Deterioration degree determination method, information processing apparatus, program, learning method, and apparatus

Abstract

To improve technique for determining a degree of deterioration of a steel sheet roof of a building.SOLUTION: A deterioration degree determination method to be executed by an information processing apparatus includes the steps of: acquiring a captured image of a steel sheet roof; inputting the captured image to a trained model that has been trained by machine learning using subdivided images of steel sheet roofs divided by a predetermined section and degrees of deterioration of the roofs, as training data, to estimate a degree of deterioration of the steel sheet roof related to the captured image; and outputting the degree of deterioration of the steel sheet roof related to the captured image. The degrees of deterioration in the training data are determined, on the basis of the subdivided images with luminance before adjustment, and adjusted subdivided images with luminance after adjustment.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a deterioration degree determination method for a steel roof of a building, an information processing device, a program, and a learning method and device. [Background technology]

[0002] Conventional roof deterioration diagnosis involves measuring and identifying the occurrence of rust and loss of gloss on steel roofs from satellite images (see, for example, Patent Document 1). There is also a method for identifying the difference in color value of steel roofs from aerial images (see, for example, Patent Document 2). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-21268 A [Patent Document 2] JP 2019-218681 A Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional methods had the problem that they could not accurately judge the deterioration level of steel roofs in the shadowed areas of aerial photography, satellite images, etc. In addition, there was a problem that they could not judge the deterioration level of steel roofs that were painted red or that had dust that looked like rust on them due to the surrounding environment. In other words, there was room for improvement in the technology for judging the deterioration level of steel roofs of buildings.

[0005] An object of the present disclosure, made in consideration of the above circumstances, is to improve the technology for determining the degree of deterioration of steel roofs of buildings. [Means for solving the problem]

[0006] (1) A deterioration level determination method according to an embodiment of the present disclosure is a deterioration level determination method executed by an information processing device, comprising: A step of acquiring a photographed image of a steel roof; A step of estimating the degree of deterioration of the steel roof related to the photographed image by inputting the photographed image into a trained model that has been machine-learned using images of the steel roof divided into predetermined sections and the deterioration degree of the roof as training data; A step of outputting a deterioration degree of the steel roof related to the photographed image; Including, The degree of degradation in the training data is determined based on an image before the brightness of the partition image is adjusted, and an image after the brightness of the partition image is adjusted.

[0007] (2) A deterioration level determination method according to an embodiment of the present disclosure is the deterioration level determination method according to (1), If the degree of deterioration estimated by the trained model is greater than or equal to a threshold, the estimated degree of deterioration is output, and if the degree of deterioration estimated by the trained model is less than the threshold, the degree of deterioration is output based on the repair history.

[0008] (3) A deterioration level determination method according to an embodiment of the present disclosure is the deterioration level determination method according to (1), The trained model includes a first trained model machine-learned using a blue roof section image and a deterioration level of the blue roof as training data, and a second trained model machine-learned using a red roof section image and a deterioration level of the red roof as training data, When the steel roof in the photographed image is blue, in the estimating step, the photographed image is input to a first trained model to estimate a deterioration degree of the steel roof in the photographed image, and in the outputting step, the deterioration degree is output; If the steel roof in the captured image is red or a dusty roof, in the estimating step, the captured image is input into a second trained model to estimate the degree of deterioration of the steel roof in the captured image, and if the estimated degree of deterioration is equal to or greater than a threshold, the estimated degree of deterioration is output in the outputting step, and if the degree of deterioration estimated by the second trained model is less than the threshold, the degree of deterioration is output based on the repair history in the outputting step.

[0009] (4) A deterioration level determination method according to an embodiment of the present disclosure is the deterioration level determination method according to (1), The trained model includes a first trained model trained by machine learning using a section image of a blue roof and a deterioration level of the blue roof as training data, a second trained model trained by machine learning using a section image of a red roof and a deterioration level of the red roof as training data, and a third trained model trained by machine learning using a section image of a dusty roof and a deterioration level of the dusty roof as training data, When the steel roof in the photographed image is blue, in the estimating step, the photographed image is input to a first trained model to estimate a deterioration degree of the steel roof in the photographed image, and in the outputting step, the deterioration degree is output; When the steel roof in the captured image is red, in the estimating step, the captured image is input into a second trained model to estimate a degree of deterioration of the steel roof in the captured image, and when the estimated degree of deterioration is equal to or greater than a threshold, the estimated degree of deterioration is output in the outputting step, and when the degree of deterioration estimated by the second trained model is less than the threshold, the degree of deterioration is output based on a repair history in the outputting step; If the steel roof in the captured image is a dusty roof, in the estimating step, the captured image is input into a third trained model to estimate the degree of deterioration of the steel roof in the captured image, and if the estimated degree of deterioration is equal to or greater than the threshold, the estimated degree of deterioration is output in the outputting step, and if the degree of deterioration estimated by the third trained model is less than the threshold, the degree of deterioration is output based on the repair history in the outputting step.

