A method and system for dynamically monitoring a construction zone

Abstract

The application relates to a method and system for dynamically monitoring a construction area, wherein the method comprises the following steps: collecting real-time construction images of the construction area by using a UAV; slicing the real-time construction images according to a preset size; inputting N real-time tile images obtained through the slicing into a pre-trained construction ground object detection model to perform construction ground object detection, obtaining a construction ground object detection result, and converting the real-time tile images whose construction fence detection result is a construction fence and whose construction ground object detection result is an under-construction building into vector polygons respectively, merging the vector polygons in space, obtaining a real-time core construction area, buffering the real-time core construction area outward according to a preset distance to generate a potential influence area, merging the potential influence area and the core construction area in space, obtaining a construction monitoring range of the construction area, and dynamically monitoring the construction area based on the construction monitoring range. Therefore, the application can meet the real-time monitoring requirement, accurately and objectively generate the core construction area, and improve the monitoring efficiency.

Description

Technical Field

[0001] This invention relates to the field of urban management technology, and in particular to a method and system for dynamic monitoring of construction areas. Background Technology

[0002] In the process of urban management, the monitoring of construction areas mainly relies on manual ground patrols. However, due to its limited coverage, long response time, and strong subjectivity, it is difficult to adapt to large-scale and highly dynamic monitoring scenarios.

[0003] In recent years, drone aerial photography technology has been gradually applied to monitoring scenarios. However, the images of construction areas captured by drones are usually high-resolution images. When the construction area is large, the corresponding image data volume becomes excessive. When analyzing construction area images, memory overflow due to excessive data volume can easily occur, resulting in low analysis efficiency, long response time, and an inability to meet the needs of real-time monitoring. Furthermore, traditional monitoring of construction areas mainly focuses on the core construction area. The division of the core construction area is usually based on manual drawing, which involves significant subjectivity, affecting the accuracy of the generated core construction area and impacting the monitoring efficiency of the construction area. Summary of the Invention

[0004] The technical problem to be solved by the present invention is: the present invention provides a method and system for dynamic monitoring of construction areas, which can not only meet the needs of real-time monitoring, but also eliminate the dependence on manual labor, generate core construction areas in a precise and objective manner, improve the accuracy of the generated core construction areas, and improve monitoring efficiency.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a method for dynamically monitoring a construction area, comprising: Real-time construction images of the current construction area are collected by drones. The real-time construction images are sliced ​​according to a preset size to obtain N real-time tile images. The N real-time tile images are then input into a pre-trained construction feature detection model to detect construction features and obtain the construction feature detection results. The real-time tile images of construction fences and buildings under construction are converted into vector polygons and merged in space to obtain the real-time core construction area. The real-time core construction area is buffered outward at a preset distance to generate a potential impact area. The construction features of the potential impact area are extracted by a visual recognition algorithm, and a construction association area is generated based on the construction features. The construction association area is then spatially merged with the core construction area to obtain the construction impact area of ​​the current construction area.

[0006] The beneficial effects of this invention are as follows: Real-time construction images collected by drones are sliced ​​according to preset sizes, and construction features are detected in the resulting N real-time tile images. Compared with the traditional method of directly analyzing the entire real-time construction image, this significantly reduces computational power, avoids memory overflow problems caused by excessive data volume, shortens response time, and meets the needs of real-time monitoring. The real-time core construction zone is constructed using the construction feature detection results as the basis for real-time tile images of construction fences and buildings under construction. Compared with the traditional method of manual drawing, this eliminates reliance on manual labor, reduces subjectivity, and generates the core construction zone in a precise and objective manner, improving the accuracy of the generated core construction zone. Furthermore, a potential impact zone is generated by caching the core construction zone outwards. The construction monitoring range is generated by merging the potential impact zone with the core construction zone, breaking through the limitation of traditional methods that only monitor the core construction zone while ignoring surrounding related areas. This avoids safety hazards caused by missing related areas, achieving dynamic monitoring of the construction area and improving monitoring efficiency.

