Object distance estimation device, object distance estimation method, and program

Abstract

The present invention enables the estimation of the distance to an object even when a part of the object is hidden, without increasing the processing load. An acquisition unit (11) acquires an image captured by a camera. A recognition unit (12) recognizes an object included in the acquired image. A detection unit (13) detects the size of a predetermined portion of the recognized object. A determination unit (14) determines whether a part of the object is hidden in the acquired image. A distance estimation unit (15) estimates the distance to the recognized object. When the determination unit (14) determines that a portion of the object in contact with the road is hidden, the distance estimation unit (15) estimates the distance to the object on the basis of the distance to the object estimated when a part of the object is not hidden and the size of the predetermined portion of the object, and the size of the predetermined portion of the object detected by the detection unit (13).

Description

Object distance estimation apparatus, object distance estimation method, and program

[0001] The present invention relates to an object distance estimation apparatus, an object distance estimation method, and a program.

[0002] As a related art, Patent Document 1 discloses an object detection apparatus. The object detection apparatus described in Patent Document 1 detects the position and region of an object represented in an image acquired from a camera. The object detection apparatus detects whether or not a predetermined part of the outer shape of each object represented in the image is hidden by another object. When it is detected that a predetermined part of the outer shape of the detected object is hidden, the object detection apparatus detects the position and size of the object based on the position of the part of the outer shape of the object that is hidden by another object. In particular, the object detection apparatus estimates the outer shape of the object when assuming that the entire object is not hidden based on the position of the hidden part of the object, and detects the position and size of the target object based on the estimated outer shape.

[0003] Japanese Unexamined Patent Publication No. 2020-86545

[0004] Consider a case where a camera is installed at an intersection or the like, and the distance from the position of the camera to a road object such as a vehicle or a pedestrian is estimated from an image captured by the camera. When the camera is fixed, the distance between the camera and the object can be estimated based on the position of the object on the image. However, in an image, if a part of the object, particularly a portion where the object is in contact with the ground, that is, the road, is hidden by another object or the like, the distance cannot be estimated correctly.

[0005] In Patent Document 1, the object detection apparatus estimates the outer shape of an object when assuming that the entire object is not hidden, and can detect the position of the target object based on the estimated outer shape. However, when estimating the distance to an object, estimation of the outer shape of the object is not necessarily required. When estimating a distance, since it is only required to provide information on the distance of the object at each time, the process of estimating the outer shape is redundant, and increases the amount of computation and processing time.

[0006] In view of the above circumstances, one of the objectives of the present invention is to provide an object distance estimation device, an object distance estimation method, and a program that can estimate the distance to an object even when part of the object is hidden, without increasing the processing load.

[0007] The object distance estimation device according to the first aspect of this embodiment includes an acquisition unit that acquires an image captured by a camera, a recognition unit that recognizes an object included in the acquired image, a determination unit that determines whether or not a part of the object is hidden in the acquired image, a detection unit that detects the size of a predetermined part of the recognized object, and a distance estimation unit that estimates the distance to the recognized object. The distance estimation unit stores the estimated distance and the detected size, and if the determination unit determines that a part of the object that is in contact with the road is hidden, the distance estimation unit estimates the distance to the object based on the stored distance and size and the size detected by the detection unit.

[0008] A second aspect of this embodiment of the object distance estimation method includes acquiring an image captured by a camera, recognizing an object included in the acquired image, determining whether a part of the object is hidden in the acquired image, detecting the size of a predetermined part of the recognized object, estimating the distance to the recognized object, storing the estimated distance and the detected size, and, if it is determined that a part of the object that is in contact with the road is hidden, estimating the distance to the object based on the stored distance and size and the detected size.

[0009] The program according to the third aspect of this embodiment acquires an image captured by a camera, recognizes an object included in the acquired image, determines whether a part of the object is hidden in the acquired image, detects the size of a predetermined part of the recognized object, estimates the distance to the recognized object, stores the estimated distance and the detected size, and if it is determined that the part of the object that is in contact with the road is hidden, causes the computer to perform a process to estimate the distance to the object based on the stored distance and size and the detected size.

