Program, information processing method and information processing device

Abstract

To provide a program and the like capable of detecting a desired object from a drawing image.SOLUTION: A program causes a computer to execute the processing of acquiring a drawing image, inputting the acquired drawing image into a model that has been subjected to learning to detect a predetermined object when the drawing image is input, and detecting the object. Preferably, the drawing image is a map image. More preferably, the map image is a road register map image, and a road object including an intersection is detected by inputting the acquired road register map image into the model.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to a program, an information processing method, and an information processing device. [Background technology]

[0002] There is a technology for detecting objects such as intersections from a bird's-eye view image of a road. For example, Patent Document 1 discloses a map generating device that detects an intersection and a plurality of roads connected to the intersection from a bird's-eye view image, detects an entrance lane for entering the intersection and an exit lane for exiting the intersection for each detected road, and generates a lane network that represents the connection relationship between lanes at an intersection such that the entrance lane is connected to the exit lane of another road on which a vehicle can proceed. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-107986 A Summary of the Invention [Problem to be solved by the invention]

[0004] In one aspect, an object of the present invention is to provide a program or the like capable of detecting a desired object from a drawing image. [Means for solving the problem]

[0005] In one aspect, the program causes a computer to execute a process of acquiring a drawing image, inputting the acquired drawing image into a model that has been trained to detect a specified object when a drawing image is input, and detecting the object. Effect of the Invention

[0006] In one aspect, a desired object can be detected from a drawing image. [Brief description of the drawings]

[0007] [Figure 1] FIG. 1 is an explanatory diagram illustrating an example of the configuration of a drawing recognition system. [Diagram 2] FIG. 2 is a block diagram showing an example of the configuration of a server. [Diagram 3] FIG. 2 is a block diagram showing a configuration example of a terminal. [Figure 4] FIG. 1 is an explanatory diagram showing an overview of a first embodiment. [Diagram 5] FIG. 11 is an explanatory diagram relating to a process for creating a reference drawing. [Figure 6] FIG. 11 is an explanatory diagram showing another example of the reference diagram. [Figure 7] 13 is a flowchart showing a procedure for generating a detection model. [Figure 8] 13 is a flowchart showing a procedure for creating a reference map. [Figure 9] 11 is a flowchart showing a procedure for generating a detection model according to the second embodiment. [Figure 10] 13 is a flowchart showing a procedure for creating a reference map according to the second embodiment. [Figure 11] FIG. 13 is an explanatory diagram showing an overview of a third embodiment. [Figure 12] 13 is a flowchart showing the procedure of a segmentation model generation process. [Figure 13] 13 is a flowchart showing a procedure for creating a reference map according to the third embodiment. [Figure 14] 13 is a flowchart showing a procedure for creating a reference map according to the fourth embodiment. [Figure 15] FIG. 13 is a block diagram showing an example of the configuration of a server according to the fifth embodiment. [Figure 16] FIG. 2 is an explanatory diagram illustrating an example of a record layout of a map DB. [Figure 17] 13 is a flowchart showing a procedure for creating a reference map according to the fifth embodiment. [Figure 18] 13 is a flowchart showing a procedure for a map search process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] The present invention will now be described in detail with reference to the drawings showing embodiments thereof. (Embodiment 1) FIG. 1 is an explanatory diagram showing an example of the configuration of a plan recognition system. In this embodiment, a plan recognition system is described that uses a detection model 50 (see FIG. 2) constructed by machine learning to detect objects such as intersections from a map image, which is a representative plan image, and automatically creates a road construction safety facility installation reference map (hereinafter, for brevity, referred to as a "reference map") based on the detection result of the object. The plan recognition system includes an information processing device 1 and a terminal 2. Each device is communicatively connected via a network N such as the Internet.

[0009] As described later, the image targeted in this embodiment may be any drawing image expressed by figures or symbols, and the drawing image is not limited to a map image.

[0010] The reference map (see Figures 5 and 6) is a drawing created for security measures at a construction site. Specifically, as shown in Figures 5 and 6, the reference map shows the parking positions of work vehicles, the positions of traffic controllers, and the locations of color cones (registered trademark) to define detours for pedestrians and passing vehicles. The reference map is created, for example, as an attachment to construction application documents.