[0007] Optionally, the step of inputting N real-time tile images into a pre-trained construction feature detection model to perform construction feature detection and obtain construction feature detection results includes: The pre-trained construction feature detection model is a semantic segmentation model built on a deep learning network. N real-time tile images are input into the semantic segmentation model to detect construction features, generating N construction feature detection results and corresponding N real-time confidence scores. Determine whether there is a real-time confidence level below the first confidence threshold. If so, filter the construction feature detection results corresponding to the real-time confidence level below the first confidence threshold from the N construction feature detection results to obtain M construction feature detection results. Calculate the density and regularity of the real-time tile images corresponding to the M construction feature detection results, and filter the construction feature detection results corresponding to the real-time tile images whose density is lower than a first density threshold and whose regularity is lower than a first regularity threshold from the M construction feature detection results to obtain L construction feature detection results. Perform contextual semantic association on the real-time tile images corresponding to the L construction feature detection results, and determine whether the construction feature detection results corresponding to adjacent real-time tile images are empty. If so, filter the construction feature detection results corresponding to adjacent real-time tile images from the L construction feature detection results to obtain P construction feature detection results.

[0008] As described above, the detection results of N construction features are first filtered based on real-time confidence levels to reduce the risk of misjudgment at the source. Then, a second filtering is performed based on the density and regularity of the real-time tile images to accurately eliminate interference from non-construction features and natural features, further reducing the misjudgment rate. Finally, contextual semantic association is performed on the real-time tile images to consider the spatial continuity of the construction feature detection results and avoid fragmented misjudgments and omissions. This progressive filtering method improves the accuracy of the final construction feature detection results.

[0009] Optionally, the spatial merging of the potential impact area and the core construction area to obtain the construction monitoring range of the construction area includes: Construction features of the potential impact area are extracted using a visual recognition algorithm, and a construction association area is generated based on the construction features. The construction association area is then spatially merged with the core construction area to obtain the construction monitoring range of the construction area.

[0010] As described above, by extracting the construction features of the potential impact area to generate the construction association area, misjudgment caused by the generalization of the potential impact area can be avoided, and ineffective monitoring of the construction area can be reduced.

[0011] Optionally, the dynamic monitoring of the construction area based on the construction monitoring range includes: The construction equipment detection results are obtained by using a pre-trained target detection model to detect construction equipment within the construction monitoring area. The construction equipment detection results include the construction equipment and the equipment type of the construction equipment. The total number of construction equipment of all types is obtained by counting the number of construction equipment of the same type. The construction equipment of the equipment type of tower crane, excavator, bulldozer and crane is taken as the core equipment. The number of core equipment of each core equipment within a preset radius is counted to obtain the total number of core equipment. The total number of core equipment and the total number of construction equipment are input into a first construction activity formula for calculation to obtain a first activity value. This first activity value is used as the construction activity level within the construction monitoring range to dynamically monitor the construction area. The first construction activity formula is: ; in, Indicates the first active value. Indicates the first weight. This indicates the second weight.

[0012] As described above, the construction activity level is calculated within the construction monitoring area. This activity level is obtained by detecting construction equipment within the monitoring area and using a dual-weight formula based on the total number of construction equipment and the total number of core equipment. The total number of core equipment is generated based on the spatial clustering degree of the core equipment, thus achieving an accurate and objective assessment of the construction activity level.

[0013] Optionally, the step of dynamically monitoring the construction area by using the first activity value as the construction activity level within the construction monitoring range includes: Obtain historical construction images of the construction area within a preset period, and obtain the historical core construction area based on the historical construction images; The change rate of the core construction area is calculated based on the historical core construction area and the real-time core construction area. This change rate is then input into a second construction activity formula along with the first activity value to obtain a second activity value. This second activity value is used as the construction activity level within the construction monitoring range for dynamic monitoring of the construction area. The second construction activity level is: ; in, Indicates the second most active value. Indicates the first active value. Indicates the first weight. This indicates the third weight.

[0014] As described above, when calculating construction activity, the change rate of the core construction area between the historical core construction area and the real-time core construction area is further taken into account. By integrating the time dimension, the dynamic trend of construction can be reflected, which facilitates the subsequent capture of trend anomalies and early risk warnings.

