Apparatus and method for predicting traffic information

Abstract

An apparatus and method for predicting traffic information are provided to improve traffic information prediction accuracy. The apparatus includes a memory storing a plurality of probe data generation models based on characteristics of roads, and a communication device receiving probe data from a probe vehicle traveling on a target road. A controller detects a probe data generation model corresponding to a characteristic of the target road among the plurality of probe data generation models, generates a preset number of probe data based on the detected probe data generation model, and predicts traffic information of the target road based on the generated probe data and the received probe data.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to Korean Patent Application No. 10-2021-0044210, filed with the Korean Intellectual Property Office on April 5, 2021, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This disclosure relates to techniques for predicting traffic information about roads using a learning model based on generated probe data. Background Technology

[0004] Typically, navigation systems respond to user requests by providing real-time traffic information for a specific area or the optimal route to the destination. In this context, real-time traffic information refers to traffic information at the time the user's traffic information request was generated.

[0005] Because this traffic information changes over time, when a user uses real-time traffic information to travel along the optimal route and arrive at a specific point, the real-time traffic information at that point differs from the real-time traffic information at the time the traffic information request was generated. Therefore, the traffic information initially provided to the user is less reliable. To prevent this, methods have been proposed to predict the traffic information at a specific point at the time the user expects to arrive, using both past and real-time traffic information.

[0006] In this regard, real-time traffic information (e.g., ETA: estimated time of arrival) is predicted based on detection data (e.g., GPS data) received from detection vehicles traveling on the road. To predict accurate traffic information (e.g., time spent traversing the road), the number of detection vehicles traversing the road (or a reference segment of the road) within a reference time (e.g., 5 minutes) must exceed a reference value (e.g., 30), but the number of detection vehicles is finite. Finally, conventional traffic information prediction techniques use fewer detection data (e.g., 30) than the reference number to predict road traffic information, and therefore their accuracy is significantly reduced.

[0007] The description in this background section is intended to enhance the understanding of the background technology of the present invention, which may include content other than the prior art known to those skilled in the art. Summary of the Invention

[0008] This disclosure has been made to address the aforementioned problems in the prior art while maintaining the advantages achieved by the prior art. One aspect of this disclosure provides an apparatus and method for predicting traffic information. The apparatus and method have multiple detection data generation models that have been learned for each characteristic of a road; detect a detection data generation model among the multiple detection data generation models that corresponds to a characteristic of a target road; generate predetermined detection data based on the detected detection data generation model; and predict traffic information for the target road based on the generated predetermined detection data and detection data received from a detection vehicle traveling on the target road, thereby improving the accuracy of traffic information prediction.

[0009] The technical problems to be solved by the present invention are not limited to those described above, and any other technical problems not mentioned herein will be clearly understood by those skilled in the art from the following description.

[0010] According to one aspect of this disclosure, an apparatus for predicting traffic information may include: a memory for storing a plurality of detection data generation models based on road characteristics; a communication device configured to receive detection data from a detection vehicle traveling on a target road; and a controller configured to: detect a detection data generation model among the plurality of detection data generation models that corresponds to the characteristics of the target road; generate a preset number of detection data based on the detected detection data generation model; and predict traffic information of the target road based on the generated detection data and the received detection data.

[0011] In one implementation, the detection data can be road transit time. The controller can be configured to generate a preset number of road transit times based on the detected detection data; and to calculate the transit time of the target road based on the generated road transit times and the road transit times received from the detection vehicle. Alternatively, the controller can be configured to calculate the average of the generated and received road transit times as the transit time of the target road.

[0012] The characteristics of a road may include at least one of the following: the number of detected vehicles, the type of road, the number of lanes, the length of the road, and the shape of the road. In one embodiment, the controller may be configured to calculate a similarity to each characteristic of the target road based on the characteristics of the target road; and detect the detection data generation model corresponding to the characteristic of the road with the highest similarity as the detection data generation model of the target road. The controller may be configured to, when no detection data generation model of the target road is detected based on the calculated similarity, detect the detection data generation model of the road characteristic having the smallest difference in the number of detected vehicles with respect to the number of detected vehicles of the target road as the detection data generation model of the target road.