[0010] (5) A deterioration level determination method according to an embodiment of the present disclosure is the deterioration level determination method according to any one of (1) to (4), The training data includes attribute information of a steel roof corresponding to the section image, In the estimating step, the degree of deterioration of the steel roof related to the photographed image is further estimated by inputting the photographed image and attribute information of the steel roof corresponding to the photographed image into the trained model.

[0011] (6) A deterioration level determination method according to an embodiment of the present disclosure is the deterioration level determination method according to any one of (1) to (5), In the outputting step, the deterioration degree of the steel roof related to the photographed image is output in association with polygon data of buildings corresponding to a plurality of sections.

[0012] (7) An information processing device according to an embodiment of the present disclosure, An information processing device including a control unit, The control unit is Obtaining images of the steel roof, The photographed image is input to a trained model that has been machine-learned using images of a steel roof divided into predetermined sections and the deterioration degree of the steel roof as training data, thereby estimating the deterioration degree of the steel roof related to the photographed image; outputting a deterioration level of the steel roof related to the photographed image; The degree of degradation in the training data is determined based on an image before the brightness of the partition image is adjusted, and an image after the brightness of the partition image is adjusted.

[0013] (8) A program according to an embodiment of the present disclosure, On the computer, Acquiring a photographed image of a steel roof; Inputting the photographed image into a trained model that has been machine-learned using images of a steel roof divided into predetermined sections and the deterioration degree of the steel roof as training data, thereby estimating the deterioration degree of the steel roof related to the photographed image; outputting a deterioration degree of the steel roof related to the photographed image; Run the command, The degree of degradation in the training data is determined based on an image before the brightness of the partition image is adjusted, and an image after the brightness of the partition image is adjusted. Effect of the Invention

[0014] According to one embodiment of the present disclosure, the technology for determining the deterioration level of a steel roof of a building can be improved. [Brief description of the drawings]

[0015] [Figure 1] 1 is a block diagram showing a schematic configuration of an information processing device according to an embodiment of the present disclosure. [Diagram 2] 1 is a flowchart illustrating a method of brightness adjustment and creating a trained model according to an embodiment of the present disclosure. [Diagram 3] 1A and 1B are diagrams illustrating an example of an image before and after brightness adjustment according to an embodiment of the present disclosure. [Figure 4] 4 is a flowchart illustrating a deterioration level determination method according to an embodiment of the present disclosure. [Diagram 5] 13 illustrates an example of a display output of a deterioration level according to an embodiment of the present disclosure. [Figure 6] 10 is a flowchart showing a deterioration level determining method according to Modification 1. [Figure 7] 10 is a flowchart showing a deterioration level determining method according to Modification 2. [Figure 8] 13 is a flowchart showing a deterioration level determination method according to Modification 3. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Hereinafter, a deterioration level determination technique according to an embodiment of the present disclosure will be described with reference to the drawings.

[0017] In each drawing, the same or corresponding parts are denoted by the same reference numerals. In the description of this embodiment, the description of the same or corresponding parts will be omitted or simplified as appropriate.

[0018] First, an overview of the present embodiment will be described. The deterioration level determination technique according to the embodiment of the present disclosure is executed by an information processing device 10. The deterioration level determination technique according to the embodiment of the present disclosure can be used, for example, in determining the deterioration level of a steel roof of a factory such as a steelworks.

[0019] More specifically, the deterioration degree determination technique according to the embodiment of the present disclosure includes a step of acquiring a photographed image of a steel roof. The deterioration degree determination technique according to the embodiment of the present disclosure also includes a step of estimating the deterioration degree of the steel roof related to the photographed image by inputting the photographed image into at least one trained model that has been machine-trained using an image divided into a predetermined size of a predetermined section (hereinafter also referred to as a section image) and the deterioration degree of the steel roof as training data. The deterioration degree determination technique according to the embodiment of the present disclosure also includes a step of outputting the deterioration degree of the steel roof related to the photographed image. Here, in the deterioration degree determination technique according to the embodiment of the present disclosure, the deterioration degree in the training data is determined based on an image before the brightness of the section image is adjusted and an image after the brightness of the section image is adjusted.

[0020] In this way, according to the deterioration degree determination technology of the present embodiment, the deterioration degree in the teacher data is determined based on the image before the brightness of the section image is adjusted and the image after the brightness of the section image is adjusted. Therefore, the possibility of an accurate deterioration degree determination result being assigned to the teacher data is increased, and the accuracy of the trained model is improved, thereby improving the deterioration degree determination technology for steel roofs of buildings.

