Estimation system, model learning system, estimation method, model learning method, and program

Abstract

To provide an estimation system, a model learning system, an estimation method, a model learning method, and a program capable of easily estimating the number of objects existing in each of a plurality of areas in a predetermined space.SOLUTION: An estimation system of the present invention includes obtaining means 51 for obtaining information related to a radio signal based on a radio signal received from each of at least one terminal present in any area of a plurality of areas in a predetermined space, and estimation means 52 for estimating the number of objects present in any area based on the obtained information related to the radio signal and a learned model based on mechanical learning using training data, wherein the training data includes information related to a radio signal obtained within a first predetermined period in a predetermined area of the plurality of areas as a feature amount, and includes the number of objects present in the predetermined area within the first predetermined period detected using a predetermined device as a correct answer label.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an estimation system, a model learning system, an estimation method, a model learning method, and a program. [Background technology]

[0002] Conventionally, there is known a technique for estimating the number of objects (e.g., the number of people) present in a predetermined space in order to analyze the degree of congestion, the flow of people, and other behaviors in the space (for example, Patent Document 1). The technique described in Patent Document 1 counts (estimates) the number of people present in a predetermined space based on video captured by a camera. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2015-70359 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in order to estimate the number of objects (number of people) in detail, it is necessary to place many cameras in a given space, which may result in increased costs for estimating the number of objects (number of people).

[0005] The present invention has been made in consideration of the above-mentioned problems, and aims to provide an estimation system, a model learning system, an estimation method, a model learning method, and a program that can easily estimate the number of objects present in each of multiple areas within a specified space. [Means for solving the problem]

[0006] In order to solve the above problems, firstly, the present invention provides an estimation system comprising: an acquisition means for acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in any one of a plurality of regions within a predetermined space; and an estimation means for estimating the number of objects present in any one of the regions based on information about the acquired wireless signal and a trained model based on machine learning using training data, wherein the training data includes features based on information about wireless signals acquired in a predetermined region of the plurality of regions within a first predetermined period, and a ground truth label based on the number of objects present in the predetermined region within the first predetermined period, detected using a predetermined device (Invention 1).

[0007] Here, the number of objects (e.g., number of people, etc.) present in a certain area and the number of terminal devices present in that certain area may be equal or different. For example, if one terminal device is provided for each of one or more objects (e.g., people, etc.) present in that area, the number of objects present in that certain area is considered to be equal to the number of terminal devices present in that certain area. On the other hand, if at least one of the one or more objects present in that area is not provided with a terminal device, or if at least one object is provided with multiple terminal devices, the number of objects present in that certain area may be different from the number of terminal devices present in that certain area.

[0008] According to this invention (Invention 1), it is possible to estimate the number of objects (e.g., the number of people) present in one of multiple areas within a predetermined space using wireless signals received from one or more terminal devices present in the area and a trained model. Furthermore, according to this invention (Invention 1), the trained model is generated using training data including features based on information about wireless signals acquired in a predetermined area of ​​the multiple areas within a first predetermined period and ground truth labels based on the number of objects present in the predetermined area within the first predetermined period, as detected using a predetermined device. Therefore, by inputting information about wireless signals in one of the areas into this trained model, it is possible to estimate the number of objects present in the area. This makes it possible to easily estimate the number of objects present in each of multiple areas within the predetermined space based on wireless signals received from terminal devices present in each of the multiple areas, even if, for example, an imaging device is not installed in each of the multiple areas.

[0009] In the above invention (invention 1), the predetermined device may be an imaging device (invention 2).

[0010] According to this invention (Invention 2), a trained model can be generated using training data including correct labels based on the number of objects detected from an image captured by an imaging device, and based on this trained model, it becomes possible to easily estimate the number of objects present in any area.

[0011] In the above invention (Invention 1), the acquisition means may acquire information about the wireless signals based on the wireless signals received from each of the one or more terminal devices present in any of the areas within a second predetermined period, and the estimation means may estimate the number of objects present in any of the areas within the second predetermined period (Invention 3).

[0012] According to this invention (Invention 3), it becomes possible to easily estimate the number of objects present in any area within the second predetermined period.

[0013] In the above invention (Invention 1), the feature may include any of information regarding the transmission interval of the wireless signal, information regarding the packet length of the wireless signal, information regarding the identification information of the terminal device, information regarding the received signal strength of the wireless signal, information regarding the amount of data that the terminal device can transmit within a third predetermined period, and information regarding the transmission frequency of the wireless signal (Invention 4).

[0014] According to this invention (Invention 4), it becomes possible to estimate the number of objects present in any area within a specified space using any of the following information: information regarding the transmission interval of the radio signal; information regarding the packet length of the radio signal; information regarding the identification information of the terminal device; information regarding the received signal strength of the radio signal; information regarding the amount of data that the terminal device can transmit within a third specified period; and information regarding the transmission frequency of the radio signal.

[0015] In the above invention (invention 4), the identification information of the terminal device may be a MAC address of the terminal device (invention 5).

[0016] According to this invention (Invention 5), it becomes possible to estimate the number of objects present in any area within a predetermined space using a wireless signal including a MAC address.

[0017] In the above invention (Invention 1), the wireless signal may be a probe request signal (Invention 6).

[0018] According to this invention (Invention 6), it becomes possible to estimate the number of objects present in any area within a predetermined space using probe request signals.

[0019] In the above invention (Invention 1), if an entrance / exit is provided for entering and exiting the specified space, the specified area may be an area including the entrance / exit (Invention 7).

[0020] According to this invention (Invention 7), it becomes possible to detect the number of objects present in a predetermined area by using a predetermined device to measure the number of objects passing through a doorway.

[0021] In the above invention (Invention 1), when a payment area where a predetermined payment process is performed is provided in the predetermined space, the predetermined area may be an area that includes the payment area (Invention 8).

[0022] According to this invention (Invention 8), it becomes possible to detect the number of objects present in a specified area by using a specified device to measure the number of objects present in a payment area in order to perform a specified payment process.

