JP7493545B2 - Position estimation device, position estimation method, and program - Google Patents

Description

This disclosure relates to a position estimation device, a position estimation method, and a program.

There are various vegetation, including weeds and trees, on the side of the road. When this vegetation grows and the number of leaves and stems increases, it grows taller, and the trunks and stems become thicker, it may hinder the progress of vehicles traveling on the road. For this reason, in road management work, the presence or absence of vegetation protruding onto the road surface and its positional relationship are visually determined, and then the extent of cutting down the vegetation protruding onto the road surface is determined. For example, Patent Document 1 describes how vegetation is photographed multiple times with a near-infrared camera over a period of time, and the rate of change of the vegetation is calculated from the multiple captured images to select the vegetation to be cut down.

JP 2021-108582 A

However, there is a need to easily detect the relative positions of vegetation to be cut down.

In order to solve the above-mentioned problems and achieve the object, the position estimation device according to the present disclosure includes an area detection unit that detects an object area including an object from captured image data, a distance detection unit that detects depth information of the range captured in the captured image, and an area distance acquisition unit that acquires information on the distance to the object area based on the depth information, and the area detection unit detects a first object area including a first object and a second object area including a second object, and further includes a determination unit that determines whether the first object area and the second object area are on the same perpendicular line in the image and whether the difference in the depth information between the depth of the first object area and the depth of the second object area is within a predetermined range.

In order to solve the above-mentioned problems and achieve the objective, the position estimation method according to the present disclosure includes the steps of detecting a target area including a target from captured image data, detecting depth information of the range captured in the captured image, and acquiring information on the distance to the target area based on the depth information, and the step of detecting the target area further includes the steps of detecting a first target area including a first target and a second target area including a second target, determining whether the first target area and the second target area are on the same perpendicular line in the image and whether the difference in the depth information between the depth of the first target area and the depth of the second target area is within a predetermined range.

In order to solve the above-mentioned problems and achieve the object, the program of the present disclosure causes a computer to execute the steps of detecting a target area including an object from captured image data, detecting depth information of the range captured in the captured image, and acquiring information on the distance to the target area based on the depth information, and the step of detecting the target area further includes the steps of detecting a first target area including a first object and a second target area including a second object, determining whether the first target area and the second target area are on the same perpendicular line of the image and whether the difference in the depth information between the depth of the first target area and the depth of the second target area is within a predetermined range.

According to this disclosure, it is possible to easily detect the positional relationship to an object captured in a captured image.

FIG. 1 is a schematic block diagram of a position estimation system according to the present embodiment. FIG. 2 is a schematic diagram of a vehicle. FIG. 3 is a schematic block diagram of a position estimation device. FIG. 4 is a diagram showing an example of a captured image. FIG. 5 is a diagram showing an example of a result of detection of a target region by the region detection unit. FIG. 6 is a diagram showing an example of a result of detection of depth information by the distance detection unit. FIG. 7 is a diagram showing an example of a result of the area distance acquisition unit acquiring information on the distance to the target object. FIG. 8 is a flowchart illustrating a process flow of the position estimation device.

Embodiments of the present disclosure are described in detail below with reference to the drawings. Note that the present disclosure is not limited to the embodiments described below.

(Location Estimation System)
Fig. 1 is a schematic block diagram of a position estimation system according to this embodiment. As shown in Fig. 1, the position estimation system 1 according to this embodiment includes a vehicle 4, a measurement data acquisition device 6, and a position estimation device 10. The position estimation system 1 calculates a distance to an area including an object captured in a captured image by the position estimation device 10. Hereinafter, the area including the object will be referred to as the target area as appropriate.

In the position estimation system 1, the vehicle 4 captures images of the surroundings while traveling on a road, and transmits the captured image data to the measurement data acquisition device 6. The measurement data acquisition device 6 is, for example, a device (computer) managed by an entity that manages the road. The measurement data acquisition device 6 transmits the image data transmitted from the vehicle 4 to the position estimation device 10. In this way, the position estimation device 10 acquires image data via the measurement data acquisition device 6 and calculates the distance to an object depicted in the image data. However, this is not limited to this, and for example, the position estimation system 1 may not be provided with the measurement data acquisition device 6, and the position estimation device 10 may acquire image data from the vehicle 4. Furthermore, the position estimation device 10 is not limited to acquiring image data captured by the vehicle 4, and may acquire any image data captured and calculate the distance to an object depicted in the image data.