[0015] Secondly, the present invention provides a system for dynamically monitoring a construction area, comprising: The construction feature detection module is used to collect real-time construction images of the construction area by drone, slice the real-time construction images according to a preset size to obtain N real-time tile images, input the N real-time tile images into a pre-trained construction feature detection model to detect construction features and obtain construction feature detection results. The real-time core construction area generation module is used to convert the real-time tile images of the construction site objects that are detected as construction fences and the real-time tile images of the construction site objects that are detected as buildings under construction into vector polygons respectively and merge them in space to obtain the real-time core construction area. The dynamic monitoring module is used to buffer the real-time core construction area outward at a preset distance to generate a potential impact area, merge the potential impact area with the core construction area in space to obtain the construction monitoring range of the construction area, and perform dynamic monitoring of the construction area based on the construction monitoring range.

[0016] The beneficial effects of this invention are as follows: Real-time construction images collected by drones are sliced ​​according to preset sizes, and construction features are detected in the resulting N real-time tile images. Compared with the traditional method of directly analyzing the entire real-time construction image, this significantly reduces computational power, avoids memory overflow problems caused by excessive data volume, shortens response time, and meets the needs of real-time monitoring. The real-time core construction zone is constructed using the construction feature detection results as the basis for real-time tile images of construction fences and buildings under construction. Compared with the traditional method of manual drawing, this eliminates reliance on manual labor, reduces subjectivity, and generates the core construction zone in a precise and objective manner, improving the accuracy of the generated core construction zone. Furthermore, a potential impact zone is generated by caching the core construction zone outwards. The construction monitoring range is generated by merging the potential impact zone with the core construction zone, breaking through the limitation of traditional methods that only monitor the core construction zone while ignoring surrounding related areas. This avoids safety hazards caused by missing related areas, achieving dynamic monitoring of the construction area and improving monitoring efficiency.

[0017] Optionally, the construction site feature detection module specifically comprises: The pre-trained construction feature detection model is a semantic segmentation model built on a deep learning network. N real-time tile images are input into the semantic segmentation model to detect construction features, generating N construction feature detection results and corresponding N real-time confidence scores. Determine whether there is a real-time confidence level below the first confidence threshold. If so, filter the construction feature detection results corresponding to the real-time confidence level below the first confidence threshold from the N construction feature detection results to obtain M construction feature detection results. Calculate the density and regularity of the real-time tile images corresponding to the M construction feature detection results, and filter the construction feature detection results corresponding to the real-time tile images whose density is lower than a first density threshold and whose regularity is lower than a first regularity threshold from the M construction feature detection results to obtain L construction feature detection results. Perform contextual semantic association on the real-time tile images corresponding to the L construction feature detection results, and determine whether the construction feature detection results corresponding to adjacent real-time tile images are empty. If so, filter the construction feature detection results corresponding to adjacent real-time tile images from the L construction feature detection results to obtain P construction feature detection results.

[0018] As described above, the detection results of N construction features are first filtered based on real-time confidence levels to reduce the risk of misjudgment at the source. Then, a second filtering is performed based on the density and regularity of the real-time tile images to accurately eliminate interference from non-construction features and natural features, further reducing the misjudgment rate. Finally, contextual semantic association is performed on the real-time tile images to consider the spatial continuity of the construction feature detection results and avoid fragmented misjudgments and omissions. This progressive filtering method improves the accuracy of the final construction feature detection results.

[0019] Optionally, the dynamic monitoring module specifically comprises: Construction features of the potential impact area are extracted using a visual recognition algorithm, and a construction association area is generated based on the construction features. The construction association area is then spatially merged with the core construction area to obtain the construction monitoring range of the construction area.

[0020] As described above, by extracting the construction features of the potential impact area to generate the construction association area, misjudgment caused by the generalization of the potential impact area can be avoided, and ineffective monitoring of the construction area can be reduced.

[0021] Optionally, the dynamic monitoring module includes: The construction activity calculation module is used to detect construction equipment in the construction monitoring range through a pre-trained target detection model and obtain construction equipment detection results, which include construction equipment and the equipment type of construction equipment. The total number of construction equipment of all types is obtained by counting the number of construction equipment of the same type. The construction equipment of the equipment type of tower crane, excavator, bulldozer and crane is taken as the core equipment. The number of core equipment of each core equipment within a preset radius is counted to obtain the total number of core equipment. The total number of core equipment and the total number of construction equipment are input into a first construction activity formula for calculation to obtain a first activity value. This first activity value is used as the construction activity level within the construction monitoring range to dynamically monitor the construction area. The first construction activity formula is: ; in, Indicates the first active value. Indicates the first weight. This indicates the second weight.