[0010] According to this embodiment, the distance to an object can be estimated even when part of the object is hidden, without increasing the processing load.

[0011] This is a block diagram showing an example configuration of the object distance estimation device according to the present invention. This is a schematic diagram showing an example of the object recognition result in the recognition unit. This is a schematic diagram showing another example of the object recognition result in the recognition unit. This is a flowchart showing the operation procedure of the object distance estimation device. This is a schematic diagram showing an example of the object recognition result in the recognition unit.

[0012] Hereinafter, exemplary embodiments of the present invention will be described with reference to the drawings. Note that the following description and drawings have been omitted and simplified as appropriate for clarity of explanation. Furthermore, in the following drawings, the same elements and similar elements are denoted by the same reference numerals, and redundant explanations have been omitted where necessary.

[0013] A first embodiment will be described. Figure 1 is a block diagram showing an example configuration of an object distance estimation device 10 according to the present invention. The object distance estimation device 10 includes an acquisition unit 11, a recognition unit 12, a detection unit 13, a determination unit 14, a distance estimation unit 15, and an output unit 18. The object distance estimation device 10 is configured as a device including, for example, one or more memories and one or more processors. At least a part of the functions of the object distance estimation device 10 can be realized by the processor executing processing according to instructions read from the memory.

[0014] The program described above, when loaded into a computer, includes a set of instructions (or software code) for causing the computer to perform one or more of the functions described in the embodiments. The program may be stored on a non-temporary computer-readable medium or a physical storage medium. Examples, but not limited to, include Random Access Memory (RAM), Read Only Memory (ROM), flash memory, Solid State Drive (SSD), or other memory technologies, Compact Disc (CD), Digital Versatile Disc (DVD), Blu-ray® disc, or other optical disc storage, magnetic cassette, magnetic tape, magnetic disk storage, or other magnetic storage devices. The program may be transmitted over a temporary computer-readable medium or a communication medium. Examples, but not limited to, include temporary computer-readable medium or a communication medium that includes electrically, optically, acoustically, or otherwise propagating signals.

[0015] The object distance estimation device 10 can be used to estimate the position of an object, for example, the distance from the camera 30 to the object, using an image captured by the camera 30, and to provide the estimated distance to an external device. The object distance estimation device 10 uses an image captured by the camera 30 to recognize an object on the road, such as a vehicle or a person, and estimates the distance from the camera 30 to the recognized object. In the following embodiments, an example in which the object whose distance is estimated is a vehicle will be mainly described. However, in the present invention, the object whose distance is estimated by the object distance estimation device 10 is not limited to a vehicle.

[0016] Camera 30 photographs objects that are the subject of distance estimation. Camera 30 is installed, for example, near an intersection. Camera 30 photographs, for example, vehicles approaching or moving away from an intersection on a road, for example, from the front or rear of the vehicle. The object distance estimation device 10 aims to provide vehicle information as far away as possible, that is, at an early stage, in order to ensure traffic safety. Camera 30 is installed, for example, to photograph distant vehicles, so as to prevent vehicles from being hidden by obstacles such as buildings and so as to prevent multiple vehicles from overlapping.

[0017] The camera 30 may be installed, for example, on a pole installed along the roadside. The pole may be a multi-functional pole on which equipment having multiple functions other than the camera 30 is installed. In one embodiment, the object distance estimation device 10 may be installed on a multi-functional pole. In that case, the multi-functional pole has the function of providing distance estimation results for objects such as vehicles. The camera 30 is configured as, for example, a far-infrared camera. The camera 30 may be a near-infrared camera or a visible-light camera.