[0021] Here, any method and format for taking the image of the steel roof can be adopted. For example, the photograph of the steel roof may be taken by an aircraft equipped with a GNSS receiver and an inertial navigation system (hereinafter also referred to as GNSS / IMU). The aircraft may be, but is not limited to, an airplane, a helicopter, a drone, etc. The photograph is given the photographed position and attitude information acquired by the GNSS / IMU, and orthographic transformation is performed using altitude information acquired separately. In other words, the photograph is converted from central projection to orthographic projection. Next, a plurality of images are mosaicked (combined) to create an orthographic projection that fits a map divided into a size that is easy to handle in a geographic information system (hereinafter also referred to as GIS). Using the divided orthographic projection and a location information file (hereinafter also referred to as world file), a tile image is created and input to the GIS. The photographed image of the steel roof is divided into images (section images) of a predetermined size of a given section, and is managed as data for learning or evaluation. Data for each section is created based on the unit by which the roof manager manages the degree of deterioration, for example, the area to be repaired in one repair.

[0022] In this case, the imaging plan is created taking into consideration the overlap rate and the side lap rate so that an orthographic projection image can be easily created. For example, the imaging conditions may be an overlap rate of 80% and a side lap rate of 80%.

[0023] (Configuration of information processing device) Next, a detailed description will be given of each component of the information processing device 10. The information processing device 10 is any device used by a user. For example, a personal computer, a server computer, a general-purpose electronic device, or a dedicated electronic device can be adopted as the information processing device 10.

[0024] As shown in FIG. 1, the information processing device 10 includes a control unit 11, a storage unit 12, a communication unit 13, an input unit 14, and a display unit 15.

[0025] The control unit 11 includes at least one processor, at least one dedicated circuit, or a combination of these. The processor is a general-purpose processor such as a central processing unit (CPU) or a graphics processing unit (GPU), or a dedicated processor specialized for a specific process. The dedicated circuit is, for example, a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC). The control unit 11 executes processes related to the operation of the information processing device 10 while controlling each part of the information processing device 10.

[0026] The storage unit 12 includes at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The semiconductor memory is, for example, a random access memory (RAM) or a read only memory (ROM). The RAM is, for example, a static random access memory (SRAM) or a dynamic random access memory (DRAM). The ROM is, for example, an electrically erasable programmable read only memory (EEPROM). The storage unit 12 functions as, for example, a main storage device, an auxiliary storage device, or a cache memory. The storage unit 12 stores data used in the operation of the information processing device 10 and data obtained by the operation of the information processing device 10. For example, the storage unit 12 stores a trained model, data related to GIS, and the like.

[0027] The communication unit 13 includes at least one external communication interface. The communication interface may be either a wired communication interface or a wireless communication interface. In the case of wired communication, the communication interface is, for example, a LAN (Local Area Network) interface or a USB (Universal Serial Bus). In the case of wireless communication, the communication interface is, for example, an interface compatible with mobile communication standards such as LTE (Long Term Evolution), 4G (4th generation), or 5G (5th generation), or an interface compatible with short-range wireless communication such as Bluetooth (registered trademark). The communication unit 13 receives data used in the operation of the information processing device 10, and transmits data obtained by the operation of the information processing device 10.

[0028] The input unit 14 includes at least one input interface. The input interface is, for example, a physical key, a capacitive key, a pointing device, or a touch screen integrally provided with a display. The input interface may also be, for example, a sound sensor that accepts voice input, or a camera that accepts gesture input. The input unit 14 accepts an operation to input data used for the operation of the information processing device 10. The input unit 14 may be connected to the information processing device 10 as an external input device instead of being provided in the information processing device 10. As a connection method, any method such as USB (Universal Serial Bus), HDMI (registered trademark) (High-Definition Multimedia Interface), or Bluetooth (registered trademark) can be used.

[0029] The display unit 15 includes at least one output interface. The output interface is, for example, a display that outputs information as a video. The display is, for example, a liquid crystal display (LCD) or an organic EL (electro luminescence) display. The display unit 15 displays and outputs data obtained by the operation of the information processing device 10. The display unit 15 may be connected to the information processing device 10 as an external output device instead of being provided in the information processing device 10. As a connection method, any method such as USB, HDMI (registered trademark), or Bluetooth (registered trademark) can be used.

[0030] The functions of the information processing device 10 are realized by executing a program according to this embodiment on a processor equivalent to the information processing device 10. That is, the functions of the information processing device 10 are realized by software. The program causes a computer to execute the operations of the information processing device 10, thereby causing the computer to function as the information processing device 10. That is, the computer functions as the information processing device 10 by executing the operations of the information processing device 10 in accordance with the program.

