Pothole prediction system, pothole prediction method and program

Abstract

A pothole prediction system according to the present disclosure comprises: an acquisition means that acquires a road surface image in which a road surface is captured; an analysis means that analyzes the state of cracks on the road surface from the road surface image; a calculation means that uses a prediction model in which data showing a relationship between the state of cracks and the occurrence of a pothole has been trained as teaching data, and calculates the probability of pothole occurrence predicted from the analysis results obtained by the analysis means; and an output means that outputs information indicating the calculated probability of occurrence of a pothole.

Description

Pothole prediction system, pothole prediction method, and recording medium

[0001] The present disclosure relates to a pothole prediction system and the like.

[0002] Paved roads are subject to deterioration such as cracks caused by vehicle traffic and rainfall. In order to understand the state of road deterioration and plan road repairs, analysis of the state of road deterioration is carried out.

[0003] Patent Document 1 discloses a method for quantitatively analyzing the risk of potholes occurring in drainage pavements. In Patent Document 1, the risk of potholes occurring is predicted using the amount of local settlement calculated from road surface property data, the G / R value, which is the ratio of green to red obtained from image data, and the value of the average profile depth calculated from the road surface property data.

[0004] Patent Document 2 discloses a crack analysis device that detects cracks of a specific shape from an image of a road surface and displays the crack detection results. Patent Document 3 discloses a deterioration prediction system that predicts the level of road deterioration at a future point in time and superimposes the predicted deterioration level on a map in a display format according to the deterioration level for each prediction point in time.

[0005] JP 2018-028486 A JP 2018-040666 A International Publication No. 2021 / 192790

[0006] According to Patent Document 1, the amount of local subsidence and the average profile depth are used to predict the risk of potholes occurring. Therefore, it is not possible to predict the risk of potholes occurring without using an optical section photography device that emits a slit laser.

[0007] An object of the present disclosure is to provide a pothole prediction system and the like that can determine the probability of pothole occurrence with a simple configuration.

[0008] The pothole prediction system according to the present disclosure comprises an acquisition means for acquiring a road surface image obtained by photographing the road surface, an analysis means for analyzing the state of cracks on the road surface from the road surface image, a calculation means for calculating the probability of pothole occurrence predicted from the analysis results by the analysis means using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data, and an output means for outputting information indicating the calculated probability of pothole occurrence.

[0009] The pothole prediction method according to the present disclosure acquires a road surface image by photographing the road surface, analyzes the state of cracks on the road surface from the road surface image, calculates the predicted probability of pothole occurrence from the results of the analysis using a prediction model trained using data indicating the relationship between the state of the cracks and the occurrence of potholes as training data, and outputs information indicating the calculated probability of pothole occurrence.

[0010] A program according to the present disclosure causes a computer to execute a process of acquiring a road surface image obtained by photographing a road surface, analyzing the state of cracks on the road surface from the road surface image, calculating a predicted probability of pothole occurrence from the results of the analysis using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data, and outputting information indicating the calculated probability of pothole occurrence. The program may be stored in a computer-readable non-transitory recording medium.

[0011] According to the present disclosure, it is possible to determine the probability of pothole occurrence with a simple configuration.

[0012] FIG. 1 is a diagram illustrating an overview of devices connected to the pothole prediction system. FIG. 1 is a block diagram illustrating an example configuration of the pothole prediction system according to the first embodiment. FIG. 2 is a diagram illustrating an example of crack detection results. FIG. 3 is a diagram illustrating an example of training data. FIG. 4 is a diagram illustrating an example of a prediction model. FIG. 5 is a flowchart illustrating an example operation of the pothole prediction system according to the first embodiment. FIG. 6 is a diagram illustrating an example of a displayed screen. FIG. 7 is a flowchart illustrating an example operation of an output unit that displays a scale. FIG. 7 is a block diagram illustrating an example configuration of a pothole prediction system according to a second embodiment. FIG. 8 is a flowchart illustrating an example operation of the pothole prediction system according to the second embodiment. FIG. 9 is a diagram illustrating an example of a displayed screen. FIG. 10 is a block diagram illustrating an example hardware configuration of a computer.

[0013] Cracks in road surfaces begin as linear cracks that increase and spread, eventually leading to the pavement peeling away and turning into potholes, which collapse into cave-ins. To prevent accidents caused by potholes, road managers repair the road surface. Having information to base road repair plans on can help create efficient plans.

[0014] The pothole prediction system according to the present disclosure is a system that predicts the probability of pothole occurrence using the state of cracks on the road surface analyzed from road surface images and a prediction model that has learned the relationship between the state of cracks and the occurrence of potholes.

[0015] It should be noted that the road surfaces targeted by the pothole prediction system according to the present disclosure are not limited to ordinary roads on which vehicles and people pass, but also include vehicle test courses, airport runways and taxiways, etc. In other words, the pothole prediction system can target a wide range of paved road surfaces.

[0016] 1 is a diagram showing an overview of devices communicably connected to a pothole prediction system 100 via a communication network 30, either wired or wirelessly. The pothole prediction system 100 is connected to, for example, a camera 10, a display 20, an input device 21, and a database 40.

