Parking Lot Management System

Abstract

Provided are a two-wheeler identification system and a parking management system that can classify and identify types of two-wheelers. 【Solution means】The two-wheeler identification system 20 includes a first camera 212 for acquiring a first image that is an image of the side surface of the two-wheeler 12a, a second camera 214 for acquiring a second image that is an image of the license plate of the two-wheeler, and an identification device 220 that identifies at least the legally defined classification of the two-wheeler 12a based on the first image and the second image.

Description

Technical Field

[0001] The present invention relates to a parking lot management system.

Background Art

[0002] Patent Document 1 discloses a vehicle type discrimination device that can be easily installed and can easily discriminate the vehicle type. This vehicle type discrimination device includes temperature detection means for detecting the temperature at a predetermined position of a vehicle, and discrimination means for discriminating the vehicle type of the vehicle based on the temperature detected by the temperature detection means.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of the present invention is to provide a parking lot management system that can classify and identify the types of two-wheeled vehicles.

Means for Solving the Problems

[0005] The invention according to claim 1 comprises: a first camera for acquiring a first image which is an image of the side surface of a two-wheeled vehicle; a second camera for acquiring a second image which is an image of the license plate of the two-wheeled vehicle; a vehicle type specifying unit for specifying at least whether the type of the two-wheeled vehicle is a bicycle, a motorcycle or a kick scooter based on the first image; a plate specifying unit for specifying the type of the license plate from the second image; and an identifying device having an identifying unit for identifying the classification of the two-wheeled vehicle based on the combination of the type of the two-wheeled vehicle specified by the vehicle type specifying unit and the type of the license plate specified by the plate specifying unit. The vehicle type specifying unit includes: a primary vehicle type processing unit for detecting the two-wheeled vehicle by using a first machine learning model used for object detection based on the first image and outputting information on an area A where the detected two-wheeled vehicle is located; and a secondary vehicle type processing unit for specifying the type of the two-wheeled vehicle by using a second machine learning model used for category classification based on a cut-out image cut out from the first image so as to include the area A. The plate specifying unit includes: a primary plate processing unit for detecting the license plate by using a third machine learning model used for object detection based on the second image and outputting information on an area B where the detected license plate is located; and a secondary plate processing unit for specifying the type of the license plate by using a fourth machine learning model used for category classification based on a cut-out image cut out from the second image so as to include the area B. It is a parking lot management system.

[0006] The invention according to claim 2 is the parking lot management system according to claim 1, wherein Before the first machine learning model specifies the type of the detected two-wheeled vehicle Do .

[0007] The invention according to claim 3 is the parking lot management system according to claim 2, wherein when the primary vehicle type processing unit fails to detect the two-wheeled vehicle, it determines that there is no two-wheeled vehicle.

[0008] The invention according to claim 4 is the parking lot management system according to any one of claims 1 to 3, wherein the one identified by the identifying deviceDivision It further includes an admission reception machine that issues an admission ticket on which Division is printed.

Advantages of the Invention

[0009] According to the present invention, a parking lot management system capable of classifying and identifying the types of two-wheeled vehicles can be provided.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Embodiments for Carrying Out the Invention

[0011] Subsequently, with reference to the attached drawings, embodiments embodying the present invention will be described to facilitate understanding of the present invention. Note that parts not relevant to the description may be omitted from the illustration.

[0012] A parking lot management system 10 according to an embodiment of the present invention can set a parking fee according to the classification of two-wheeled vehicles specified by the combination of the type of two-wheeled vehicle and the type of license plate. The identified classifications include at least the classifications legally defined by the Road Transportation Vehicle Act, etc. For example, as shown in FIG. 1, 1) bicycle, 2) moped type 1 (bicycle with a first type of prime mover), 3) moped type 2 (bicycle with a second type of prime mover), 4) small two-wheeler (small motorcycle), 5) specific small moped (bicycle with a specific small prime mover), 6) kick scooter (without a prime mover), and 7) light two-wheeler.

[0013] Note that "White (Small)" and "White (Large)" shown in FIG. 1 respectively indicate small and large white number plates. "Yellow" indicates a yellow number plate. "Pink" indicates a pink number plate. "White + Green Frame Line" indicates a number plate edged with a green line on a white background. "New Number" indicates a number plate that is 10 centimeters both vertically and horizontally and is to be attached to a specific small motorized bicycle. Also, the reason for classifying a motorcycle without a number plate as a bicycle (1) is that motorcycles and bicycles cannot be distinguished with a sufficiently high degree of accuracy, so classifying it as a bicycle can make the parking fee setting cheaper and reduce claims from users.

