Difference recognition method and difference recognition system

Abstract

To provide a difference recognition method and a difference recognition system that correctly recognizes the latest progress status.SOLUTION: A difference recognition system 10 recognizes the latest progress status of a construction site on the basis of a difference, which indicates progress of a construction site. The difference recognition system 10 includes: a data management unit 31 that acquires first captured-image data captured in a first state at a first time point and second captured-image data captured in a second state at a second time point that is subsequent to the first time point; and a process determination unit 33 serving as a difference recognition unit that recognizes a newly installed building element, on the basis of the first captured-image data and second captured-image data, as a difference between the first state and the second state. The process determination unit 33 determines the latest progress status on the basis of the recognized difference.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a difference recognition method and a difference recognition system for recognizing a difference, which is the progress at a construction site of a construction work. [Background technology]

[0002] The progress of construction work at a construction site is generally managed by a construction manager patrolling the site to check whether the work is progressing as planned, or by the construction manager requesting each person in charge of each type of work to report the progress on a daily basis. Patent Document 1 discloses a system for determining the progress of construction work based on an image (current image) of the current construction site as a system for determining the progress. In Patent Document 1, image features of architectural elements contained in an image are calculated from an image (past image) of a construction site taken in the past, and training data is generated based on the calculated image features. Then, the image features of the current image and the training data are used to determine the progress of construction work. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2017-107443 A Summary of the Invention [Problem to be solved by the invention]

[0004] On the other hand, the progress status of a construction site is not uniform for each site. In addition, in some cases, changes in the progress status may occur due to rework. Therefore, the technology of Patent Document 1 may not be able to correctly recognize the latest progress status. [Means for solving the problem]

[0005] A difference recognition method that solves the above problem is a difference recognition method that recognizes a difference, which is progress at a construction site, and includes a first state acquisition step of acquiring first photographed data that captures a first state at a first point in time, a second state acquisition step of acquiring second photographed data that captures a second state at a second point in time after the first point in time, and a difference recognition step of recognizing a difference corresponding to a change from the first state to the second state based on the first photographed data and the second photographed data.

[0006] The difference recognition system that solves the above problem is a difference recognition system that recognizes differences, which are progress at a construction site, and includes a data management unit that acquires first photographed data that captures a first state at a first point in time and second photographed data that captures a second state at a second point in time after the first point in time, and a difference recognition unit that recognizes the difference between the first state and the second state based on the first photographed data and the second photographed data.

[0007] According to the above configuration, the progress at the construction site is recognized based on the first image data and the second image data arranged in chronological order, so that it is possible to correctly recognize the latest progress situation, including rework.

[0008] In the difference recognition method having the above configuration, the difference recognition step may obtain a change in the recognition probability of the difference when the context of the first photographed data and the second photographed data is swapped, and recognize rework as the difference according to the change in the recognition probability. In this way, it is possible to recognize rework according to the change in the recognition probability of the difference when the context of the first photographed data and the second photographed data is swapped.

[0009] In the difference recognition method having the above configuration, the difference recognition step may recognize rework as a difference when it is determined that an element included in the first photographed data has been removed in the second photographed data. In this way, it is possible to recognize rework when an element included in the first photographed data does not exist in the second photographed data and is determined to have been removed.

[0010] In the difference recognition method having the above configuration, the difference recognition step may recognize rework by comparing a change in the process from the first photographed data to the second photographed data with a chronological order of the process prepared in advance. In this way, it is possible to recognize rework by comparing a change in the process from the first photographed data to the second photographed data with a construction work process prepared in advance.

[0011] In the difference recognition method having the above configuration, in the difference recognition step, the difference may be recognized using a model that has learned the difference. According to the above configuration, differences are recognized using a model that has learned differences, so that even if there is a discrepancy between the first image data and the second image data, such as a difference in the shooting location, it is possible to correctly recognize the latest progress status.

[0012] In the difference recognition method having the above configuration, in the difference recognition step, a changed portion between the first photographed data and the second photographed data may be extracted, and then the difference may be recognized using the model.