[0017] The camera 10 captures road surface images including the road surface. The road surface images captured by the camera 10 are stored in a database 40. The camera 10 is realized, for example, by a drive recorder mounted on a vehicle. However, the type of camera is not limited to this, and various types of cameras may be used. For example, the road surface images may be captured by a camera mounted on another mobile object such as a bicycle or a drone, a camera carried by a person, or a fixed camera installed on the road. The road surface images may be still images or moving images continuously captured by the camera 10 while the mobile object is moving. The road surface images may be captured by a person or automatically.

[0018] The display 20 displays information to the user. The display 20 includes, for example, a display or a tablet. The display 20 displays various information in accordance with the output from the pothole prediction system 100. The information to be displayed will be described later.

[0019] The input device 21 receives operations from a user. The input device 21 includes, for example, a mouse, a keyboard, etc. If the display 20 is a touch panel display, the display 20 may be configured as the input device 21.

[0020] The database 40 stores maps. The database 40 may further store road surface images captured by the camera 10. The database 40 that stores maps and the database 40 that stores road surface images may be provided separately.

[0021] 2 is a block diagram showing an example of the configuration of a pothole prediction system 100 according to the first embodiment. The pothole prediction system 100 according to the first embodiment includes an acquisition unit 110, an analysis unit 120, a calculation unit 130, and an output unit 140. The calculation unit 130 includes a prediction model storage unit 131 and a calculation unit 132.

[0022] The acquisition unit 110 acquires road surface images obtained by photographing the road surface. For example, the acquisition unit 110 acquires the road surface images from the database 40. In another example, the acquisition unit 110 may acquire the road surface images from the camera 10 via the communication network 30. In this case, the pothole prediction system 100 is connected to the camera 10 so as to be able to communicate with it as necessary.

[0023] The acquisition unit 110 may acquire location information of the point where the road surface image was captured together with the road surface image. The location information includes, for example, latitude and longitude, location information based on a Global Navigation Satellite System (GNSS) or a Global Positioning System (GPS), or a location on a map.

[0024] Furthermore, the acquisition unit 110 may acquire the date and time when the road surface image was captured together with the road surface image.

[0025] The analysis unit 120 analyzes the state of cracks from the road surface image acquired by the acquisition unit 110. The analysis unit 120, for example, detects cracks and analyzes the state of the detected cracks.

[0026] For example, the analysis unit 120 detects cracks using a known image recognition technique on the road surface image. The analysis unit 120 may detect cracks using a trained model. The analysis unit 120 may determine whether or not each pixel in the road surface image is a deteriorated road surface.

[0027] FIG. 3 is a diagram showing an example of the detection result of cracks on a road from a road surface image captured of the road. The capture range of the road surface image is not limited to the example of FIG. 3 and may be narrow or wide, for example, in the vertical or horizontal direction. For example, the road surface image may include the sky, sidewalks on both sides of the road, and buildings. The analysis unit 120 may detect road surface deterioration included in a detection area F1 in the road surface image, for example. The detection area F1 is the area in which road surface deterioration is to be detected.

[0028] The analysis unit 120 may, for example, divide the road surface image into predetermined units. Then, the analysis unit 120 may detect and analyze cracks for each divided unit. The analysis unit 120 may divide the detection area F1 of the road surface image, where the detection of road surface deterioration is performed, into blocks of a predetermined size.

[0029] The crack state indicated by the analysis result by the analysis unit 120 is data indicating the progression state of cracks occurring on the road surface. The crack state includes, for example, the crack rate, crack length, crack width, crack area, crack shape, and the presence or absence of cracks.

[0030] The crack rate is expressed, for example, by 100 x (crack area / road surface area). The crack area can be calculated by any method. Note that the method for calculating the crack rate is not particularly limited, and other known calculation methods can be applied in addition to the above.

[0031] The width of a crack may be represented by the width of the widest crack in a predetermined range, or by the average width of the crack in a predetermined range.

[0032] The shape of the crack may include, for example, whether the detected crack is a straight crack or a tortoiseshell crack. The shape of the crack may be expressed by a numerical value according to the presence or absence of a crack of a predetermined shape. For example, the presence of tortoiseshell cracks in the road surface image may be expressed as 1, and the absence of tortoiseshell cracks may be expressed as 0.

[0033] The state of cracks analyzed by the analysis unit 120 may include the amount of hexagonal cracks. The amount of hexagonal cracks indicates the amount of intersecting cracks. The amount of hexagonal cracks may be expressed by the number of units that contain hexagonal cracks when a road surface image is divided into predetermined units. For example, the amount of hexagonal cracks is expressed by the number of hexagonal crack blocks, which is the number of blocks that contain cracks that make up hexagonal cracks when a road surface image is divided into blocks. The amount of hexagonal cracks may also be the area of ​​the blocks that contain hexagonal cracks. Alternatively, the amount of hexagonal cracks may be expressed by the area of ​​the cracks that make up the hexagonal cracks.

[0034] The calculation unit 130 calculates the predicted probability of pothole occurrence from the analysis results by the analysis unit 120 using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data. The probability of pothole occurrence indicates the likelihood that a pothole will occur within a predetermined period of time. The predetermined period can be set as appropriate, for example, one month, six months, one year, etc. The occurrence probability is expressed as a numerical value between 0 and 1. A higher calculated occurrence probability indicates a higher likelihood that a pothole will occur on the road surface analyzed by the analysis unit 120. The occurrence probability may also be expressed as a percentage between 0% and 100%.