[0018] The acquisition unit 11 acquires images captured by the camera 30. The recognition unit 12 recognizes objects, such as vehicles, included in the images acquired by the acquisition unit 11. The recognition unit 12 recognizes one or more vehicles included in the images using known object detection or object recognition methods. The recognition unit 12 detects the bounding rectangle of the recognized object as a recognition frame. The recognition unit 12 generates coordinate information of the recognition frame as position information of the recognized object within the image. The coordinate information includes, for example, the coordinates of the top left and bottom right of the recognition frame. If the same object was included in past images, the recognition unit 12 assigns the same identifier to that object and tracks the object in chronological order.

[0019] The detection unit 13 detects the size of a predetermined portion of the recognized object. In the first embodiment, the detection unit 13 detects the overall width of an object, such as a vehicle, as the size of a predetermined portion of the object. For example, the detection unit 13 detects the length of one side of the object's recognition frame as the size of a predetermined portion of the object. For example, the detection unit 13 detects the length of the side of the recognition frame in the width direction of the object as the size of the object.

[0020] The determination unit 14 determines whether a part of an object is hidden in the image acquired by the acquisition unit 11. The determination unit 14 determines whether the part of an object that is in contact with the ground or road, i.e., the contact point, such as the tires of a vehicle, is hidden by an obstacle such as another object. For example, if the lower area of ​​an object is obstructed by another object on the camera 30 side and the contact point of the object is not visible in the image, the determination unit 14 determines that a part of the object is hidden. If multiple objects are recognized by the recognition unit 12, the determination unit 14 determines whether the contact point is hidden for each object.

[0021] The distance estimation unit 15 estimates the distance to the object recognized by the recognition unit 12. The distance estimation unit 15 includes a first estimation unit 16 and a second estimation unit 17. The first estimation unit 16 calculates or estimates the distance from the camera 30 to the recognized object based on the object's ground contact position in the image. The first estimation unit 16 has, for example, a correspondence table (not shown) that defines the correspondence between the position in the image and the actual distance to the road surface. The first estimation unit 16 estimates the distance using the correspondence table from the object's ground contact position in the image, that is, the position in the image where the lower end of the object is in contact with the ground.

[0022] The first estimation unit 16 estimates the distance from the camera 30 to the object when the determination unit 14 determines that a part of the object, particularly the ground contact point, is not obscured. The first estimation unit 16 stores the position and size of the object's recognition frame and the estimated distance to the object based on the ground contact point of the object's recognition frame in a storage unit (not shown). For example, for each object, the first estimation unit 16 stores, for example, the position of the recognition frame, the length of one side of the recognition frame detected by the detection unit 13, and the estimated distance. The first estimation unit 16 only needs to be able to estimate the distance from an image in which the ground contact point is not obscured, and the algorithm used for distance estimation in the first estimation unit 16 is not limited to a specific algorithm.

[0023] The second estimation unit 17 estimates the distance to the object based on the past distance estimation results in the image, the size of a predetermined part of the object, and the size detected by the detection unit 13, when the determination unit 14 determines that a part of the object, particularly the ground contact point, is hidden. In the first embodiment, the second estimation unit 17 estimates the distance to the object based on the length of one side of the recognition frame detected by the detection unit 13 and the length of one side of the recognition frame in the image used for distance estimation, which is stored by the first estimation unit 16. In other words, the second estimation unit 17 estimates the distance to the object based on the length of one side of the detected recognition frame and the length of one side of the recognition frame of the object recognized in the image in which the determination unit 14 determined that a part of the object is not hidden.

[0024] Figure 2 is a schematic diagram showing an example of the object recognition result by the recognition unit 12. In this example, the object to be recognized is a vehicle 50. The recognition unit 12 detects the vehicle 50 from the image and generates a bounding rectangle surrounding the area in which the vehicle 50 is detected as the recognition frame 60. In Figure 2, the recognition frame 60 surrounds the entire vehicle 50. In the example shown in Figure 2, the determination unit 14 determines that the ground contact position 55 of the vehicle 50 is not hidden. The detection unit 13 detects the length of the sides in the width direction of the recognition frame of the vehicle 50, i.e., the width W1, as the width of the vehicle 50. Note that the shape of the recognition frame 60 is not limited to a bounding rectangle; it may be a circle, an ellipse, or any other shape.