[0023] Secondly, the present invention provides a model learning system comprising: a first information acquisition means for acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in a predetermined area among a plurality of areas within a predetermined space; a second information acquisition means for acquiring information about the number of objects present in the predetermined area detected using the predetermined device; and a learning means for learning a model used to estimate the number of objects present in any of the plurality of areas by machine learning using learning data, wherein the learning data includes features based on the information about the wireless signal and a correct answer label based on the number of objects present in the predetermined area (Invention 9).

[0024] Third, the present invention provides an estimation method in which a computer executes the following steps: acquiring information about a wireless signal based on a wireless signal received from each of one or more terminal devices present in one of a plurality of regions within a predetermined space; and estimating the number of objects present in one of the regions based on the information about the acquired wireless signal and a trained model based on machine learning using training data, wherein the training data includes features based on information about wireless signals acquired in a predetermined region of the plurality of regions within a first predetermined period, and a ground truth label based on the number of objects present in the predetermined region within the first predetermined period, as detected using a predetermined device (Invention 10).

[0025] Fourth, the present invention provides a model learning method in which a computer executes the following steps: acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in a predetermined area among a plurality of areas within a predetermined space; acquiring information about the number of objects present in the predetermined area detected using the predetermined device; and learning a model used to estimate the number of objects present in any of the plurality of areas by machine learning using learning data, wherein the learning data includes features based on the information about the wireless signal and a ground truth label based on the number of objects present in the predetermined area (Invention 11).

[0026] Fifth, the present invention provides a program for causing a computer to realize the following functions: a function of acquiring information about wireless signals based on wireless signals received from each of one or more terminal devices present in any one of a plurality of regions within a predetermined space; and a function of estimating the number of objects present in any one of the regions based on information about the acquired wireless signals and a trained model based on machine learning using training data, wherein the training data includes features based on information about wireless signals acquired in a predetermined region of the plurality of regions within a first predetermined period, and a correct answer label based on the number of objects present in the predetermined region within the first predetermined period, detected using a predetermined device (Invention 12).

[0027] Sixth, the present invention provides a program for causing a computer to realize the following functions: a function for acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in a predetermined area among a plurality of areas within a predetermined space; a function for acquiring information about the number of objects present in the predetermined area detected using the predetermined device; and a function for learning a model used to estimate the number of objects present in any of the plurality of areas by machine learning using learning data, wherein the learning data includes features based on the information about the wireless signal and a correct answer label based on the number of objects present in the predetermined area (Invention 13). [Effects of the Invention]

[0028] According to the estimation system, model learning system, estimation method, model learning method, and program of the present invention, it is possible to easily estimate the number of objects present in each of a plurality of regions within a predetermined space. [Brief explanation of the drawings]

[0029] [Figure 1] 1 is a diagram illustrating a basic configuration of an estimation system according to an embodiment of the present invention. [Figure 2] FIG. 1 is a block diagram showing a configuration of a communication device. [Figure 3] FIG. 1 is a block diagram showing a configuration of an estimation device. [Figure 4] FIG. 1 is a functional block diagram for explaining functions that play a major role in an estimation system and a model learning system. [Figure 5] FIG. 10 is a diagram illustrating an example of the configuration of acquired data. [Figure 6] FIG. 10 is a diagram illustrating an example of the configuration of learning data. [Figure 7] 1 is a flowchart illustrating an example of main processing of a model learning system according to an embodiment of the present invention. [Figure 8] 3 is a flowchart illustrating an example of main processing of the estimation system according to one embodiment of the present invention. [Figure 9] 10A and 10B are diagrams illustrating an example of how the functions of the estimation system and the model learning system are shared between an estimation device and a learning device. DETAILED DESCRIPTION OF THE INVENTION

[0030] Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings. However, this embodiment is an example and the present invention is not limited to this embodiment.

[0031] (1) Basic configuration of the estimation system and model learning system Fig. 1 is a diagram illustrating a schematic diagram of the basic configuration of an estimation system and a model learning system according to an embodiment of the present invention. As shown in Fig. 1, in the estimation system according to this embodiment, wireless signals transmitted from one or more terminal devices 10 present in each of a plurality of areas (in the illustrated example, prepared food area, raw meat area, fresh fish area, beverage area, daily necessities area 1, daily necessities area 2, daily necessities area 3, vegetable area, entrance / exit area 1, checkout area, and entrance / exit area 2) within a predetermined space SP of a store or facility such as a supermarket are received by a communication device 20 provided in each area. An estimation device 40 estimates the number of objects (e.g., the number of people) present in each area based on information related to the wireless signals received by each communication device 20.

[0032] Furthermore, in the model learning system according to this embodiment, the estimation device 40 learns a model used to estimate the number of objects present in one of the multiple areas by machine learning using learning data including information about wireless signals received within a first predetermined period (e.g., 60 seconds) by a communication device 20 installed in a predetermined area (in the example shown in the figure, the checkout area) among the multiple areas, and the number of objects present in the predetermined area within the first predetermined period, detected using an imaging device 30 installed in the predetermined area.

[0033] In the example shown in Fig. 1, at least one communication device 20 is provided in each of a plurality of areas. Note that in Fig. 1, to avoid cluttering the diagram, the communication devices 20 are only shown in the checkout area. Also, in Fig. 1, the communication ranges of the communication devices 20 provided in each area are indicated by circular dashed lines. Furthermore, in this embodiment, the communication devices 20, the imaging device 30, and the estimation device 40 are communicably connected via a communication network NW (network) such as the Internet or a LAN (Local Area Network).

[0034] Each terminal device 10 is configured to be able to perform wireless communication with a communication device 20 provided in one of the multiple areas within the space SP using a wireless LAN (for example, Wi-Fi (registered trademark)). Furthermore, each terminal device 10 is configured to transmit a wireless signal (for example, a probe request signal) including its own identification information (for example, a MAC address) at predetermined intervals (for example, every few seconds) in order to perform wireless communication with the communication device 20. Each terminal device 10 may be a terminal device operated by an individual user, such as a mobile terminal, a smartphone, a PDA (Personal Digital Assistant), a personal computer, or a television receiver with a two-way communication function (including so-called multi-function smart televisions).