[0031] In this embodiment, the program can be recorded on a computer-readable recording medium. The computer-readable recording medium includes a non-transitory computer-readable medium, such as a magnetic recording device, an optical disc, a magneto-optical recording medium, or a semiconductor memory. The program is distributed, for example, by selling, transferring, or lending a portable recording medium such as a DVD (digital versatile disc) or a CD-ROM (compact disc read only memory) on which the program is recorded. The program may also be distributed by storing the program in the storage of an external server and transmitting the program from the external server to another computer. The program may also be provided as a program product.

[0032] A part or all of the functions of the information processing device 10 may be realized by a dedicated circuit corresponding to the control unit 11. In other words, a part or all of the functions of the information processing device 10 may be realized by hardware.

[0033] In the deterioration level determination technique according to the embodiment of the present disclosure, the deterioration level in the training data is determined based on an image before the brightness of the section image is adjusted and an image after the brightness of the section image is adjusted. A method of brightness adjustment and learning model creation according to the embodiment of the present disclosure is shown with reference to the flowchart in FIG.

[0034] Step S10: The control unit 11 of the information processing device 10 acquires a photographed image of the steel roof. Various methods can be adopted to acquire the photographed image of the steel roof. For example, the control unit 11 may acquire the photographed image taken by an aircraft by the communication unit 13 via the aircraft or a network via the cloud or the like.

[0035] Step S20: The control unit 11 executes a process for adjusting the brightness of the captured image (hereinafter, also referred to as a brightness adjustment process). An image of a steel roof may be shaded due to the change in the position of the sun during the day and the influence of adjacent buildings. Therefore, even within the same roof, the degree of influence of the shadow may differ depending on the section. Also, the degree of influence of the shadow may differ depending on the day or time of the image capture. In this way, there are cases where the brightness and contrast differ greatly for each section image. For this reason, in this embodiment, the control unit 11 performs a brightness adjustment process on the captured image. Specifically, for example, the brightness adjustment process includes a process for acquiring brightness distribution information, a process for determining a correction coefficient, and a tone mapping process. Note that the brightness adjustment process may be performed by automatically correcting the brightness level, brightening only images with a low brightness level, or the like.

[0036] In the process of acquiring the luminance distribution information, the control unit 11 measures the distribution of luminance of each section image. In this embodiment, in order to grasp the difference in luminance of each section image of the steel roof, the luminance distribution information of each section is acquired. This makes it possible to accurately correct images with large differences in luminance.

[0037] In the process of determining the correction coefficient, the control unit 11 determines the correction coefficient in the tone mapping process. The tone mapping process is performed to make the image easier to see by confining the brightness of the high dynamic range image to the range of the low dynamic range image. In this embodiment, the correction coefficient is determined based on the luminance distribution information of each section. This correction coefficient makes it possible to create a tone map image in which the luminance difference between the images of each section is minimized.

[0038] In tone mapping, the control unit 11 converts a high dynamic range image into a low dynamic range image. This adjusts the brightness of the image to make it easier to see. In this embodiment, a correction coefficient is determined and tone mapping is performed to create a tone map image that minimizes the luminance difference between each section image. The tone map image obtained in this manner (hereinafter also referred to as the adjusted image) has little difference in brightness and contrast, and can improve the accuracy of visual assessment of the deterioration level of a building.

[0039] The brightness adjustment process described above makes it possible to give the correct degradation level to roof images that include shaded areas. Specifically, it becomes possible to correctly judge dark areas that were previously difficult to judge by eye, and learning can be performed using the correct answer label.

[0040] 3 shows an image 20 before the brightness is adjusted by the above method, and an image 30 after the brightness is adjusted. As shown in images 20 and 30, image 30, in which the brightness has been adjusted, makes it easier to visually determine the degree of deterioration of each section image in terms of brightness, contrast, etc.

[0041] Step S30: The control unit 11 executes annotation processing based on the user's input operation to the input unit 14. The annotation processing is a process of selecting each section image, determining a level according to the degree of deterioration, and labeling it. The degree of deterioration is classified into five levels (R1-R5), for example. The level of deterioration is the least deteriorated level R1, and the level of deterioration is the most deteriorated level R5. The control unit 11 assigns a label of the deterioration level of one of the five levels R1-R5 to each section image based on the user's input operation to the input unit 14. Note that the degree of deterioration is not limited to five levels, and may be less than five levels or six levels or more.

[0042] During annotation processing, an image before the brightness of the section image is adjusted and an image after the brightness of the section image is adjusted are used. That is, when a user visually determines the degree of deterioration of a steel roof of a certain section, the user uses the image before the brightness of the section image is adjusted or the image after the brightness of the section image is adjusted. In other words, annotation processing is performed using a section image that has been made to have a substantially uniform brightness contrast by the brightness adjustment processing.