[0025] The first estimation unit 16 estimates the distance L1 from the camera 30 to the vehicle 50 on the road using the vehicle's ground contact position and a correspondence table. The first estimation unit 16 stores the detected width W1 and the distance L1 estimated based on the position of the ground contact position 55.

[0026] Figure 3 is a schematic diagram showing another example of the object recognition result by the recognition unit 12. The recognition unit 12 detects the vehicle 50 from the image and generates a bounding rectangle surrounding the area where the vehicle 50 is detected as the recognition frame 61. In the example in Figure 3, the vehicle 50 is located behind another vehicle 51 that is closer to the camera 30, and the vehicle 50's ground contact point is not included in the recognition frame 61 of the vehicle 50. In the example shown in Figure 3, the determination unit 14 determines that the ground contact point of the vehicle 50 is hidden.

[0027] Even if the lower part of vehicle 50 is hidden by another vehicle 51 and the ground contact point of vehicle 50 is not visible, the upper end, which is the roof portion of vehicle 50, is visible in the image. Therefore, the width of vehicle 50 in the image can be obtained. The detection unit 13 detects the length of the side in the width direction of the recognition frame 61 of vehicle 50, i.e., the width W2, as the width of vehicle 50. The second estimation unit 17 estimates the distance from camera 30 to vehicle 50 using a geometric method, using the width W2 of vehicle 50 in the image, the width W1 of vehicle 50 in the image that is stored when the entire vehicle is visible, and the estimated distance L1 to vehicle 50. Specifically, the second estimation unit 17 estimates the current distance L2 to vehicle 50 using the following equation 1. (Equation 1) L2 = L1 × (W1 / W2)

[0028] The output unit 18 outputs distance information. The distance information includes the distance estimated by the distance estimation unit 15. The distance information may also include, for each object recognized by the recognition unit 12, the coordinate information of the recognition frame and the estimated distance to the object. The output unit 18 outputs the distance information to a device that will use the distance information.

[0029] For example, the output unit 18 outputs distance information to a monitoring device used for monitoring road conditions. The monitoring device displays, for example, a recognition frame and the estimated distance on the image captured by the camera 30. The output unit 18 may also transmit the estimated distance for an object, such as a vehicle entering the intersection, to a vehicle or person on a road intersecting the road where the camera 30 is installed, using wireless communication such as V2X (Vehicle to Everything) communication. A device using the distance information can use the distance from the camera to the object included in the distance information to estimate the speed of the object, such as a vehicle, and calculate the estimated time it will take for the object to reach the intersection.

[0030] Next, the operating procedure will be explained. Figure 4 is a flowchart showing the operating procedure of the object distance estimation device 10. The acquisition unit 11 acquires an image from the camera 30 (step A1). The recognition unit 12 determines whether or not an object has been recognized in the image acquired in step A1 (step A2). If the image does not contain an object, the recognition unit 12 determines that no object has been recognized (NO in step A2) and terminates the process.

[0031] If the detection unit 13 determines that the recognition unit 12 has recognized an object (YES in step A2), it detects the size of a predetermined portion of the recognized object (step A3). In step A3, the detection unit 13 detects, for example, the length of the object in the width direction of the recognition frame as the size of the predetermined portion.

[0032] The determination unit 14 determines whether the ground contact point of the recognized object is hidden or not (step A4). If the determination unit 14 determines that the ground contact point is not hidden (NO in step A4), the distance estimation unit 15, specifically the first estimation unit 16, calculates the distance to the recognized object based on the ground contact point (step A5). The first estimation unit 16 stores the position of the object's recognition frame, the size of the predetermined portion detected in step A3, and the distance estimated in step A4 as past information (step A6). The output unit 18 outputs distance information including the distance estimated by the distance estimation unit 15 (step A10).