[0014] The classification is not limited to those shown in FIG. 1. For example, as one of the classifications, a mini-car registered vehicle with a light blue number plate may be added. Furthermore, the parking lot management system 10 is a system for classifying and identifying the types of two-wheeled vehicles as needed, and is not for identifying according to the legally defined classifications.

[0015] As shown in FIGS. 2 to 5, the parking lot management system 10 includes a two-wheeled vehicle identification system 20, an entrance reception device 50, and a flap gate 60.

[0016] As shown in FIGS. 2 and 5, the two-wheeled vehicle identification system 20 has a side camera 212, a rear camera 214, and an identification device 220, and can identify the above-mentioned classifications of two-wheeled vehicles, such as the bicycle 12a shown in FIG. 3 and the first type of motorized bicycle 12b shown in FIG. 4.

[0017] The side camera (an example of the first camera) 212 is a camera for acquiring an image of the side of the vehicle body (an example of the first image), and is disposed, for example, at a height position of 2,000 to 2,500 mm. The reason for setting the lower limit of the height position of the side camera 212 is that since the side camera 212 is disposed at a position where a child's hand cannot reach, the possibility of being tampered with is reduced. The reason for setting the upper limit of the height position is that since the side camera 212 is disposed at a position where an image capturing the characteristics of a two-wheeled vehicle can be acquired, the type of the two-wheeled vehicle can be specified with higher accuracy.

[0018] The rear camera (an example of the second camera) 214 is a camera for acquiring an image of the number plate provided at the rear of the vehicle body (an example of the second image). When there are restrictions on setting the route of the two-wheeled vehicle entering from the road lane, the rear camera 214 does not necessarily have to be provided on the rear side of the vehicle body. For example, the rear camera 214 may be provided at the upper part in front of the vehicle body, and may image the number plate image through a mirror installed on the rear side of the vehicle body.

[0019] The identification device 220 can identify at least the legally defined classification of the two-wheeled vehicle based on the side image of the two-wheeled vehicle captured by the side camera 212 and the image of the number plate captured by the rear camera 214. The identification device 220 includes a vehicle type specifying unit 222, a plate specifying unit 224, and an identification unit 226.

[0020] The vehicle type specifying unit 222 is composed of a primary vehicle type processing unit 222a and a secondary vehicle type processing unit 222b, and can specify at least which of a bicycle, a motorcycle, and a kick scooter the type of the two-wheeled vehicle is from the side image of the two-wheeled vehicle captured by the side camera 212.

[0021] The primary vehicle type processing unit 222a detects the two-wheeled vehicle by using a first machine learning model learned to be able to detect the two-wheeled vehicle based on the side image of the two-wheeled vehicle, specifies the type of the detected two-wheeled vehicle, and can output the position information of the region A where the two-wheeled vehicle is located. In addition, when the plurality of motorcycles are included in the image acquired by the side camera 212, the primary vehicle type processing unit 222a can determine the vehicle to be identified. The first machine learning model is a machine learning model used for object detection.

[0022] The secondary vehicle type processing unit 222b can identify and classify the type of motorcycle by using a second machine learning model that has been trained to classify motorcycles based on the cropped image cropped from the side image of the motorcycle to include the region A. The second machine learning model is a machine learning model used for category classification. Since the type of the motorcycle is specified based on the region A where the motorcycle already exists by the second machine learning model, the accuracy of the specified result is improved.

[0023] The license plate specifying unit 224 is composed of a primary license plate processing unit 224a and a secondary license plate processing unit 224b, and can specify the type of the license plate from the image of the license plate captured by the rear camera 214. Incidentally, the license plate specifying unit 224 does not read the characters of the license plate.

[0024] The primary license plate processing unit 224a can detect the license plate by using a third machine learning model that has been trained to detect the license plate based on the image of the license plate, and can output the information on the position of the region B where the detected license plate exists. The third machine learning model is a machine learning model used for object detection.

[0025] The secondary license plate processing unit 224b can identify and classify the type of the license plate by using a fourth machine learning model that has been trained to classify the license plate based on the cropped image cropped from the license plate image to include the region B. In addition, the secondary license plate processing unit 224b can remedy the false detection of the primary license plate processing unit 224a and classify it as having no license plate. The fourth machine learning model is a machine learning model used for category classification. Based on the area B that already has a license plate, the fourth machine learning model can identify the type of the license plate, thus improving the accuracy of the identified result.