[0022] As described above, the construction activity level is calculated within the construction monitoring area. This activity level is obtained by detecting construction equipment within the monitoring area and using a dual-weight formula based on the total number of construction equipment and the total number of core equipment. The total number of core equipment is generated based on the spatial clustering degree of the core equipment, thus achieving an accurate and objective assessment of the construction activity level.

[0023] Optionally, the construction activity calculation module specifically comprises: Obtain historical construction images of the construction area within a preset period, and obtain the historical core construction area based on the historical construction images; The change rate of the core construction area is calculated based on the historical core construction area and the real-time core construction area. This change rate is then input into a second construction activity formula along with the first activity value to obtain a second activity value. This second activity value is used as the construction activity level of the construction monitoring range. The second construction activity level is: ; in, Indicates the second most active value. Indicates the first active value. Indicates the first weight. This indicates the third weight.

[0024] As described above, when calculating construction activity, the change rate of the core construction area between the historical core construction area and the real-time core construction area is further taken into account. By integrating the time dimension, the dynamic trend of construction can be reflected, which facilitates the subsequent capture of trend anomalies and early risk warnings. Attached Figure Description

[0025] Figure 1 This is a flowchart of a method for dynamically monitoring a construction area provided in this embodiment; Figure 2 This is a schematic diagram of the overall process of a method for dynamically monitoring a construction area provided in this embodiment. Figure 3 This is a schematic diagram of the structure of a system for dynamically monitoring a construction area, as provided in this embodiment.

[0026] [Explanation of Labels in the Attached Image] 1. A system for dynamic monitoring of a construction area; 2. Construction Site Feature Detection Module; 3. Real-time core construction area generation module; 4. Dynamic monitoring module; 41. Construction activity calculation module. Detailed Implementation

[0027] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0028] Example 1 Please refer to Figures 1 to 2 This invention provides a method for dynamic monitoring of a construction area, comprising the following steps: S1. Collect real-time construction images of the construction area using a drone, slice the real-time construction images according to a preset size to obtain N real-time tile images, input the N real-time tile images into a pre-trained construction feature detection model to detect construction features, and obtain the construction feature detection results. In this embodiment, as Figure 2 As shown, the real-time construction impact of the construction area is collected by drone, and the real-time construction images are sliced ​​according to a preset size, which is 512×512, to obtain N real-time tile images. The N real-time tile images are then input into a pre-trained construction feature detection model to detect construction features and obtain the construction feature detection results. The construction features include, but are not limited to: buildings under construction, construction fences, bare soil, foundation pits, etc.

[0029] At this point, the step S1 of inputting N real-time tile images into the pre-trained construction feature detection model for construction feature detection, and obtaining the construction feature detection results includes: S11. The pre-trained construction site detection model is a semantic segmentation model built based on a deep learning network. N real-time tile images are input into the semantic segmentation model to detect construction sites, generating N construction site detection results and corresponding N real-time confidence scores. S12. Determine whether there is a real-time confidence level below the first confidence threshold. If so, filter the construction feature detection results corresponding to the real-time confidence level below the first confidence threshold from the N construction feature detection results to obtain M construction feature detection results. S13. Calculate the density and regularity of the real-time tile images corresponding to the M construction feature detection results, and filter the construction feature detection results corresponding to the real-time tile images whose density is lower than the first density threshold and whose regularity is lower than the first regularity threshold from the M construction feature detection results to obtain L construction feature detection results. S14. Perform contextual semantic association on the real-time tile images corresponding to the L construction feature detection results, and determine whether the construction feature detection results corresponding to adjacent real-time tile images are empty. If so, filter the construction feature detection results corresponding to adjacent real-time tile images from the L construction feature detection results to obtain P construction feature detection results.

