Data Acquisition Equipment

Abstract

To make it possible to properly transmit data from a data acquisition device to a server while suppressing power consumption.SOLUTION: A data collection device collects data necessary for use or learning of a machine learning model that is used only when the number of people in a prescribed target area is greater than or equal to a prescribed reference value. The data collection device has a number-of-people estimation unit 231 for estimating the number of people in an object area and a transmission control unit 232 for controlling data transmission from a data acquisition device for collecting data necessary for use or learning of the machine learning model. When the number of people estimated by the number-of-people estimation unit is less than a reference value, the transmission control unit causes the data acquisition device to transmit no data, or causes the data acquisition device to reduce an amount of data transmitted per unit time to the data collection device compared with a case where the number of people estimated by the number-of-people estimation unit is equal to or larger than the reference value.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present disclosure relates to a data collection device. [Background technology]

[0002] In smart cities, it has been proposed to collect data from multiple entities within the community. In particular, Patent Document 1 proposes collecting corrected data to eliminate the uncertainty of data obtained from the information systems of different business entities, since such data is uncertain. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2013-69084 A Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, in a smart city or the like, it is conceivable that various processes in the smart city are performed using a machine learning model that is used only when the number of people in the smart city is equal to or greater than a predetermined reference value. In order to train such a machine learning model, a data collection device is required to collect data necessary for training the machine learning model into a server.

[0005] Here, when collecting data, it is necessary to transmit the data acquired by the data acquisition device to a server, but transmitting data from the data acquisition device to the server requires power, so if the data acquisition device is allowed to blindly transmit data to the server, there is a possibility that excessive power will be consumed.

[0006] In view of the above problems, an object of the present disclosure is to provide a data collection device that can appropriately transmit data from a data acquisition device to a server while suppressing power consumption. [Means for solving the problem]

[0007] The gist of the present disclosure is as follows.

[0008] (1) A data collection device that collects data necessary for use or learning of a machine learning model that is used only when the number of people in a specified target area is equal to or greater than a specified reference value, A number of people estimation unit that estimates the number of people in the target area; A transmission control unit that controls transmission of data from a data acquisition device that acquires data necessary for use or learning of the machine learning model, The transmission control unit of the data collection device does not cause the data acquisition device to transmit data when the number of people estimated by the number of people estimation unit is less than the reference value, or reduces the amount of data transmitted per unit time by the data acquisition device to the data collection device compared to when the number of people estimated by the number of people estimation unit is equal to or greater than the reference value. Effect of the Invention

[0009] According to the present disclosure, a data collection device is provided that can appropriately transmit data from a data acquisition device to a server while suppressing power consumption. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a schematic diagram of a machine learning system. [Diagram 2] FIG. 2 is a diagram illustrating a schematic hardware configuration of the server. [Diagram 3] FIG. 3 is a functional block diagram of the processor of the server. [Figure 4] FIG. 4 is a flowchart showing the flow of image data collection processing performed by the processor of the server. [Diagram 5] FIG. 5 is a flowchart showing the flow of a transportation vehicle dispatch process using a machine learning model, which is performed in the processor of the server. [Figure 6] FIG. 6 is a flowchart showing the flow of the learning process of the machine learning model performed in the processor of the server. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, the embodiments will be described in detail with reference to the drawings. In the following description, like elements are designated by like reference numerals.

[0012] <Machine learning system configuration> The configuration of a machine learning system 1 according to an embodiment will be described with reference to Figures 1 and 2. Figure 1 is a schematic configuration diagram of the machine learning system 1. The machine learning system 1 trains a machine learning model used in a server. The machine learning system also functions as a data collection system that collects data necessary for using and training the machine learning model.

[0013] As shown in FIG. 1, the machine learning system 1 has a plurality of fixed data acquisition devices 10, a mobile data acquisition device 15, and a server 20 that can communicate with the fixed data acquisition devices 10 and the mobile data acquisition devices 15. Each of the fixed data acquisition devices 10 and the server 20 are configured to be able to communicate with each other via a communication network 4 and, if necessary, a wireless base station 5. Each of the mobile data acquisition devices 15 and the server 20 are configured to be able to communicate with each other via the communication network 4 and the wireless base station 5. The communication network 4 is, for example, configured with an optical communication line, and includes the Internet network, a carrier network, and the like. The wireless base station 5 is connected to the communication network 4 via a gateway (not shown). Various wide-area wireless communications with long communication distances can be used for communication between the fixed data acquisition devices 10 and the mobile data acquisition devices 15 and the wireless base station 5, and for example, communications conforming to any communication standard such as 4G, LTE, 5G, WiMAX, etc., established by 3GPP and IEEE, are used.