[0013] According to the above configuration, since the changed portion between the first image data and the second image data is extracted before the difference is recognized by the model, it is possible to reduce the calculation load when recognizing the difference using the model.

[0014] In the difference recognition method having the above configuration, a label indicating the difference may be added in learning the model. According to the above configuration, a label indicating a difference is assigned in model learning, so that the progress status can be recognized for each label.

[0015] In the difference recognition method having the above configuration, in learning the model, photographed data in which a region showing the difference is distinguished may be added. According to the above configuration, in learning the model, the photographed data in which the area showing the difference is distinguished is added, whereby the difference area between the first photographed data and the second photographed data can be visually recognized.

[0016] In the difference recognition method having the above configuration, it is preferable that the difference is classified in the difference recognition step. According to the above configuration, the difference between the first photographed data and the second photographed data can be classified.

[0017] The difference recognition method having the above configuration may include a method in which an intermediate time point exists between the first time point and the second time point, and the difference between the first time point and the intermediate time point is recognized based on the first shooting data and the second shooting data.

[0018] According to the above configuration, by recognizing the difference between the first time point and the intermediate time point, it is possible to recognize what kind of difference exists between the first time point and the second time point. [Brief description of the drawings]

[0019] [Figure 1] 1 is a block diagram showing a schematic configuration of an embodiment of a difference recognition system; [Diagram 2] FIG. 2 is a block diagram showing an example of a hardware configuration of an information processing device. [Diagram 3] FIG. 2 is a diagram showing an example of shooting data arranged in chronological order and an example of labels added between each piece of shooting data. [Figure 4] FIG. 10 is a diagram illustrating a learning process. [Diagram 5] FIG. 2 is a diagram illustrating a data structure of a schedule information storage unit. [Figure 6] 2 is a diagram illustrating a data structure of a site photography information storage unit. FIG. [Figure 7] 4 is a diagram illustrating a data configuration of a progress information storage unit. FIG. [Figure 8] 13 is a flowchart showing an example of a process determination process. [Figure 9] FIG. 10 is a diagram illustrating a learning process according to a modified example. [Figure 10] FIG. 13 is a diagram illustrating correct answer data of teacher data in a modified example. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] An embodiment of a difference recognition method and a difference recognition system will be described with reference to FIGS. As shown in Fig. 1, the difference recognition system 10 is a system that supports management of the progress of construction work by recognizing differences, which are progress at a construction site, using photographic data obtained by photographing the construction site. The difference recognition system 10 includes a photographing device 11, a manager terminal 12, and a difference recognition device 20.

[0021] (Hardware configuration) 2, a hardware configuration of an information processing device H10 that constitutes the photographing device 11, the administrator terminal 12, and the difference recognition device 20 will be described. The information processing device H10 includes a communication device H11, an input device H12, a display device H13, a storage device H14, and a processor H15. Note that this hardware configuration is an example, and it can also be realized by other hardware.

[0022] The communication device H11 is an interface that establishes a communication path with another device and transmits and receives data, and is, for example, a network interface or a wireless interface.

[0023] The input device H12 is a device that accepts input from a user, etc., and is, for example, a mouse, a keyboard, etc. The display device H13 is, for example, a display that displays various information. A touch panel display or the like functions as the input device H12 and the display device H13.

[0024] The storage device H14 is a storage device that stores data and various programs for executing various functions of the photographing device 11, the administrator terminal 12, and the difference recognition device 20. Examples of the storage device H14 include a ROM, a RAM, and a hard disk.

[0025] The processor H15 controls each process using programs and data stored in the storage device H14. Examples of the processor H15 include a CPU and an MPU. The processor H15 loads programs stored in a ROM or the like into a RAM and executes various processes for each service.

[0026] The processor H15 is not limited to a processor that performs software processing for all the processes it executes. For example, the processor H15 may include a dedicated hardware circuit (e.g., an application specific integrated circuit (ASIC)) that performs hardware processing for at least a part of the processes it executes. That is, the processor H15 may be configured as follows:

[0027] [1] One or more processors operating according to a computer program (software); [2] One or more dedicated hardware circuits that perform at least some of the various processes; or [3] Circuits that include combinations of these The processor includes a CPU and memory, such as RAM and ROM, that stores program codes or instructions configured to cause the CPU to execute processes. Memory, i.e., computer-readable media, includes any available media that can be accessed by a general-purpose or special-purpose computer.