[0011] Conventionally, reference maps have been created manually, with the creator checking the locations of intersections, pedestrian crossings, nearby parking lots, etc. on a road register map, and then filling in notations indicating the scope of construction and the location of construction work to create the reference map. In this embodiment, this process is automated, and the reference map is created using a detection model 50 constructed by machine learning.

[0012] The information processing device 1 is an information processing device capable of various information processing and information transmission and reception, such as a server computer, a personal computer, etc. In this embodiment, the information processing device 1 is a server computer, and will be referred to as server 1 below for simplicity. The server 1 generates a detection model 50 that detects a predetermined road object (e.g., an intersection, a pedestrian crossing, a parking lot, etc.) when a road register map image is input by learning predetermined training data. Then, the server 1 uses the generated detection model 50 to detect road objects including an intersection from a road register map image of a construction site. As described later, the server 1 specifies the distance from the detected intersection to the construction site, specifies the construction range according to the distance to the specified construction site and the type of construction, and creates a reference map by superimposing an object representing the specified construction range and the construction site on the road register map image.

[0013] In this embodiment, the map image is described as being a road register map image, but the map image is not limited to a map related to the road register and may be other map images (for example, a residential map or Google Maps (registered trademark)).

[0014] The terminal 2 is a terminal device used by the creator of the reference map, and is, for example, a personal computer, a tablet terminal, etc. The creator uploads a map image to the server 1 via the terminal 2, causes object detection to be performed, and obtains a map image (reference map) on which the object is superimposed.

[0015] In this embodiment, the server 1 is mainly described as performing a series of processes such as detecting objects using the detection model 50 and creating a reference map, but the embodiment is not limited to this, and some or all of the processes performed by the server 1 may be performed by the client terminal 2.

[0016] 2 is a block diagram showing an example of the configuration of the server 1. The server 1 includes a control unit 11, a main storage unit 12, a communication unit 13, and an auxiliary storage unit . The control unit 11 has one or more arithmetic processing devices such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), a GPU (Graphics Processing Unit), etc., and performs various information processing, control processing, etc. by reading and executing a program P1 stored in the auxiliary storage unit 14. The main storage unit 12 is a temporary storage area such as an SRAM (Static Random Access Memory) or a DRAM (Dynamic Random Access Memory), and temporarily stores data required for the control unit 11 to execute arithmetic processing. The communication unit 13 is a communication module for performing processing related to communication, and transmits and receives information to and from the outside.

[0017] The auxiliary storage unit 14 is a non-volatile storage area such as a large-capacity memory or a hard disk, and stores a program P1 (program product) and other data necessary for the control unit 11 to execute processing. The auxiliary storage unit 14 also stores a detection model 50. The detection model 50 is a machine learning model that has learned predetermined training data, and is a model that has learned to detect a predetermined object (road object) when a map image is input.

[0018] The auxiliary storage unit 14 may be an external storage device connected to the server 1. The server 1 may be a multi-computer consisting of a plurality of computers, or may be a virtual machine virtually constructed by software.

[0019] In the present embodiment, the server 1 is not limited to the above configuration, and may include, for example, an input unit for accepting operation input, a display unit for displaying images, etc. The server 1 may also include a reading unit for reading a portable storage medium 1a such as a CD (Compact Disk)-ROM or a DVD (Digital Versatile Disc)-ROM, and may read and execute the program P1 from the portable storage medium 1a.

[0020] 3 is a block diagram showing an example of the configuration of the terminal 2. The terminal 2 includes a control unit 21, a main memory unit 22, a communication unit 23, a display unit 24, an input unit 25, and an auxiliary memory unit . The control unit 21 has one or more processors such as CPUs, and performs various information processing by reading and executing the program P2 stored in the auxiliary storage unit 26. The main storage unit 22 is a temporary storage area such as RAM, and temporarily stores data necessary for the control unit 21 to execute arithmetic processing. The communication unit 23 is a communication module for performing processing related to communication, and transmits and receives information to and from the outside. The display unit 24 is a display screen such as a liquid crystal display, and displays images. The input unit 25 is an operation interface such as a touch panel, and accepts operation input from the user. The auxiliary storage unit 26 is a non-volatile storage area such as a hard disk, and stores the program P2 (program product) and other data necessary for the control unit 21 to execute processing.