[0014] In particular, in this embodiment, the server 20 communicates with the fixed data acquisition device 10 and the mobile data acquisition device 15 located within a predetermined target area. The target area is an area surrounded by a predetermined boundary, and is, for example, a smart city defined as a "sustainable city or region that solves the various problems faced by the city or region and continues to create new value through advanced management (planning, maintenance, management, operation, etc.) while utilizing new technologies such as ICT." Alternatively, the target area is a part of the smart city. The server 20 may also communicate with the data acquisition devices 10 and 15 outside the predetermined target area via a wireless base station 5.

[0015] The data acquisition devices 10 and 15 acquire data necessary for use or learning of a machine learning model, which will be described later. In particular, in this embodiment, the data acquisition devices 10 and 15 acquire information within a predetermined target area.

[0016] The fixed data acquisition devices 10 each include a device that acquires information on a specific area within the target area. Specifically, the fixed data acquisition devices 10 each include a photographing device that photographs an arbitrary area within the target area to generate image data, for example, a plurality of surveillance cameras arranged within the target area. Such surveillance cameras photograph people located within the arbitrary area within the target area. The fixed data acquisition devices 10 may also include devices arranged within the target area to acquire information other than images. Such devices include, for example, an infrared sensor that detects the temperature of an arbitrary area within the target area, an external server other than the server 20 of the machine learning system 1, and the like. The infrared sensor detects the body temperature of a person within the arbitrary area within the target area. The external server acquires weather information, such as the weather, temperature, and humidity of the target area.

[0017] The fixed data acquisition device 10 includes a communication module directly connected to the communication network 4, and data acquired by the fixed data acquisition device 10 is transmitted from the fixed data acquisition device 10 to the server 20 via the communication network 4. Alternatively, the fixed data acquisition device 10 may include a communication module capable of communicating with the wireless base station 5. In this case, the data acquired by the fixed data acquisition device 10 is transmitted from the fixed data acquisition device 10 to the server 20 via the wireless base station 5 and the communication network 4.

[0018] The mobile data acquisition device 15 includes a device that acquires information on people in a predetermined target area. Specifically, the mobile data acquisition device 15 includes a terminal device carried by a person in the target area. Such terminal devices include, for example, a mobile terminal and a wearable terminal (such as a smart watch or smart glasses). Such terminal devices acquire information on each person, such as, for example, location information of each person in the target area, vital signs (body temperature, heart rate, blood pressure, respiratory rate), blood oxygen concentration, blood glucose level, etc.

[0019] The mobile data acquisition device 15 includes a communication module capable of communicating with the wireless base station 5. The data acquired by the mobile data acquisition device 15 is transmitted from the mobile data acquisition device 15 to the server 20 via the wireless base station 5 and the communication network 4.

[0020] The server 20 is connected to a plurality of data acquisition devices 10, 15 via the communication network 4. The server 20 performs processing using a machine learning model, which will be described later, based on data acquired by the plurality of data acquisition devices 10, 15. In addition, the server 20 also functions as a learning device that trains the machine learning model used in the server 20. The server 20 also functions as a data collection device that collects data necessary for using and learning the machine learning model from the plurality of data acquisition devices 10, 15.

[0021] Fig. 2 is a diagram illustrating a schematic hardware configuration of the server 20. As illustrated in Fig. 2, the server 20 includes a communication module 21, a storage device 22, and a processor 23. The server 20 may also include input devices such as a keyboard and a mouse, and output devices such as a display and a speaker.

[0022] The communication module 21 is an example of a communication device that communicates with devices outside the server 20. The communication module 21 includes an interface circuit for connecting the server 20 to the communication network 4. The communication module 21 is configured to be able to communicate with each of the multiple data acquisition devices 10 and 15 via the communication network 4 and the wireless base station 5.

[0023] The storage device 22 is an example of a storage device that stores data. The storage device 22 has, for example, a hard disk drive (HDD), a solid state drive (SSD), or an optical recording medium. The storage device 22 may also have a volatile semiconductor memory (for example, RAM), a non-volatile semiconductor memory (for example, ROM), or the like. The storage device 22 stores computer programs for executing various processes by the processor 23, and various data used when the various processes are executed by the processor 23. In particular, the storage device 22 stores data received from the data acquisition devices 10 and 15, and data used for processing by the machine learning model and for learning the machine learning model.