[0035] The communication device 20 is provided in each area at a position where it can perform wireless communication with each terminal device 10 using a wireless LAN. The communication device 20 may be configured to measure a received signal strength indicator (RSSI) when receiving a wireless signal transmitted from each terminal device 10. The communication device 20 may be a device that relays wireless communication between each terminal device 10 existing in the same area, a device that relays wireless communication between each terminal device 10 and a terminal device (not shown) existing in another area within the space SP, or a device that relays communication between each terminal device 10 and another device connected via a communication network NW. Furthermore, the communication device 20 may be a packet capture device.

[0036] The imaging device 30 may be, for example, an imaging device (e.g., a digital camera or digital video camera) that captures moving images and / or still images, and is installed to capture an image of a predetermined area (here, the checkout area) at a predetermined position within the space SP. The imaging device 30 may be installed on a predetermined moving object and capture an image of a predetermined imaging range while moving in accordance with the movement of the moving object. The imaging device 30 is configured to perform imaging processing at a predetermined frame rate (e.g., 30 fps (flames per second)) and transmit the captured images to the estimation device 40 via the communication network NW. The imaging device 30 may perform imaging processing when it receives a predetermined imaging instruction signal from the estimation device 40.

[0037] Here, the imaging device 30 may be an imaging device (for example, an omnidirectional camera) that captures an omnidirectional image (for example, an image of 360° around the center of a predetermined area (for example, 360° around in the horizontal direction)). In this case, it is possible to capture an image of an object (for example, a user of the terminal device 10) over a wide range of the predetermined area.

[0038] Furthermore, the imaging device 30 may be an imaging device (for example, an infrared camera) that captures infrared images. This allows the object (for example, a user of the terminal device 10) to be detected in the captured image even when the object is present in an environment with poor visibility (for example, at night, in a dark place, in bad weather, etc.).

[0039] Furthermore, the imaging device 30 may be an imaging device that captures stereo images (for example, a stereo camera, etc.). Here, a stereo image may be, for example, a set of two images having a predetermined parallax.

[0040] The estimation device 40 is configured to acquire, via the communication network NW, information about wireless signals transmitted from a communication device 20 provided in any area within the space SP, and estimate the number of objects present in that area based on the acquired information about the wireless signals. The estimation device 40 may be, for example, a general-purpose personal computer.

[0041] (2) Configuration of communication device The configuration of the communication device 20 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing the internal configuration of the communication device 20. As shown in Fig. 2, the communication device 20 includes a CPU (Central Processing Unit) 21, a ROM (Read Only Memory) 22, a RAM (Random Access Memory) 23, a storage device 24, and a communication interface unit 25, and is provided with a bus 20a for transmitting control signals or data signals between the respective units.

[0042] When the power supply to the communication device 20 is turned on, the CPU 21 loads various programs stored in the ROM 22 or the storage device 24 into the RAM 23 and executes them.

[0043] The storage device 24 may be, for example, a non-volatile storage device such as a flash memory, an SSD (Solid State Drive), a magnetic storage device (for example, an HDD (Hard Disk Drive), a floppy disk (registered trademark), a magnetic tape, etc.), an optical disk, or a volatile storage device such as RAM, and stores programs executed by the CPU 21 and data referenced by the CPU 21.

[0044] The communication interface unit 25 includes an interface circuit for performing communication using the above-mentioned wireless communication method, and an interface circuit for performing communication via the communication network NW. The interface circuit for performing communication using the wireless communication method may be provided with an RSSI circuit for detecting received signal strength indicator (RSSI) when receiving a wireless signal transmitted from each terminal device 10.

[0045] Here, every time the CPU 21 receives (acquires) a wireless signal from any of the terminal devices 10 via the communication interface unit 25, the CPU 21 may store, in the RAM 23 or the storage device 24, acquisition date and time information (e.g., year, month, day, hour, minute, second, etc.) when the wireless signal was received (acquired) and the value of the received signal strength of the wireless signal detected by the RSSI circuit in association with the data of the wireless signal. Then, every time the CPU 21 receives a wireless signal from the terminal device 10, the CPU 21 may transmit information about the wireless signal, including the acquisition date and time information when the wireless signal was received and the value of the received signal strength of the wireless signal, to the estimation device 40 via the communication interface unit 25. Note that the information about the wireless signal may include, for example, identification information (e.g., MAC address, etc.) of the terminal device 10 that transmitted the wireless signal, the packet length of the wireless signal, the amount of data (HT Capability) that the terminal device 10 that transmitted the wireless signal can transmit within a third predetermined period (e.g., 1 second, etc.), the transmission frequency of the wireless signal, etc. Furthermore, the CPU 21 may transmit information about all wireless signals received within a second predetermined period (for example, 60 seconds) to the estimation device 40 via the communication interface unit 25.

[0046] (3) Configuration of the estimation device The configuration of the estimation device 40 will be described with reference to Fig. 3. Fig. 3 is a block diagram showing the internal configuration of the estimation device 40. As shown in Fig. 3, the estimation device 40 includes a CPU 41, a ROM 42, a RAM 43, a storage device 44, a display processing unit 45, a display unit 46, an input unit 47, and a communication interface unit 48, and is provided with a bus 40a for transmitting control signals or data signals between the respective units.

[0047] When power is applied to the estimation device 40, the CPU 41 loads various programs stored in the ROM 42 or the storage device 44 into the RAM 43 and executes them. In this embodiment, the CPU 41 reads out and executes the programs stored in the ROM 42 or the storage device 44, thereby realizing the functions of an acquisition means 51, an estimation means 52, a first information acquisition means 53, a second information acquisition means 54, and a learning means 55 (shown in FIG. 4 ), which will be described later.

[0048] The storage device 44 may be a non-volatile storage device such as a flash memory, an SSD, a magnetic storage device (e.g., an HDD, a floppy disk (registered trademark), a magnetic tape, or the like), an optical disk, or a volatile storage device such as a RAM, and stores programs executed by the CPU 41 and data referenced by the CPU 31. The storage device 44 also stores acquired data (shown in FIG. 5) and learning data (shown in FIG. 6), which will be described later.