[0021] The terminal 2 may be provided with a reading unit for reading a portable storage medium 2a such as a CD-ROM, and may read and execute the program P2 from the portable storage medium 2a.

[0022] Fig. 4 is an explanatory diagram showing an overview of the first embodiment. Fig. 4 shows how an object is detected (illustrated by a bounding box with a dashed line) from a map image (drawing image). The outline of a detection model 50 will be described with reference to Fig. 4.

[0023] The detection model 50 is a model that has been trained to detect a predetermined object when a map image (drawing image) is input, and is, for example, an SNN (Spiking Neural Network). Note that the detection model 50 is not limited to an SNN, and may be another machine learning model, such as a CNN (Convolutional Neural Network).

[0024] The server 1 generates a detection model 50 by using training data in which the correct position and type of an object to be detected are associated with a training map image. Objects to be detected in this embodiment include, for example, intersections, crosswalks, parking lots, sidewalk drop-offs (vehicle entrances and exits), road markings painted on roads, etc. Note that these are only examples of objects, and other objects may be detected from a map image.

[0025] The server 1 detects the position and type of an object by inputting a training map image into the detection model 50. The server 1 updates parameters such as weights between neurons so that the detection result approximates the correct position and type. The server 1 repeats this process for each training map image to generate a detection model 50 with optimized parameters.

[0026] When actually detecting objects from a map image using the detection model 50, the server 1 acquires a map image of a construction site from the terminal 2. The server 1 detects various objects by inputting the acquired map image into the detection model 50. As a result, road objects such as intersections, crosswalks, and parking lots are detected from the map image, as shown in Fig. 4.

[0027] In this embodiment, as an example of the application of object detection using the above-mentioned detection model 50, a form in which a reference map is automatically created from a map image will be described.

[0028] Fig. 5 is an explanatory diagram of the process of creating a reference map. Fig. 5 conceptually illustrates how objects representing traffic cones, etc. are superimposed according to the positions of the intersection detected above and the construction site. The process of creating a reference map will be described with reference to Fig. 5.

[0029] First, the server 1 acquires information on the construction site (shown as a hatched rectangular area in FIG. 5) from the terminal 2. The construction site may be, for example, a utility pole, an electric wire, an underground power distribution cable, etc., but is not limited thereto. The server 1 identifies the position of the construction site on the map image based on the acquired information. Then, the server 1 identifies the distance from the intersection detected by the detection model 50 to the construction site.

[0030] Next, the server 1 receives the type of construction work to be carried out from the terminal 2. The server 1 identifies the scope of the construction work based on the received type of construction work and the distance from the intersection identified above to the construction work site.

[0031] For example, the server 1 predefines the length of the work area required for the work along the road according to the type of work. The server 1 then identifies the length of the work area required for the work according to the type of work received, and compares the identified length of the work area with the distance from the intersection to the work location. If the length of the work area is shorter than the distance from the intersection to the work location, the server 1 sets (identifies) the work area continuously, centered on the work location. On the other hand, if the length of the work area is longer than the distance from the intersection to the work location, the server 1 sets the work area intermittently across the intersection.

[0032] For example, in the example of Fig. 5, the distance from the intersection to the construction site is sufficiently long and the length of the construction range is shorter than the distance from the intersection to the construction site where the construction was successful, so the server 1 sets the construction range continuously with the construction site at the center without spanning the intersection. In contrast, Fig. 6 is an explanatory diagram showing another example of the reference map. In the example of Fig. 6, the distance from the intersection to the construction site is short and the length of the construction range is longer than the distance from the intersection to the construction site, so the server 1 sets the construction range on both sides of the intersection, spanning the intersection.

[0033] Returning to Fig. 5, the explanation will continue. Finally, the server 1 superimposes objects representing the identified construction area and construction site on the map image in accordance with predetermined rules. For example, the server 1 superimposes (places) objects representing traffic cones so as to surround the identified construction area. The server 1 also superimposes objects representing traffic controllers, work vehicles, electric display vehicles, etc. within the construction area. In this way, the server 1 creates a reference map.