[0033] If the determination unit 14 determines that the ground contact position is hidden (YES in step A4), the distance estimation unit 15, in turn, has the second estimation unit 17 acquire the past recognition frame position of the object, the size of a predetermined part of the object, and the previously estimated distance stored in step A6 (step A7). The second estimation unit 17 compares the position of the recognition frame of the object recognized from the current image with the past recognition frame position acquired in step A7 to determine whether the recognition frame has moved or not (step A8). If the second estimation unit 17 determines that the recognition frame has not moved (NO in step A8), the process ends. Note that after the processing in step A7, the process of determining whether the recognition frame has moved in step A8 may be omitted, and the process may proceed to step A9.

[0034] If the second estimation unit 17 determines that the recognition frame has moved (YES in step A8), it estimates the distance to the object using a geometric method based on the size of the predetermined part in the past, the distance estimated in the past, and the size of the predetermined part detected in step A3 (step A9). The process then proceeds to step A10, where the output unit 18 outputs distance information including the distance estimated by the distance estimation unit 15.

[0035] Furthermore, if the first estimation unit 16 performs distance estimation for the same object multiple times, the first estimation unit 16 does not necessarily have to store the size of the predetermined portion and the estimated distance in step A6 each time distance estimation is performed. For example, the first estimation unit 16 compares the distance already stored with the distance estimated in the current estimation to determine whether the object is approaching the camera 30. The farther the distance to the object, the smaller the size of the object in the image becomes, and the larger the error in the distance estimated by the second estimation unit 17 is thought to be. The first estimation unit 16 may update the size of the predetermined portion and the estimated distance stored when the object approaches the camera 30. In this case, the second estimation unit 17 can use the size of the predetermined portion and the estimated distance when the object is close to the camera 30 to perform distance estimation, and the accuracy of distance estimation can be improved.

[0036] For example, if traffic congestion occurs and the ground contact point of the recognition frame of an object, such as a vehicle, is obscured, the first estimation unit 16 cannot estimate the distance to that object. In the first embodiment, if the ground contact point of the recognition frame is obscured and the distance cannot be calculated by the first estimation unit 16, the second estimation unit 17 estimates the distance to the object using the ratio of the acquired width of the object to the width of the object at the time of past distance estimation. In the first embodiment, even if the ground contact point is obscured, if the width of the object, for example the length of one side of the roof of a vehicle, can be obtained, the second estimation unit 17 can estimate the distance to the object. Therefore, even if multiple objects overlap in the image captured by the camera 30 due to traffic congestion, distance information can be obtained for more objects compared to when only the first estimation unit 16 is used. In the first embodiment, since the distance can be estimated according to the width ratio, the amount of computation required is less and the distance can be estimated with less processing load compared to Patent Document 1, which estimates the obscured part of the recognition frame.

[0037] Next, a second embodiment will be described. In the second embodiment, the recognition unit 12 recognizes an object and further recognizes a predetermined area of ​​the recognized object. If the object to be recognized is a vehicle, the recognition unit 12 recognizes, for example, the area of ​​the vehicle's side mirror as the predetermined area. The detection unit 13 detects the size of the side mirror area recognized by the recognition unit 12 as the size of a predetermined part of the object. The detection unit 13 detects, for example, the length of at least one side of the recognition frame of the side mirror area, or the area of ​​the recognition frame, as the size of the predetermined part.

[0038] In the second embodiment, when the first estimation unit 16 estimates the distance to an object from an image in which the ground contact position is not obscured, it stores the position of the object's recognition frame, the size of a predetermined area, such as the side mirror area, and the estimated distance. When the determination unit 14 determines that the ground contact position of the object is obscured, the second estimation unit 17 estimates the distance to the object based on the size of the predetermined area and the ratio of the size of the predetermined area stored. The distance estimation in the second estimation unit 17 may be the same as the distance estimation in the first embodiment, except that the size of a predetermined area, such as the side mirror area, is used instead of the width of the object.