[0028] (System Configuration) As shown in Fig. 1, the photographing device 11 is a photographing means for photographing a construction site in a construction work. For example, the photographing device 11 is a camera that acquires image data as photographed data of the construction site. In addition, for example, the photographing device 11 is a three-dimensional laser scanner that acquires point cloud data as photographed data of the construction site. The photographing device 11 is not limited to being carried by a worker, but may be mounted on a robot that patrols the construction site by autonomous control or an unmanned aerial vehicle that can be operated by a worker.

[0029] The photographing device 11 associates the photographing data with an identifier for identifying the photographing date and time and the location where the photograph was taken, and stores the data. The photographing date and time can be acquired from a built-in timer of the photographing device 11. The photographing location can be acquired by manually inputting the photographing location via an input means of the photographing device 11, or by using a location identification means such as a Global Positioning System (GPS). The photographing data is uploaded to the difference recognition device 20. The imaging data may be uploaded directly from the photographing device 11 to the difference recognition device 20, or may be uploaded via the administrator terminal 12.

[0030] The manager terminal 12 is a computer terminal used by a manager who manages the progress of work at a construction site. The difference recognition device 20 is a computer that supports management of the progress of construction work using photographed data of the construction site. The difference recognition device 20 includes a control unit 21, a model storage unit 22, a teacher information storage unit 23, a schedule information storage unit 24, a site photographed information storage unit 25, and a progress information storage unit 26.

[0031] The control unit 21 functions as a data management unit 31, a learning processing unit 32, a process determination unit 33, and the like by executing programs for various processes described later. The data management unit 31 executes a process of acquiring teacher data from the administrator terminal 12. In this process, the data management unit 31 stores the acquired teacher data in the teacher information storage unit 23.

[0032] The data management unit 31 executes a process of acquiring schedule management data from the manager terminal 12. In this process, the data management unit 31 stores the acquired schedule management data in the schedule information storage unit 24.

[0033] The data management unit 31 executes a process of acquiring the image capturing data captured by the image capturing device 11. In this process, the data management unit 31 stores the acquired image capturing data in the on-site image capturing information storage unit 25.

[0034] The learning processing unit 32 executes a learning process using the teacher data stored in the teacher information storage unit 23 and the machine learning model stored in the model storage unit 22. The training data includes learning photographic data in chronological order, which shows the state of each process at a certain time in the construction site. The certain time can be the time when each process is completed, or any time between the start and the completion of the process. The training data also includes labels indicating the architectural elements installed in each process as correct answer data.

[0035] More specifically, the teacher data includes multiple patterns of combinations of first learning image data at time t (first time point) showing a first state of the construction site, second learning image data at time t+1 (second time point) showing a second state of the construction site, and labels showing building elements (hereinafter referred to as new elements) newly installed between time t and time t+1. The labels are correct answer data corresponding to the first learning image data and the second learning image data.

[0036] An example of a pattern contained in the training data will be described with reference to Fig. 3. In Fig. 3, photographed data from a certain construction site is shown arranged in chronological order from left to right. As shown in FIG. 3, a plasterboard 43 and an LGS (Light Gauge Steel) 44 are recognized as new elements between the learning photographed data 41 and the learning photographed data 42. In this case, the learning photographed data 41 is the first learning photographed data, and the learning photographed data 42 is the second learning photographed data. Then, "plasterboard" and "LGS" are given as labels. The corresponding process is the plasterboard / LGS installation process. In the learning photographed data 42, the area where the plasterboard 43 is installed is shaded, and the area where the LGS 44 is installed is dotted.

[0037] Glass wool 46 and plasterboard 47 are recognized as new elements between learning photographed data 42 and learning photographed data 45. In this case, learning photographed data 42 is the first learning photographed data, and learning photographed data 45 is the second learning photographed data. Then, "glass wool" and "plasterboard" are given as labels. The corresponding process is the glass wool / plasterboard installation process. In learning photographed data 45, the area where glass wool 46 is installed is shaded, and the area where plasterboard 47 is installed is dotted.