[0034] As described above, according to this embodiment, objects such as intersections can be automatically detected from a map image, and a reference map can be created using the detection results.

[0035] Fig. 7 is a flowchart showing the procedure of the generation process of the detection model 50. The process content when generating the detection model 50 by machine learning will be described with reference to Fig. 7. The control unit 11 of the server 1 acquires training data for generating the detection model 50 (step S11). The training data is data in which the correct position and type of an object are associated with a training map image (drawing image).

[0036] Based on the training data, the control unit 11 generates a detection model 50 that detects an object when a map image (drawing image) is input (step S12). Specifically, as described above, the control unit 11 generates an SNN as the detection model 50. The control unit 11 detects the position and type of the object by inputting a training map image into the detection model 50, and generates the detection model 50 by optimizing parameters such as the weights between neurons so that the detection result approximates the correct position and type. The control unit 11 ends the series of processes.

[0037] 8 is a flow chart showing the procedure of the process of creating the base map. The process of creating the base map will be described with reference to FIG. The control unit 11 of the server 1 acquires a map image of the construction site from the terminal 2 (step S31). As described above, the map image is a map image related to a road register. The control unit 11 inputs the acquired map image into the detection model 50 to detect a predetermined object (step S32). Specifically, the control unit 11 detects road objects including intersections.

[0038] The control unit 11 acquires information on the construction site from the terminal 2 (step S33). The control unit 11 identifies the distance from the intersection detected in step S32 to the construction site (step S34).

[0039] The control unit 11 receives the type of construction work to be carried out via the terminal 2 (step S35). The control unit 11 specifies the construction range based on the received type of construction work and the distance from the intersection specified in step S34 to the construction work location (step S36). Specifically, the control unit 11 specifies the length of the construction range required according to the type of construction work, and compares the specified length of the construction range with the distance from the intersection to the construction work location. If the length of the construction range is shorter than the distance from the intersection to the construction work location, the control unit 11 sets (specifies) the construction range continuously with the construction work location as the center. On the other hand, if the length of the construction range is longer than the distance from the intersection to the construction work location, the control unit 11 sets the construction range intermittently across the intersection.

[0040] The control unit 11 superimposes objects representing the identified construction area and construction site on the map image (step S37). Specifically, as described above, the control unit 11 superimposes (places) objects representing traffic cones so as to surround the identified construction area in accordance with a predetermined rule, and also superimposes objects representing traffic controllers, work vehicles, electric display vehicles, etc. within the construction area. The control unit 11 ends the series of processes.

[0041] In this embodiment, the image input to the detection model 50 is described as a map image, but the embodiment is not limited to this. For example, the server 1 may input some layout drawing, design drawing, etc., other than a map, to the detection model 50 to detect an object in the drawing. As such a use case, for example, a case where a figure or the like is read from a drawing and the detection result is output to a 3D printer, thereby processing from the drawing, is considered. Another use case is a case where the location of a package is detected from a layout drawing of the package in a warehouse and the package is automatically located.

[0042] In this way, the image input to the detection model 50 may be any drawing image expressed with figures or symbols, and the drawing image is not limited to a map image.

[0043] As described above, according to the first embodiment, a desired object can be detected from a drawing image.

[0044] (Embodiment 2) In this embodiment, a description will be given of an embodiment in which the accuracy of detecting an object is improved by enlarging a map image (drawing image). Note that the same reference numerals are used to designate the same contents as in the first embodiment, and the description thereof will be omitted.

[0045] First, an outline of the present embodiment will be described. In the present embodiment, the server 1 does not input a map image (drawing image) directly to the detection model 50, but enlarges the map image and then inputs it to the detection model 50, thereby improving the detection accuracy of the object.

[0046] When generating (training) the detection model 50, the server 1 enlarges the training map image to a predetermined magnification before learning the detection model 50 using training data. Then, the server 1 generates the detection model 50 based on the enlarged training map image and the correct position and type of the object. That is, the server 1 detects the position and type of the object by inputting the enlarged map image into the detection model 50, and generates the detection model 50 by optimizing parameters such as the weights between neurons so that the detection result approximates the correct position and type.