[0013] According to another aspect of this disclosure, a method for predicting traffic information may include: storing a plurality of detection data generation models based on road characteristics in a memory; receiving detection data from a detection vehicle traveling on a target road by a communication device; detecting a detection data generation model among the plurality of detection data generation models that corresponds to the characteristics of the target road by a controller; generating a preset number of detection data based on the detected detection data generation models by the controller; and predicting traffic information of the target road based on the generated detection data and the received detection data.

[0014] In one implementation, predicting traffic information for a target road may include: generating a preset number of road travel times based on the detected detection data; and calculating the travel time of the target road based on the generated road travel times and the road travel times received from the detection vehicles. Calculating the travel time of the target road may include: calculating the average of the generated road travel times and the received road travel times as the travel time of the target road.

[0015] Furthermore, detecting the detection data generation model corresponding to the characteristics of the target road may include: calculating the similarity between the target road and each characteristic of the road; and detecting the detection data generation model corresponding to the characteristic with the highest similarity to the road as the detection data generation model for the target road. Detecting the detection data generation model corresponding to the characteristics of the target road may further include: when no detection data generation model for the target road is detected based on the calculated similarity, detecting the detection data generation model with the smallest difference in the number of detection vehicles compared to the number of detection vehicles on the target road as the road characteristic as the detection data generation model for the target road. Attached Figure Description

[0016] The above and other objects, features and advantages of this disclosure will become more apparent from the following detailed description taken in conjunction with the accompanying drawings:

[0017] Figure 1 This is a block diagram of a traffic information prediction device according to an embodiment of the present disclosure;

[0018] Figure 2 This is an exemplary diagram illustrating the structure of a detection data generation model used in a traffic information prediction apparatus according to an embodiment of the present disclosure;

[0019] Figure 3 This is an exemplary diagram illustrating the operation of a generator in a detection data generation model for a traffic information prediction device according to an embodiment of the present disclosure;

[0020] Figure 4This is an exemplary diagram illustrating the operation of a discriminator in a detection data generation model for a traffic information prediction device according to embodiments of the present disclosure; and

[0021] Figure 5 This is a flowchart of an embodiment of a traffic information prediction method according to an embodiment of the present disclosure. Detailed Implementation

[0022] In the following, some embodiments of the present disclosure will be described in detail with reference to the exemplary accompanying drawings. When adding reference numerals to the components in each drawing, it should be noted that the same numerals are used to denote the same or equivalent components even when the same or equivalent components are shown in other drawings. Furthermore, in describing embodiments of the present disclosure, detailed descriptions of relevant known configurations or functions will be omitted if it is determined that such configurations or functions interfere with the understanding of the embodiments of the present disclosure.

[0023] In describing components according to embodiments of this disclosure, terms such as first, second, A, B, (a), (b), etc., may be used. These terms are intended only to distinguish a component from other components, and they do not limit the nature, order, or sequence of the components. Unless otherwise defined, all terms used herein (including technical or scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It will be further understood that terms (such as those defined in common dictionaries) shall be interpreted as having the meaning consistent with their meaning in the context of the relevant art and shall not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0024] It should be understood that the terms “vehicle” or “vehicular” or other similar terms used herein generally include motor vehicles, such as passenger cars including sport utility vehicles (SUVs), buses, trucks, and various commercial vehicles; water vehicles including various boats and vessels; aircraft, etc., and include hybrid vehicles, electric vehicles, internal combustion engine vehicles, plug-in hybrid vehicles, hydrogen-powered vehicles, and other alternative fuel vehicles (e.g., fuels derived from resources other than petroleum).

[0025] Although the exemplary embodiments are described as using multiple units to perform the exemplary processes, it should be understood that the exemplary processes may also be performed by one or more modules. Furthermore, it should be understood that the term controller / control unit refers to a hardware device including a memory and a processor and specifically programmed to perform the processes described herein. The memory is configured to store modules, and the processor is specifically configured to execute said modules to perform one or more processes further described below.

[0026] Furthermore, the control logic of this disclosure can be embodied on a non-transitory computer-readable medium containing executable program instructions that are executed by a processor, controller / control unit, etc. Examples of computer-readable media include, but are not limited to, ROM, RAM, optical disc (CD)-ROM, magnetic tape, floppy disk, flash drive, smart card, and optical data storage device. The computer-readable medium can also be distributed across a network-coupled computer system, such that the computer-readable medium is stored and executed in a distributed manner, for example, via a telematics server or a controller area network (CAN).