[0038] For example, if a reinforcing bar region is recognized in the first learning image data and concrete is recognized in the reinforcing bar region in the second learning image data, the label "concrete" is assigned. The corresponding process is a concrete pouring process.

[0039] As shown in Fig. 4, in the learning process, the learning processing unit 32 performs learning on a machine learning model 53 stored in the model storage unit 22 using the first learning shooting data 51 and the second learning shooting data 52 as input data and the label 54 as output data. The learning processing unit 32 performs learning using each pattern included in the teacher data. That is, the learning processing unit 32 performs learning to recognize a new element that is a difference between the first learning shooting data and the second learning shooting data based on the first learning shooting data and the second learning shooting data. Note that a known learning method can be used for the learning processing unit 32.

[0040] The process determination unit 33 executes a process of determining the progress of the construction site based on the schedule management data stored in the schedule information storage unit 24, the photographed data stored in the site photographed information storage unit 25, and the machine learning model 53 stored in the model storage unit 22. Details of this process will be described later.

[0041] As shown in Fig. 5, the schedule information storage unit 24 records schedule management data 56 related to the schedule of the construction site. The schedule management data 56 is recorded through the manager terminal 12 when new construction elements to be installed and the planned completion date are decided for each process at the construction site. The schedule management data 56 includes data that associates the location, process, planned completion date, and architectural elements for each construction site. In other words, the schedule management data 56 includes information indicating the construction site and data indicating the location, process, planned completion date, and architectural elements associated therewith.

[0042] The location data area records data related to an identifier for identifying the location of the construction site. For example, data for identifying a work area or a room is used as location data. The process data area records data related to an identifier for identifying each process to be carried out at the construction site. The planned completion date data area records data related to the planned date for completion of each process at the construction site. Based on the planned completion dates, the order of the processes at the construction site can be identified. The building element data area records data related to new elements for each process.

[0043] 6, image management data 57 including image data obtained by photographing a construction site is stored in the site photography information storage unit 25. The image management data 57 is stored when the photography data is acquired.

[0044] The image management data 57 includes an identification code, shooting date and time, location, and image-related data. The identification code data area stores data related to an identifier for identifying the shooting data. The shooting date and time data area stores data related to the date and time when the image was captured by the camera device 11. The location data area stores data related to an identifier for identifying the location of the construction site captured by the camera device 11. The shooting data area stores shooting data of the construction site.

[0045] 7, progress management data 58 for managing the progress status of a construction site is recorded in the progress information storage unit 26. The progress management data 58 is recorded when a process determination process is performed.

[0046] The progress management data 58 includes data on locations, identification codes, processes, and progress status. The location data area records data on identifiers for identifying each location on the construction site. The identification code data area records data on identifiers for identifying the photographing data of the location. The process data area records data on identifiers for identifying the process recognized based on the first photographing data and the second photographing data. The progress status data area records data indicating the progress status against the schedule of the construction site.

[0047] (Process Judgment Processing) The process determination process executed by the control unit 21 of the difference recognition device 20 will be described with reference to FIG. First, the control unit 21 of the difference recognition device 20 executes a process of acquiring photographed data of a construction site (step S101: second state acquisition step). In this process, the data management unit 31 assigns an identification code to the photographed data uploaded to the difference recognition device 20 and stores it in the site photographed information storage unit 25. This photographed data is associated with an identifier related to the photographed date and time and location.

[0048] Next, the control unit 21 of the difference recognition device 20 repeats the following process for each piece of photographic data, using the photographic data newly stored in the on-site photographic information storage unit 25 as judgment photographic data. First, the process determination unit 33 executes a process of acquiring the first photographed data (step S102: first state acquisition step). In this process, the process determination unit 33 acquires, as the first photographed data, the photographed data of the previous photographed date and time at the location based on the photographed date and time and the location associated with the photographed data for judgment from the on-site photographed information storage unit 25.