(vehicle)
FIG. 2 is a schematic diagram of a vehicle. As shown in FIG. 2, the vehicle 4 includes a camera 4A and a measuring device 4B. The camera 4A is a camera that captures an image of the surroundings of the vehicle 4. More specifically, in this embodiment, the camera 4A captures an image of an area including a road on which the vehicle 4 is moving. The camera 4A may be realized by a camera including an optical element and an image sensor. The optical element is an element that constitutes an optical system, such as a lens, a mirror, a prism, or a filter. The image sensor is an element that converts light incident through the optical element into an image signal that is an electrical signal. The image sensor may be realized by, for example, a CCD (Charge Coupled Device) sensor or a CMOS (Complementary Metal Oxide Semiconductor) sensor.

The vehicle 4 may also have a position sensor that acquires position information of the vehicle 4. The position information of the vehicle 4 is information indicating the position of the vehicle 4 in a specified coordinate system that is a coordinate system that can define the positions of the vehicle 4 and the road. The specified coordinate system here may be any coordinate system that can define the position of the vehicle 4 and the road, and may be the Earth coordinate system in this embodiment, for example. In this case, the position sensor may be a module for GNSS (Global Navigation Satellite System).

The measuring device 4B is a device that controls the camera 4A to capture an image and records the image data captured by the camera 4A. In other words, the measuring device 4B functions as a data logger that records image data. The measuring device 4B can also be said to be a computer, and includes a control unit 4B1, a memory unit 4B2, and a communication unit 4B3. The control unit 4B1 is a calculation device and includes a calculation circuit such as a CPU (Central Processing Unit). The memory unit 4B2 is a memory that stores various information such as the calculation contents, programs, and image data of the control unit 4B1, and includes at least one of a main memory device such as a RAM (Random Access Memory) and a ROM (Read Only Memory), and a non-volatile memory device such as a flash memory or a HDD (Hard Disk Drive). The program for the control unit 4B1 stored in the memory unit 4B2 may be stored in a recording medium that can be read by the measurement device 4B. The communication unit 4B3 is a communication module that communicates with external devices, and is realized by, for example, a NIC (Network Interface Card) or a wireless LAN (Local Area Network) card.

The control unit 4B1 reads out the program stored in the memory unit 4B2 and executes control of the camera 4A. While the vehicle 4 is traveling on a road, the control unit 4B1 causes the camera 4A to capture an image of an area including the road, and acquires the captured image data.

(Location Estimation Device)
FIG. 3 is a schematic block diagram of the position estimation device. As shown in FIG. 3, the position estimation device 10 is, for example, a computer, and includes a storage unit 12, a control unit 13, a communication unit 14, and a display unit 15. The storage unit 12 is a memory that stores the calculation contents and programs of the control unit 13, and includes at least one of a RAM, a main storage device such as a ROM, and a non-volatile storage device such as a flash memory or a HDD. The program for the control unit 13 stored in the storage unit 12 may be stored in a recording medium that can be read by the position estimation device 10. The communication unit 14 is a communication module that communicates with an external device, and is realized by, for example, a NIC, a wireless LAN card, or the like. The display unit 15 is a display that displays an image. The display unit 15 is controlled by an output unit 25 described later, and the output unit 25 outputs image data to the display unit 15, thereby causing the display unit 15 to display an image corresponding to the image data.

The control unit 13 is a calculation device and includes a calculation circuit such as a CPU. The control unit 13 includes an area detection unit 21, a distance detection unit 22, an area distance acquisition unit 23, a determination unit 24, and an output unit 25. The control unit 13 realizes the area detection unit 21, the distance detection unit 22, the area distance acquisition unit 23, the determination unit 24, and the output unit 25 by reading and executing a program (software) from the storage unit 12, and executes these processes. The control unit 13 may execute these processes using one CPU, or may be provided with multiple CPUs and execute the processes using the multiple CPUs. Also, at least a part of the area detection unit 21, the distance detection unit 22, the area distance acquisition unit 23, the determination unit 24, and the output unit 25 may be realized by hardware.

In this embodiment, the position estimation device 10 detects the distance to an object shown in image data using a first AI (Artificial Intelligence) model that has been machine-learned to learn the correspondence between image data and an area containing the object, and a second AI model that has been machine-learned to learn the correspondence between image data and the depth for each position on the image. In the following, a learning method for the first AI model and the second AI model will be described, and then a distance detection method using these trained AI models will be described. In the following, an example will be given in which the position estimation device 10 trains these AI models by machine learning, but this is not limited thereto, and the position estimation device 10 may acquire these AI models that have been machine-learned by other devices.