[0035] The prediction model storage unit 131 included in the calculation unit 130 stores the trained prediction model. The calculation unit 130 includes an arithmetic unit 132 that inputs the analysis results from the analysis unit 120 into the trained prediction model and calculates the probability of pothole occurrence.

[0036] The calculation unit 130 calculates the probability of a pothole occurring using, for example, logistic regression. The learning phase of the prediction model will now be described. The prediction model in logistic regression can be expressed by assigning the value x obtained from the linear regression equation of Equation 1 to the sigmoid function of Equation 2.

[0037] (Equation 1) x = explanatory variable 1 × w1 + explanatory variable 2 × w2 + ... + explanatory variable z × w ｚ

[0038] (Equation 2) Equation 1 is a linear regression equation in which each explanatory variable is multiplied by a weight. y in Equation 2 is the objective variable. There is no particular limit to the number of explanatory variables. Prediction is possible with at least one explanatory variable. By providing an x ​​value to the sigmoid function in Equation 2, an output value y between 0 and 1 is obtained. Weight learning is then performed using labels of 0 or 1 assigned to the explanatory variables of the training data. For example, if no potholes have occurred, a label of 0 is assigned, and if a pothole has occurred, a label of 1 is assigned. Learning of the prediction model is completed when a weight is obtained that minimizes the error between the output value y and the label.

[0039] The more cracks in the road surface progress, the easier it is for rainwater and other water to penetrate into the road surface. This causes the pavement to deteriorate due to water, making it more likely for potholes to occur. Therefore, there is a relationship between the state of the cracks and the occurrence of potholes.

[0040] Figure 4 is a diagram showing an example of training data showing the relationship between crack conditions and the occurrence of potholes. For example, values ​​of the crack rate, crack width, and number of tortoiseshell crack blocks at multiple points may be used as explanatory variables. The training data in Figure 4 includes labels indicating whether or not a pothole has occurred at each point.

[0041] A prediction model when using the training data of Fig. 4 will be described with reference to Fig. 5. When using the training data of Fig. 4, Equation 1 can be expressed as Equation 3 below.

[0042] (Equation 3) x = crack rate × w1 + crack width × w2 + number of hexagonal crack blocks × w3 For example, the value of x is calculated by inputting the values ​​of the crack rate of 56.7, crack width of 5.2, and number of hexagonal crack blocks of 8 at point 1 into Equation 3. The calculated value of x is input to the sigmoid function of Equation 2, and an output value y of, for example, 0.7 is obtained. Here, since the label at point 1 is 1, the weights w1, w2, and w3 are adjusted so that the output value y approaches 1. Similarly, the weights w1, w2, and w3 are adjusted using the values ​​of the crack rate, crack width, and number of hexagonal crack blocks at points 2 and 3, etc. Thus, weights w1, w2, and w3 that can accurately predict the probability of pothole occurrence are learned from the crack rate, crack width, and number of hexagonal crack blocks observed at various points.

[0043] The learning of the prediction model described above may be performed in the calculation unit 130 or in another device (not shown).

[0044] The prediction model storage unit 131 stores the learned prediction model. In the phase of inference using the prediction model, the calculation unit 132 inputs the analysis results by the analysis unit 120 as explanatory variables to the prediction model stored in the prediction model storage unit 131. The calculation unit 132 outputs the calculation result of the probability of pothole occurrence for the input explanatory variables.

[0045] For example, when using the prediction model shown in Figure 5, the analysis unit 120 analyzes the state of cracks on the road surface from a road surface image and outputs the crack rate, crack width, and number of tortoiseshell crack blocks as the analysis results. The calculation unit 132 obtains the value of x in Equation 3 from the value of the analysis result obtained from the analysis unit 120. The calculation unit 132 obtains the predicted value y of the occurrence probability by applying the value of x to the sigmoid function of Equation 2.

[0046] In the above example, the crack rate, crack width, and number of tortoiseshell crack blocks were used as explanatory variables. However, the type of explanatory variable can be selected as appropriate. For example, explanatory variables including at least one of the crack rate, crack length, crack width, crack area, crack shape, amount of tortoiseshell cracks, and presence or absence of cracks may be used as explanatory variables. If the accuracy of the value predicted using one explanatory variable is insufficient, two or more explanatory variables can be used. Even when it is difficult to predict the probability of pothole occurrence from one explanatory variable, such as the crack rate, it is possible to predict the probability of pothole occurrence by combining multiple explanatory variables that indicate the state of the cracks.

[0047] The training data may include road information as an explanatory variable in addition to the state of cracks. The calculation unit 130 may use a prediction model to calculate the probability of pothole occurrence based on the analysis results of the analysis unit 120 and road information about the road surface. The road information is information indicating the characteristics of the road on which vehicles travel. The road information includes, for example, traffic volume, lane width, or number of lanes. Traffic volume represents, for example, the number of vehicles traveling on the road surface within a specified period. The traffic volume may also be the number of vehicles weighing a certain amount or more. The greater the traffic volume, the faster the deterioration rate of the road surface. The narrower the lane width, the more likely load is placed on the same position on the road surface, making it more susceptible to deterioration. Furthermore, the fewer the number of lanes, the more traffic volume is concentrated and the more likely deterioration is likely. Therefore, the higher the traffic volume, the narrower the lane width, or the fewer the number of lanes, the higher the predicted probability of pothole occurrence.

