Information Processing Systems

Abstract

To appropriately ensure thermal comfort when the number of temperature sensors that can be installed in a facility is limited. [Solution] The information processing system 10 includes a photographing unit, a congestion determination unit, a storage device 120, and a determination unit 250. The photographing unit generates a plurality of photographed images each depicting a plurality of target locations within a facility 1 by photographing the target locations. The congestion determination unit determines, for each of the plurality of target locations, whether a congestion of people is occurring at the target location based on the photographed images depicting the target location. The storage device 120 stores congestion history data 124 indicating a history of the results of the determination by the congestion determination unit, and complaint history data 126 indicating a history of complaints about temperature. The determination unit 250 determines, based on the congestion history data 124 and the complaint history data 126, a recommended installation location within the facility for a temperature sensor that outputs a measurement value to an air conditioning system.

Description

[Technical Field]

[0001] The present disclosure relates to an information processing system. [Background technology]

[0002] Japanese Patent Publication No. 2020-186850 (Patent Document 1) discloses a prediction system for predicting whether or not complaints about air conditioning will occur. Complaints about air conditioning are expressions of a user's opinion that the air-conditioned environment in a facility equipped with an air conditioning system is uncomfortable, such as complaints that the facility is "too hot" or "too cold". The prediction system comprises a measuring device and a prediction device. The measuring device outputs measured values ​​of the indoor temperature in the facility. The prediction device includes a storage unit and a prediction unit. The storage unit stores a trained model that represents the relationship between past measured values ​​and whether or not complaints occurred when those measured values ​​were obtained. The prediction unit uses the trained model to predict whether or not complaints about air conditioning will occur according to the current measured values. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-186850 Summary of the Invention [Problem to be solved by the invention]

[0004] Temperatures in large facilities such as commercial buildings vary depending on the location. On the other hand, the number of temperature sensors that can be installed in such facilities may be limited to the minimum necessary due to budget constraints. Basically, air conditioning systems operate by receiving measurements from installed temperature sensors and adjusting each measurement to reach the set temperature or a temperature close to it (optimal temperature). When the number of temperature sensors is limited, the temperature in areas where temperature sensors are installed is adjusted to the optimal temperature because the temperature measurement is output (feedback) from the temperature sensors to the air conditioning system. On the other hand, the temperature in areas where temperature sensors are not installed may not be adjusted to the optimal temperature because the temperature information is not output to the air conditioning system.

[0005] The temperature at each location within a facility also changes depending on the flow of people. For example, in areas where people flow smoothly, heat is less likely to accumulate, so the temperature is less likely to deviate from the optimum temperature even if a temperature sensor is not installed.

[0006] On the other hand, in places where people are stagnating, heat tends to accumulate in those places, and if no temperature sensors are installed, the temperature is particularly likely to deviate from the optimum temperature. This can lead to complaints about air conditioning and a loss of thermal comfort.

[0007] The present disclosure has been made to solve the above-mentioned problems, and its purpose is to provide an information processing system for appropriately ensuring thermal comfort when the number of temperature sensors that can be installed in a facility is limited. [Means for solving the problem]

[0008] The information processing system disclosed herein includes a photographing unit, a stagnation determination unit, a storage unit, and a determination unit. The photographing unit generates a plurality of photographed images each depicting a plurality of target locations within a facility equipped with an air conditioning system by photographing the plurality of target locations. The stagnation determination unit determines, for each of the plurality of target locations, whether a stagnation of people is occurring at the target location based on the photographed images depicting the target location. The storage unit stores first history data indicating a history of the results of the determination by the stagnation determination unit, and second history data indicating, for each of the plurality of target locations, a history of complaints about the temperature at the target location. The determination unit determines, based on the first history data and the second history data, a recommended installation location within the facility for installing a temperature sensor that outputs a measurement value to the air conditioning system. [Effects of the Invention]