[0043] Step S40: The control unit 11 generates at least one trained model. The control unit 11 generates at least one trained model using the section images of the steel roof divided into predetermined sections and the deterioration level of the roof as teacher data. Specifically, the trained model is trained using the section images of the steel roof as input and the labels corresponding to the section images, that is, the deterioration level, as output. The learning model algorithm is, for example, CNN (convolutional neural network), but may be any other algorithm capable of image classification. For example, the learning model algorithm may be Resnet, FCN, etc.

[0044] Next, a degradation level determination method according to an embodiment of the present disclosure will be described with reference to the flowchart of FIG.

[0045] Step S110: The control unit 11 of the information processing device 10 acquires a photographed image of the steel roof.

[0046] Step S120: The control unit 11 estimates the degree of deterioration of the steel roof related to the photographed image by inputting the photographed image into the trained model.

[0047] Step S130: The control unit 11 outputs the deterioration degree estimated by the trained model. Any method can be adopted for outputting the deterioration degree. For example, the control unit 11 may display and output the deterioration degree of the steel roof related to the photographed image in association with polygon data of the building corresponding to a plurality of sections. The polygon data of the building may be generated by any method based on map data, GIS data, etc. FIG. 5 shows an example of the display output of the deterioration degree. In FIG. 5, the deterioration degree of the steel roof is represented by a pattern corresponding to each section image. The display mode of the deterioration degree is not limited to this, and may be represented by a color according to the deterioration degree. In this way, by displaying and outputting the deterioration degree in association with the polygon data of the building, it is possible to intuitively show which position of the steel roof is deteriorated. Note that each section image in FIG. 5 may be displayed selectable, and section 50 indicates that it is in a selected state. For example, attribute information of a section image selected from among the section images may be displayed.

[0048] In this way, according to the deterioration degree determination technology of the present embodiment, the deterioration degree in the teacher data is determined based on the image before the brightness of the section image is adjusted and the image after the brightness of the section image is adjusted. Therefore, the possibility of an accurate determination result being assigned to the teacher data is increased, and the accuracy of the trained model is improved, thereby improving the deterioration degree determination technology for steel roofs of buildings.

[0049] The training data may include attribute information of the steel roof corresponding to the section image. The attribute information may include, for example, the type of roof, material, construction year, repair history, etc. The type of roof includes a blue steel roof (hereinafter also referred to as a blue roof), a red roof (hereinafter also referred to as a red roof), or a roof with dust on it (hereinafter also referred to as a dusty roof). The repair history includes information related to painting, covering, re-roofing, etc. In this case, in the estimation step of the above step S120, the deterioration degree of the steel roof related to the photographed image may be estimated by inputting the photographed image and attribute information of the steel roof corresponding to the photographed image into the trained model.

[0050] In this embodiment, the case where the brightness adjustment process is performed for each section image has been described, but this is not limited to the above. The brightness adjustment process may be performed for the entire roof image. For example, if the degree of influence of the shadow varies depending on the section, the brightness adjustment process may be performed for each section image. On the other hand, if the degree of influence of the shadow does not vary depending on the section, overall processing may be used. By performing the brightness adjustment process for each section, correct evaluation is possible even when the influence of the shadow is partially present on the entire roof.

[0051] Although the present disclosure has been described based on the drawings and examples, it should be noted that those skilled in the art can easily make various modifications and corrections based on the present disclosure. Therefore, it should be noted that these modifications and corrections are included in the scope of the present disclosure. For example, the functions included in each means or step can be rearranged so as not to be logically inconsistent, and multiple means or steps can be combined into one or divided. In addition, multiple means or steps may be omitted.

[0052] (Variation 1) For example, the deterioration level estimated by the trained model may be changed depending on the type of steel roof and the estimation result of the trained model. FIG. 6 shows a deterioration level determination method according to the first modification. In the first modification, the deterioration level is changed and output depending on the type of steel roof and the estimation result of the trained model. In the first modification, the deterioration level is determined and output based on the repair history of the steel roof (year of reroofing, year of painting) and the most recent year of construction. The repair history may be, for example, the number of years since the most recent repair history. The type of steel roof in the captured image and the repair history may be managed by, for example, a database. For example, the database may be stored in the memory unit 12 of the information processing device 10.

[0053] Step S210: The control unit 11 of the information processing device 10 acquires a photographed image of the steel roof.

[0054] Step S220: The control unit 11 estimates the deterioration degree of the steel roof related to the photographed image by inputting the photographed image to the trained model.