[0049] Next, the process determination unit 33, as a difference recognition unit, executes a process of inputting the first photographed data and the second photographed data, that is, the photographed data for determination, to the machine learning model 53, and acquiring the label output from the machine learning model 53 as a result of the input as a new element (step S103). Note that the subsequent processes including step S103 constitute a difference recognition step.

[0050] Next, the process determination unit 33 executes a process for searching for a process at the time when the determination shooting data was captured (step S104). In this process, the process determination unit 33 searches for a process at the time when the determination shooting data was captured, based on the location associated with the determination shooting data, the new element output from the machine learning model 53, and the schedule management data 56 stored in the schedule information storage unit 24. More specifically, the process determination unit 33 acquires the schedule management data 56 corresponding to the location associated with the determination shooting data from the schedule information storage unit 24. Then, the process determination unit 33 searches the schedule management data 56 for a process including a building element corresponding to the new element.

[0051] Next, the process determination unit 33 determines whether or not the process at the time of capturing the determination-use imaging data can be identified as a result of the search (step S105). If the process cannot be identified (step S105: NO), the process determination unit 33 executes a process selection process (step S106). Specifically, the process determination unit 33 identifies a process including at least one of the newly-installed elements in the determination-use photographed data as a process candidate, and then outputs the identified process candidate to the manager terminal 12. Then, the process determination unit 33 acquires the process candidate designated in the manager terminal 12 as a process corresponding to the determination-use photographed data.

[0052] If the process can be identified (step S105: YES), the process determination unit 33 acquires the identified process as the process corresponding to the determination-use shooting data, and skips the process of step S106.

[0053] Next, the process determination unit 33 executes a process recording process (step S107). Specifically, the process determination unit 33 of the control unit 21 records the progress management data 58 corresponding to the determination-use photographed data in the progress information storage unit .

[0054] In this process, the process determination unit 33 records the location of the determination-use photographed data, the identification code, and the acquired process in the progress information storage unit 26. Furthermore, when the process determination unit 33 determines that the acquired process is completed, it compares the planned completion date recorded in the schedule management data 56 acquired at the time of the search with the photographing date and time recorded in the site photographing information storage unit 25 to calculate the difference in days. Then, the process determination unit 33 records the progress status in the progress management data 58. Specifically, when the photographing date is later than the planned completion date, the process determination unit 33 records "delay of the difference in days" in the progress status data area. On the other hand, when the photographing date is earlier than the planned completion date, the process determination unit 33 records "advancement of the difference in days" in the progress status data area. Furthermore, when the photographing date and the planned completion date match, the process determination unit 33 records "as planned" in the progress status data area. Note that the process completion determination by the process determination unit 33 can be performed based on, for example, the photographing data for determination, the new element output from the machine learning model 53, and data indicating the position of the new element at the construction site. Data indicating the location of the new element is recorded as part of the schedule management data 56 in association with information indicating the construction site.

[0055] The process determination unit 33 repeats the above process until it is completed for all of the photographed data newly stored in the site photographed information storage unit 25. Then, the process determination unit 33 records the progress status for all of the photographed data.

[0056] (action) In the difference recognition system 10, when the location for which the progress status is to be checked is input into the manager terminal 12, the control unit 21 of the difference recognition device 20 outputs the progress status corresponding to that location to the manager terminal 12 based on the progress management data 58 recorded in the progress information storage unit 26.

[0057] According to the difference recognition method and the difference recognition system of the present embodiment, the following effects can be obtained. (1) The difference recognition system 10 recognizes new elements indicating the difference between the first photographed data and the second photographed data, based on the first photographed data and the second photographed data. Then, the process is determined based on the new elements. Therefore, when the construction is progressing smoothly, the newly installed building elements in each process are recognized as new elements. Even if there is rework, building elements that did not exist in the photographed data before the rework but exist in the photographed data after the rework are recognized as new elements. As a result, the progress of the construction site can be correctly recognized.

[0058] (2) The difference recognition system 10 recognizes the new element using a machine learning model trained based on training data. This makes it possible to more reliably recognize the correct new element even if there is a difference between the shooting location of the first shooting data and the shooting location of the second shooting data.