(Learning process of first AI model)
The position estimation device 10 trains the first AI model by machine learning using image data and information indicating the target area as teacher data. The information indicating the target area is information indicating the position (coordinates) of the target area in the image. The position estimation device 10 inputs a data set in which the image data is used as an input value and the information indicating the target area is used as an output value to the first AI model as teacher data. It is preferable that the position estimation device 10 prepares a plurality of data sets consisting of image data and information indicating the target area, and inputs each of the plurality of data sets to the first AI model. As a result, the first AI model trains the correspondence between the image data and the target area by machine learning, and becomes a model (program) that outputs information indicating the target area when the image data is input. In this embodiment, multiple types of target areas are set. Therefore, when image data is input, the first AI model outputs, as information indicating the target area, the type of the target area included in the image (a label indicating the type of the target area) and information indicating the position of the target area in the image, among the multiple types of target areas. Any image data may be used for the machine learning of the first AI model, and for example, image data of an area including a road captured in advance by the camera 4A may be used. Information specifying the target area may be appropriately set, for example, by a user.

Thus, the first AI model is a so-called supervised AI model, but is not limited to this and may also be an unsupervised AI model.

The first AI model may be any AI model that can determine the label of the data based on the input data, for example, a CNN (Conventional Neural Network) model. The first AI model may be realized by semantic segmentation, which is a method of classifying pixels of image data according to which object class they belong to, instant segmentation, which is a method of classifying pixels of image data according to which object class and which example they belong to, or panoptic segmentation, which is a combination of semantic segmentation and instant segmentation, in which all pixels of image data are assigned a label and, for countable objects, each of them is recognized individually.

(Learning process of second AI model)
The position estimation device 10 trains the second AI model by machine learning using image data and depth information as teacher data. Depth information is the distance from the detection source to the target. The position estimation device 10 inputs a data set in which image data is used as an input value and depth information for each pixel (position on the image) in the image data is used as an output value to the second AI model. It is preferable that the position estimation device 10 prepares a plurality of data sets consisting of image data and depth information for each pixel, and inputs each of the plurality of data sets to the second AI model. As a result, the second AI model trains the correspondence between image data and depth information for each pixel by machine learning, and becomes a model (program) that outputs depth information for each pixel when image data is input.

The image data used for the machine learning of the second AI model may be any image data, but for example, image data of an area including a road that has been captured in advance by the vehicle 4 may be used. Also, the depth information is detected in advance. In this case, for example, a depth detection sensor is provided in the vehicle 4, and while the camera 4A captures an image, the depth detection sensor detects depth information of the range captured in the image captured by the camera 4A. The depth information detected by the depth detection sensor can also be said to be the distance from the vehicle 4 (depth detection sensor) that is the detection source to the target. The depth detection sensor may be realized, for example, using LiDAR (Light Detection and Ranging). LiDAR irradiates a laser light to the target and measures the time until the reflected light reflected by the laser light hitting the target is detected, thereby measuring the distance to the target and the direction in which the target is located. Also, the depth detection sensor may be a stereo camera. The stereo camera captures images of the target using two cameras and measures the distance to the target using triangulation. The second AI model can also be any AI model.

The following describes how the position estimation device 10 detects the distance to an object.

(Detection of target area)
FIG. 4 is a diagram showing an example of a captured image, and FIG. 5 is a diagram showing an example of a result of the area detection unit detecting the target area. The area detection unit 21 detects the target area from the captured image data. Specifically, the area detection unit 21 acquires image data captured by the camera 4A of the vehicle 4 from the measurement data acquisition device 6, and detects the target area from the acquired image data. In this embodiment, the area detection unit 21 detects the target area included in the image captured by the camera 4A by inputting the image data captured by the camera 4A to a first AI model that has been machine-learned. In the first AI model, the image data is input as input data, a calculation is performed, and information on the target area included in the image data is output as output data. More specifically, the first AI model outputs the type of the target area included in the input image data and the position of the target area in the image as information on the target area. The area detection unit 21 detects the target area by acquiring information on the target area output from the first AI model.