[0049] The display processing unit 45 displays the display data provided by the CPU 41 on the display unit 46. The display unit 46 is, for example, an LCD (Liquid Crystal Display) monitor including thin film transistors arranged in a matrix on a pixel-by-pixel basis, and displays the data to be displayed on the display screen by driving the thin film transistors based on the display data.

[0050] When the estimation device 40 is a button input type terminal device, the input unit 47 has a button group including a plurality of instruction input buttons such as direction buttons and a decision button for accepting user operation input, and a button group including a plurality of instruction input buttons such as a numeric keypad, and includes an interface circuit for recognizing the press (operation) input of each button and outputting it to the CPU 41.

[0051] When the estimation device 40 is a terminal device that uses a touch panel input method, the input unit 47 mainly accepts inputs via the touch panel method by touching the display screen with a fingertip or a pen. The touch panel input method may be a known method such as a capacitance method.

[0052] Furthermore, when the estimation device 40 is a device capable of receiving voice input, the input unit 47 may be configured to include a microphone for voice input, or may be provided with an interface circuit for outputting voice data input via an external microphone to the CPU 41. Furthermore, when the estimation device 40 is a device capable of receiving moving images and / or still images, the input unit 47 may be configured to include a digital camera or digital video camera for image input, or may be provided with an interface circuit for receiving image data captured by an external digital camera or digital video camera and outputting it to the CPU 41.

[0053] The communication interface unit 48 includes an interface circuit for communicating with other devices (for example, the communication device 20 and the imaging device 30) via the communication network NW.

[0054] (4) Overview of the functions of the estimation system and model learning system The functions realized by the estimation system and model learning system of this embodiment will be described with reference to Fig. 4. Fig. 4 is a functional block diagram for explaining the functions that play a major role in the estimation system and model learning system of this embodiment. In the functional block diagram of Fig. 4, acquisition means 51 and estimation means 52 correspond to the main components of the estimation system of the present invention, and first information acquisition means 53, second information acquisition means 54, and learning means 55 correspond to the main components of the model learning system of the present invention.

[0055] In describing the functions of the estimation system of this embodiment, it is assumed that each terminal device 10 transmits a wireless signal (e.g., a probe request signal) including its own MAC address as its own identification information at a predetermined interval (e.g., every few seconds) in any of the multiple areas in the space SP. All terminal devices 10 present in any of the multiple areas in the space SP may be configured so that the MAC address included in the wireless signal does not change over time (having a fixed MAC address), or may be configured so that the MAC address changes over time (having a dynamic MAC address). Furthermore, each of the multiple areas in the space SP may contain a mixture of terminal devices 10 whose MAC addresses do not change over time and terminal devices 10 whose MAC addresses change over time.

[0056] The acquisition means 51 has a function of acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices 10 present in any one of a plurality of areas (e.g., a prepared food area) within a predetermined space SP.

[0057] Furthermore, the acquisition unit 51 may acquire information about wireless signals based on wireless signals received from one or more terminal devices 10 present in an area (e.g., a prepared food area) during a second predetermined period. Here, the second predetermined period may be a period following the first predetermined period, and the length of the second predetermined period may be the same as or different from the length of the first predetermined period. This makes it possible to easily estimate the number of objects (e.g., the number of people) present in an area during the second predetermined period using wireless signals received from one or more terminal devices 10 present in the area during the second predetermined period.

[0058] Furthermore, the radio signal may be a probe request signal, which can be used to estimate the number of objects present in any region within a given space SP.

[0059] The function of the acquisition means 51 is realized, for example, as follows. First, the CPU 21 of the communication device 20 provided in one of the areas (for example, the prepared food area) may transmit information about all wireless signals received within the second predetermined period to the estimation device 40 each time the second predetermined period elapses. Meanwhile, the CPU 41 of the estimation device 40 stores the information about the received wireless signal in the acquired data shown in FIG. 5 each time the CPU 41 receives (acquires) information about the wireless signal from the communication device 20 provided in one of the areas (here, the prepared food area) via the communication interface unit 48. The acquired data shown in FIG. 5 is data in which, for each wireless signal received by the communication device 20 from each terminal device 10, the acquisition date and time of the wireless signal, the packet length of the wireless signal, identification information (here, the MAC address), a received signal strength indicator (RSSI) value of the wireless signal, the amount of data that the terminal device 10 can transmit within a third predetermined period (HT Capability), and the transmission frequency of the wireless signal are described in association with each other.

[0060] Here, an example is described in which the information about the wireless signal includes the acquisition date and time of the wireless signal, the packet length of the wireless signal, identification information (MAC address) of the wireless signal, the received signal strength indicator (RSSI) value of the wireless signal, the amount of data that the terminal device 10 can transmit within a third predetermined period (HT Capability), and the transmission frequency of the wireless signal, but the information about the wireless signal may be configured to include at least one of these pieces of information.

[0061] Of the information related to the wireless signal, the "packet length of the wireless signal," "identification information (MAC address) of the wireless signal," "amount of data that the terminal device 10 can transmit within a third predetermined period (HT Capability)," and "transmission frequency of the wireless signal" may be included in the wireless signal transmitted from the terminal device 10. Also, of the information related to the wireless signal, the "acquisition date and time of the wireless signal" and the "received signal strength indicator (RSSI) value of the wireless signal" may be measured by the communication device 20.

[0062] Furthermore, when multiple wireless signals satisfy a predetermined condition, the CPU 41 of the estimation device 40 may determine that the multiple wireless signals are wireless signals transmitted from the same terminal device 10. This makes it possible to determine that multiple wireless signals that satisfy the predetermined condition are transmitted from the same terminal device 10, even if, for example, the identification information (here, MAC address) of the terminal device 10 included in the wireless signals changes over time.