[0024] The processor 23 has one or more CPUs and their peripheral circuits. The processor 23 may further have an arithmetic circuit such as a GPU, a logical arithmetic unit, or a numerical arithmetic unit. The processor 23 executes various processes based on computer programs stored in the storage device 22. Specific processes executed by the processor 23 of the server 20 will be described later.

[0025] <Machine learning model> In this embodiment, when performing a predetermined process in the server 20, a machine learning model that has been machine-learned is used. In particular, in this embodiment, the machine learning model is used only when the number of people in a predetermined target area is equal to or greater than a predetermined reference value. Various models are possible as such machine learning models, but the following will be described using as an example a machine learning model that predicts the number of transportation vehicles (such as taxis) that will be required in the future in each region of the target area based on the flow of people in the target area. Such a machine learning model is used only when the number of people in the target area is greater than or equal to a reference value, because the error becomes large and optimization is not possible unless the number of people in the area within the target area is relatively large. Therefore, the reference value here is a value such that if the number of people in the target area decreases any further, the estimation accuracy in the machine learning model will be significantly reduced to a certain level or less.

[0026] Specifically, image data captured by a plurality of surveillance cameras included in the fixed data acquisition device 10 is input as input parameters to the machine learning model. Weather information such as the weather, temperature, and humidity of the target area may also be input to the machine learning model. In addition, time information such as the current time and date may also be input to the machine learning model. The machine learning model recognizes people included in the image represented by the image data, and estimates the gender, age, moving direction, and number of people in the target area of ​​each person. Then, based on these recognition results and estimation results, the machine learning model outputs the number of transportation vehicles required in each region of the target area after a predetermined time (hereinafter referred to as the "expected required number") as an output parameter. Note that the machine learning model may be a model that directly outputs the expected required number of vehicles in each region of the target area when the above-mentioned input parameters are input, without performing the above-mentioned recognition and estimation.

[0027] Various machine learning algorithms can be used for the machine learning model. In this embodiment, the machine learning model is a model that is learned by supervised learning, such as a neural network (NN), a support vector machine (SVM), or a decision tree (DT). In particular, in this embodiment, the machine learning model is preferably a recurrent neural network (RNN) model that inputs image data in a time series as an input parameter.

[0028] The machine learning model may be any model that is used only when the number of people in the target area is equal to or greater than a reference value. Therefore, the machine learning model may be, for example, a model that predicts future demand for a shop in the target area based on the flow of people in the target area and the state of each person. In this case, image data captured by a plurality of surveillance cameras included in the fixed data acquisition device 10 is input as input parameters to the machine learning model. Weather information and time information for the target area may also be input to the machine learning model. Furthermore, information such as vital signs of people in the target area may be input as input parameters to the machine learning model. Then, when such input parameters are input, the machine learning model outputs drinking water, food, etc. that will be needed in the future at the shop in the target area as output parameters.

[0029] Such a machine learning model is trained using a training data set including data used as input parameters and values ​​of output parameters (correct values ​​or correct labels) corresponding to the data. In this embodiment, the training data set includes data acquired by the data acquisition devices 10 and 15 at a certain point in time and the number of vehicles required after a predetermined time from that point in time (correct value).

[0030] The correct answer value is calculated based on the number of transportation vehicles actually dispatched to each area after a predetermined time, the number of people waiting for transportation vehicles to be dispatched in each area after a predetermined time, and the number of transportation vehicles waiting to be boarded in each area after a predetermined time, etc. For example, the correct answer value is calculated by the server 20 based on people and vehicles shown in an image captured by the surveillance camera of the fixed data acquisition device 10.

[0031] In machine learning of an RNN model, for example, the learning parameters of the RNN model (e.g., weights and biases) are repeatedly updated by a known backpropagation method so that the difference between the output value of the RNN model and the correct value of the output parameter included in the learning dataset becomes small. As a result, the RNN model is trained, and a trained RNN model is generated. Note that any known method can be used for machine learning in the machine learning model. In addition, in this specification, the learning parameters refer to parameters whose values ​​are repeatedly updated by learning.

[0032] <Data collection> Next, a data collection process in the server 20 will be described with reference to Figures 3 and 4. The server 20 collects data from the data acquisition devices 10 and 15 when using and learning the machine learning model.