[0027] Figure 1 This is a block diagram of a traffic information prediction device according to an embodiment of the present disclosure. Figure 1 As shown, the traffic information prediction device 100 according to an embodiment of the present disclosure may include a memory 10, a communication device 20, an output device 30, and a controller 40. In this regard, according to the method for implementing the traffic information prediction device 100 according to an embodiment of the present disclosure, components may be combined with each other to form a single component, and some components may be omitted.

[0028] Each component will be described. First, the memory 10 can be configured to store multiple probe data generation models that have been learned for each feature of the road. In this regard, the probe data generation model, which generates false road travel times based on the true road travel time (time spent traveling through the road) of the probe vehicle 200 and the latent vector "z", can be implemented, for example, using a conditional generative adversarial network (CGAN) that has been learned. In this regard, the CGAN can be configured to perform learning for generating travel times for each road based on the designer's intent, or more specifically, to perform learning for generating travel times for each segment of each road.

[0029] The memory 10 can be configured to store various logics, algorithms, and programs required in the process of detecting multiple detection data generation models corresponding to the characteristics of the target road, generating predetermined detection data (a preset number of detection data) based on the detected detection data generation models, and predicting traffic information of the target road (e.g., the time spent crossing the target road or the time spent passing through a reference section of the target road) based on the generated predetermined detection data and detection data (e.g., GPS data) received from the detection vehicle 200 traveling on the target road. In this regard, the GPS data includes time data and coordinate data.

[0030] The memory 10 may include at least one type of recording medium (storage medium) such as flash memory, hard disk, micro, card (e.g., security digital card (SD card) or eXtream digital card (XD card)), as well as random access memory (RAM), static RAM (SRAM), read-only memory (ROM), programmable ROM (PROM), electrically erasable PROM (EEPROM), magnetic RAM (MRAM), disk and optical disk type memory.

[0031] The communication device 20, serving as a module providing a communication interface with the detection vehicle 200 traveling on the road, can be configured to periodically receive detection data from the detection vehicle 200. In this regard, the detection vehicle 200 may have a telematics terminal as a vehicle terminal. The communication device 20 may include at least one of a mobile communication module, a wireless internet module, and / or a short-range communication module to communicate with the detection vehicle 200.

[0032] The mobile communication module can be configured to communicate with the detection vehicle 200 through a mobile communication network built on technical standards or communication schemes for mobile communication (e.g., Global System for Mobile Communications (GSM), Code Division Multiple Access (CDMA), Code Division Multiple Access 2000 (CDMA2000), Enhanced Voice Data Optimized or Enhanced Voice Data Only (EV-DO), Wideband CDMA (WCDMA), High-Speed ​​Downlink Packet Access (HSDPA), High-Speed ​​Uplink Packet Access (HSUPA), Long Term Evolution (LTE), Advanced Long Term Evolution (LTEA), etc.), fourth-generation mobile communication (4G), and fifth-generation mobile communication (5G).

[0033] The wireless internet module, used for wireless internet access, can be configured to communicate with the detection vehicle 200 via wireless LAN (WLAN), Wi-Fi, Wi-Fi Direct, Digital Living Network Alliance (DLNA), WiBro, Global Microwave Access Interoperability (WiMAX), High-Speed ​​Downlink Packet Access (HSDPA), High-Speed ​​Uplink Packet Access (HSUPA), Long Term Evolution (LTE), and Advanced Long Term Evolution (LTE-A). The short-range communication module can support short-range communication using at least one of the following technologies: Bluetooth™, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra Wideband (UWB), ZigBee, Near Field Communication (NFC), and Wireless Universal Serial Bus (Wireless USB).

[0034] For example, output device 30 can provide the user with the time required to travel through the target road or the time required to travel through a reference section of the target road (this is traffic information for the target road predicted by controller 40). Controller 40 can be configured to perform overall control, enabling each component to perform its function correctly. Controller 40 can be implemented in hardware, software, or a combination of hardware and software. Preferably, controller 40 can be implemented as a microprocessor, but it is not limited to this.