[0063] Here, when the identification information (MAC address) of the terminal device 10 included in the wireless signals transmitted from the same terminal device 10 changes over time, the timing of changing the MAC address may differ depending on the model, OS, OS version, etc. of the terminal device 10. For example, depending on the model of the terminal device 10, the same MAC address may be used for several seconds, or the MAC address may change every time several wireless signals are transmitted. Therefore, the above-mentioned predetermined condition may include the identification information of the terminal device 10 included in each of the multiple wireless signals being the same. In this case, the CPU 41 may determine that, among the data of the wireless signals in the acquired data, the data of multiple wireless signals including the same MAC address and for which the difference in the acquisition dates and times of the wireless signals is within a predetermined range (e.g., 2 seconds) is the data of the wireless signals transmitted from the same terminal device 10.

[0064] Furthermore, even if the identification information (MAC address) of the terminal device 10 included in the wireless signal transmitted from the same terminal device 10 changes over time, the packet length of the wireless signal differs depending on, for example, the model, OS, and OS version of the terminal device 10. For example, if the packet lengths of two wireless signals are equal, these two wireless signals can be considered to have been transmitted from the same terminal device 10. Therefore, the above-mentioned predetermined condition may include the packet lengths of the multiple wireless signals being the same. In this case, the CPU 41 may determine that, among the data of the wireless signals in the acquired data, the data of multiple wireless signals having the same packet length and for which the difference in the acquisition dates and times of the wireless signals is within a predetermined range (for example, 2 seconds) is the data of the wireless signal transmitted from the same terminal device 10.

[0065] The estimation means 52 has a function of estimating the number of objects present in one of the areas (e.g., the prepared food area) based on information about the acquired wireless signals and a trained model based on machine learning using the training data. Here, the training data includes features based on information about wireless signals acquired within a first predetermined period in a predetermined area (e.g., the checkout area) among the multiple areas, and ground truth labels based on the number of objects present in the predetermined area within the first predetermined period, detected using the imaging device 30 (the predetermined device).

[0066] The predetermined device may be the imaging device 30. This makes it possible to generate a trained model using training data including correct labels based on the number of objects detected from an image captured by the imaging device 30, and to easily estimate the number of objects present in any area (for example, a prepared food area) based on this trained model.

[0067] Furthermore, estimation means 52 may estimate the number of objects present in any area (for example, the prepared food area) within the second predetermined period. This makes it possible to easily estimate the number of objects present in any area (here, the prepared food area) within the second predetermined period.

[0068] Furthermore, the feature may include any of information on the transmission interval of the wireless signal, information on the packet length of the wireless signal, information on identification information (e.g., MAC address) of the terminal device 10, information on the received signal strength (RSSI) of the wireless signal, information on the amount of data that the terminal device 10 can transmit within a third predetermined period (HT Capability), and information on the transmission frequency of the wireless signal. This makes it possible to estimate the number of objects present in any area (here, the prepared food area) within the predetermined space SP using any of information on the transmission interval of the wireless signal, information on the packet length of the wireless signal, information on identification information of the terminal device 10, information on the received signal strength of the wireless signal, information on the amount of data that the terminal device 10 can transmit within a third predetermined period, and information on the transmission frequency of the wireless signal.

[0069] Furthermore, the identification information of the terminal device 10 may be the MAC address of the terminal device 10. This makes it possible to estimate the number of objects present in any area (here, the prepared food area) within the predetermined space SP using a wireless signal including the MAC address.

[0070] Furthermore, if a payment area (here, a checkout area) where a predetermined payment process is performed is provided in a predetermined space SP, the predetermined area may be an area that includes the payment area (checkout area). This makes it possible to detect the number of objects present in the predetermined area (checkout area) by using a predetermined device (imaging device 30) to measure the number of objects present in the payment area (checkout area) in order to perform the predetermined payment process.

[0071] The function of estimation means 52 is realized, for example, as follows: CPU 41 of estimation device 40 may obtain feature amounts from information related to wireless signals within a second predetermined period in any area (for example, the prepared food area) acquired based on the function of acquisition means 51, and may estimate the number of objects (for example, the number of people) present in any area (the prepared food area) within the second predetermined period by inputting the obtained feature amounts into a trained model (described later) generated using training data.

[0072] Here, the feature amount may include, as information on the transmission interval of the wireless signal, the variation (number of types of transmission intervals) (i.e., the number of different transmission intervals) of the wireless signal of each of one or more terminal devices 10 within a second predetermined period in any of the areas (prepared food areas), or may include, as information on the packet length of the wireless signal, the number of types of packet lengths (i.e., the number of different packet lengths) of the wireless signal of each of one or more terminal devices 10 within a second predetermined period in any of the areas (prepared food areas). Furthermore, the feature amount may include, as information on the identification information (e.g., MAC addresses) of each of one or more terminal devices 10 within a second predetermined period in any of the areas (prepared food areas), the number of types (i.e., the number of different MAC addresses) of the identification information (e.g., MAC addresses) of the terminal devices 10, or may include, as information on the received signal strength indicator (RSSI), the maximum, minimum, and average values ​​of the received signal strength indicator (RSSI) of one or more wireless signals within a second predetermined period in any of the areas (prepared food areas). Furthermore, the feature may include the number of types of HT Capability (i.e., the number of different HT Capabilities) of each of one or more terminal devices 10 within a second predetermined period in any area (prepared food area) as information regarding the amount of data (HT Capability) that the terminal device 10 can transmit within a third predetermined period, or may include the number of types of frequencies (i.e., the number of different frequencies) of each of one or more wireless signals within a second predetermined period in any area (prepared food area) as information regarding the transmission frequency of the wireless signal.

[0073] The CPU 41 may access the acquired data and determine the transmission interval of the wireless signals from the time difference between the acquisition dates and times of the wireless signals transmitted from the same terminal device 10. Furthermore, if the identification information (MAC address) of the terminal device 10 changes over time, the CPU 41 may aggregate multiple pieces of identification information (MAC addresses) that are determined to be the same terminal device 10 into one piece of identification information (MAC address), and then determine the number of types of identification information (MAC addresses).