[0026] Based on the type of the two-wheeler identified by the vehicle type identification unit 222 and the type of the license plate identified by the plate identification unit 224, the identification unit 226 can identify whether the classification of the two-wheeler is one of the following: 1) bicycle, 2) moped type 1, 3) moped type 2, 4) small two-wheeler, 5) specific small moped, 6) kick scooter, and 7) light two-wheeler.

[0027] Specifically, the identification unit 226 identifies the classification according to the matrix shown in FIG. 1. For example, when the vehicle type identification unit 222 identifies the two-wheeler as a bicycle and the plate identification unit 224 identifies the type of the license plate as a new number, the identification unit 226 identifies the bicycle as a specific small moped. Also, when the vehicle type identification unit 222 identifies the two-wheeler as a kick scooter and the plate identification unit 224 identifies the type of the license plate as "white + green border line", the identification unit 226 identifies the kick scooter as a small two-wheeler. The information of the identified two-wheeler is transmitted to the entrance reception machine 50.

[0028] Note that if the identification device 220 has at least the vehicle type identification unit 222, it can identify whether the two-wheeler is a motorcycle or a kick scooter. That is, when it is not necessary to identify the more detailed classification as shown in FIG. 1 and it is sufficient to identify the type of the two-wheeler based on the shape of the vehicle body seen from the side, the two-wheeler identification system 20 may not include the rear camera 214, the plate identification unit 224, and the identification unit 226. In this case, the information of the two-wheeler identified by the vehicle type identification unit 222 is transmitted from the identification device 220 to the entrance reception machine 50.

[0029] Such a two-wheeler identification system 20 has the following advantages, for example, compared with a conventional loop coil type vehicle identification device. (1) It can identify a kick scooter. (2) It can reduce the false detection of motorcycles with a small amount of metal in the vehicle body. (3) It can reduce the false detection of bicycles with a large amount of metal in the vehicle body (such as electric bicycles and bicycles equipped with child seats, etc.). (4) It can identify children's bicycles with a small amount of metal in the vehicle body.

[0030] The admission reception machine 50 is connected to the identification device 220 and can issue admission tickets. The admission ticket is issued when the ticket issuing button provided on the admission reception machine 50 is pressed, and the classification identified by the identification device 220 is printed on the issued admission ticket.

[0031] The flap type gate (an example of a gate) 60 is arranged at the entrance of the access route to the parking space and is configured to allow one two-wheeler to enter at a time. The opening and closing of the flap type gate 60 are controlled by the admission reception machine 50.

[0032] Next, the operation of the parking lot management system 10 will be described along the flow of a user riding a two-wheeler using the parking lot. A user riding a two-wheeler enters the parking lot and stops in front of the flap type gate 60 and in front of the admission reception machine 50. The stopped two-wheeler comes into the view of the side camera 212 and the rear camera 214.

[0033] At a predetermined timing, an image of the side of the two-wheeler is acquired by the side camera 212, and an image of the license plate is acquired by the rear camera 214. Here, this predetermined timing is as follows, for example. (1) When the user presses the ticket issuing button of the admission reception machine 50 (2) When there is a reaction to a loop coil (not shown) buried at the entrance (3) The side camera 214 has a moving object detection function, and when an event occurs due to this moving object detection function However, the image may be constantly acquired by the side camera 212 at any time rather than at a predetermined timing.

[0034] Based on the image of the side of the two-wheeler, the primary vehicle type processing unit 222a detects the presence of the two-wheeler using the first machine learning model and identifies the type of the detected two-wheeler. Further, it outputs information on the position of the area A where the two-wheeler is located. Based on the cropped image cropped from the image of the side of the two-wheeler so as to include the area A, the secondary vehicle type processing unit 222b identifies and classifies the type of the two-wheeler using the second machine learning model.

[0035] Here, since the first machine learning model is a machine learning model used for object detection as described above, the accuracy of identifying the type of the two-wheeler tends to be lower compared to the second machine learning model specialized for category classification. Therefore, when the result R1 which is the type of the two-wheeler identified by the primary vehicle type processing unit 222a and the result R2 which is the type of the two-wheeler identified by the secondary vehicle type processing unit 222b match, the vehicle type identification unit 222 identifies the result as the final type of the two-wheeler.