[0030] In this embodiment, as Figure 2 As shown, the pre-trained construction feature detection model is a semantic segmentation model built on a deep learning network. The semantic segmentation model performs construction feature detection on each input tile image, generating N corresponding construction feature detection results and N corresponding real-time confidence scores. In this embodiment, a multi-layer filtering mechanism is set up to initially filter the construction feature detection results based on the real-time confidence scores. Specifically, the construction feature detection results corresponding to real-time confidence scores below a first confidence threshold are filtered from the N construction feature detection results, resulting in M ​​construction feature detection results, where M≤N.

[0031] Considering that in real-world scenarios, bare soil and foundation pits formed by certain natural factors or non-construction activities may be mistakenly identified as construction feature detection results, the geometric features of the real-time tile images corresponding to the M construction feature detection results are used for filtering. Specifically, the compactness and regularity of the real-time tile images corresponding to the M construction feature detection results are calculated, and the construction feature detection results corresponding to real-time tile images with compactness below the first compactness threshold and regularity below the first regularity threshold are filtered out from the M construction feature detection results to obtain L construction feature detection results, where L≤M.

[0032] In real-world scenarios, construction features actually possess spatial continuity. For example, construction fences do not exist in isolation within a single tile image but are continuously distributed across multiple adjacent tile images. Construction fences and other associated construction features often surround buildings under construction. Therefore, contextual semantic association is performed on the real-time tile images corresponding to L construction feature detection results. If the construction feature detection results corresponding to adjacent real-time tile images are empty, then they do not possess spatial continuity. That is, the construction feature detection results corresponding to adjacent real-time tile images are filtered from the L construction feature detection results to obtain P construction feature detection results, where P ≤ L.

[0033] S2. Convert the real-time tile images of the construction site where the detection result is a construction fence and the real-time tile images of the construction site where the detection result is a building under construction into vector polygons respectively, and merge them in space to obtain the real-time core construction area. In this embodiment, as Figure 2As shown, the real-time tile images of construction fences and buildings under construction are converted into vector polygons and then merged in space. When converting to vector polygons, instead of simply extracting pixel contours, coordinate mapping is performed based on the pos data corresponding to the real-time construction images to generate vector polygons. When merging in space, vector polygons within a preset merging range of 2 meters are actually merged to obtain the real-time core construction area.

[0034] S3. Buffer the real-time core construction area outwards at a preset distance to generate a potential impact area, merge the potential impact area with the core construction area in space to obtain the construction monitoring range of the construction area, and perform dynamic monitoring of the construction area based on the construction monitoring range.

[0035] At this point, step S3, which involves spatially merging the potential impact area and the core construction area to obtain the construction monitoring range of the construction area, includes: S31. Extract the construction features of the potential impact area through a visual recognition algorithm, generate a construction association area based on the construction features, and merge the construction association area with the core construction area in space to obtain the construction monitoring range of the construction area.

[0036] In this embodiment, as Figure 2 As shown, the real-time core construction area is buffered outward to generate a potential impact area at a preset distance of 15 meters. The construction features of the potential impact area are extracted by a visual recognition algorithm. The construction features include, but are not limited to, construction signs, construction materials, construction personnel, and construction equipment. Based on the construction features, a construction association area is generated and spatially merged with the core construction area to obtain the construction monitoring range of the construction area.

[0037] At this point, the dynamic monitoring of the construction area based on the construction monitoring range in step S3 includes: S32. The construction equipment is detected in the construction monitoring area by a pre-trained target detection model to obtain the construction equipment detection results, which include the construction equipment and the equipment type of the construction equipment. S33. Count the number of construction equipment of the same type to obtain the total number of construction equipment of all types. S34. Take the construction equipment of the equipment type as tower crane, the equipment type as excavator, the equipment type as bulldozer and the equipment type as crane as core equipment, count the number of core equipment of each core equipment within a preset radius, and obtain the total number of core equipment; S35. Input the total number of core equipment and the total number of construction equipment into the first construction activity formula for calculation to obtain the first activity value. Use the first activity value as the construction activity level of the construction monitoring range to dynamically monitor the construction area. The first construction activity formula is: ; in, Indicates the first active value. Indicates the first weight. This indicates the second weight.