[0048] The above describes a case where the probability of a pothole occurring is calculated using logistic regression. However, the calculation unit 130 may calculate the probability of a pothole occurring using other prediction models that predict the probability of an event occurring. For example, the calculation unit 130 may use a light GBM (Gradient Boosting Machine).

[0049] The calculation unit 130 may further predict the size of a pothole that will occur. In this case, the prediction model storage unit 131 may store a trained model that predicts the size of a pothole based on the state of the crack. The calculation unit 132 predicts the size of a pothole based on the state of the crack analyzed by the analysis unit 120 and the trained model. The state of the crack that serves as the explanatory variable is, for example, the crack rate, the length of the crack, or the amount of tortoiseshell cracks. The size of the pothole that serves as the objective variable is, for example, the area, width, length, or depth of the pothole, or a combination of these.

[0050] The output unit 140 outputs information indicating the probability of pothole occurrence calculated by the calculation unit 130. The output unit 140 may be a display control unit that controls the display on the display 20. The output unit 140 may, for example, cause the display 20 to display a numerical value of the probability of pothole occurrence.

[0051] The output unit 140 may also display the estimated time of pothole occurrence according to the probability of pothole occurrence. The output unit 140 displays a period such as within one month, within three months, or within one year as the time of pothole occurrence. The correspondence between the probability of pothole occurrence and the estimated time of pothole occurrence may be determined in advance. For example, a location where the probability of pothole occurrence is calculated to be 80% is estimated to have a pothole occur within one month, and a location where the probability of pothole occurrence is calculated to be 60% to 70% is estimated to have a pothole occur within two to three months. In this way, the output unit 140 displays the time of pothole occurrence based on the predetermined correspondence.

[0052] The output unit 140 may display an icon indicating the calculated probability of pothole occurrence on a map showing the road surface where the image used to predict the probability of pothole occurrence was taken. For example, the output unit 140 acquires map data from the database 40. Furthermore, the output unit 140 acquires, for example, location information of the point where the road surface image was taken from the acquisition unit 110. Then, the output unit 140 displays an icon indicating the probability of pothole occurrence on the map as information indicating the probability of pothole occurrence.

[0053] The output unit 140 may display an icon at a point on the map where the calculated value of the probability of pothole occurrence is equal to or greater than a predetermined value. The threshold value for displaying the icon may be changeable by the user. For example, the user inputs, via the input device 21, a value that serves as a threshold value for determining whether or not to display the icon. In this case, the pothole prediction system 100 may further include a receiving unit (not shown) that receives the threshold value for the probability of pothole occurrence. The output unit 140 displays, on the map, an icon indicating the probability of pothole occurrence that is equal to or greater than the threshold value received by the receiving unit.

[0054] 6 is a flowchart showing an example of the operation of the pothole prediction system 100. The pothole prediction system 100 may start the operation of FIG.

[0055] The acquisition unit 110 acquires a road surface image obtained by capturing a road surface (step S11). The acquisition unit 110 provides the acquired image to the analysis unit 120.

[0056] The analysis unit 120 analyzes the state of cracks on the road surface from the road surface image acquired by the acquisition unit 110 (step S12). The analysis unit 120 provides the analyzed state of cracks to the calculation unit 130.

[0057] The calculation unit 130 calculates the probability of pothole occurrence predicted from the analysis results by the analysis unit 120 using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data (step S13). The calculation unit 130 provides the calculated probability of pothole occurrence to the output unit 140.

[0058] The output unit 140 outputs information indicating the probability of pothole occurrence calculated by the calculation unit 130 (step S14). For example, as shown in Fig. 12 (described later), the output unit 140 outputs the numerical value of the probability of pothole occurrence to the display 20. Alternatively, as shown in Fig. 8 (described later), the output unit 140 displays an icon at a point on the map where the probability of pothole occurrence is, for example, 30% or higher, as a process of outputting the information indicating the probability of pothole occurrence.

[0059] With the above, the pothole prediction system 100 ends the operation shown in FIG. 6 .

[0060] The manner in which the output unit 140 displays information indicating the probability of occurrence of a pothole on the display 20 will be described in more detail.

[0061] The output unit 140 may change the color of the icon on the map depending on the probability of pothole occurrence. For example, the icon may be displayed in blue for an occurrence probability of 0% to 39%, in yellow for an occurrence probability of 40% to 69%, and in red for an occurrence probability of 70% or higher. The types of colors and the stages of color change can be designed as appropriate.

[0062] 7 is a diagram illustrating the display mode of an icon according to the probability of pothole occurrence. For example, a map pin as shown in FIG. 7 can be used as an icon indicating a location where the probability of pothole occurrence is predicted. However, the shape of the icon is not limited to a map pin. For example, the icon may be displayed in a lighter color as the occurrence probability decreases, and in a darker color as the occurrence probability increases, corresponding to the colors of the color scale bar shown in FIG. 7.

[0063] Fig. 8 is a diagram showing an example of a screen displayed by the output unit 140. The screen in Fig. 8 includes an operation menu on the left side and a map on the right side of the operation menu. The operation menu includes a display D1 of the target period, a pothole display switching button D2, and a future prediction function switching button D3. The future prediction function will be described in the second embodiment.