In other words, the area detection unit 21 detects the object (target area) from the image by detecting the object along the boundary of the target area included in the image data. The area detection unit 21 also classifies the type of the detected object by detecting each type of object. The area detection unit 21 may also generate an extracted image that shows only the detected target area by extracting image data of the detected target area from the image data captured by the camera 4A. For example, in this case, the extracted image may be generated by performing a binarization process that sets the gradation value of the target area to greater than zero while setting the gradation value of the area other than the target area to zero, or by performing a process that sets the gradation value of the area other than the target area to zero while varying the gradation value for each type of target area.

As shown in FIG. 4, an example is taken in which image data of an image PA including a road A1 and vegetation A2 is acquired. In this case, for example, the first AI model may have already machine-learned the feature amount of a first target area whose type of object is a road and the feature amount of a second target area whose type of object is vegetation. In this case, the area detection unit 21 detects information on the first target area including the road and information on the second target area including vegetation from the image data of the image PA, and generates an extracted image PB (extracted image data) that includes only the first target area CL1a corresponding to the road A1 and the second target area CL2a corresponding to the vegetation A2, as shown in FIG. 4. Note that in FIG. 4, vegetation and roads are detected as the types of target areas, but the types of target areas are not limited to these and may be any. In addition, the number of types of target areas detected (classified) is not limited to two and may be any, and may be one or three or more.

The area detection unit 21 is not limited to detecting the target area using the first AI model, but may detect the target area from the image data using any method.

(Depth information detection)
FIG. 6 is a diagram showing an example of the result of the distance detection unit detecting depth information. The distance detection unit 22 detects depth information of the range captured in the image captured by the camera 4A (the image in which the area detection unit 21 performed the object detection). Specifically, the distance detection unit 22 detects depth information for each position (pixel) of the image captured by the camera 4A. In this embodiment, the distance detection unit 22 detects depth information using image data captured by the camera 4A (the image data in which the area detection unit 21 performed the object detection). More specifically, the distance detection unit 22 detects depth information for each position of the image data by inputting the image data captured by the camera 4A into a second AI model that has been machine-learned. In the second AI model, the image data is input as input data, a calculation is performed, and the depth information for each position of the image data is output as output data. The area detection unit 21 detects depth information by acquiring the depth information output from the second AI model.

The distance detection unit 22 may generate a distance image in which the depth information for each detected position is mapped in a matrix. In this case, for example, the distance detection unit 22 generates the distance image by varying the gradation value for each depth magnitude. The example of FIG. 6 shows a distance image PC in which the output depth information is set for each position of the image data so that the hatching becomes lighter (the gradation value becomes higher) from the farthest depth to the closer depth. In other words, the darkly hatched areas in the center of FIG. 6 indicate that the distance from the vehicle 4 is far, and the lightly hatched areas at the bottom of FIG. 6 indicate that the distance from the vehicle 4 is close. In this way, the distance detection unit 22 estimates the depth information for each position of the image data.

The distance detection unit 22 is not limited to detecting depth information based on image data using the second AI model, and may detect depth information in any manner. For example, a depth detection sensor may be provided in the vehicle 4, and while the camera 4A is capturing an image, the depth detection sensor may detect depth information of the range captured in the image captured by the camera 4A. The distance detection unit 22 may acquire the depth information detected by the depth detection sensor.

(Calculating the distance to the object)
FIG. 7 is a diagram showing an example of the result of the area distance acquisition unit acquiring information on the distance to the target object. The area distance acquisition unit 23 acquires information on the distance to the target area (target object) based on the depth information. The information on the distance to the target area indicates the distance from the image capture source (vehicle 4) to the target object. Specifically, the area distance acquisition unit 23 acquires information on the distance to the target area (target object) by extracting depth information at the position of the target area from the depth information for each position of the image data detected by the distance detection unit 22. That is, the area distance acquisition unit 23 acquires depth information at the same position as the position of the target area detected by the area detection unit 21 from the depth information for each position of the image data as information on the distance to the target area. When multiple target areas are detected, the area distance acquisition unit 23 acquires information on the distance to the target area for each target area.

The area distance acquisition unit 23 may generate target image data in a display mode in which areas other than the target area are not displayed and the target area indicates the distance to the target area. For example, the area distance acquisition unit 23 may generate target image data such that the gradation value at the position of the detected target area is the gradation value set in the distance image set by the distance detection unit 22, and the gradation value outside the target area is constant (for example, zero). In the example of FIG. 7, a target image P is generated in which the first target area CL1a and the second target area CL2a have gradation values for each depth.

