs to be built for workers to work on the higher levels. This platform is called a construction layer. Each construction layer is marked with a level label: the level label on the ground is 1, the level label on the construction layer adjacent to the ground is 2, and so on, until all construction layers are marked.

[0126] If the current construction layer level is 1, it indicates that the construction layer is located on the ground, so there is no risk of injury from falling objects. Therefore, it is only necessary to perform a risk assessment on each item using the built-in hazard assessment model. Conversely, if the current construction layer level is not 1, for example, if the construction layer level is 2, it indicates that there is a possibility of injury from falling objects. Therefore, it is not possible to directly assess the hazard of each item within the monitoring range using the hazard assessment model. Instead, it is necessary to divide the items within the range, perform a hazard assessment model assessment on items with fixed locations, and perform a fall risk assessment and hazard assessment model assessment on moving items. This makes the final assessment results more accurate and improves the accuracy of the construction site safety assessment.

[0127] Based on the description of the above embodiments of the integrated intelligent monitoring method for smart construction sites, this invention also discloses an integrated intelligent monitoring system for smart construction sites:

[0128] like Figure 2 As shown, an integrated intelligent monitoring system for smart construction sites, by applying the integrated intelligent monitoring method for smart construction sites as described above, includes: a data acquisition and planning module 1, a first risk assessment module 2, a second risk assessment module 3, a hazard assessment module 4, and an alarm module 5;

[0129] The data acquisition planning module 1 is used to deploy surveillance cameras based on the construction scene of the construction site, and to acquire images of the construction site based on the deployed surveillance cameras to obtain the acquired image data.

[0130] The first risk assessment module 2, based on image recognition technology, performs image recognition on the collected image data to obtain item data, and performs structural risk assessment on the item data, marking items with risks to obtain the first item data;

[0131] The second risk assessment module 3 determines the location information of the item data based on the collected image data, and performs a status risk assessment on the item based on the location information, marking the risky item as the second item data;

[0132] Hazard assessment module 4 determines the hazard of an item based on the data of the first item and the data of the second item, and marks items with hazards as hazardous materials;

[0133] Alarm module 5, when it is determined that there is a dangerous object, will mark the dangerous object with an alarm.

[0134] Compared with existing integrated intelligent monitoring systems and methods for smart construction sites, this invention improves the accuracy of construction site safety monitoring.

[0135] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An integrated intelligent monitoring method for a smart construction site, characterized in that, The method comprises the following steps: arranging monitoring cameras based on the construction scene of the construction site, and collecting images of the construction site based on the arranged monitoring cameras to obtain collected image data; performing image recognition on the collected image data based on image recognition technology to obtain article data, and performing structural risk judgment on the article data to mark articles with risks to obtain first article data; performing image recognition on the collected image data based on image recognition technology to determine all articles contained in the monitoring area to obtain article item data; determining the position and shape of each article based on the collected image data to obtain first position information and shape information; screening articles based on the first position information to obtain articles whose positions are within the built-in spatial range, and marking the corresponding articles as first article items; determining the action area of the corresponding article item based on the first article item; obtaining the material properties of each article in the first article item, and substituting the material properties and the action area into the built-in first evaluation formula to obtain the structural risk value of the article, and matching the structural risk value with the built-in first evaluation index, marking the article with a structural risk value greater than the built-in first evaluation index as the first article data; determining the position information of the article data based on the collected image data, and performing state risk assessment on the article according to the position information, and marking the article with risks as second article data; determining the danger of the article according to the first article data and the second article data, and marking the article with danger as a dangerous article; when it is determined that there is a dangerous article, marking the dangerous article for alarm. 2.The integrated intelligent monitoring method for smart construction site according to claim 1, characterized in that: The method comprises the following steps: determining the construction scene according to the construction characteristics of the construction site; the construction scene comprises plane construction and three-dimensional construction; if the construction scene is three-dimensional construction, each construction layer is divided into plane construction corresponding to the spatial relationship to obtain a construction area; obtaining the monitoring range of the monitoring camera, and taking the monitoring range as the distance between the monitoring cameras, and arranging the monitoring cameras in the construction area according to the distance between the monitoring cameras; collecting images of the monitoring range based on the monitoring camera to obtain collected image data. 3.The integrated intelligent monitoring method for smart construction site according to claim 2, characterized in that: The method comprises the following steps: performing arbitrary image collection on the construction area based on the monitoring camera to obtain to-be-analyzed image data; performing article recognition on the to-be-analyzed image data based on image recognition technology to obtain to-be-judged article data; comparing the to-be-judged article data with the actual article data to determine the recognition accuracy, and comparing the recognition accuracy with the preset accuracy value to determine whether the recognition accuracy meets the standard; if the recognition accuracy does not meet the standard, adjusting the monitoring range of the monitoring camera until the recognition accuracy of the image recognition technology on the collected image is greater than the preset accuracy value.