[0064] The target period indicates the period during which the road surface images used for analysis were captured. For example, the target period indicates that the road surface images were captured within the past 90 days from the reference date entered on the screen.

[0065] When the user turns on the pothole display switch button D2, the output unit 140 displays an icon indicating the occurrence probability calculated by the calculation unit 130 on the map. In FIG. 8 , the pothole occurrence probability is displayed on the map using icons in three levels of color. The output unit 140 may display an icon only for a portion of the map. For example, the output unit 140 may display an icon for an area selected by the user on the map.

[0066] The operation menu in FIG. 8 further includes a user interface D4 for narrowing down the icons to be displayed. In FIG. 8, the locations for which icons are to be displayed are narrowed down to locations where the probability of pothole occurrence is "30% or higher." The user can appropriately set a threshold value for whether or not to display an icon via the input device 21. For example, the user may input the threshold value as a numerical value. The user may also set the threshold value by moving an arrow D5 indicating the value on a scale D6 in FIG. 8 left or right. By setting the threshold value in this way, the user can instantly check locations where the probability of pothole occurrence is high.

[0067] Furthermore, when the output unit 140 receives a selection of an icon on the map, it may display a graphic representing the probability of pothole occurrence indicated by the selected icon. For example, the output unit 140 may display a graphic representing a standard indicating the magnitude of the probability of pothole occurrence as a graphic representing the probability of pothole occurrence. The graphic representing the standard of the probability of pothole occurrence is also called a scale. Standard values ​​of the probability of pothole occurrence (e.g., 0% and 100%) are set at the reference points of the scale (e.g., both ends of the scale).

[0068] For example, the output unit 140 may display a scale on the map such that an icon on the map indicates a value on the scale, whereby the icon on the map indicates a value on the scale, and the scale represents a value of the probability of occurrence of a pothole.

[0069] The scale displayed by the output unit 140 may be a color scale legend that indicates the probability of pothole occurrence indicated by the color of an icon on the map. The output unit 140 may display the color scale legend at a position where the color of the icon on the map corresponds to the color on the scale.

[0070] The output unit 140 may also display a scale indicating a value on the scale using an icon displayed separately from the selected icon on the map. As shown in FIG. 8 , the output unit 140 may pop up a scale D9 and an icon D8 indicating a value on the scale for a selected icon D10 on the map. In a pop-up area D7, for example, an icon D8 of the same color as the selected icon D10 indicates the value on the scale D9. Note that when the output unit 140 displays an icon indicating a value on the scale separately from the icon on the map, the icon indicating the value on the scale may be a shape different from the icon on the map, such as an arrow or a line.

[0071] When the probability of pothole occurrence is displayed in a low-level color, the output unit 140 displays a scale corresponding to the icon, allowing the user to understand the predicted occurrence probability in more detail than the icon on the map. Also, when the probability of pothole occurrence is displayed in multiple levels of color, the output unit 140 displays a scale, allowing the user to check the occurrence probability indicated by the color of the icon on the scale.

[0072] 9 is a flowchart showing an example of the operation of the output unit 140 that displays the scale. For example, after step S13 in FIG. 6, the output unit 140 receives the probability of occurrence of a pothole and starts the operation shown in FIG.

[0073] The output unit 140 displays an icon indicating the probability of pothole occurrence superimposed on the map (step S21), and then receives a selection of an icon on the map selected by the user using the input device 21 (step S22).

[0074] Then, the output unit 140 displays the scale by associating the occurrence probability indicated by the selected icon with the position of the occurrence probability on the scale (step S23). With this, the output unit 140 ends the operation of FIG. 9.

[0075] Furthermore, the output unit 140 may display the probability of pothole occurrence on the map by a method other than using icons. For example, the output unit 140 may display a mesh-like area on the map or an area of ​​a road divided into predetermined sections in a color corresponding to the probability of pothole occurrence in that area.

[0076] The output unit 140 may further display the degree of road surface deterioration for each road section on the map in addition to the probability of pothole occurrence. For example, the output unit 140 may display an icon such as an arrow of a different color for each road section depending on the degree of deterioration.

[0077] The output unit 140 may display an overview of information indicating the occurrence probabilities of multiple locations, and may display more detailed information about a selected location. The output unit 140 may further display, as detailed information, a road surface image of the location, the date and time the road surface image was captured, the analysis results of the crack state, and the calculated value of the pothole occurrence probability. The output unit 140, for example, displays the road surface image used for analysis by the analysis unit 120. By displaying the road surface image and the pothole occurrence probability side by side, the user can easily understand the degree of cracks indicated by the numerical value of the pothole occurrence probability. Furthermore, the output unit 140 may display, for example, the date and time the image was captured acquired by the acquisition unit 110. When the calculation unit 130 predicts the size of a pothole that will occur, the output unit 140 may further display the predicted pothole size.

[0078] The analysis unit 120 according to the first embodiment analyzes the state of cracks on the road surface from road surface images. The calculation unit 130 then calculates the predicted probability of pothole occurrence from the analysis results using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data. Therefore, according to the first embodiment, the probability of pothole occurrence can be determined with a simple configuration.

[0079] For example, according to the first embodiment, the probability of pothole occurrence can be calculated based on crack information analyzed from road surface images captured by a drive recorder. Therefore, there is no need to measure the amount of local subsidence or the average profile depth using a slit laser. Therefore, the probability of pothole occurrence can be calculated with a simple configuration.