[0047] When creating a reference map, the server 1 first enlarges the map image acquired from the terminal 2 to a predetermined magnification. The server 1 then inputs the enlarged map image to the detection model 50 to detect objects including intersections. The subsequent processing is the same as in the first embodiment, and the server 1 identifies the distance from the intersection to the construction site, identifies the construction area, and superimposes the object.

[0048] 9 is a flowchart showing a procedure of the process of generating the detection model 50 according to the embodiment 2. After acquiring training data for generating the detection model 50 (step S11), the server 1 executes the following process. The control unit 11 of the server 1 enlarges the training map image (drawing image) to a predetermined magnification (step S201). The control unit 11 generates a detection model 50 based on the enlarged training map image and the correct position and type of the object (step S202). That is, the control unit 11 detects the position and type of the object by inputting the enlarged map image to the detection model 50, and generates the detection model 50 by optimizing parameters such as weights between neurons so that the detection result approximates the correct position and type. The control unit 11 ends the series of processes.

[0049] 10 is a flowchart showing the procedure of the process of creating a base map according to embodiment 2. After acquiring a map image from the terminal 2 (step S31), the server 1 executes the following process. The control unit 11 of the server 1 enlarges the acquired map image to a predetermined magnification (step S221). The control unit 11 inputs the enlarged map image to the detection model 50 to detect a predetermined object (step S222). The control unit 11 shifts the process to step S33.

[0050] As described above, according to the second embodiment, the accuracy of detecting an object can be improved by enlarging the map image (drawing image).

[0051] (Embodiment 3) In this embodiment, a segmentation process is performed to color a map image (drawing image) before detecting an object using the detection model 50.

[0052] Fig. 11 is an explanatory diagram showing an overview of the third embodiment. Fig. 11 conceptually shows how a specific object (e.g., a sidewalk) on a map image is colored (shown by hatching in Fig. 11) by inputting the map image into a segmentation model 40 (second model) different from the detection model 50. The overview of this embodiment will be described with reference to Fig. 11.

[0053] As described above, in this embodiment, a road register map image is used as the map image. However, a road register map image is an image expressed in two values, black and white, and it is difficult to detect objects such as intersections from a black and white image. Therefore, in this embodiment, a segmentation process is performed to color objects such as sidewalks and roadways on the road map image in advance, making it easier to detect the objects.

[0054] The segmentation model 40 is a model that identifies image regions corresponding to objects such as sidewalks, roadways, etc. when a map image (drawing image) is input, and is, for example, a semantic segmentation model, which is a type of CNN. For example, the segmentation model 40 identifies an image region on a map image that corresponds to a sidewalk, as shown in FIG.

[0055] The server 1 generates a segmentation model 40 based on training data in which information indicating the correct image area of ​​the sidewalk (or roadway) is associated with a training map image. For example, the server 1 acquires training data by accepting an operation to color the sidewalk portion of the map image from an operator via the terminal 2. The server 1 inputs the training map image into the segmentation model 40 to identify the image area corresponding to the sidewalk, and optimizes parameters such as the weights between neurons so that the identification result approximates the correct image area. In this way, the server 1 generates the segmentation model 40.

[0056] When creating a reference map, the server 1 performs a segmentation process using the segmentation model 40 before inputting a map image to the detection model 50. That is, the server 1 inputs the map image to the segmentation model 40 to identify an image area corresponding to a sidewalk, and generates a map image in which the identified image area, i.e., the sidewalk, is colored. The server 1 inputs the colored map image to the detection model 50 to detect objects including intersections from the map image.

[0057] Fig. 12 is a flowchart showing the procedure of the process of generating the segmentation model 40. The process of generating the segmentation model 40 by machine learning will be described with reference to Fig. 12. The control unit 11 of the server 1 acquires training data for generating the segmentation model 40 (step S301). The training data is data in which information indicating the correct image area of ​​the sidewalk or roadway is associated with a training map image.

[0058] Based on the training data, the control unit 11 generates a segmentation model 40 (second model) that identifies image areas corresponding to specific objects such as sidewalks and roadways when a map image (drawing image) is input (step S302). Specifically, as described above, the control unit 11 generates a semantic segmentation model as the segmentation model 40. The control unit 11 generates the segmentation model 40 by inputting a training map image to the segmentation model 40 to identify image areas corresponding to sidewalks or roadways, and optimizing parameters such as weights between neurons so that the identification result approximates the correct image area. The control unit 11 ends the series of processes.