[0055] Step S221: The control unit 11 determines the type of the steel roof related to the photographed image. Specifically, the control unit 11 determines whether the steel roof related to the photographed image is a blue roof, a red roof, or a dusty roof. Any method can be adopted for determining the type of roof. For example, the control unit 11 may separately obtain attribute information related to the steel roof related to the photographed image and determine the type of the steel roof based on the attribute information. Alternatively, the control unit 11 may determine the type of the steel roof based on the selection of the type of the steel roof to be determined from the user. If the steel roof related to the selected photographed image is a blue roof, the process proceeds to step S230. On the other hand, if the steel roof related to the photographed image is a red roof or a dusty roof, the process proceeds to step S231.

[0056] Step S230: If the steel roof related to the captured image is a blue roof, the control unit 11 outputs the deterioration degree estimated by the trained model.

[0057] Step S231: If the steel roof related to the captured image is a red roof or a dusty roof, the control unit 11 determines whether the deterioration degree estimated by the trained model is less than a threshold value. For example, the threshold value may be R4. If the deterioration degree estimated by the trained model is less than the threshold value, the process proceeds to step S232. On the other hand, if the deterioration degree estimated by the trained model is less than the threshold value, the process proceeds to step S233.

[0058] Step S232: If the deterioration degree estimated by the trained model is less than the threshold, the control unit 11 determines and outputs the deterioration degree based on the repair history. For example, the control unit 11 may determine and output the deterioration degree based on a table showing the correspondence between the repair history and the deterioration degree (hereinafter, also referred to as a correspondence table). An example of the correspondence table is shown in Table 1.

[0059] [Table 1]

[0060] For example, the correspondence table may be stored in the storage unit 12. The control unit 11 determines the deterioration degree by referring to the correspondence table. Years A, B, C, and D may be determined appropriately.

[0061] Step S233: If the deterioration level estimated by the trained model is equal to or greater than the threshold, the control unit 11 outputs the deterioration level estimated by the trained model.

[0062] In this way, when the steel roof in the captured image is a red roof or a dusty roof and the deterioration level estimated by the trained model is less than the threshold, the deterioration level may be determined and output based on the repair history, rather than the estimation result by the trained model. In this way, it is possible to prevent the overlooking of steel roofs with a certain level of danger or higher for red roofs or dusty roofs, which are generally difficult to judge using the trained model.

[0063] (Variation 2) For example, the trained model may be generated according to the type of steel roof. The trained model used to estimate the deterioration level may be changed according to the type of steel roof. Specifically, the trained model may include a first trained model trained by machine learning using a blue roof section image and the deterioration level of the blue roof as training data, and a second trained model trained by machine learning using a red roof section image and the deterioration level of the red roof as training data. FIG. 7 shows a deterioration level determination method according to the second modification.

[0064] Step S310: The control unit 11 of the information processing device 10 acquires a photographed image of the steel roof.

[0065] Step S311: The control unit 11 determines the type of the steel roof related to the captured image. Specifically, the control unit 11 determines whether the steel roof related to the captured image is a blue roof, a red roof, or a dusty roof. Any method can be adopted to determine the type of roof, as described above. If the steel roof related to the captured image is a blue roof, the process proceeds to step S320. On the other hand, if the steel roof related to the captured image is a red roof or a dusty roof, the process proceeds to step S340.

[0066] Step S320: The control unit 11 estimates the deterioration degree of the steel roof related to the photographed image by inputting the photographed image to the first trained model.

[0067] Step S330: The control unit 11 outputs the deterioration level estimated by the first learned model.

[0068] Step S340: The control unit 11 estimates the deterioration degree of the steel roof related to the photographed image by inputting the photographed image to the second trained model.

[0069] Step S341: The control unit 11 determines whether the deterioration level estimated by the second trained model is less than a threshold. For example, the threshold may be R4. If the deterioration level estimated by the second trained model is less than the threshold, the process proceeds to step S342. On the other hand, if the deterioration level estimated by the second trained model is less than the threshold, the process proceeds to step S343.

[0070] Step S342: If the deterioration degree estimated by the second trained model is less than the threshold, the control unit 11 determines and outputs the deterioration degree based on the repair history. For example, the control unit 11 may determine and output the deterioration degree based on the correspondence table shown in Table 1.

[0071] Step S343: If the deterioration level estimated by the second trained model is equal to or greater than the threshold, the control unit 11 outputs the deterioration level estimated by the second trained model.

[0072] In this way, when the steel roof in the captured image is a red roof or a dusty roof and the deterioration level estimated by the second trained model is less than the threshold, the deterioration level may be determined and output based on the repair history, rather than the estimation result by the second trained model. In this way, it is possible to prevent the overlooking of steel roofs with a certain level of danger or higher for red roofs or dusty roofs, which are generally difficult to judge using trained models.