[0074] An example of training data is shown in Figure 6. The training data shown in Figure 6 is data in which feature quantities of information related to wireless signals (in the example shown in the figure, "variation in transmission intervals (number of different transmission intervals)," "number of packet length types (number of different packet lengths)," "number of MAC address types (number of different MAC addresses)," "maximum, minimum, and average RSSI values," "number of HT capability types (number of different HT capabilities)," and "number of frequency types (number of different frequencies)") are described in association with the number of objects (people in this case) (correct answer labels) (in the example shown in the figure, "1 person," "2 people," "3 people," "4 people," etc.). As a result of machine learning using this training data, a trained model is constructed that shows the relationship between the feature quantities of information related to wireless signals and the number of objects (people in this case).

[0075] Therefore, the CPU 41 can estimate the number of objects present in any area (prepared food area) during the second predetermined period by inputting features obtained from information regarding wireless signals in any area (for example, the prepared food area) during the second predetermined period into the trained model (here, it can estimate that the number of objects (people) present in any area during the second predetermined period is one of "1 person," "2 people," "3 people," "4 people," etc.).

[0076] The CPU 41 may present the result of estimating the number of objects present in any area (prepared food area) within the second predetermined period. For example, when the CPU 41 estimates the number of objects based on the function of the estimation means 52, the CPU 41 may display information regarding the result of estimating the number of objects on, for example, the display unit 46. Here, the information regarding the result of estimating the number of objects may be composed of text data or image data, etc. Furthermore, if the information regarding the result of estimating the number of objects is composed of audio data, the CPU 41 may output the information regarding the result of estimating the number of objects from an audio output device such as a speaker. Furthermore, the CPU 41 may transmit the information regarding the result of estimating the number of objects to another computer (e.g., a server, etc.) connected to the estimation device 40 via the communication network NW.

[0077] The first information acquisition means 53 has a function of acquiring information regarding wireless signals based on wireless signals received from each of one or more terminal devices 10 present in a predetermined area (e.g., a checkout area) among multiple areas within a predetermined space SP.

[0078] The function of the first information acquisition means 53 is realized, for example, as follows. First, the CPU 21 of the communication device 20 provided in a predetermined area (here, the transaction area) may transmit information about all wireless signals received within the first predetermined period to the estimation device 40 each time a first predetermined period elapses. Meanwhile, each time the CPU 41 of the estimation device 40 receives (acquires) information about wireless signals from the communication device 20 provided in the predetermined area (transaction area) via the communication interface unit 48, the CPU 41 obtains a feature amount (here, at least one of "variation in transmission intervals (the number of different transmission intervals)," "number of packet length types (the number of different packet lengths)," "number of MAC address types (the number of different MAC addresses)," "maximum, minimum, and average RSSI," "number of HT capability types (the number of different HT capabilities)," and "number of frequency types (the number of different frequencies)") based on the information about the received wireless signals. The CPU 41 then stores the obtained feature amount in the learning data shown in FIG. 6.

[0079] The second information acquisition means 54 has a function of acquiring information relating to the number of objects present in a predetermined area (for example, a checkout area) detected using a predetermined device (for example, the imaging device 30).

[0080] The function of the second information acquisition means 54 is realized, for example, as follows. First, the imaging device 30 may transmit captured images (here, images of the checkout area during the first predetermined period) to the estimation device 40 every time a first predetermined period has elapsed. Meanwhile, the CPU 41 of the estimation device 40 detects the number of objects (e.g., people) included in the captured image every time it receives (acquires) a captured image from the imaging device 30 via the communication interface unit 48. Here, the CPU 41 may detect the number of objects (here, people) included in the captured image using a well-known object detection technique, for example, object detection using machine learning. Then, the CPU 41 stores the number of detected objects as a correct answer label in the learning data shown in FIG. 6 in association with the feature amount stored based on the function of the first information acquisition means 53.

[0081] The learning means 55 has a function of learning a model used to estimate the number of objects present in one of the multiple areas (e.g., a prepared food area) through machine learning using learning data. Here, the learning data includes features based on information about wireless signals and ground truth labels based on the number of objects present in a predetermined area (e.g., a checkout area).

[0082] The function of the learning means 55 is realized, for example, as follows. For example, when a predetermined model learning instruction is input using the input unit 47, the CPU 41 of the estimation device 40 learns a model using the learning data shown in FIG. 6 . The CPU 41 may perform learning using, for example, a time-series adaptive neural network model. Here, as the time-series adaptive neural network, for example, a recurrent neural network (RNN) or a long short-term memory (LSTM), which is an advanced version of an RNN, can be applied. Furthermore, the CPU 41 may perform learning using any one of a plurality of models, for example, a graph neural network (GNN) model, a convolutional neural network (CNN) model, a support vector machine (SVM) model, a fully connected neural network (FNN) model, a gradient boosting (HGB) model, a wave net (WN) model, a CNN-LSTM model, etc. Furthermore, the CPU 41 may perform learning using any of the graph convolutional neural network (GCN) model, the graph attention network (GAT) model, and the graph convolutional LSTM (GC-LSTM) model, which are derivatives of GNN.

[0083] (5) Main processing flow of the model learning system of this embodiment Next, an example of the flow of main processes performed by the model learning system of this embodiment will be described with reference to the flowchart of FIG.

[0084] Based on the function of the first information acquisition means 53, the CPU 41 of the estimation device 40 acquires information about the wireless signals based on the wireless signals received from each of one or more terminal devices 10 present in a predetermined area (e.g., a checkout area) among multiple areas within the predetermined space SP (step S100).

[0085] Next, the CPU 41 of the estimation device 40 acquires information regarding the number of objects present in a specified area (checkout area) detected using a specified device (here, the imaging device 30) based on the function of the second information acquisition means 54 (step S102).

[0086] Next, the CPU 41 of the estimation device 40, based on the function of the learning means 56, learns a model used to estimate the number of objects present in one of the multiple areas (for example, the prepared food area) through machine learning using the learning data (step S104). Here, the learning data includes features based on information about the wireless signal and ground truth labels based on the number of objects present in a predetermined area (checkout area).