[0059] 13 is a flowchart showing the procedure of the process of creating a base map according to embodiment 3. After acquiring a map image from the terminal 2 (step S31), the server 1 executes the following process. The control unit 11 of the server 1 performs a segmentation process (step S321) in which the acquired map image (drawing image) is input to the segmentation model 40 to identify image areas corresponding to specific objects such as sidewalks and roadways, and color the identified image areas, i.e., objects such as sidewalks and roadways on the map image. The control unit 11 detects objects by inputting the colored map image (drawing image) to the detection model 50 (step S322). The control unit 11 transitions the process to step S33.

[0060] As described above, according to the third embodiment, by coloring a specific object on a drawing image (for example, a sidewalk or roadway on a map image), the accuracy of object detection by the detection model 50 can be improved.

[0061] (Embodiment 4) In this embodiment, an embodiment will be described in which annotations are received for undetected objects, and the detection model 50 is re-learned to improve the detection accuracy of the objects.

[0062] First, an overview of this embodiment will be described. When creating a reference map, the server 1 inputs a map image (drawing image) to a detection model 50 to detect objects, as in the first embodiment.

[0063] Here, consider a case where an object is not detected. In this embodiment, the server 1 re-learns the detection model 50 so as to be able to detect the object that was not detected.

[0064] For example, the server 1 outputs a map image (see FIG. 4) showing the object detection results to the terminal 2 and displays it to the creator. The server 1 then accepts an annotation operation from the creator via the terminal 2 to add information showing undetected objects to the map image. For example, the server 1 accepts a drawing operation to enclose an arbitrary coordinate range on the map image, and also accepts an input to specify the type of object (e.g., an intersection, a crosswalk, etc.) located within the drawn coordinate range.

[0065] The server 1 may receive annotations from an administrator of the system who creates and manages the detection model 50, in addition to the creator who is a user of the system, and may perform re-learning.

[0066] The server 1 re-learns the detection model 50 according to the annotation. That is, the server 1 updates the detection model 50 based on the map image (drawing image) and the information indicating the object added by the creator. The server 1 optimizes parameters such as the weights between neurons so that the detection result of the object approximates the position and type of the object added by the creator.

[0067] 14 is a flowchart showing the procedure of the process of creating a base map according to embodiment 4. After detecting an object from the map image (step S32), the server 1 executes the following process. The control unit 11 of the server 1 accepts an annotation operation from the creator via the terminal 2 to add information indicating an undetected object to the map image (step S401). For example, the control unit 11 accepts a drawing that encloses an arbitrary coordinate range on the map image, and also accepts an input that specifies the type of object located within the drawn coordinate range. The control unit 11 updates the detection model 50 based on the map image and the information indicating the object added by the creator (step S402). The control unit 11 transitions the process to step S33.

[0068] As described above, according to the fourth embodiment, the detection accuracy of the object can be improved through the operation of this system.

[0069] (Embodiment 5) In this embodiment, a form will be described in which a map search is possible using an object such as an intersection as a search query.

[0070] 15 is a block diagram showing a configuration example of the server 1 according to the embodiment 5. The auxiliary storage unit 14 of the server 1 according to the embodiment stores a map DB 141. The map DB 141 is a database that stores the detection results of the object by the detection model 50 (the position and type of the object).

[0071] 16 is an explanatory diagram showing an example of a record layout of the map DB 141. The map DB 141 includes a map ID column, a coordinate column, and a type column. The map ID column stores a map ID for identifying each map image. The coordinate column and type column store the coordinate range of an object on the map image and the type of the object, respectively, in association with the map ID.

[0072] Here, an outline of this embodiment will be described. As described above, the server 1 inputs a map image (drawing image) in which the positions and types of objects are unknown to the detection model 50, and detects various objects from the map image. In this embodiment, the server 1 stores the detection results of the objects in the map DB 141 in association with the map image. Specifically, as shown in Fig. 16, the server 1 stores the coordinate range of the object on the map image and the type of the object in the map DB 141 in association with the map ID.