[0035] Specifically, the controller 40 may include multiple detection data generation models that have been learned for each characteristic of the road, and performs various controls in the following processes: detecting the detection data generation model among the multiple detection data generation models that corresponds to the characteristics of the target road; generating predetermined detection data based on the detected detection data generation model; and predicting traffic information of the target road based on the generated predetermined detection data and detection data received from the detection vehicle 200 traveling on the target road. In this regard, the characteristics of the road may include the number of detection vehicles 200, the type of road, the number of lanes, the length of the road, the shape of the road, etc.

[0036] As an example, controller 40 can be configured to generate a preset number of road travel times (time spent traveling on the road) based on the detected detection data generation model, and calculate the target road travel time based on the generated road travel times and the road travel times received from the detection vehicle 200 traveling on the target road. In this respect, controller 40 can be configured to calculate the average of the road travel times generated based on the detection data generation model and the road travel times received from the detection vehicle 200 as the travel time of the target road.

[0037] Since the controller 40 can periodically receive GPS data (including time data) from the detection vehicle 200 via the communication device 20, it can identify the location of the detection vehicle 200 in real time. Therefore, the controller 40 can be configured to identify the entry time of the detection vehicle 200 onto a target road or a reference segment of the target road, and calculate the time required to travel across the target road (required time) or the time required to travel across the reference segment of the target road (required time) as traffic information based on the entry time and the calculated travel time on the target road. The controller 40 can be configured to identify the location of the detection vehicle 200 in real time in association with a navigation system (not shown). In other words, the controller 40 can be configured to detect the location of the detection vehicle 200 on the road based on GPS data received from the detection vehicle 200.

[0038] The controller 40 can use a generally known similarity calculation algorithm during the process of detecting a probe data generation model corresponding to the characteristics of the target road among multiple probe data generation models corresponding to the characteristics of the road. In other words, the controller 40 can be configured to calculate the similarity to each characteristic of the road based on the characteristics of the target road, and detect the probe data generation model corresponding to the characteristic of the road with the highest similarity as the probe data generation model of the target road. In this regard, when there is no probe data generation model with a similarity exceeding the reference value, the controller 40 can be configured to determine the probe data generation model of the target road as the number of probe vehicles 200 that are most similar to the number of probe vehicles 200 of the target road (the number of probe data received from the probe vehicles 200) as the number of probe data generation models for the characteristics of the road.

[0039] In the following text, reference will be made to Figures 2 to 4 Describe the structure of the probe data generation model and the process of training the probe data generation model using the controller 40. Figure 2 This is an exemplary diagram illustrating the structure of a detection data generation model used in a traffic information prediction apparatus according to an embodiment of the present disclosure.

[0040] like Figure 2 As shown, the detection data generation model for a traffic information prediction device according to embodiments of this disclosure can be implemented, for example, using a conditional generative adversarial network (CGAN). Such a CGAN may include a generator 210 and a discriminator 220. In this regard, to make it difficult for the discriminator 220 to determine whether the detection data is real or fake, the generator 210, which attempts to generate fake detection data that is as realistic as possible, and the discriminator 220, which attempts to distinguish between real and fake detection data with high accuracy, learn in an adversarial manner.

[0041] The controller 40 can be configured to repeatedly perform adversarial learning, which is a process of first training the discriminator 220 and then training the generator 210 by reflecting the learning results of the discriminator 220. The training of the discriminator 220 consists of two main processes. The first process is to input true probe data into the discriminator 220 and train it to identify the true probe data as true. The second process is to input false probe data generated by the generator 210 and train the discriminator 220 to identify the false probe data as false. Through this process, the discriminator 220 can be configured to distinguish between true probe data and false probe data. After training the discriminator 220 in this way, it is necessary to train the generator 210 in the direction that deceives the discriminator 220. In other words, the controller 40 can be configured to train the generator 210 to generate false probe data similar to the true probe data, which is sufficient to be identified as true by the discriminator 220.

[0042] When this training process is repeated, discriminator 220 and generator 210 identify each other as adversarial competitors, and both evolve. Therefore, generator 210 can be configured to generate false probe data that is completely similar to the true probe data. Consequently, discriminator 220 cannot distinguish between true and false probe data. In other words, generator 210 and discriminator 220 compete with each other in such a way that generator 210 attempts to reduce the discriminator 220's probability of successful discrimination, while discriminator 220 attempts to increase its probability of successful discrimination, thus causing generator 210 and discriminator 220 to evolve together.