[0033] 3 is a functional block diagram of the processor 23 of the server 20. As shown in FIG. 3, the processor 23 has a number of people estimation unit 231, a transmission control unit 232, a data receiving unit 233, a model execution unit 234, a command transmission unit 235, a data set creation unit 236, and a model learning unit 237. These functional blocks of the processor 23 of the server 20 are, for example, functional modules realized by a computer program running on the processor 23. Alternatively, these functional blocks of the processor 23 may be dedicated arithmetic circuits provided in the processor 23. In particular, the number of people estimation unit 231 and the transmission control unit 232 are used to collect data from the data acquisition devices 10 and 15.

[0034] The number of people estimation unit 231 estimates the number of people in the target area. The number of people estimation unit 231 estimates the number of people in the target area based on, for example, the number of mobile data acquisition devices 15 located in the target area, in particular, the number of terminal devices located in the target area.

[0035] The number of people estimation unit 231 may estimate the number of people in the target area by other methods. For example, the number of people estimation unit 231 estimates the number of people in the target area based on image data (image data from a surveillance camera) acquired by the fixed data acquisition device 10. In this case, the number of people estimation unit 231 inputs the image data received from the fixed data acquisition device 10 into a known model for object detection and calculates the number of detected objects (people). Then, the number of people estimation unit 231 estimates the number of people in the target area based on the number of objects (people) calculated in this way.

[0036] The transmission control unit 232 controls the transmission of data from the data acquisition devices 10 and 15. The transmission control unit 232 controls, for example, the frequency of data transmission from each of the data acquisition devices 10 and 15. In other words, the transmission control unit 232 controls the ratio of data acquired by the data acquisition devices 10 and 15 to be transmitted to the server 20. Therefore, when the data transmission frequency is controlled to be high, all data acquired by the data acquisition devices 10 and 15 (all data used in the machine learning model) is transmitted to the server 20. On the other hand, when the data transmission frequency is controlled to be low, only a portion of the data acquired by the data acquisition devices 10 and 15 (a portion of the data used in the machine learning model) is transmitted to the server 20. In order to control the frequency of data transmission from each of the data acquisition devices 10 and 15, the transmission control unit 232 transmits, for example, to each of the data acquisition devices 10 and 15, an instruction regarding the frequency of image data transmission.

[0037] The transmission control unit 232 may control the transmission speed of image data from the data acquisition devices 10, 15 to the server 20. In this case, the transmission control unit 232 transmits, for example, to each of the data acquisition devices 10, 15, a command regarding the upper limit transmission speed of data to the server 20. In any case, the transmission control unit 232 controls the amount of data transmission per unit time that the data acquisition devices 10, 15 transmit to the server 20.

[0038] FIG. 4 is a flowchart showing the flow of image data collection processing performed by the processor 23 of the server 20.

[0039] As shown in Fig. 4, first, the number of people estimation unit 231 estimates the number of people in the target area (step S11). In this embodiment, the number of people estimation unit 231 estimates the number of mobile data acquisition devices 15 located in the target area. Specifically, the number of people estimation unit 231 receives location information of each terminal device from each terminal device, and calculates the number of terminal devices identified as being located in the target area from the received location information. Then, the number of people estimation unit 231 estimates the number of people in the target area assuming that the greater the number of terminal devices calculated in this way, the greater the number of people in the target area.

[0040] When the number of people is estimated in step S11, the transmission control unit 232 determines whether the estimated number of people is equal to or greater than a predetermined reference value (step S12). Here, the reference value is the same or substantially the same as the reference value used when determining whether or not a machine learning model is used.

[0041] When it is determined in step S12 that the estimated number of people is equal to or greater than the reference value, the transmission control unit 232 transmits to the data acquisition devices 10, 15 a command to permit data transmission from the data acquisition devices 10, 15 (step S13). That is, the transmission control unit 232 controls the frequency of data transmission from each of the data acquisition devices 10, 15 to the maximum. In this case, the data acquisition devices 10, 15 transmit all acquired data (all data used in the machine learning model) to the server 20.

[0042] On the other hand, when it is determined in step S12 that the estimated number of people is less than the reference value, the transmission control unit 232 transmits a command to the data acquisition devices 10, 15 to prohibit data transmission from the data acquisition devices 10, 15 (step S14). That is, the transmission control unit 232 controls the frequency of data transmission from each of the data acquisition devices 10, 15 to zero. In this case, the data acquisition devices 10, 15 do not transmit the acquired data to the server 20.

[0043] In this embodiment, data is transmitted from the data acquisition devices 10 and 15 when the estimated number of people is equal to or greater than the reference value, and data is not transmitted from the data acquisition devices 10 and 15 when the estimated number of people is less than the reference value. In addition, when the estimated number of people is less than the reference value, the machine learning model is not used, and learning of the machine learning model using data in such a case is not performed. Therefore, data transmission from the data acquisition devices 10 and 15 can be stopped without affecting the use or learning of the machine learning model, and therefore data can be appropriately transmitted while suppressing power consumption.