[0073] (Variation 3) Furthermore, a trained model may be generated for the case where the steel roof is a dusty roof. Specifically, the trained model may include a first trained model trained by machine learning using a section image of a blue roof and a deterioration level of the blue roof as training data, a second trained model trained by machine learning using a section image of a red roof and a deterioration level of the red roof as training data, and a third trained model trained by machine learning using a section image of a dusty roof and a deterioration level of the dusty roof as training data. FIG. 8 shows a deterioration level determination method according to the third modification.

[0074] Step S410: The control unit 11 of the information processing device 10 acquires a photographed image of the steel roof.

[0075] Step S411: The control unit 11 determines the type of the steel roof related to the captured image. Specifically, the control unit 11 determines whether the steel roof related to the captured image is a blue roof, a red roof, or a dusty roof. Any method can be adopted to determine the type of roof, as described above. If the steel roof related to the captured image is a blue roof, the process proceeds to step S420. If the steel roof related to the captured image is a red roof, the process proceeds to step S440. If the steel roof related to the captured image is a dusty roof, the process proceeds to step S450.

[0076] Step S420: The control unit 11 estimates the deterioration degree of the steel roof related to the photographed image by inputting the photographed image to the first trained model.

[0077] Step S430: The control unit 11 outputs the deterioration level estimated by the first learned model.

[0078] Step S440: The control unit 11 estimates the deterioration degree of the steel roof related to the photographed image by inputting the photographed image to the second trained model.

[0079] Step S441: The control unit 11 determines whether the deterioration level estimated by the second trained model is less than a threshold. For example, the threshold may be R4. If the deterioration level estimated by the second trained model is less than the threshold, the process proceeds to step S442. On the other hand, if the deterioration level estimated by the second trained model is less than the threshold, the process proceeds to step S443.

[0080] Step S442: If the deterioration degree estimated by the second trained model is less than the threshold, the control unit 11 determines and outputs the deterioration degree based on the repair history. For example, the control unit 11 may determine and output the deterioration degree based on the correspondence table shown in Table 1.

[0081] Step S443: If the deterioration level estimated by the second trained model is equal to or greater than the threshold, the control unit 11 outputs the deterioration level estimated by the second trained model.

[0082] Step S450: The control unit 11 estimates the deterioration degree of the steel roof related to the photographed image by inputting the photographed image to the third trained model.

[0083] Step S451: The control unit 11 determines whether the deterioration level estimated by the third trained model is less than a threshold. For example, the threshold may be R4. If the deterioration level estimated by the third trained model is less than the threshold, the process proceeds to step S452. On the other hand, if the deterioration level estimated by the third trained model is less than the threshold, the process proceeds to step S453.

[0084] Step S452: If the deterioration degree estimated by the third trained model is less than the threshold, the control unit 11 determines and outputs the deterioration degree based on the repair history. For example, the control unit 11 may determine and output the deterioration degree based on the correspondence table shown in Table 1.

[0085] Step S453: If the deterioration level estimated by the third trained model is equal to or greater than the threshold, the control unit 11 outputs the deterioration level estimated by the third trained model.

[0086] In this way, when the steel roof related to the captured image is a red roof or a dusty roof and the deterioration level estimated by the second trained model or the third trained model is less than the threshold, the deterioration level may be determined and output based on the repair history, rather than the estimation result by the second trained model or the third trained model. In this way, it is possible to prevent the overlooking of steel roofs with a certain level of danger or higher for red roofs or dusty roofs, which are generally difficult to judge using trained models. [Explanation of symbols]

[0087] 10. Information processing device 11 Control section 12 Storage section 13. Communications Department 14 Input section 15 Display section 20 images 30 images 50 Selected Pane

Claims

1. A deterioration degree determination method executed by an information processing device, comprising: A step of acquiring a photographed image of a steel roof; A step of estimating the degree of deterioration of the steel roof related to the photographed image by inputting the photographed image into a trained model that has been machine-learned using images of the steel roof divided into predetermined sections and the deterioration degree of the roof as training data; A step of outputting a deterioration degree of the steel roof related to the photographed image; Including, A deterioration degree determination method in which the deterioration degree in the training data is determined based on an image before the brightness of the partition image is adjusted and an adjusted image after the brightness of the partition image is adjusted.

2. 2. The deterioration degree determination method according to claim 1, A deterioration degree determination method, comprising: outputting an estimated deterioration degree when the deterioration degree estimated by the trained model is equal to or greater than a threshold; and outputting a deterioration degree based on a repair history when the deterioration degree estimated by the trained model is less than the threshold.