[0087] In this way, by machine learning using as learning data information about wireless signals received from each of one or more terminal devices 10 present in a predetermined area (e.g., a checkout area) among multiple areas within a predetermined space SP, and information about the number of objects present in the predetermined area (checkout area) detected using an imaging device 30 (predetermined device), it becomes possible to learn a model that can be used to estimate the number of objects present in any of multiple areas.

[0088] (6) Main processing flow of the estimation system of this embodiment Next, an example of the flow of main processes performed by the estimation system of this embodiment will be described with reference to the flowchart of FIG.

[0089] First, CPU 21 of communication device 20 provided in one of the areas (for example, the prepared food area) may transmit information about all wireless signals received within a second predetermined period to estimation device 40 each time a second predetermined period elapses. Meanwhile, CPU 41 of estimation device 40, based on the function of acquisition means 51, acquires information about the wireless signals based on the wireless signals received from each of one or more terminal devices 10 present in one of the areas (prepared food area) among the multiple areas within predetermined space SP (step S200). Specifically, each time CPU 41 receives (acquires) information about the wireless signals from communication device 20 provided in one of the areas (prepared food area) via communication interface unit 48, it stores the information about the received wireless signals in acquisition data.

[0090] Next, the CPU 41 of the estimation device 40 estimates the number of objects present in one of the areas (prepared food area) based on information about the acquired wireless signals and a trained model based on machine learning using the training data, using the function of the estimation means 52 (step S202). Here, the training data includes features based on information about wireless signals acquired within a first predetermined period in a predetermined area (checkout area) among the multiple areas, and ground truth labels based on the number of objects present in the predetermined area within the first predetermined period, detected using the imaging device 30 (prescribed device).

[0091] As described above, the estimation system, estimation method, and program of this embodiment make it possible to estimate the number of objects (e.g., the number of people) present in one of a plurality of areas within a predetermined space SP (e.g., the prepared food area) using wireless signals received from each of one or more terminal devices 10 present in the area and a trained model. Furthermore, the estimation system, estimation method, and program of this embodiment generate the trained model using training data that includes, as features, information about wireless signals acquired within a first predetermined period in a predetermined area (e.g., the checkout area) among a plurality of areas, and includes, as correct labels, the number of objects present in the predetermined area (checkout area) within the first predetermined period, as detected using the imaging device 30 (predetermined device). Therefore, by inputting information about the wireless signals in one of the areas (predetailed food area) into this trained model, it becomes possible to estimate the number of objects present in the area (predetailed food area). This makes it possible to easily estimate the number of objects present in each of the multiple areas based on the radio signals received from the terminal devices 10 present in each of the multiple areas within a specified space SP, even if, for example, an imaging device 30 is not placed in each of the multiple areas.

[0092] Furthermore, according to the model learning system, model learning method, and program of this embodiment, it is possible to learn a model that can be used to estimate the number of objects present in any one of multiple areas by machine learning using, as learning data, information regarding wireless signals received from each of one or more terminal devices 10 present in a specified area (e.g., a checkout area) among multiple areas within a specified space SP, and information regarding the number of objects present in the specified area (checkout area) detected using an imaging device 30 (specified device).Therefore, by using this model, the number of objects present in each of the multiple areas can be easily estimated.

[0093] The program of the present invention may be stored in a computer-readable storage medium. The storage medium on which this program is recorded may be the ROM 22, RAM 23, or storage device 24 of the communication device 20 shown in Fig. 2, or the ROM 42, RAM 43, or storage device 44 of the estimation device 40 shown in Fig. 3. The storage medium may also be a CD-ROM or the like that can be read by being inserted into a program reading device such as a CD-ROM drive. Furthermore, the storage medium may be a magnetic tape, cassette tape, flexible disk, MO / MD / DVD, or a semiconductor memory.

[0094] The above-described embodiments have been described to facilitate understanding of the present invention, and are not intended to limit the present invention. Therefore, each element disclosed in the above embodiments is intended to include all design modifications and equivalents that fall within the technical scope of the present invention.

[0095] For example, in the above-described embodiment, the case where the object is a person has been described as an example, but the present invention is not limited to this case. The object may be anything that can be counted, such as an animal or other moving object (e.g., a construction machine, a vehicle, an aircraft, etc.). Furthermore, at least one of the multiple objects may be provided with a terminal device 10.

[0096] Furthermore, in the above-described embodiment, the predetermined area within the space SP is described as an area including a payment area (accounting area) where a predetermined payment is made, but the present invention is not limited to this case. For example, if a gateway is provided for entering and exiting the predetermined space SP, the predetermined area may be an area including the gateway (e.g., gateway area 1 and / or gateway area 2, etc.). In this case, it is possible to detect the number of objects present in the predetermined area by using a predetermined device (e.g., imaging device 30) to measure the number of objects (e.g., the number of people) passing through the gateway.

[0097] Furthermore, in the above-described embodiment, the predetermined device is described as an imaging device 30 as an example, but the present invention is not limited to this. For example, the predetermined device may be an entrance / exit gate provided in a predetermined area (e.g., entrance / exit area 1 and / or entrance / exit area 2, etc.). In this case, the entrance / exit gate may detect the number of objects (e.g., the number of people) passing through the entrance / exit gate within a first predetermined period as the number of objects present in the predetermined area within the first predetermined period. Furthermore, the predetermined device may be a Point of Sales (POS) register provided in a predetermined area (e.g., the checkout area). In this case, the POS register may detect the number of objects (e.g., the number of people) that have undergone payment processing using the POS register within the first predetermined period as the number of objects present in the predetermined area within the first predetermined period.

[0098] Furthermore, in the above-described embodiment, a case where one communication device 20 communicates with each of one or more terminal devices 10 has been described as an example, but the present invention is not limited to this case. For example, each of multiple communication devices 20 may communicate with each of one or more terminal devices 10.

[0099] In the above-described embodiment, a case where wireless communication is performed between each terminal device 10 and the communication device 20 using a wireless LAN has been described as an example, but the present invention is not limited to this case. For example, other wireless communication methods such as Bluetooth (registered trademark), ZigBee (registered trademark), UWB, and optical wireless communication (e.g., infrared) may also be used.