4. The integrated intelligent monitoring method for a smart construction site according to claim 3, characterized in that: The image recognition technology is used to recognize the collected image data, determine all the items contained in the monitoring area, and obtain item data, specifically as follows: A three-dimensional space is constructed based on the monitoring area of the monitoring camera and the image perspective relationship; The three-dimensional space is matched with the area monitored by the monitoring camera to determine the items in the monitoring area that are beyond the surface of the three-dimensional space, and the items are marked as to-be-identified items; The to-be-identified items are identified based on the image recognition technology to obtain item data. 5.The integrated intelligent monitoring method for smart construction site according to claim 4, characterized in that: The position information of the item data is determined based on the collected image data, and the state risk of the item is evaluated based on the position information, and the item with risk is marked as second item data, specifically as follows: The position of the item is marked based on the item data to obtain position data; The position data of the same item in multiple collected image data of the same monitoring camera are compared with each other to determine whether the position of the item changes; If it is determined that the position of the item changes, the reason for the change of the position of the item is judged to determine whether there is human intervention; If it is determined that there is human intervention, the position of the item is re-marked and judged; If it is determined that there is no human intervention, the relationship between the displacement distance and the time of the item is determined based on the collected image data of the item to obtain a state change index, and the risk of the item is determined based on the state change index. 6.The integrated intelligent monitoring method for smart construction site according to claim 5, characterized in that: If it is determined that the position of the item changes, the reason for the change of the position of the item is judged to determine whether there is human intervention, specifically as follows: The time period corresponding to the change of the position of the item is obtained, and the image recognition is performed on the collected image data in the corresponding time period to determine whether there is a human image on the corresponding collected image data; If it is determined that there is a human image on the collected image data, the position movement data of the human image is determined based on the collected image data, and the position movement data of the human image is matched with the position change data when the position of the item changes to determine whether the position movement data of the human image coincides with the position change data when the position of the item changes; If it is determined that the position movement data of the human image coincides with the position data before the change of the position of the item, it is determined that the change of the position of the item is caused by human intervention; otherwise, it is determined that it is not caused by human intervention.

7. The integrated intelligent monitoring method for a smart construction site according to claim 6, characterized in that: The risk of the item is determined based on the first item data and the second item data, and the item with risk is marked as a dangerous item, specifically as follows: The hierarchical relationship of the construction scene is established based on the construction layer, and the hierarchical data of each monitoring camera is determined based on the hierarchical relationship; The hierarchical data and the area position information of the current monitoring area are obtained, and the collected image data located in the same area position and having hierarchical data smaller than the current monitoring range are obtained based on the hierarchical data, the area position information, and the construction scene, and the collected image data and the collected image data corresponding to the current monitoring range are statistically sorted based on the hierarchical relationship to obtain a to-be-analyzed group; If the hierarchical data of the current monitoring range is the first layer, the first item data and the second item data are evaluated based on the built-in danger evaluation model to determine whether the item has risk. If the current monitoring range is not the first layer, the position state of the first article data is judged, and the first article data is divided into the first fixed article whose position does not change and the first mobile article whose position changes according to the position state of each article; Based on the built-in danger assessment model, the article judgment is performed on the first fixed article to determine whether there is a danger judgment standard of the article in the danger assessment model, if there is a danger judgment standard of the article, the danger assessment model is used to assess the danger of the article to determine whether the article is dangerous; If there is no danger judgment standard of the article in the danger assessment model, an auxiliary signal is output, and a human danger mark result is received after the human danger mark, the human danger mark result is used as the danger assessment result of the article, and the corresponding article training is performed on the danger assessment model based on the human danger mark result to obtain the danger judgment standard of the corresponding article; Based on the to-be-analyzed group, the portrait recognition is performed on the to-be-analyzed group to determine whether there is a human trajectory in the monitoring range below the current monitoring range, if it is determined that there is a human trajectory, the risk of falling of the corresponding article is determined according to the state change index of each article in the first mobile article and the second article data to obtain the first index data; Based on the built-in danger assessment model, the danger assessment is performed to obtain the second index data, and whether the article is dangerous is determined based on the first index data and the second index data.

8. An integrated intelligent monitoring system for a smart construction site, characterized in that, The system is used to realize the integrated intelligent monitoring method for intelligent construction site in any one of claims 1-7: including: a collection planning module, a first risk judgment module, a second risk judgment module, a danger judgment module and an alarm module; The collection planning module arranges the monitoring camera based on the construction scene of the construction site, and collects the image of the construction site based on the arranged monitoring camera to obtain the collected image data; The first risk judgment module performs image recognition on the collected image data based on image recognition technology to obtain article data, and judges the structural risk of the article data to mark the article with risk to obtain first article data; The second risk judgment module determines the position information of the article data based on the collected image data, and performs state risk assessment on the article according to the position information to mark the article with risk as second article data; The danger judgment module determines the danger of the article according to the first article data and the second article data, and marks the article with danger as dangerous article; The alarm module marks the dangerous article when it is determined that there is a dangerous article.

Images