[0043] More specifically, CGAN is trained using an objective function V(D,G) in a scheme that solves the “minmax problem” as shown in Equation 1 below.

[0044] Equation 1

[0045]

[0046] In this respect, x ~ pdata(x) represents data sampled from the probability distribution of the true probe data, and z ~ p z (z) typically represents data sampled from random noise using a Gaussian distribution, and "z" represents the latent vector (the vector in the latent space). D(x|y) is the discriminator 220, and is 1 when the probe data is true and 0 when the probe data is false. D(G(z|y)) is 1 when the probe data generated by generator 210 is determined to be true, and 0 when the probe data is determined to be false.

[0047] First, in maximizing V(D,G) through D as the discriminator 220 to maximize Equation 1, both the first and second terms on the right-hand side must be maximized, such that both log D(x|y) and log(1-D(G(z|y))) should be maximized. Therefore, D(x|y) should be 1, meaning that training D will classify true probes as true. Similarly, since 1-D(G(z|y)) should be 1, D(G(z|y)) should be 0, meaning that training the discriminator 220 will classify false probes generated by the generator 210 as false. Finally, training that allows V(D,G) to become the maximum D is the process of training the discriminator 220 to distinguish true probes as true and false probes as false.

[0048] Next, regarding minimizing V(D,G) through G as generator 210, the first term can be omitted since G is not included in the first term on the right-hand side of Equation 1, as it is irrelevant to generator 210. To minimize the second term, log(1-D(G(z|y))) must be minimized. Therefore, log(1-D(G(z|y))) should be 0, and D(G(z|y)) should be 1. This means training generator 210 to generate sufficiently perfect false probe data to be identified as true by discriminator 220. Therefore, training discriminator 220 in the direction that maximizes V(D,G) and training generator 210 in the direction that minimizes V(D,G) is called the "min-maximization problem".

[0049] Figure 3 This is an exemplary diagram illustrating the operation of a generator in a detection data generation model for a traffic information prediction device according to an embodiment of the present disclosure. Figure 3 As shown, the generator 210 in the detection data generation model for a traffic information prediction device according to an embodiment of the present disclosure can be configured to receive true detection data "y" and a potential vector "z" from the detection vehicle 200, and generate false detection data G(z|y) according to the distribution of the true detection data "y". In this respect, the generator 210 can be configured to generate multiple false detection data (G(z|y)).

[0050] Figure 4 This is an exemplary diagram illustrating the operation of a discriminator in a detection data generation model for a traffic information prediction device according to an embodiment of the present disclosure. Figure 4 As shown, the discriminator 220 in the detection data generation model for a traffic information prediction device according to an embodiment of the present disclosure can be configured to receive true detection data “y” from the detection vehicle 200 and false detection data G(z|y) generated by the generator 210, determine that the true detection data “y” is true (D(y)), and determine that the false detection data G(z|y) is false (D(G(z|y))).

[0051] Figure 5 This is a flowchart of an embodiment of the traffic information prediction method according to an embodiment of the present disclosure. First, the memory 10 may be configured to store multiple detection data generation models based on road characteristics (501). Then, the communication device 20 may be configured to receive detection data from a detection vehicle traveling on a target road (502).

[0052] Subsequently, controller 40 can be configured to detect a detection data generation model (503) among multiple detection data generation models that corresponds to the characteristics of the target road. Then, controller 40 can be configured to generate a preset number of detection data based on the detected detection data generation model, and predict traffic information for the target road based on the generated and received detection data (504). In this respect, controller 40 can be configured to predict the time required to travel through the target road as the traffic information for the target road.

[0053] The above description is merely an illustration of the technical concept of this disclosure, and those skilled in the art can make various modifications and changes without departing from the basic characteristics of this disclosure. Therefore, the embodiments disclosed in this disclosure are not intended to limit the technical concept of this disclosure but are used to illustrate this disclosure, and the scope of the technical concept of this disclosure is not limited by the embodiments. The scope of this disclosure should be interpreted as being covered by the scope of the appended claims, and all technical concepts falling within the scope of the claims should be interpreted as being included within the scope of this disclosure.