3. 2. The deterioration degree determination method according to claim 1, The trained model includes a first trained model machine-learned using a section image of a blue roof and a deterioration level of the blue roof as training data, and a second trained model machine-learned using a section image of a red roof and a deterioration level of the red roof as training data, When the steel roof in the photographed image is blue, in the estimating step, the photographed image is input to a first trained model to estimate a deterioration degree of the steel roof in the photographed image, and in the outputting step, the deterioration degree is output; A deterioration degree determination method, in which, if the steel roof in the captured image is red or a dusty roof, in the estimating step, the captured image is input into a second trained model to estimate the degree of deterioration of the steel roof in the captured image, and if the estimated degree of deterioration is equal to or greater than a threshold, the estimated degree of deterioration is output in the outputting step, and if the degree of deterioration estimated by the second trained model is less than the threshold, the degree of deterioration is output based on repair history in the outputting step.

4. 2. The deterioration degree determination method according to claim 1, The trained model includes a first trained model trained by machine learning using a section image of a blue roof and a deterioration level of the blue roof as training data, a second trained model trained by machine learning using a section image of a red roof and a deterioration level of the red roof as training data, and a third trained model trained by machine learning using a section image of a dusty roof and a deterioration level of the dusty roof as training data, When the steel roof in the photographed image is blue, in the estimating step, the photographed image is input to a first trained model to estimate a deterioration degree of the steel roof in the photographed image, and in the outputting step, the deterioration degree is output; When the steel roof related to the photographed image is red, in the estimating step, the photographed image is input into a second trained model to estimate a degree of deterioration of the steel roof related to the photographed image, and when the estimated degree of deterioration is equal to or greater than a threshold, the estimated degree of deterioration is output in the outputting step, and when the degree of deterioration estimated by the second trained model is less than the threshold, the degree of deterioration is output based on a repair history in the outputting step; A deterioration degree determination method, in which, when the steel roof in the captured image is a dusty roof, in the estimating step, the photographed image is input into a third trained model to estimate the degree of deterioration of the steel roof in the captured image, and when the estimated degree of deterioration is equal to or greater than the threshold, the estimated degree of deterioration is output in the outputting step, and when the degree of deterioration estimated by the third trained model is less than the threshold, the degree of deterioration is output based on the repair history in the outputting step.

5. 2. The deterioration degree determination method according to claim 1, The training data includes attribute information of a steel roof corresponding to the section image, A deterioration degree determination method, in which in the estimating step, the deterioration degree of the steel roof related to the photographed image is estimated by inputting attribute information of the photographed image and the steel roof corresponding to the photographed image into the trained model.

6. 2. The deterioration degree determination method according to claim 1, A deterioration degree determination method, wherein in the outputting step, the deterioration degree of the steel roof related to the photographed image is output in association with polygon data of a building corresponding to a plurality of sections.

7. An information processing device including a control unit, The control unit is Obtaining images of the steel roof, The photographed image is input to a trained model that has been machine-learned using images of a steel roof divided into predetermined sections and the deterioration degree of the steel roof as training data, thereby estimating the deterioration degree of the steel roof related to the photographed image; outputting a deterioration level of the steel roof related to the photographed image; An information processing device, wherein the degree of degradation in the training data is determined based on an image before the brightness of the partition image is adjusted and an image after the brightness of the partition image is adjusted.

8. On the computer, Acquiring a photographed image of a steel roof; Inputting the photographed image into a trained model that has been machine-learned using images of a steel roof divided into predetermined sections and the deterioration degree of the steel roof as training data, thereby estimating the deterioration degree of the steel roof related to the photographed image; outputting a deterioration degree of the steel roof related to the photographed image; Run the command, A program in which the degree of degradation in the training data is determined based on an image before the brightness of the section image is adjusted and an image after the brightness of the section image is adjusted.

9. A learning method for a learning model executed by an information processing device, comprising: A step of acquiring a photographed image of a steel roof; A step of training a learning model that estimates the deterioration level of a steel roof related to a photographed image by inputting a photographed image using a section image of the steel roof divided into predetermined sections and the deterioration level of the roof as teacher data, and generating a learned model; A learning method in which the degree of degradation in the training data is determined based on an image before the brightness of the region image is adjusted and an image after the brightness of the region image is adjusted.

10. An apparatus including a control unit, The control unit is Obtaining images of the steel roof, A learning model is trained to estimate the degree of deterioration of a steel roof related to a photographed image by inputting the photographed image using a section image of the steel roof divided into predetermined sections and the deterioration degree of the roof as teacher data, thereby generating a trained model; The degree of degradation in the training data is determined based on an image before the brightness of the region image is adjusted and an image after the brightness of the region image is adjusted.

Images