[0080] Furthermore, according to the first embodiment, by predicting the occurrence of potholes based on information about cracks, which are one of the main causes of pothole occurrence, it becomes possible to predict the occurrence of potholes with high accuracy.

[0081] Furthermore, according to the first embodiment, the output unit 140 outputs information indicating the calculated probability of pothole occurrence, so that the user can efficiently consider road surface repair plans in accordance with the output information.

[0082] Second Embodiment Fig. 10 is a block diagram showing an example of the configuration of a pothole prediction system 200 according to a second embodiment. The pothole prediction system 200 differs from the pothole prediction system 100 according to the first embodiment in that it includes a crack prediction unit 121. With regard to the configuration of the second embodiment, some of the description of the configuration that is the same as that of the first embodiment will be omitted.

[0083] The crack prediction unit 121 predicts the future state of cracks on the road surface from which the road surface image was captured, based on the analysis results by the analysis unit 120. The crack prediction unit 121 predicts the state of cracks after a predetermined period of time has passed since the road surface image was captured. The predetermined period is set as appropriate, for example, six months, one year, or two years. The crack prediction unit 121 may predict the state of cracks at a future time point specified by the user. Furthermore, the crack prediction unit 121 may predict the state of cracks at multiple future time points.

[0084] The method for predicting the future crack state is not particularly limited. The crack prediction unit 121 may predict the future crack state using existing technology. For example, the crack prediction unit 121 predicts the future crack rate, crack width, crack area, or amount of tortoiseshell cracks based on the crack rate, crack width, crack area, or amount of tortoiseshell cracks analyzed by the analysis unit 120. The crack prediction unit 121 may predict the future crack state based on other information such as road surface information and weather information in addition to the analysis results by the analysis unit 120.

[0085] The calculation unit 130 calculates the probability of pothole occurrence predicted from the prediction results of the crack prediction unit 121 using, for example, the same prediction model as the prediction model according to the first embodiment.

[0086] The output unit 140 outputs information indicating the calculated probability of pothole occurrence. The output unit 140 may, for example, display on the display 20 a numerical value of the probability of pothole occurrence calculated based on the future state of cracks. The output unit 140 may also display an icon indicating the calculated probability of pothole occurrence on a map. For example, in FIG. 8 , when the future prediction function switch button D3 is pressed, the output unit 140 displays an icon indicating the probability of pothole occurrence based on the future state of cracks. The output unit 140 may further display the future state of cracks predicted by the crack prediction unit 121 for the selected point.

[0087] The output unit 140 may display a graph showing the change over time in the probability of pothole occurrence. In this case, the pothole prediction system 200 may include a graph generation unit (not shown). The graph generation unit obtains the pothole occurrence probability calculated from the output unit 140 and plots it. The graph generation unit then provides the generated graph to the output unit 140.

[0088] The output unit 140 may also display a predicted image showing the future state of the crack. In this case, the pothole prediction system 200 may include an image generation unit (not shown). The image generation unit generates a predicted image showing the progression of the crack using the road surface image acquired by the acquisition unit 110. The image generation unit generates the predicted image according to the future state of the crack predicted by the crack prediction unit 121, for example.

[0089] 11 is a flowchart showing an example of the operation of the pothole prediction system 200 according to the second embodiment. For example, the pothole prediction system 200 performs the operations from step S11 to step S14 shown in FIG.

[0090] After step S14, if the user turns on the future prediction function switch button D3 (step S31: Yes), the crack prediction unit 121 predicts the future state of cracks in the road surface from which the road surface image was captured based on the analysis result by the analysis unit 120 (step S32). The crack prediction unit 121 provides the prediction result to the calculation unit 130.

[0091] The calculation unit 130 uses the prediction model to calculate the probability of pothole occurrence predicted from the prediction result of the crack prediction unit 121 (step S33). The calculation unit 130 outputs the calculated probability of pothole occurrence to the output unit 140.

[0092] The output unit 140 outputs information indicating the probability of pothole occurrence predicted from the future state of the crack (step S34). For example, the output unit 140 causes the display 20 to display the information indicating the probability of pothole occurrence.

[0093] With the above, the pothole prediction system 200 ends the operation shown in FIG.

[0094] 12 and 13 are diagrams showing examples of screens displayed by the output unit 140. FIG. 12 is a screen that displays information based on a photographed road surface image. FIG. 13 is a screen that displays information based on the future state of cracks. The screen of FIG. 12 is displayed, for example, when the user selects a specific point on a map. When "One year later" is selected from the pull-down list that is displayed by pressing the "Future prediction" button in FIG. 12, the screen of FIG. 13 is displayed.

[0095] 12, the value of the probability of pothole occurrence calculated from the analysis result of analysis unit 120 is displayed as the current probability of pothole occurrence. The screen of Fig. 12 also includes the analysis result of analysis unit 120, a road surface image showing the analyzed road surface, and a graph plotting the current probability of pothole occurrence. In another example, output unit 140 may further plot the probability of pothole occurrence calculated based on past road surface images.

[0096] On the screen of Fig. 13, the value of the probability of pothole occurrence calculated from the future crack state predicted by the crack prediction unit 121 is displayed as the probability of pothole occurrence one year from now. The screen of Fig. 13 also includes the future crack state predicted by the crack prediction unit 121 and a predicted image one year from now. Furthermore, the screen of Fig. 13 includes a graph plotting the current probability of pothole occurrence and the probability of pothole occurrence one year from now.