In this way, the position estimation device 10 according to this embodiment acquires information on the distance to the target area based on the target area detected from the image data and the depth information of the range captured in the image. Therefore, it is possible to easily and accurately detect the distance to the target of interest captured in the captured image. Furthermore, in this embodiment, since the depth information is detected from the image data, the distance to the target of interest can be easily calculated from the image data alone without using a separate sensor for detecting depth. In addition, since a separate sensor is not used, the device can be made smaller and less expensive. Furthermore, since the distance can be calculated from the depth information obtained by the monocular camera, versatility is improved compared to the case of using LiDAR or a stereo camera.

(Determination of Positional Relationship Between Target Areas)
Furthermore, the position estimation device 10 may determine the positional relationship between the target areas by the determination unit 24. In the following, the determination of the positional relationship between the first target area and the second target area will be described as an example.

The determination unit 24 determines whether a position in the first target area and a position in the second target area are on the same perpendicular line of the image, and the difference in depth between the position in the first target area and the position in the second target area is within a predetermined range. For example, the determination unit 24 extracts positions in the first target area that are on the same perpendicular line to the positions in the second target area. On the same perpendicular line here refers to positions being lined up along the vertical direction in the coordinate system of the image. However, this is not limited to positions strictly overlapping when viewed vertically, and positions may be determined to be on the same perpendicular line even if they are lined up vertically and the horizontal deviation between the positions is within a predetermined value.

The determination unit 24 determines whether the difference between the depth at the extracted position in the first target area and the depth at the position in the second target area on the same perpendicular line as that position is within a predetermined range. If the difference in depth is within the predetermined range, the determination unit 24 determines that the positions are lined up vertically in the real world. On the other hand, if the difference in depth is not within the predetermined range, the determination unit 24 determines that the positions are lined up vertically in the real world. The determination unit 24 performs the same process for each position in the first target area. The example in FIG. 7 shows an example in which it is determined whether there is a position in the second target area on the same perpendicular line for each of positions A, B, and C in the first target area CL1a, and whether the difference in depth is within a predetermined range.

In the above description, positions in the first and second target areas that are on the same perpendicular line are extracted, and then it is determined whether the depth difference is within a predetermined range. However, the order of processing may be reversed. That is, the determination unit 24 may extract pairs of positions in the first and second target areas whose depth difference is within a predetermined range, and determine whether these positions are on the same perpendicular line. In this case, if positions whose depth difference is within the predetermined range are on the same perpendicular line, the determination unit 24 determines that the positions are lined up vertically in the real world. On the other hand, if positions whose depth difference is within the predetermined range are not on the same perpendicular line, the determination unit 24 determines that the positions are not lined up vertically in the real world.

The output unit 25 outputs the determination result by the determination unit 24. For example, the output unit 25 may output positional relationship information indicating the first target area and the second target area that include the positions that are determined to be on the same perpendicular line. The positional relationship information may be, for example, image data indicating the first target area and the second target area that include the positions that are determined to be on the same perpendicular line, or data indicating the positions (earth coordinates) of the first target area and the second target area that include the positions that are determined to be on the same perpendicular line. In this way, by outputting the positional relationship information indicating the first target area and the second target area, the user can be made to recognize the positional relationship between the first target area and the second target area. For example, the positional relationship between the first target area and the second target area makes it possible to recognize that there is vegetation vertically above the road, and therefore, for example, vegetation that is on the road and is to be cut down can be appropriately detected.

The determination unit 24 may also calculate the distance from a position on the first target area, which is assumed to be on the same perpendicular line, to a position on the second target area. In this case, for example, the determination unit 24 acquires the correspondence between the distance between pixels in the coordinate system of the image, calculated from the imaging conditions of the camera 4A (e.g., the angle of view, etc.), and the actual distance. Then, the determination unit 24 calculates the distance from a position on the first target area to a position on the second target area based on the distance from a position on the first target area to a position on the second target area in the coordinate system of the image and this correspondence. The determination unit may also output this distance information as positional relationship information. This allows, for example, the distance between a tree on a road and the road to be grasped, so that, for example, a tree to be cut down can be more appropriately selected. In other words, when trees whose distance from the road is equal to or less than a predetermined value are to be cut down, the tree to be cut down can be more easily selected from this distance information.

In the above example, the positional relationship between a tree and a road was determined, but the subject of the positional relationship determination is not limited to this and may be anything.