[0009] According to the present disclosure, both the history of whether people are lingering in each target location and the history of complaints about temperature are reflected in the recommended installation location. This allows the installation location of a temperature sensor to be determined based on the flow of people, and the user is prompted to install the temperature sensor in such a location. The temperature sensor then outputs a measurement value to the air conditioning system, which then operates to adjust the temperature at the location to an appropriate temperature. This reduces complaints about air conditioning and ensures appropriate thermal comfort. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 2 is a diagram for illustrating a temperature sensor that can be installed in a facility to which an information processing system according to a first embodiment is applied, and functions of an air conditioning system provided in the facility. [Figure 2] FIG. 1 is a diagram illustrating an example of a floor of a facility. [Figure 3] This is a diagram for explaining that the temperature at each location within a facility changes depending on whether or not there is a stagnation of people within the facility. [Figure 4] FIG. 2 is a diagram illustrating a hardware configuration of the information processing system according to the first embodiment. [Figure 5]10 shows an example of retention determination data. [Figure 6] An example of complaint record data is shown below. [Figure 7] FIG. 10 is a diagram illustrating an example of retention history data. [Figure 8] FIG. 10 is a diagram illustrating an example of complaint history data. [Figure 9] FIG. 2 is a functional block diagram of a terminal device. [Figure 10] FIG. 10 is a diagram illustrating an example of a screen displayed on a display device. [Figure 11] FIG. 10 is a diagram illustrating an example of aptitude data. [Figure 12] 4 is a flowchart illustrating a process executed by a control device in the first embodiment. [Figure 13] FIG. 10 is a diagram illustrating an overall configuration of an information processing system according to a second embodiment. [Figure 14] FIG. 2 is a diagram illustrating an example of each piece of learning data included in a learning dataset. [Figure 15] FIG. 2 is a diagram showing the configuration of a neural network used by a model generation unit. [Figure 16] This shows the configuration of an information processing system for utilizing a trained model. [Figure 17] FIG. 10 is a diagram illustrating an overall configuration of an information processing system according to a first modified example. [Figure 18] FIG. 2 is a diagram illustrating an example of each piece of learning data included in a learning dataset. [Figure 19] This shows the configuration of an information processing system for utilizing a trained model. [Figure 20] FIG. 10 is a diagram illustrating an overall configuration of an information processing system according to a second modification. [Figure 21] An example of retention and customer determination data is shown below. [Figure 22] FIG. 2 is a diagram illustrating an example of each piece of learning data included in a learning dataset. [Figure 23] FIG. 10 is a diagram illustrating a method by which the model generation unit generates a trained model. [Figure 24]FIG. 1 is a diagram showing the configuration of an information processing system for utilizing a trained model. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. The same or corresponding parts in the drawings will be denoted by the same reference numerals, and the description thereof will not be repeated. The embodiments and their modifications may be combined with each other as appropriate.

[0012] [Embodiment 1] FIG. 1 is a diagram illustrating functions of a temperature sensor that can be installed in a facility to which an information processing system according to a first embodiment is applied, and an air conditioning system provided in the facility.

[0013] Referring to FIG. 1, facility 1 is a large commercial facility such as a shopping mall. Temperature sensors 3-1, 3-2, ... 3-n can be installed (retrofitted) at various locations within facility 1 and can also be removed from those locations. Each temperature sensor is also referred to as a "temperature sensor 3." When installed within facility 1, a temperature sensor 3 measures the temperature at the installation location and outputs the measured value MV to air conditioning system 5. The number (n) of temperature sensors 3 that can be installed in facility 1 is limited to the minimum necessary due to budgetary and other considerations. Therefore, it is difficult to install temperature sensors 3 everywhere in facility 1 and measure the temperatures at all of these locations.

[0014] The air conditioning system 5 receives the measured values ​​MV from each temperature sensor 3 and operates to keep each measured value MV within a predetermined temperature range. In other words, the air conditioning system 5 adjusts the temperature at the location in the facility 1 where the temperature sensor 3 is installed to an appropriate temperature.

[0015] Fig. 2 is a diagram showing an example of a floor of facility 1. Referring to Fig. 2, facility 1 has a plurality of entrances (for example, entrances G1 to G4) and various stores.