[0097] According to the second embodiment, the crack prediction unit 121 predicts the future state of cracks in the road surface, and the calculation unit 130 calculates the predicted probability of pothole occurrence from the prediction result of the crack prediction unit 121. Therefore, according to the second embodiment, the probability of pothole occurrence can be determined based on the future state of cracks. This allows the user to consider the need for longer-term repairs, taking into account the progression of future cracks. For example, the user can create a repair plan for the current period based on the output according to the first embodiment, and create a plan for the next period based on the output according to the second embodiment.

[0098] According to the second embodiment, the output unit 140 outputs the probability of pothole occurrence based on the state of cracks analyzed from the road surface image and the probability of pothole occurrence based on the future state of cracks. Therefore, the user can plan road surface repairs taking into account the degree of increase in the probability of pothole occurrence.

[0099] This concludes the description of each embodiment.

[0100] [Modifications] Each embodiment may be modified and used.

[0101] For example, the pothole prediction system 100 may further include a repair point determination unit. The repair point determination unit may determine, for example, points where the probability of pothole occurrence exceeds a predetermined threshold as points requiring repair. The repair point determination unit may also determine, as areas requiring repair, areas where the number of points where the probability of pothole occurrence exceeds a predetermined threshold exceeds a predetermined threshold. The output unit 140 outputs information indicating the determined points.

[0102] For example, the repair point determination unit acquires a repair plan that includes predetermined locations where repairs will be performed. The repair point determination unit then determines locations that are not included in the repair plan even though the probability of pothole occurrence exceeds a predetermined threshold. This allows the user to consider repairs for locations that were not included in the repair plan.

[0103] The pothole prediction system 100 may further include a repair priority determination unit that determines the repair priority of the road surface based on the pothole occurrence probability calculated by the calculation unit 130 and other parameters. The output unit 140 displays points with high repair priorities on a map.

[0104] The repair priority determination unit determines that a point where the probability of pothole occurrence exceeds a predetermined threshold has a high priority. The repair priority determination unit may further determine the repair priority of the road surface based on the traffic volume on the road surface as another parameter. For example, if there are points with the same probability of pothole occurrence, the repair priority determination unit may determine that the repair priority of the point with the higher traffic volume is higher.

[0105] Furthermore, as other parameters, the repair priority determination unit may determine the repair priority based on the results of an analysis of the crack condition, information on the width of the road, or the presence or absence of a detour.

[0106] [Hardware Configuration] In the above-described embodiments, each component of the pothole prediction systems 100 and 200 represents a functional block. Some or all of the components of the pothole prediction systems 100 and 200 may be realized by any combination of a computer 500 and a program.

[0107] Fig. 14 is a block diagram showing an example of the hardware configuration of a computer 500. Referring to Fig. 14, the computer 500 includes, for example, a processor 501, a read only memory (ROM) 502, a random access memory (RAM) 503, a program 504, a storage device 505, a drive device 507, a communication interface 508, an input device 509, an input / output interface 511, and a bus 512.

[0108] The processor 501 controls the entire computer 500. The processor 501 may be, for example, a CPU (Central Processing Unit). The number of processors 501 is not particularly limited, and there may be one or more processors 501.

[0109] The program 504 includes instructions for realizing each function of the pothole prediction systems 100, 200. The program 504 is stored in advance in the ROM 502, RAM 503, and storage device 505. The processor 501 executes the instructions included in the program 504 to realize each function of the pothole prediction systems 100, 200. The RAM 503 may also store data to be processed in each function of the pothole prediction systems 100, 200. For example, a road surface image may be stored in the RAM 503 of the computer 500.

[0110] The drive device 507 reads and writes data from and to the recording medium 506. The communication interface 508 provides an interface with a communication network. The input device 509 is, for example, a mouse or a keyboard, and receives information input from a user or the like. The output device 510 is, for example, a display, and outputs (displays) information to a user or the like. The input / output interface 511 provides an interface with peripheral devices. The bus 512 connects these hardware components. The program 504 may be supplied to the processor 501 via a communication network, or may be stored in advance on the recording medium 506, read by the drive device 507, and supplied to the processor 501.

[0111] It should be noted that the hardware configuration shown in FIG. 14 is an example, and other components may be added, or some components may not be included.

[0112] There are various modified examples of the method for realizing the pothole prediction systems 100, 200. For example, the pothole prediction systems 100, 200 may be realized by any combination of different computers and programs for each component. Furthermore, multiple components included in the pothole prediction systems 100, 200 may be realized by any combination of a single computer and program.

[0113] Furthermore, at least a part of the pothole prediction systems 100 and 200 may be provided in a software as a service (SaaS) format. That is, at least a part of the functions for realizing the pothole prediction systems 100 and 200 may be executed by software that is executed via a network.

[0114] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, the configurations in the respective embodiments can be combined with each other without departing from the scope of the present disclosure.

[0115] Some or all of the above embodiments may be described as, but are not limited to, the following supplementary notes.

[0116] [Supplementary Note 1] A pothole prediction system comprising: an acquisition means for acquiring a road surface image obtained by photographing the road surface; an analysis means for analyzing the state of cracks on the road surface from the road surface image; a calculation means for calculating the probability of pothole occurrence predicted from the analysis results by the analysis means using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data; and an output means for outputting information indicating the calculated probability of pothole occurrence.