[0073] The server 1 provides a map search function based on this map DB 141. For example, the server 1 accepts a search query specifying a type of object from the terminal 2. The server 1 searches the map DB 141 for map images in which the specified type of object exists, and outputs the search results to the terminal 2.

[0074] In this way, by accumulating the results of object detection by the detection model 50, a database for map searches can be constructed.

[0075] 17 is a flowchart showing the procedure of the process of creating a base map according to embodiment 5. After executing the process of step S37, the server 1 executes the following processes. The control unit 11 of the server 1 stores the detection result of the object in step S32 in the map DB 141 in association with the map image (drawing image) (step S501). Specifically, the control unit 11 stores the coordinate range of the object detected from the map image and the type of the object in association with the map image (map ID) in the map DB 141. The control unit 11 ends the series of processes.

[0076] 18 is a flowchart showing the procedure of the map search process. The contents of the map search process based on an object will be described with reference to FIG. The control unit 11 of the server 1 accepts an input from the terminal 2 specifying the type of object to be used as a search query (step S521). The control unit 11 searches the map DB 141 for a map image (drawing image) including the specified type of object (step S522). The control unit 11 outputs the searched map image to the terminal 2 (step S523), and ends the series of processes.

[0077] As described above, according to the fifth embodiment, the detection model 50 can also be used to build a database for map searches.

[0078] The embodiments disclosed herein are illustrative in all respects and should not be considered as limiting. The scope of the present invention is defined by the claims, not by the above meaning, and is intended to include all modifications within the scope and meaning equivalent to the claims.

[0079] The matters described in each embodiment can be combined with each other. In addition, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations regardless of the citation format. Furthermore, the claims use a format in which a claim cites two or more other claims (multi-claim format), but this is not limited to this. They may also be written using a format in which a multiple claim cites at least one other multiple claim (multi-multi claim). [Explanation of symbols]

[0080] 1. Server (information processing device) 11 Control section 12 Main memory 13. Communications Department 14 Auxiliary storage P1 Program 40 Segmentation model (second model) 50 Detection Models 141 Map DB 2. Terminal 21 Control section 22 Main memory 23 Communications Department 24 Display section 25 Input section 26 Auxiliary storage P2 Program

Claims

1. Get the drawing image, A model that has been trained to detect a specific object when a drawing image is input is used to detect the object by inputting the acquired drawing image. A program that causes a computer to carry out processing.

2. The drawing image is a map image. The program according to claim 1.

3. the map image is a road register map image, The acquired road register map image is input to the model to detect road objects including intersections. The program according to claim 2.

4. Obtain information on the construction site, Identifying the distance from the detected intersection to the construction site The program according to claim 3.

5. Accept the type of construction, Identifying a construction area based on the type of construction received and the distance from the identified intersection to the construction site; An object representing the identified construction area and construction site is superimposed on the road register map image. The program according to claim 4.

6. Enlarging the acquired drawing image; The enlarged image of the drawing is input to the model to detect the object. The program according to claim 1.

7. performing a segmentation process for coloring a specific object on the acquired drawing image using a second model that has been trained to identify an image region corresponding to a specific object when the drawing image is input; The colored drawing image is input to the model to detect the object. The program according to claim 1.

8. The drawing image is a road register map image, performing a segmentation process for coloring the sidewalk or roadway on the acquired road register map image using the second model that has been trained to identify an image area corresponding to the sidewalk or roadway when a road register map image is input; The colored road register map image is input to the model to detect road objects. The program according to claim 7.

9. After detecting the object, an input is received to add information indicating the undetected object to the drawing image; The model is updated based on the drawing image and the added information indicating the object. The program according to claim 1.

10. storing the detection result of the object in a storage unit in association with the drawing image; Accepting an input designating the object; The drawing image including the specified object is searched from the storage unit. The program according to claim 1.

11. Get the drawing image, A model that has been trained to detect a specific object when a drawing image is input is used to detect the object by inputting the acquired drawing image. An information processing method in which processing is performed by a computer.

12. An information processing device including a control unit, The control unit: Get the drawing image, A model that has been trained to detect a specific object when a drawing image is input is used to detect the object by inputting the acquired drawing image. Information processing device.

Images