[0016] The cameras 20-1 to 20-8 are network cameras installed within the facility 1, and each captures target locations R1 to R8 within the facility 1. Each camera is also referred to as a "camera 20." The target locations R1 to R8 are locations captured by the cameras 20-1 to 20-8, respectively. Each target location is also referred to as a "target location RX." The cameras 20-1 to 20-8 capture multiple captured images of the target locations R1 to R8, respectively. These captured images capture the target locations R1 to R8, respectively. The camera 20 can also fine-tune the target location RX by rotating itself. In this case, the target location RX can change to a nearby area of ​​any of the target locations R1 to R8. The number (N) of cameras 20 is not limited to 8 and may be, for example, 9 or more. The number of cameras 20 is assumed to be greater than the number of temperature sensors 3 (FIG. 1) (N>n).

[0017] The temperature within the facility 1 may vary depending on the location. If the number of temperature sensors 3 is limited, the temperature in locations where the temperature sensors 3 are installed will be adjusted to an appropriate temperature because the measured values ​​MV are output (feedback) from the temperature sensors 3 to the air conditioning system 5. On the other hand, the temperature in locations where the temperature sensors 3 are not installed may not be adjusted to an appropriate temperature because the measured values ​​MV are not output to the air conditioning system 5.

[0018] Figure 3 illustrates how the temperature at each location within Facility 1 changes depending on the presence or absence of pedestrian flow. Referring to Figure 3, in this example, camera 20-X is one of cameras 20-1 to 20-8, and pedestrian flow is the flow of customers in Facility 1.

[0019] Whether or not there is a stagnation of people flowing through the target location RX can affect whether or not there will be complaints about the temperature at that location. For example, when there is a smooth flow of people at the target location RX as shown in FIG. 3(A), the heat emitted from the bodies of customers CS is less likely to accumulate at the target location RX. Furthermore, if store B near the target location RX is a restaurant, the total amount of heat generated by the cooking equipment at store B is small because the number of food orders for store B is small. As a result, even if a temperature sensor 3 is not installed at or near the target location RX, the temperature at the target location RX is less likely to deviate from the set temperature of the air conditioning system 5. Therefore, complaints about the temperature at the target location RX are less likely to occur.

[0020] On the other hand, as shown in Figure 3(B), if store B is popular and there is a stagnation of people at the target location RX, the heat released from the bodies of customers CS in the target location RX tends to accumulate in that location. Furthermore, if store B is a restaurant, the total heat generated by cooking appliances is high due to the large number of food orders. As a result, if a temperature sensor 3 is not installed at or near the target location RX, the temperature at the target location RX is particularly likely to deviate from the set temperature of the air conditioning system 5. This can lead to complaints about the temperature of the air conditioning and can impair the thermal comfort of customers CS. Furthermore, when there is a stagnation of people at the target location RX, the absolute number of customers CS at the target location RX is higher than when there is no stagnation. As a result, there is a high possibility of complaints about the temperature being too hot at the target location RX. In addition, the opening and closing time of doors varies depending on whether there is a stagnation of people near the entrances and exits of facility 1. For example, when there is a stagnation of people, the opening and closing time of the doors installed at the entrances and exits is longer than when there is no stagnation. As a result, the amount of air flowing from the outside into facility 1 and from the inside to the outside increases. This may change the air flow within the facility 1, and therefore may affect the temperature within the facility 1. In this way, whether or not there is a stagnation of people flow may also affect whether or not there are complaints about temperature, and the thermal comfort of customers CS.

[0021] Employees of Facility 1 are asked to anticipate areas where air conditioning complaints are likely to occur and determine the installation locations for temperature sensors 3. However, accurately predicting areas prone to air conditioning complaints and appropriately determining the installation locations for temperature sensors 3-1 to 3-n requires extensive experience and knowledge of the facility, making it essentially a task dependent on individual expertise. As a result, it is difficult for many people to appropriately determine the installation locations for temperature sensors 3-1 to 3-n.

[0022] Therefore, the information processing system according to the first embodiment has a configuration for appropriately ensuring the thermal comfort of the customer CS and for easily determining the installation locations of the n temperature sensors 3. This point will be described below.