[0117] [Supplementary Note 2] The pothole prediction system according to Supplementary Note 1, wherein the analysis results include at least one of a crack rate, a crack length, a crack width, a crack area, a crack shape, an amount of tortoiseshell cracks, and the presence or absence of cracks.

[0118] [Supplementary Note 3] The pothole prediction system according to Supplementary Note 2, wherein the analysis results include a crack rate, a crack width, and an amount of tortoiseshell cracks.

[0119] [Supplementary Note 4] The pothole prediction system according to Supplementary Note 3, wherein the amount of the hexagonal cracks is the number of predetermined units that contain hexagonal cracks when the road surface image is divided into the predetermined units.

[0120] [Supplementary Note 5] The pothole prediction system according to any one of Supplementary Notes 1 to 4, further comprising a crack prediction means for predicting a future state of cracks in the road surface based on the analysis results, wherein the calculation means calculates a predicted probability of pothole occurrence from the prediction results of the crack prediction means.

[0121] [Supplementary Note 6] The pothole prediction system according to any one of Supplementary Notes 1 to 5, wherein the output means displays an icon indicating the calculated probability of pothole occurrence on a map showing the road surface.

[0122] [Supplementary Note 7] The pothole prediction system according to Supplementary Note 6, wherein when the output means receives a selection of the icon on the map, the output means displays a graphic representing a criterion for the probability of occurrence of the pothole indicated by the selected icon.

[0123] [Supplementary Note 8] The pothole prediction system according to Supplementary Note 6 or 7, further comprising a receiving means for receiving a threshold value for the probability of pothole occurrence, wherein the output means displays the icon indicating the probability of pothole occurrence equal to or greater than the received threshold value.

[0124] [Supplementary Note 9] The pothole prediction system according to any one of Supplementary Notes 1 to 8, wherein the training data is data that includes road information as an explanatory variable in addition to the state of the cracks, and the calculation means calculates the probability of pothole occurrence based on the analysis result and the road information of the road surface.

[0125] [Supplementary Note 10] A pothole prediction method comprising: acquiring a road surface image obtained by photographing a road surface; analyzing the state of cracks on the road surface from the road surface image; calculating a predicted probability of pothole occurrence from the results of the analysis using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data; and outputting information indicating the calculated probability of pothole occurrence.

[0126] [Supplementary Note 11] A recording medium that non-temporarily records a program that causes a computer to execute the following processes: acquiring a road surface image obtained by photographing a road surface; analyzing the state of cracks on the road surface from the road surface image; calculating a predicted probability of pothole occurrence from the results of the analysis using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data; and outputting information indicating the calculated probability of pothole occurrence.

[0127] 100 Pothole prediction system 110 Acquisition unit 120 Analysis unit 130 Calculation unit 131 Prediction model storage unit 132 Calculation unit 140 Output unit 10 Camera 20 Display 21 Input device 30 Communication network 40 Database

Claims

1. An acquisition means for acquiring a road surface image obtained by photographing a road surface; an analysis means for analyzing a state of cracks on the road surface from the road surface image; a calculation means for calculating a probability of pothole occurrence predicted from the analysis result by the analysis means, using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data; an output means for outputting information indicating the calculated probability of pothole occurrence; A pothole prediction system comprising:

2. The analysis result includes at least one of a crack rate, a crack length, a crack width, a crack area, a crack shape, an amount of hexagonal cracks, and the presence or absence of cracks. The pothole prediction system according to claim 1 .

3. The analysis results include crack rate, crack width and amount of tortoiseshell cracks. The pothole prediction system according to claim 2.

4. The amount of the tortoiseshell cracks is the number of units that contain tortoiseshell cracks when the road surface image is divided into predetermined units. The pothole prediction system according to claim 3 .

5. Further comprising a crack prediction means for predicting a future state of cracks in the road surface based on the analysis result, The calculation means calculates a probability of occurrence of a pothole predicted from a prediction result of the crack prediction means. The pothole prediction system according to any one of claims 1 to 4.

6. The output means displays an icon indicating the calculated probability of occurrence of a pothole on a map showing the road surface. The pothole prediction system according to any one of claims 1 to 4.

7. When the output means receives a selection of the icon on the map, it displays a graphic representing a standard of the occurrence probability of the pothole indicated by the selected icon. The pothole prediction system according to claim 6.

8. The method further includes receiving means for receiving a threshold value of the probability of occurrence of the pothole, The output means displays the icon indicating the probability of occurrence of the pothole equal to or greater than the received threshold value. The pothole prediction system according to claim 6.

9. Obtaining road surface images by photographing the road surface, Analyzing the state of cracks on the road surface from the road surface image; Calculating the probability of pothole occurrence predicted from the results of the analysis using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data; outputting information indicating the calculated probability of pothole occurrence; Pothole prediction methods.

10. Obtaining road surface images by photographing the road surface, Analyzing the state of cracks on the road surface from the road surface image; Calculating the probability of pothole occurrence predicted from the results of the analysis using a prediction model trained using data indicating the relationship between the state of cracks and the occurrence of potholes as training data; outputting information indicating the calculated probability of pothole occurrence; A program that causes a computer to carry out processing.