[0039] Figure 5 is a schematic diagram showing an example of the object recognition result by the recognition unit 12. In the example in Figure 5, the recognized object is a vehicle 50. The vehicle 50 is located behind another vehicle 51 and 52, which are in front of the camera 30. In this case, the ground contact point of the vehicle 50 is not included in the recognition frame 62 of the vehicle 50. Also, part of the roof of the vehicle 50 is obscured by another tall vehicle 52, such as a bus, so the overall width of the vehicle 50 cannot be obtained from the image.

[0040] If another tall vehicle 52, such as a bus, is located in front of vehicle 50, even if the entire roof of vehicle 50 is not visible, it is highly likely that one of the side mirrors of vehicle 50 is still visible. Therefore, the size of the area of ​​the side mirror of vehicle 50 can be obtained. In the second embodiment, the recognition unit 12 generates a recognition frame 70 as a bounding rectangle surrounding the area of ​​the side mirror of vehicle 50. The detection unit 13 detects the length of one side of the recognition frame 70, the sum of the lengths of the sides of the recognition frame 70, or the area of ​​the recognition frame 70 as the size of the side mirror area. For example, the detection unit 13 detects the area of ​​the recognition frame 70 of vehicle 50 as the size of the side mirror area of ​​vehicle 50.

[0041] The first estimation unit 16 estimates the distance to the vehicle 50 from an image in which the ground contact position of the vehicle 50 is not obscured, for example, an image in which the entire vehicle 50 is visible, as shown in Figure 2. The first estimation unit 16 stores the size of the side mirror area in the image in which the entire vehicle 50 is visible, and the estimated distance. For example, the first estimation unit 16 stores the area of ​​the recognition frame corresponding to the side mirror area in the image shown in Figure 2 as the size S1 of the side mirror area on the image of the vehicle 50. The first estimation unit 16 also stores the estimated distance L1 to the vehicle 50.

[0042] When the ground contact position of the vehicle 50 is hidden, the second estimating unit 17 estimates the distance from the camera 30 to the vehicle 50 by a geometric method using the size of the side mirror area of the vehicle 50 on the image, the size of the side mirror area stored when the entire vehicle is visible, and the estimated distance to the vehicle 50. For example, the second estimating unit 17 estimates the distance from the camera 30 to the vehicle 50 by a geometric method using the size S2 of the side mirror area of the vehicle 50 on the image, the size S1 of the side mirror area of the vehicle 50 on the image stored when the entire vehicle is visible, and the estimated distance L1 to the vehicle 50. Specifically, the second estimating unit 17 estimates the current distance L2 to the vehicle 50 according to the following formula 2. (Formula 2) L2 = L1 × √(S1 / S2)

[0043] In the first embodiment, the second estimating unit 17 cannot estimate the distance to an object when both the width and height of the recognition frame of the object are not detected. In the second embodiment, the second estimating unit 17 estimates the distance to the object by using the size of a partial area of the object such as a side mirror. In the second embodiment, for an object such as a vehicle, even if the entire roof of the object is not visible, the second estimating unit 17 can estimate the distance to the object as long as the size of a predetermined area such as a side mirror can be detected.

[0044] It should be noted that the first embodiment and the second embodiment can be appropriately combined. For example, the detection unit 13 may detect both the width of the recognition frame of an object and the size of a predetermined area on the object, for example, the size of a side mirror area. When the first estimating unit 16 estimates the distance to an object such as a vehicle from an image in which the entire object is visible, the first estimating unit 16 may store both the width of the recognition frame and the size of the side mirror area.

[0045] The second estimating unit 17 determines whether or not the width of the recognition frame of the object has been detected in an image where the ground contact position is hidden. If the width of the recognition frame of the object is detected, the second estimating unit 17 may estimate the distance to the object by using the detected width of the recognition frame and the stored width of the recognition frame. If the width of the recognition frame of the object is not detected and the size of the side mirror region is detected, the second estimating unit 17 may estimate the distance to the object by using the detected size of the side mirror region and the stored size of the side mirror region.