[0059] (3) The machine learning model is trained based on training data that includes labels indicating newly installed architectural elements, allowing the difference recognition system 10 to determine the detailed progress status for each newly installed architectural element.

[0060] (4) The machine learning model is trained based on teacher data including multiple labels for a set of learning photographic data, which allows the process determination unit 33 to recognize the differences as multiple classified new elements.

[0061] The above embodiment may be modified as follows. 9, in the learning process, the learning processing unit 32 may input extracted data 60 obtained by extracting a changed portion between the first learning shooting data 51 and the second learning shooting data 52 to the machine learning model 53. In this case, the process determination unit 33 inputs the data obtained by extracting a changed portion between the first shooting data and the second shooting data to the machine learning model 53.

[0062] In addition, the learning processing unit 32 may perform feature matching using the first learning shooting data 51 and the second learning shooting data 52, and generate extracted data 60 after aligning the positions of the first learning shooting data 51 and the second learning shooting data 52.

[0063] 3 and 10, the teacher data may be data obtained by distinguishing the area of ​​the new element between time t and time t+1 using the second learning shooting data as the correct answer data corresponding to the first learning shooting data and the second learning shooting data. In FIG. 10, the area of ​​the new element (LGS part of the ceiling) is marked with dots.

[0064] Both image data and point cloud data may be used as the photographic data. In the learning process, the learning processing unit 32 may perform learning by using the shooting data at time t as the first learning shooting data and the shooting data at time t+k (k is an integer of 2 or more) as the second learning shooting data. In other words, the learning processing unit 32 may perform learning by skipping the newly added element between time t and time t+k.

[0065] Furthermore, the learning processing unit 32 may perform learning using, for example, the photographed data at time t+1 as the first learning photographed data and the photographed data at time t as the second learning photographed data. For example, in the case of the learning photographed data 42, 45 shown in FIG. 3, the learning processing unit 32 performs learning using labels such as removal of glass wool and plasterboard as correct answer data. In this way, the learning processing unit 32 may perform learning taking rework into consideration. With this configuration, the latest progress taking rework into consideration can be recognized more accurately.

[0066] The process determination unit 33 may recognize a difference between an intermediate point between the first time point and the second time point and the first time point, based on the first photographed data and the second photographed data. For example, when the photographing times of the photographed data 41, 42, and 45 shown in FIG. 3 are the first time point, the intermediate time point, and the second intermediate time point, respectively, the process determination unit 33 may recognize the difference between the photographed data 41 and the photographed data 42, with the photographed data 41 as the first photographed data and the photographed data 45 as the second photographed data. Specifically, the process determination unit 33 recognizes the gypsum board 43 and the LGS 44 as new elements at the intermediate time point based on the recognition of the glass wool 46 and the gypsum board 47 as new elements between the photographed data 41 and the photographed data 45. That is, when the process determination unit 33 recognizes a skipped process based on the latest progress status and the progress management data 58, it records the skipped process as completed in the progress management data 58. Thereby, even if a return to the skipped process occurs, the return can be recognized more reliably.

[0067] In the process of acquiring a new element (step S103), even if the first photographed data and the photographed data for judgment (second photographed data) are inputted in chronological order to the machine learning model 53, the new element may not be acquired due to a low recognition probability. In such a case, in step S103, the process determination unit 33 may further execute a process of switching the time series of the first photographed data and the photographed data for judgment and inputting them to the machine learning model 53. Then, if the recognition probability is increased by switching the time series and the new element can be acquired, the process determination unit 33 may identify rework in the next step S104 (search for a process). In this way, it is possible to identify with a high degree of certainty that a rework has occurred by switching the time series of the photographed data and acquiring a new element (higher recognition probability). As a result, the progress of the construction site can be correctly recognized.

[0068] Furthermore, for example, if the new element cannot be acquired in step S103, the process determination unit 33 may determine that the new element in the first captured data has been removed, and may specify a rework in the next step S104.