[0044] In this embodiment, when it is determined in step S12 that the estimated number of people is less than the reference value, the transmission control unit 232 does not allow the data acquisition devices 10 and 15 to transmit data to the server 20. However, even in this case, the transmission control unit 232 may permit the data transmission to the server 20 if the amount of data transmission per unit time transmitted by the data acquisition devices 10 and 15 to the server 20 is reduced compared to when the estimated number of people is equal to or greater than the reference value. Here, as described above, when the number of people estimation unit 231 estimates the number of people in the target area based on image data acquired by the fixed data acquisition device 10, it is necessary for the server 20 to receive the image data in order to estimate the number of people. In this way, if the transmission control unit 232 reduces the amount of data transmission even when the number of people in the target area is small, the number of people estimation unit 231 estimates the number of people, while preventing the data acquisition devices 10 and 15 from consuming excessive power.

[0045] <Use of machine learning models> Next, a process using the machine learning model will be described with reference to Fig. 3 and Fig. 5. In this embodiment, the server 20 calculates the expected number of vehicles required in each region of the target area using the machine learning model. In addition, the server 20 issues a dispatch command to vehicles around each region based on the expected number of vehicles required in each region. In calculating the expected number of vehicles required and issuing an operation command to surrounding vehicles, the data receiving unit 233, the model executing unit 234, and the command transmitting unit 235 are used.

[0046] The data receiving unit 233 receives data acquired by the data acquiring devices 10, 15 from the data acquiring devices 10, 15 that can communicate with the server 20. When the data receiving unit 233 receives data from the data acquiring devices 10, 15, it stores the received data in the storage device 22.

[0047] The model execution unit 234 inputs the values ​​of the input parameters to the above-mentioned machine learning model, and causes the machine learning model to output the values ​​of the output parameters. In particular, the model execution unit 234 inputs the data received from the data acquisition devices 10 and 15, which are stored in the storage device 22, as input parameters to the machine learning model. When the values ​​of such input parameters are input, the machine learning model outputs the expected number of vehicles required (output parameters) in each region of the target area.

[0048] The command transmission unit 235 transmits a dispatch command to transportation vehicles around each area based on the expected number of vehicles required in each area. The command transmission unit 235 transmits a dispatch command to transportation vehicles that can communicate via the communication network 4. If the transportation vehicle is an autonomous vehicle, the transportation vehicle automatically travels toward the designated area according to the received dispatch command. On the other hand, if the transportation vehicle is a vehicle manually driven by a driver, the transportation vehicle notifies the driver of the received dispatch command by displaying it on a display or the like.

[0049] FIG. 5 is a flowchart showing the flow of a transportation vehicle dispatch process using a machine learning model, which is performed by the processor 23 of the server 20.

[0050] When performing the vehicle dispatch process, first, the data receiving unit 233 receives data from the data acquisition devices 10, 15 (step S21) and stores the received data in the storage device 22. As described above, when the number of people in the target area is small, data is not transmitted from the data acquisition devices 10, 15. Therefore, the data receiving unit 233 receives data from the data acquisition devices 10, 15 only when the number of people in the target area is large.

[0051] When the data is stored in the storage device 22 by the data receiving unit 233, the model executing unit 234 inputs the data stored in the storage device 22 to the machine learning model (step S22). The machine learning model to which the data has been input outputs the predicted number of vehicles required for each region.

[0052] When the model execution unit 234 outputs the estimated number of vehicles required for each region, the command transmission unit 235 transmits a dispatch command to the surrounding transport vehicles according to the estimated number of vehicles required for each region. The command transmission unit 235 compares the current number of transport vehicles in each region with the estimated number of vehicles required for each region after a predetermined time. Then, for a region where the estimated number of vehicles required after a predetermined time is smaller than the current number of transport vehicles, the command transmission unit 235 transmits a command to some of the transport vehicles so that transport vehicles head from other regions to the region. On the other hand, for a region where the estimated number of vehicles required after a predetermined time is larger than the current number of transport vehicles, the command transmission unit 235 transmits a command to some of the transport vehicles so that transport vehicles head from the region to other regions that require vehicle dispatch.