[0038] In this embodiment, a pre-trained target detection model is used to detect construction equipment within the construction monitoring area. The resulting equipment detection results include not only the construction equipment itself but also its type, which includes tower cranes, excavators, bulldozers, cranes, concrete mixer trucks, and dump trucks. The number of construction equipment of the same type is counted to obtain the total number of all equipment types. Tower cranes, excavators, bulldozers, and cranes are designated as core equipment. The number of other core equipment within a preset radius for each core equipment is counted. Specifically, a circle is drawn with the center point of each core equipment as the center, and the number of other core equipment within each circle is calculated. These counts are then summed to obtain the total number of core equipment. The preset radius is 25 meters. The total number of core equipment and the total number of construction equipment are input into a first construction activity formula to calculate a first activity value. This first activity value is used as the construction activity level within the construction monitoring area for dynamic monitoring.

[0039] In this embodiment, to further reflect the indicative significance of construction equipment on the construction status, the first construction activity formula was updated to obtain an updated first activity value. This updated first activity value is then used as the construction activity level within the construction monitoring range for dynamic monitoring of the construction area. The first construction activity formula is: ; in, This indicates the first active value after the update. Indicates the first weight. This indicates the second weight, and n represents the total number of device types. This indicates the number of construction equipment of type i. This indicates the fourth weight.

[0040] At this point, step S35, which involves using the first activity value as the construction activity level within the construction monitoring range to dynamically monitor the construction area, includes: S351. Obtain historical construction images of the construction area within a preset period, and obtain the historical core construction area based on the historical construction images; S352. Calculate the core construction area change rate based on the historical core construction area and the real-time core construction area. Input the core construction area change rate and the first activity value into the second construction activity formula for calculation to obtain the second activity value. Use the second activity value as the construction activity of the construction monitoring range to dynamically monitor the construction area. The second construction activity is: ; in, Indicates the second most active value. Indicates the first active value. Indicates the first weight. This indicates the third weight.

[0041] In this embodiment, as Figure 2 As shown, the historical core construction area is obtained by acquiring and using historical construction images of the construction area within a preset period of one week. The construction area change rate is calculated based on the historical core construction area and the real-time core construction area. The construction area change rate and the first activity value are input into the second construction activity formula for calculation. The obtained second activity value is used as the construction activity of the construction monitoring range to dynamically monitor the construction area.

[0042] Example 2 Please refer to Figure 3 The present invention provides a system 1 for dynamic monitoring of a construction area, comprising: a construction site feature detection module 2, a real-time core construction area generation module 3, a dynamic monitoring module 4, and a construction activity calculation module 41.

[0043] Among them, the construction feature detection module 2 is used to collect real-time construction images of the construction area by drone, slice the real-time construction images according to a preset size to obtain N real-time tile images, input the N real-time tile images into a pre-trained construction feature detection model to perform construction feature detection, and obtain construction feature detection results. The real-time core construction area generation module 3 is used to convert the real-time tile image of the construction site detection result as a construction fence and the real-time tile image of the construction site detection result as a building under construction into vector polygons respectively and merge them in space to obtain the real-time core construction area. The dynamic monitoring module 4 is used to buffer the real-time core construction area outward at a preset distance to generate a potential impact area, merge the potential impact area with the core construction area in space to obtain the construction monitoring range of the construction area, and perform dynamic monitoring of the construction area based on the construction monitoring range.

[0044] Specifically, the construction site feature detection module 2 is as follows: The pre-trained construction feature detection model is a semantic segmentation model built on a deep learning network. N real-time tile images are input into the semantic segmentation model to detect construction features, generating N construction feature detection results and corresponding N real-time confidence scores. Determine whether there is a real-time confidence level below the first confidence threshold. If so, filter the construction feature detection results corresponding to the real-time confidence level below the first confidence threshold from the N construction feature detection results to obtain M construction feature detection results. Calculate the density and regularity of the real-time tile images corresponding to the M construction feature detection results, and filter the construction feature detection results corresponding to the real-time tile images whose density is lower than a first density threshold and whose regularity is lower than a first regularity threshold from the M construction feature detection results to obtain L construction feature detection results. Perform contextual semantic association on the real-time tile images corresponding to the L construction feature detection results, and determine whether the construction feature detection results corresponding to adjacent real-time tile images are empty. If so, filter the construction feature detection results corresponding to adjacent real-time tile images from the L construction feature detection results to obtain P construction feature detection results.