[0046] Note that the object distance estimation apparatus 10 does not necessarily need to be configured as a single apparatus. The object distance estimation apparatus 10 may, for example, be configured as a plurality of physically separated apparatuses. Further, when the object distance estimation apparatus 10 is configured as a plurality of physically separated apparatuses, the plurality of apparatuses do not necessarily need to be disposed at the same location. For example, the object distance estimation apparatus 10 may be separated into a first apparatus disposed near an intersection and a second apparatus disposed in a data center.

[0047] Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and embodiments obtained by changing or modifying the above embodiments without departing from the spirit of the present invention are also included in the present invention.

[0048] This application claims priority based on Japanese Patent Application No. 2025-53292 filed on March 27, 2025, and the entire disclosure thereof is incorporated herein.

[0049] The present disclosure can be applied to use of estimating the distance to an object using an image.

[0050] 10: Object distance estimation apparatus 11: Acquisition unit 12: Recognition unit 13: Detection unit 14: Determination unit 15: Distance estimation unit 16: First estimating unit 17: Second estimating unit 18: Output unit 30: Camera

Claims

1. An object distance estimation device comprising: an acquisition unit for acquiring an image captured by a camera; a recognition unit for recognizing an object included in the acquired image; a determination unit for determining whether or not a part of the object is hidden in the acquired image; a detection unit for detecting the size of a predetermined part of the recognized object; and a distance estimation unit for estimating the distance to the recognized object, wherein the distance estimation unit estimates the distance to the object based on the distance to the object estimated when a part of the object is not hidden and the size of a predetermined part of the object, and the size of a predetermined part of the object detected by the detection unit, when the determination unit determines that a part of the object that is in contact with the road is hidden.

2. The object distance estimation device according to claim 1, wherein the detection unit detects the length of one side of the recognition frame of the object recognized by the recognition unit as the size of a predetermined part of the object, and the distance estimation unit estimates the distance to the object based on the detected length of one side of the recognition frame and the length of one side of the recognition frame of the object recognized in an image in which the determination unit has determined that a part of the object is not hidden.

3. The object distance estimation device according to claim 1, wherein the detection unit detects the size of a predetermined area of ​​the object recognized by the recognition unit as the size of a predetermined part of the object, and the distance estimation unit estimates the distance to the object based on the size of the detected predetermined area and the size of the predetermined area of ​​the object recognized in an image in which the determination unit has determined that a part of the object is not hidden.

4. The object distance estimation device according to claim 1, wherein the determination unit determines whether or not the ground contact position in the recognition frame of the object is hidden as part of the object.

5. The object distance estimation device according to claim 1, wherein when the determination unit determines that the portion of the object in contact with the road is hidden, and when it is determined that the recognition frame of the object recognized in an image in which a portion of the object is not hidden has moved, the distance to the object estimated when a portion of the object was not hidden and the size of a predetermined portion of the object, and the size of a predetermined portion of the object detected by the detection unit, the distance estimation unit estimates the distance to the object.

6. An object distance estimation method comprising: acquiring an image captured by a camera; recognizing an object contained in the acquired image; determining whether a part of the object is hidden in the acquired image; detecting the size of a predetermined part of the recognized object; estimating the distance to the recognized object; and, if it is determined that the part of the object that contacts the road is hidden, estimating the distance to the object based on the distance to the object estimated when no part of the object is hidden, the size of a predetermined part of the object, and the detected size.

7. A program that causes a computer to perform the following processes: acquire an image captured by a camera; recognize an object contained in the acquired image; determine whether a part of the object is hidden in the acquired image; detect the size of a predetermined part of the recognized object; estimate the distance to the recognized object; and, if it is determined that the part of the object that contacts the road is hidden, estimate the distance to the object based on the distance to the object estimated when no part of the object is hidden, the size of a predetermined part of the object, and the detected size.

Images