[0053] <Learning machine learning models> Next, the learning process of the machine learning model in the server 20 will be described with reference to Fig. 3 and Fig. 6. In this embodiment, the server 20 learns a machine learning model that calculates the expected required number of units based on the data received from the data acquisition devices 10 and 15. For learning the machine learning model, a data set creation unit 236 and a model learning unit 237 are used (see Fig. 3).

[0054] The data set creation unit 236 creates a learning data set used to train the machine learning model. In this embodiment, the learning data set includes data acquired by the data acquisition devices 10 and 15 at a certain point in time and the number of vehicles required after a predetermined time from that point in time (correct answer value).

[0055] The data acquired by the data acquisition devices 10, 15 at each time point is transmitted from the data acquisition devices 10, 15 to the server 20 and stored in the storage device 22. Therefore, the dataset creation unit 236 acquires from the storage device 22 the data acquired by the data acquisition devices 10, 15 at each time point regarding the input parameters used in the learning dataset.

[0056] In this embodiment, the number of vehicles required after a certain time (correct answer value) is calculated based on the number of transportation vehicles actually dispatched to each area after the specified time, the number of people waiting for transportation vehicles to be dispatched in each area after the specified time, and the number of transportation vehicles waiting to be boarded in each area after the specified time. For example, when the number of people waiting for transportation vehicles to be dispatched in each area after the specified time is large, the number of vehicles required after the specified time is set to a number greater than the number of transportation vehicles actually dispatched to each area after the specified time. On the other hand, when the number of transportation vehicles waiting to be boarded in each area after the specified time is large, the number of vehicles required after the specified time is set to a number less than the number of transportation vehicles actually dispatched to each area after the specified time. The correct answer value is calculated by the server 20 based on people and vehicles shown in the image captured by the surveillance camera of the fixed data acquisition device 10. Then, the data acquired by the data acquisition devices 10 and 15 at a certain point in time is combined with the calculated number of vehicles required a certain time after the certain point in time (correct answer value) to create one learning data set. This operation is then repeated to create a learning data set.

[0057] The model learning unit 237 learns the machine learning model based on the learning dataset created by the dataset creation unit 236. Specifically, as described above, the model learning unit 237 updates the learning parameters used in the machine learning model by using a known backpropagation method or the like.

[0058] FIG. 6 is a flowchart showing the flow of the learning process of the machine learning model performed by the processor 23 of the server 20.

[0059] When performing the learning process, first, similarly to step S21 shown in FIG. 5, the data receiving unit 233 receives data from the data acquisition devices 10 and 15 (step S31) and stores the received data in the storage device 22.

[0060] When the amount of data required for learning is stored in the storage device 22 by the data receiving unit 233, the dataset creating unit 236 creates a learning dataset (step S32). The dataset creating unit 236 creates a learning dataset using the data stored in the storage device 22 in step S31 as input parameters. The dataset creating unit 236 also performs object detection or the like on image data of a surveillance camera stored in the storage device 22, and detects people and vehicles included in an image represented by the image data. Then, the dataset creating unit 236 calculates the number of vehicles required at each time based on the detected people and vehicles, and creates a learning dataset using the calculated number of vehicles required as a correct answer value of an output parameter.

[0061] When a certain number of learning datasets are created by the dataset creation unit 236 in step S32, the model learning unit 237 uses the created datasets to train a machine learning model (step S33). Then, the model learning unit 237 uses the learning parameters of the trained machine learning model to update the values ​​of the learning parameters of the machine learning model used in step S22 of Fig. 5 (step S34). After the values ​​of the learning parameters of the machine learning model are updated, in step S22, the expected required number of vehicles is calculated by the machine learning model using the updated learning parameters.

[0062] Although the preferred embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and changes can be made within the scope of the claims. [Explanation of symbols]

[0063] 1. Machine Learning Systems 4. Communication Network 5. Radio base station 10 Fixed data acquisition device 15 Mobile data acquisition device 20 Servers

Claims

[Claim 1] A data collection device that collects data necessary for use or learning of a machine learning model that is used only when the number of people in a target area, which is at least a portion of a smart city, is equal to or greater than a predetermined threshold value, A number of people estimation unit that estimates the number of people in the target area; A transmission control unit that controls transmission of data from a data acquisition device that acquires data necessary for use or learning of the machine learning model, The transmission control unit of the data collection device does not cause the data acquisition device to transmit data when the number of people estimated by the number of people estimation unit is less than the reference value, or reduces the amount of data transmitted per unit time by the data acquisition device to the data collection device compared to when the number of people estimated by the number of people estimation unit is equal to or greater than the reference value.

Images