[0023] Fig. 4 is a diagram illustrating a hardware configuration of the information processing system according to the first embodiment. With reference to Fig. 4, information processing system 10 includes cameras 20-1 to 20-N, servers 30 and 40, and terminal device 100. The number (N) of cameras 20 is, for example, 8, but is not limited to this.

[0024] The server 30 includes a communication device 31, a processing device 32, and a storage device 33. The communication device 31 communicates with each camera 20 and receives captured images IMG transmitted from the camera.

[0025] The processing unit 32 includes, functionally, a counting unit 34 and a dwell determination unit 35. These functions are performed, for example, by a CPU (Central Processing Unit) built into the processing unit 32, or by a dedicated processing circuitry such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0026] The counting unit 34 performs counting processing for each camera 20 by analyzing the captured image (IMG) from the camera 20 using image processing technology. The counting processing corresponds to counting the number of customers (CS) visible in the captured image (IMG) as the number of customers (CS) at the target location (RX). The counting processing is performed at regular time intervals (e.g., every second).

[0027] The congestion determination unit 35 determines whether congestion of people is occurring at each target location RX, according to the captured image IMG showing the target location. Specifically, the congestion determination unit 35 determines whether the number of customers CS who are visible in the captured image for a standard time or longer is equal to or greater than a standard number, according to the result of the counting process. This determination process is also called the "congestion determination process." The congestion determination process is executed at regular time intervals. If the number of customers CS who are visible in the captured image for a standard time or longer is equal to or greater than a standard number, the congestion determination unit 35 determines that congestion of people is occurring at the target location RX. On the other hand, if the number of such customers CS is less than a standard number, the congestion determination unit 35 determines that there is no congestion at the target location RX. The standard time is, for example, 10 seconds. The standard number is, for example, 10.

[0028] The storage device 33 stores the database 36. The database 36 includes stagnant determination data 37-1, 37-2, 37-3, etc. Each stagnant determination data is also referred to as "stagnant determination data 37". As explained below, the stagnant determination data 37 shows the history of the determination results by the stagnant determination unit 35.

[0029] Figure 5 shows an example of dwell time determination data 37. Referring to Figure 5, the dwell time determination data 37 includes camera number 372, target location information 374, and dwell time information 376. Camera number 372 represents the identification number of camera 20. Target location information 374 indicates the target location RX captured by the corresponding camera 20. Dwell time information 376 indicates the result of the dwell time determination process for the target location RX captured by the corresponding camera 20. Since the dwell time determination process is performed at regular time intervals, the dwell time determination data 37 is generated at regular time intervals. For example, dwell time determination data 37-1, 37-2, 37-3… are generated sequentially at regular time intervals and are associated with their generation date and generation time.

[0030] Referring again to FIG. 4, the terminal device 100 includes a communication device 105, an input device 110, a display device 115, a storage device 120, and a control device .

[0031] The communication device 105 communicates with external devices such as servers 30 and 40. The input device 110 receives various user operations. User operations include inputting a record of a complaint as daily report information when a complaint about the temperature at the target location RX (or a nearby location) occurs. When such a user operation is performed, the communication device 105 transmits the input daily report information to the server 40.

[0032] Server 40 receives daily report information transmitted from terminal device 100. Server 40 stores database 44. Database 44 contains complaint record data 45-1, 45-2, 45-3, etc. Each complaint record data is also referred to as "complaint record data 45". Complaint record data 45 is generated based on the daily report information transmitted from terminal device 100 and shows the history of past complaints about temperature.

[0033] Figure 6 shows an example of complaint record data 45. Referring to Figure 6, complaint record data 45 includes date and time information 452, location information 454, and complaint information 456. Date and time information 452 represents the date and time when the complaint about temperature actually occurred. Location information 454 indicates the location where such a complaint occurred (complaint location). In the following explanation, the complaint location is assumed to be one of the target locations RX. Therefore, location information 454 is the same as the target location information 374 (Figure 5). Complaint information 456 represents the content of the complaint about temperature. Complaint record data 45-1, 45-2, 45-3… are generated at regular intervals. This interval is, for example, one month.

[0034] 4 again, various screens are displayed on the display device 115. Specific examples of the screens displayed on the display device 115 will be described in detail later.