[0069] The learning processing unit 32 may learn a standard construction process based on teacher data. A standard construction process is, for example, a chronological order of typical processes at a construction site, and indicates the general sequence of the processes.

[0070] In this case, the teacher data includes, in addition to the learning shooting data, data indicating the process that was completed when each learning shooting data was captured. Then, the learning processing unit 32 performs learning using the label indicating the change in the process between the first learning shooting data and the second learning shooting data as correct answer data.

[0071] For example, when the first learning shooting data is learning shooting data 42 and the second learning shooting data is learning shooting data 45, the learning processing unit 32 learns by using a label indicating a change from a plasterboard / LGS installation process to a glass wool / plasterboard installation process as correct answer data.

[0072] In the process determination process configured as above, in step S103, the process determination unit 33 executes a process of obtaining a change in a process instead of obtaining a new element. Next, in step S104, the process determination unit 33 searches for a process change that is the same as the process change obtained in step S103 from the schedule management data stored in the schedule information storage unit 24. Then, in step S105, if the same process change exists in the schedule management data, the process determination unit 33 identifies the process change as a difference. On the other hand, if the same process change does not exist in the schedule management data, the process determination unit 33 identifies a rework as the difference. [Explanation of symbols]

[0073] H10...information processing device, H11...communication device, H12...input device, H13...display device, H14...storage device, H15...processor, 10...difference recognition system, 11...photography device, 12...administrator terminal, 20...difference recognition device, 21...control unit, 22...model memory unit, 23...teacher information memory unit, 24...schedule information memory unit, 25...site photography information memory unit, 26...progress information memory unit, 31...data management unit, 32...learning processing unit, 33...process judgment unit functioning as difference recognition unit, 41, 42...learning photography data, 43...gypsum board, 44...LGS, 45...learning photography data, 46...glass wool, 47...gypsum board, 51...first learning photography data, 52...second learning photography data, 53...machine learning model, 54...label, 56...schedule management data, 57...image management data, 58...progress management data, 60...extracted data.

Claims

1. A method for recognizing a difference in progress at a construction site, comprising: a first state acquisition step of acquiring first photographed data obtained by photographing a first state at a first time point; a second state acquisition step of acquiring second photographed data obtained by photographing a second state at a second time point that is later than the first time point; and a difference recognition step of recognizing a difference corresponding to a change from the first state to the second state based on the first photographing data and the second photographing data. Difference recognition method.

2. In the difference recognition step, a change in a recognition probability of the difference when the first photographed data and the second photographed data are interchanged is obtained, and rework is recognized as the difference according to the change in the recognition probability. The method for detecting differences according to claim 1.

3. In the difference recognition step, when it is determined that an element included in the first photographed data has been removed in the second photographed data, a rework is recognized as a difference. The method for detecting differences according to claim 1.

4. In the difference recognition step, a change in a process from the first photographed data to the second photographed data is compared with a chronological order of the processes prepared in advance to recognize rework. The method for detecting differences according to claim 1.

5. In the difference recognition step, the difference is recognized using a model that has learned the difference. The method for identifying differences according to any one of claims 1 to 4.

6. In the difference recognition step, a change portion between the first photographed data and the second photographed data is extracted, and then the difference is recognized using the model. The method for detecting differences according to claim 5.

7. In the learning of the model, a label indicating the difference is assigned. The method for detecting differences according to claim 5.

8. In the learning of the model, photographic data that distinguishes the area showing the difference is provided. The method for detecting differences according to claim 5.

9. In the difference recognition step, the difference is classified. The method for identifying differences according to any one of claims 1 to 4.

10. An intermediate time point is between the first time point and the second time point, and a difference between the first time point and the intermediate time point is recognized based on the first photographing data and the second photographing data. The method for identifying differences according to any one of claims 1 to 4.

11. A difference recognition system that recognizes differences in progress at a construction site, a data management unit that acquires first photographing data obtained by photographing a first state at a first time point and second photographing data obtained by photographing a second state at a second time point after the first time point; a difference recognition unit that recognizes a difference corresponding to a change from the first state to the second state based on the first shooting data and the second shooting data. Differential recognition system.

Images