[0045] Specifically, the dynamic monitoring module 4 is as follows: Construction features of the potential impact area are extracted using a visual recognition algorithm, and a construction association area is generated based on the construction features. The construction association area is then spatially merged with the core construction area to obtain the construction monitoring range of the construction area.

[0046] Specifically, the dynamic monitoring module 4 includes: The construction activity calculation module 41 is used to detect construction equipment in the construction monitoring range through a pre-trained target detection model and obtain construction equipment detection results, which include construction equipment and the equipment type of construction equipment. The total number of construction equipment of all types is obtained by counting the number of construction equipment of the same type. The construction equipment of the equipment type of tower crane, excavator, bulldozer and crane is taken as the core equipment. The number of core equipment of each core equipment within a preset radius is counted to obtain the total number of core equipment. The total number of core equipment and the total number of construction equipment are input into a first construction activity formula for calculation to obtain a first activity value. This first activity value is used as the construction activity level within the construction monitoring range to dynamically monitor the construction area. The first construction activity formula is: ; in, Indicates the first active value. Indicates the first weight. This indicates the second weight.

[0047] Specifically, the construction activity calculation module 41 is as follows: Obtain historical construction images of the construction area within a preset period, and obtain the historical core construction area based on the historical construction images; The change rate of the core construction area is calculated based on the historical core construction area and the real-time core construction area. This change rate is then input into a second construction activity formula along with the first activity value to obtain a second activity value. This second activity value is used as the construction activity level within the construction monitoring range for dynamic monitoring of the construction area. The second construction activity level is: ; in, Indicates the second most active value. Indicates the first active value. Indicates the first weight. This indicates the third weight.

[0048] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0049] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0050] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0051] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0052] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0053] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0054] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A method for dynamic monitoring of a construction area, characterized in that, include: Real-time construction images of the construction area are collected by drones. These images are then sliced ​​according to a preset size to obtain N real-time tile images. These N real-time tile images are then input into a pre-trained construction feature detection model to perform construction feature detection, yielding the following results: The pre-trained construction feature detection model is a semantic segmentation model built on a deep learning network. N real-time tile images are input into the semantic segmentation model to detect construction features, generating N construction feature detection results and corresponding N real-time confidence scores. Determine whether there is a real-time confidence level below the first confidence threshold. If so, filter the construction feature detection results corresponding to the real-time confidence level below the first confidence threshold from the N construction feature detection results to obtain M construction feature detection results. Calculate the density and regularity of the real-time tile images corresponding to the M construction feature detection results, and filter the construction feature detection results corresponding to the real-time tile images whose density is lower than a first density threshold and whose regularity is lower than a first regularity threshold from the M construction feature detection results to obtain L construction feature detection results. Perform contextual semantic association on the real-time tile images corresponding to the L construction feature detection results, determine whether the construction feature detection results corresponding to adjacent real-time tile images are empty, and if so, filter the construction feature detection results corresponding to adjacent real-time tile images from the L construction feature detection results to obtain P construction feature detection results. The real-time tile images of construction fences and buildings under construction are converted into vector polygons and merged in space to obtain the real-time core construction area. The real-time core construction area is buffered outwards at a preset distance to generate a potential impact zone. This potential impact zone is then spatially merged with the core construction area to obtain the construction monitoring range of the construction area. Dynamic monitoring of the construction area is then performed based on this monitoring range, including: The construction equipment detection results are obtained by using a pre-trained target detection model to detect construction equipment within the construction monitoring area. The construction equipment detection results include the construction equipment and the equipment type of the construction equipment. The total number of construction equipment was obtained by counting the number of construction equipment. The construction equipment of the equipment type of tower crane, excavator, bulldozer and crane is taken as the core equipment. The number of core equipment of each core equipment within a preset radius is counted to obtain the total number of core equipment. The total number of core equipment and the total number of construction equipment are input into a first construction activity formula for calculation to obtain a first activity value. This first activity value is used as the construction activity level within the construction monitoring range to dynamically monitor the construction area. The first construction activity formula is: ； in, Indicates the first active value. Indicates the first weight. Indicates the second weight; Using the first activity value as the construction activity level within the construction monitoring range for dynamic monitoring of the construction area includes: Obtain historical construction images of the construction area within a preset period, and obtain the historical core construction area based on the historical construction images; The change rate of the core construction area is calculated based on the historical core construction area and the real-time core construction area. This change rate is then input into a second construction activity formula along with the first activity value to obtain a second activity value. This second activity value is used as the construction activity level within the construction monitoring range for dynamic monitoring of the construction area. The second construction activity level is: ； in, Indicates the second most active value. Indicates the first active value. Indicates the first weight. This indicates the third weight.