[0100] Furthermore, in the above-described embodiment, a case where a MAC address is used as identification information for each terminal device 10 has been described as an example, but the present invention is not limited to this case. For example, the identification information may be any information that is included in a wireless signal and is configured to be able to identify each terminal device 10.

[0101] Furthermore, in the above-described embodiment, an example has been described in which one estimation device 40 is provided, but the present invention is not limited to this. For example, a plurality of estimation devices 40 may be provided, and in this case, the operation content and processing results of any one of the estimation devices 40 may be presented in real time on the other estimation devices 40, or the processing results of any one of the estimation devices 40 may be shared among the plurality of estimation devices 40.

[0102] In the above-described embodiment, the estimation device 40 is configured to implement the functions of the acquisition means 51, the estimation means 52, the first information acquisition means 53, the second information acquisition means 54, and the learning means 55. However, this configuration is not limited to this. For example, a learning device 60 (shown in FIG. 9 ) may be provided, which is configured from a computer (e.g., a general-purpose personal computer or a server computer) communicatively connected to the estimation device 40 via a communication network such as the Internet or a LAN, and is configured to learn a model used to estimate the number of objects present in any one of multiple regions within the space SP. In this case, the estimation device 40 and the learning device 60 can have substantially the same hardware configuration, so that the learning device 60 can implement the functions of at least one of the means 51 to 55 described in the above-described embodiment. For example, the functions of the functional block diagram shown in FIG. 4 may be arbitrarily shared between the estimation device 40 and the learning device 60, as shown in FIGS. 9( a) and 9(b).

[0103] Furthermore, the functions of at least one of the above-mentioned means 51 to 55 may be realized by the terminal device 10, the communication device 20 and / or the imaging device 30. [Industrial Applicability]

[0104] The estimation system, model learning system, estimation method, model learning method, and program of the present invention as described above can easily estimate the number of objects present in each of multiple areas within a specified space, and therefore can be suitably used in systems and services that perform people flow analysis within a specified space (e.g., a store, facility, etc.), and therefore have extremely great industrial applicability. [Explanation of symbols]

[0105] 10...Terminal device 20...Communication equipment 30...imaging device 40…Estimation device 51…Acquisition means 52...Estimation means 53...First information acquisition means 54…Second information acquisition means 55...Learning methods 60...Learning device SP…Space

Claims

1. an acquisition means for acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in any one of a plurality of regions within a predetermined space; and an estimation means for estimating the number of objects present in any one of the areas based on information about acquired wireless signals and a trained model based on machine learning using training data, wherein the training data includes feature amounts based on information about wireless signals acquired in a predetermined area of ​​the plurality of areas within a first predetermined period, and ground truth labels based on the number of objects present in the predetermined area within the first predetermined period, detected using a predetermined device. Estimation system.

2. The estimation system according to claim 1 , wherein the predetermined device is an imaging device.

3. the acquiring means acquires information about the wireless signals based on the wireless signals received from each of the one or more terminal devices present in any of the areas within a second predetermined period; The estimation system according to claim 1 , wherein the estimation means estimates the number of objects present in any one of the areas within the second predetermined period.

4. 2. The estimation system according to claim 1, wherein the feature includes any one of information regarding a transmission interval of the wireless signal, information regarding a packet length of the wireless signal, information regarding identification information of the terminal device, information regarding a received signal strength of the wireless signal, information regarding an amount of data that the terminal device can transmit within a third predetermined period, and information regarding a transmission frequency of the wireless signal.

5. The estimation system according to claim 4 , wherein the identification information of the terminal device is a MAC address of the terminal device.

6. The estimation system of claim 1 , wherein the wireless signal is a probe request signal.

7. The estimation system according to claim 1 , wherein, when an entrance for entering and exiting the predetermined space is provided, the predetermined area is an area including the entrance.

8. The estimation system according to claim 1 , wherein, when a payment area where a predetermined payment process is performed is provided in the predetermined space, the predetermined area is an area that includes the payment area.

9. a first information acquisition means for acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in a predetermined area among a plurality of areas within a predetermined space; a second information acquisition means for acquiring information about the number of objects present in the predetermined area, the information being detected using a predetermined device; a learning means for learning a model used to estimate the number of objects present in any one of the plurality of regions by machine learning using learning data, the learning data including a feature based on information about the wireless signal and a ground truth label based on the number of objects present in the predetermined region; Model learning system.

10. The computer acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in any one of a plurality of regions within a predetermined space; a step of estimating the number of objects present in any of the areas based on information about the acquired wireless signals and a trained model based on machine learning using training data, wherein the training data includes features based on information about wireless signals acquired in a predetermined area among the multiple areas within a first predetermined period, and a ground truth label based on the number of objects present in the predetermined area within the first predetermined period, detected using a predetermined device; Perform each step of Estimation method.

11. The computer acquiring information about the wireless signals based on the wireless signals received from each of one or more terminal devices present in a predetermined region among a plurality of regions within a predetermined space; obtaining information about the number of objects present in the predetermined area detected using a predetermined device; a step of learning a model used to estimate the number of objects present in any one of the plurality of regions by machine learning using learning data, the learning data including features based on information about the wireless signal and a ground truth label based on the number of objects present in the predetermined region; Perform each step of Model learning methods.

12. On the computer, a function of acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in any one of a plurality of regions within a predetermined space; a function of estimating the number of objects present in any one of the areas based on information about acquired wireless signals and a trained model based on machine learning using training data, wherein the training data includes features based on information about wireless signals acquired in a predetermined area among the multiple areas within a first predetermined period, and a ground truth label based on the number of objects present in the predetermined area within the first predetermined period, detected using a predetermined device; A program to achieve this.

13. On the computer, a function of acquiring information about a wireless signal based on the wireless signal received from each of one or more terminal devices present in a predetermined area among a plurality of areas within a predetermined space; obtaining information about the number of objects present in the predetermined area detected using a predetermined device; a function of learning a model used to estimate the number of objects present in any one of the plurality of regions by machine learning using learning data, the learning data including features based on information about the wireless signal and a ground truth label based on the number of objects present in the predetermined region; A program to achieve this.

Images