[0054] The apparatus and method for predicting traffic information according to embodiments of the present disclosure as described above may have multiple detection data generation models that have been learned for each characteristic of a road, detect a detection data generation model among the multiple detection data generation models that corresponds to the characteristics of a target road, generate predetermined detection data based on the detected detection data generation model, and predict traffic information of the target road based on the generated predetermined detection data and detection data received from a detection vehicle traveling on the target road, thereby improving the accuracy of traffic information prediction.

[0055] In the foregoing, although the present disclosure has been described with reference to exemplary embodiments and accompanying drawings, the present disclosure is not limited thereto, but can be modified and altered by those skilled in the art without departing from the spirit and scope of the present disclosure as claimed in the appended claims.

Claims

1. A device for predicting traffic information, comprising: The memory is configured to store multiple detection data generation models based on road characteristics; A communication device is configured to receive detection data from a detection vehicle traveling on the target road; as well as The controller is configured as follows: Detect the detection data generation model among the multiple detection data generation models that corresponds to the characteristics of the target road; Based on the detected detection data, a model is generated to produce a preset number of detection data. and Based on the generated and received detection data, predict the traffic information of the target road. The detected data is road travel time. The controller is configured to generate a preset number of road travel times based on the detected detection data, and to calculate a target road travel time based on the generated road travel times and the road travel times received from the detection vehicle. The controller is configured to identify the entry time of the detection vehicle on the target road or a reference segment of the target road, and to calculate the time required to pass through the target road or the time required to pass through the reference segment of the target road based on the entry time and the calculated travel time on the target road as the traffic information.

2. The apparatus according to claim 1, wherein, The controller is configured to calculate the average of the generated road travel time and the received road travel time as the target road travel time.

3. The apparatus according to claim 1, wherein, The characteristics of the road include at least one of the number of the detection vehicles, the type of the road, the number of lines, the length of the road, and the shape of the road.

4. The apparatus according to claim 3, wherein, The controller is configured to: Calculate the similarity to each characteristic of the target road based on the characteristics of the target road; and The detection data generation model corresponding to the characteristic with the highest similarity to the road is used as the detection data generation model for the target road.

5. The apparatus according to claim 4, wherein, The controller is configured to, when no detection data generation model of the target road is detected based on the calculated similarity, use the number of detection vehicles with the smallest difference from the number of detection vehicles of the target road as the detection data generation model of the characteristic of the road as the detection data generation model of the target road.

6. A method for predicting traffic information, comprising the following steps: A model is generated by storing multiple detection data based on road characteristics in memory; The communication device receives detection data from the detection vehicle traveling on the target road; The controller detects the detection data generation model among the multiple detection data generation models that corresponds to the characteristics of the target road; and The controller generates a preset number of detection data based on the detected detection data, and predicts the traffic information of the target road based on the generated detection data and the received detection data. The detected data is road travel time. The method for predicting the traffic information of the target road includes the following steps: generating a preset number of road travel times based on the detected data, and calculating the target road travel time based on the generated road travel times and the road travel times received from the detected vehicles. The method of predicting the traffic information of the target road further includes the following steps: identifying the entry time of the detection vehicle on the target road or a reference segment of the target road, and calculating the time required to pass through the target road or the time required to pass through the reference segment of the target road based on the entry time and the calculated travel time of the target road as the traffic information.

7. The method according to claim 6, wherein, Calculating the travel time of the target road includes the following steps: The average of the generated road travel time and the received road travel time is calculated as the travel time of the target road.

8. The method according to claim 6, wherein, The characteristics of the road include at least one of the number of the detection vehicles, the type of the road, the number of lines, the length of the road, and the shape of the road.

9. The method according to claim 8, wherein, The detection data generation model corresponding to the characteristics of the target road includes the following steps: Calculate the similarity to each characteristic of the target road based on the characteristics of the target road; and The detection data generation model corresponding to the characteristic with the highest similarity to the road is used as the detection data generation model for the target road.

10. The method according to claim 9, wherein, The detection data generation model corresponding to the characteristics of the target road further includes the following steps: When no detection data generation model of the target road is detected based on the calculated similarity, the detection data generation model of the target road is the one that detects the number of detection vehicles with the smallest difference from the number of detection vehicles of the target road as the characteristic of the road.

Images