2. The method for dynamic monitoring of a construction area as described in claim 1, characterized in that, The spatial merging of the potential impact area and the core construction area to obtain the construction monitoring range of the construction area includes: Construction features of the potential impact area are extracted using a visual recognition algorithm, and a construction association area is generated based on the construction features. The construction association area is then spatially merged with the core construction area to obtain the construction monitoring range of the construction area.

3. A system for dynamic monitoring of a construction area, characterized in that, include: The construction feature detection module is used to collect real-time construction images of the construction area by drone, slice the real-time construction images according to a preset size to obtain N real-time tile images, input the N real-time tile images into a pre-trained construction feature detection model to detect construction features and obtain construction feature detection results. The construction site feature detection module is specifically as follows: The pre-trained construction feature detection model is a semantic segmentation model built on a deep learning network. N real-time tile images are input into the semantic segmentation model to detect construction features, generating N construction feature detection results and corresponding N real-time confidence scores. Determine whether there is a real-time confidence level below the first confidence threshold. If so, filter the construction feature detection results corresponding to the real-time confidence level below the first confidence threshold from the N construction feature detection results to obtain M construction feature detection results. Calculate the density and regularity of the real-time tile images corresponding to the M construction feature detection results, and filter the construction feature detection results corresponding to the real-time tile images whose density is lower than a first density threshold and whose regularity is lower than a first regularity threshold from the M construction feature detection results to obtain L construction feature detection results. Perform contextual semantic association on the real-time tile images corresponding to the L construction feature detection results, determine whether the construction feature detection results corresponding to adjacent real-time tile images are empty, and if so, filter the construction feature detection results corresponding to adjacent real-time tile images from the L construction feature detection results to obtain P construction feature detection results. The real-time core construction area generation module is used to convert the real-time tile images of the construction site objects that are detected as construction fences and the real-time tile images of the construction site objects that are detected as buildings under construction into vector polygons respectively and merge them in space to obtain the real-time core construction area. The dynamic monitoring module is used to buffer the real-time core construction area outward at a preset distance to generate a potential impact area, merge the potential impact area with the core construction area in space to obtain the construction monitoring range of the construction area, and perform dynamic monitoring of the construction area based on the construction monitoring range. The dynamic monitoring module includes: The construction activity calculation module is used to detect construction equipment in the construction monitoring range through a pre-trained target detection model and obtain construction equipment detection results, which include construction equipment and the equipment type of construction equipment. The total number of construction equipment was obtained by counting the number of construction equipment. The construction equipment of the equipment type of tower crane, excavator, bulldozer and crane is taken as the core equipment. The number of core equipment of each core equipment within a preset radius is counted to obtain the total number of core equipment. The total number of core equipment and the total number of construction equipment are input into a first construction activity formula for calculation to obtain a first activity value. This first activity value is used as the construction activity level within the construction monitoring range to dynamically monitor the construction area. The first construction activity formula is: ； in, Indicates the first active value. Indicates the first weight. Indicates the second weight; The construction activity calculation module is specifically as follows: Obtain historical construction images of the construction area within a preset period, and obtain the historical core construction area based on the historical construction images; The change rate of the core construction area is calculated based on the historical core construction area and the real-time core construction area. This change rate is then input into a second construction activity formula along with the first activity value to obtain a second activity value. This second activity value is used as the construction activity level within the construction monitoring range for dynamic monitoring of the construction area. The second construction activity level is: ； in, Indicates the second most active value. Indicates the first active value. Indicates the first weight. This indicates the third weight.

4. The system for dynamic monitoring of a construction area as described in claim 3, characterized in that, The dynamic monitoring module is specifically: Construction features of the potential impact area are extracted using a visual recognition algorithm, and a construction association area is generated based on the construction features. The construction association area is then spatially merged with the core construction area to obtain the construction monitoring range of the construction area.

Images