(Processing flow)
Next, the process flow of the position estimation device 10 described above will be described. FIG. 8 is a flowchart for explaining the process flow of the position estimation device. As shown in FIG. 8, the position estimation device 10 acquires image data captured by the camera 4A (step S10), detects a target area from the acquired image data by the area detection unit 21 (step S12), and detects depth information from the acquired image data by the distance detection unit 22 (step S14). The position estimation device 10 acquires information on the distance to the target area from the information on the detected target area and the depth information by the area distance acquisition unit 23 (step S16), and determines the positional relationship between the first target area and the second target area based on the information on the distance to the target area (step S18).

As described above, the position estimation device 10 according to this embodiment includes an area detection unit 21 that detects a target area including a target object from captured image data, a distance detection unit 22 that detects depth information of the range captured in the captured image, and an area distance acquisition unit 23 that acquires information on the distance to the target area based on the depth information. Therefore, according to this embodiment, it is possible to easily and accurately detect the distance to a target object of interest captured in a captured image.

The area detection unit 21 detects a first object area including a first object and a second object area including a second object. The position estimation device 10 further includes a determination unit 24 that determines whether the first object area and the second object area are on the same perpendicular line of the image and whether the difference in depth information between the depth of the first object area and the depth of the second object area is within a predetermined range. According to this embodiment, it is possible to easily and accurately detect the positional relationship between the first object area and the second object area. Furthermore, it is possible to easily and accurately detect whether the first object area and the second object area are aligned vertically.

The first object may be a road, and the second object may be vegetation. Therefore, according to this embodiment, it is possible to easily and accurately detect whether vegetation such as trees is located on the road.

The area detection unit 21 detects the target area by inputting the image data into a first AI model that has learned the correspondence between image data and areas containing the target object through machine learning. The distance detection unit 22 detects depth information by inputting the image data into a second AI model that has learned the correspondence between image data and the depth for each position in the image through machine learning. In this way, according to this embodiment, the target area and depth information are detected from the image data using an AI model. Therefore, the distance to the target object of interest can be easily detected from the image data without using a separate sensor for detecting depth.

The area distance acquisition unit 23 generates the target image data so that areas other than the target area are not displayed and the target area is displayed in a manner that indicates the distance to the target area. According to this embodiment, by generating the target image data, it is possible to allow the user to properly recognize the distance to the target object.

Although the embodiment of the present invention has been described above, the embodiment is not limited to the contents of this embodiment. The above-mentioned components include those that a person skilled in the art can easily imagine, those that are substantially the same, and those that are within the so-called equivalent range. Furthermore, the above-mentioned components can be combined as appropriate. Furthermore, various omissions, substitutions, or modifications of the components can be made without departing from the spirit of the above-mentioned embodiment.

4 Vehicle 4A Camera 10 Position estimation device 21 Area detection unit 22 Distance detection unit 23 Area distance acquisition unit 24 Determination unit 25 Output unit

Claims (5)

a region detection unit that detects a target region including a target object from the captured image data;
a distance detection unit that detects depth information of a range captured in the captured image;
a region distance acquisition unit that acquires information on a distance to the target region based on the depth information;
Including,
The area detection unit detects a first object area including a first object and a second object area including a second object,
a determination unit that determines whether the first target region and the second target region are on the same perpendicular line of an image and whether a difference in depth between the first target region and the second target region in the depth information is within a predetermined range;
Location estimation device. The area detection unit detects the object area by inputting the image data into a first AI model that has machine-learned a correspondence relationship between image data and an area including an object.
The position estimation device according to claim 1 . The distance detection unit detects the depth information by inputting the image data to a second AI model that has machine-learned a correspondence between image data and a depth for each position in the image.
The position estimation device according to claim 1 or 2. detecting a target area including a target object from the captured image data;
Detecting depth information of a range captured in the captured image;
obtaining information on a distance to the target region based on the depth information;
Including,
In the step of detecting the object region, a first object region including a first object and a second object region including a second object are detected,
The method further includes a step of determining whether the first target region and the second target region are on the same perpendicular line of an image, and whether a difference in depth between the first target region and the second target region in the depth information is within a predetermined range.
Location estimation methods. detecting a target area including a target object from the captured image data;
Detecting depth information of a range captured in the captured image;
obtaining information on a distance to the target region based on the depth information;
Run the following on your computer:
In the step of detecting the object region, a first object region including a first object and a second object region including a second object are detected,
The method further includes a step of determining whether the first target region and the second target region are on the same perpendicular line of an image, and whether a difference in depth between the first target region and the second target region in the depth information is within a predetermined range.
program.

Images