[0036] However, when the result R1 identified by the primary vehicle type processing unit 222a and the result R2 identified by the secondary vehicle type processing unit 222b do not match, the vehicle type identification unit 222 compares the accuracy of the result R2 identified by the secondary vehicle type processing unit 222b with a preset threshold value. If it is determined that the accuracy is higher than this threshold value, the result R2 identified by the secondary vehicle type processing unit 222b is adopted. Also, in the following cases, the vehicle type identification unit 222 determines that there is no two-wheeler. (1) When the primary vehicle type processing unit 222a fails to detect a two-wheeler (2) Regardless of the results R1 identified by the primary vehicle type processing unit 222a and the result R2 identified by the secondary vehicle type processing unit 222b, when the accuracies of both the result R1 and the result R2 are lower than a preset threshold value

[0037] On the one hand, the primary plate processing unit 224a detects the license plate based on the image of the license plate by means of a third machine learning model, and outputs information on the position of the area B where the detected license plate is located. The secondary plate processing unit 224b identifies and classifies the type of the license plate by means of a fourth machine learning model based on the cropped image cropped from the image of the license plate so as to include the area B. When the accuracy of the type of the license plate classified by the secondary plate processing unit 224a is higher than a preset threshold value, the plate identification unit 224 identifies this type of license plate as the final type of the license plate.

[0038] After that, based on the type of the two-wheeler identified by the vehicle type identification unit 222 and the type of the license plate identified by the plate identification unit 224, the identification unit 226 identifies the classification of the two-wheeler according to the matrix shown in FIG. 1.

[0039] The information on the classification identified by the identification unit 226 is transmitted to the entrance reception machine 50. The entrance reception machine 50 issues an entrance ticket with the received classification information printed thereon. When taking the issued entrance ticket, the flap gate 60 opens, and the user can proceed to the parking area.

[0040] After finishing using the parking area, the user settles the predetermined parking fee at the payment machine and exits from the gate provided at the exit. In the parking area managed by the parking lot management system 10, this predetermined parking fee can be set to an amount corresponding to the classification of the two-wheeler (see FIG. 1).

[0041] As described above, according to the parking lot management system 10 and the two-wheeler identification system 20, the identification device 220 includes the primary vehicle type processing unit 222a and the primary vehicle type processing unit 222a, and uses the first machine learning model and the second machine learning model with different functions for identification respectively. Therefore, the type of the two-wheeler can be identified with higher accuracy. Also, according to the parking lot management system 10 and the two-wheeled vehicle identification system 20, since the two-wheeled vehicle is specified by the combination of the type of the two-wheeled vehicle and the type of the license plate, the type of the two-wheeled vehicle can be identified in a more detailed manner.

[0042] As described above, the embodiments of the present invention have been described. However, the present invention is not limited to the above-described forms, and all changes and the like under the condition of not departing from the gist are within the scope of application of the present invention.

Explanation of Reference Numerals

[0043] 10 Parking lot management system 12a Bicycle 12b Moped type 1 20 Two-wheeled vehicle identification system 50 Entrance reception device 60 Flap gate 212 Side camera 214 Rear camera 220 Identification device 222 Vehicle type specifying unit 222a Primary vehicle type processing unit 222b Secondary vehicle type processing unit 224 Plate specifying unit 224a Primary plate processing unit 224b Secondary plate processing unit 226 Identification unit

Claims

1. A first camera for acquiring a first image that is an image of the side of a two-wheeled vehicle, A second camera for acquiring a second image that is an image of the license plate of the two-wheeled vehicle, A vehicle type specifying unit that at least specifies which of a bicycle, a motorcycle, and a kick scooter the type of the two-wheeled vehicle is from the first image, a plate specifying unit that specifies the type of the license plate from the second image, and an identification unit that identifies the classification of the two-wheeled vehicle based on the combination of the type of the two-wheeled vehicle specified by the vehicle type specifying unit and the type of the license plate specified by the plate specifying unit. An identification device having, The vehicle type specifying unit detects the two-wheeled vehicle by a first machine learning model used for object detection based on the first image, and a primary vehicle type processing unit that outputs information on an area A where the detected two-wheeled vehicle is located, A secondary vehicle type processing unit that specifies the type of the two-wheeled vehicle by a second machine learning model used for category classification based on a cut-out image cut out from the first image so as to include the area A. A parking lot management system having, The plate specifying unit detects the license plate by a third machine learning model used for object detection based on the second image, and a primary plate processing unit that outputs information on an area B where the detected license plate is located, A secondary plate processing unit that specifies the type of the license plate by a fourth machine learning model used for category classification based on a cut-out image cut out from the second image so as to include the area B.

2. The parking lot management system according to claim 1, wherein the first machine learning model specifies the type of the detected two-wheeled vehicle.

3. The parking lot management system according to claim 2, wherein when the primary vehicle type processing unit cannot detect the two-wheeled vehicle, it determines that there is no two-wheeled vehicle.

4. The parking lot management system according to any one of claims 1 to 3, further comprising an admission reception machine that issues an admission ticket printed with the classification identified by the identification device.

Images