[0035] The storage device 120 stores current installation location information 122, dwelling history data 124, complaint history data 126, and suitability data 128. The current installation location information 122 indicates the current installation location within facility 1 where temperature sensors 3-1 to 3-n are already installed. The dwelling history data 124, complaint history data 126, and suitability data 128 will be explained in detail later.

[0036] The control device 130 controls the communication device 105 and the display device 115. The control device 130 controls the communication device 105 to retrieve databases 36 and 44 from servers 30 and 40, for example. Based on the retrieved databases 36 and 44, the control device 130 generates dwell history data 124 and complaint history data 126 in the storage device 120, respectively.

[0037] Figure 7 is a diagram showing an example of dwelling history data 124. Referring to Figure 7, the dwelling history data 124 shows the history of the determination results by the dwelling determination unit 35 over a certain information collection period (e.g., the past year). In detail, the dwelling history data 124 includes camera number 372, target location information 374, and dwelling count information 125. Camera number 372 and target location information 374 are the same as those shown in Figure 5. The dwelling count information 125 includes the dwelling count for each target location RX, indicating the number of times people dwelled at that location. These dwellings occurred within the above-mentioned information collection period. The control device 130 can update the dwelling history data 124 to the latest version at a predetermined timing.

[0038] Figure 8 is a diagram showing an example of complaint history data 126. Referring to Figure 8, the complaint history data 126 shows the history of actual complaints regarding temperature at each location RX. These complaints occurred within the information collection period described above. In detail, the complaint history data 126 includes location information 454 and complaint count information 127. The complaint count information 127 includes the complaint count for each location RX, indicating the number of actual complaints regarding temperature at that location within the information collection period described above. The control device 130 can update the complaint history data 126 to the latest version at predetermined timings.

[0039] Fig. 9 is a functional block diagram of terminal device 100. With reference to Fig. 9, terminal device 100 includes a determining unit 250 and a notifying unit 260 as its functional configuration.

[0040] The determination unit 250 determines recommended installation locations for temperature sensors 3-1 to 3-n within facility 1, based on the dwell time history data 124 and the complaint history data 126. The function of the determination unit 250 is performed, for example, by a CPU built into the control device 130 or by the dedicated processing circuit described above. A specific example of how the recommended installation locations are determined will be explained in detail later.

[0041] The notification unit 260 notifies the user of information indicating the recommended installation location determined by the determination unit 250. The user is, for example, the manager of the facility 1. The function of the notification unit 260 is performed when the control device 130 controls the display device 115.

[0042] According to the determination unit 250, both the history of pedestrian congestion at each target location RX and the history of temperature complaints are reflected in the recommended installation locations. For example, not only information on locations where many temperature complaints have been made in the past, but also information on locations where pedestrian congestion was frequent (in other words, locations where temperature complaints are likely to occur in the future) is reflected in the recommended installation locations. As a result, the installation locations of temperature sensors 3-1 to 3-n that are suitable for the pedestrian flow trends (distribution of customer CS) within facility 1 are determined, and the notification unit 260 prompts the user to install temperature sensors 3-1 to 3-n in such locations. Therefore, temperature sensors 3-1 to 3-n are installed in such locations, and the measured values ​​MV from these temperature sensors 3 are output (feedback) to the air conditioning system 5. As a result, the air conditioning system 5 operates so that the temperature in such locations is adjusted to a suitable temperature. Thus, complaints regarding air conditioning can be suppressed, and the temperature comfort of customer CS can be appropriately ensured. Furthermore, since the process for determining the installation locations of the temperature sensors 3-1 to 3-n is automated, the installation locations of these temperature sensors 3 can be easily determined without much experience or knowledge about the facility 1.

[0043] The notification unit 260 determines whether the recommended installation location is different from the current installation location according to the current installation location information 122. If the recommended installation location is different from the current installation location, the notification unit 260 notifies the user by displaying information on the display device 115 prompting a change (redesign) of the installation locations of the temperature sensors 3-1 to 3-n from the current installation location to the recommended installation location.

[0044] Fig. 10 is a diagram showing an example of a screen displayed on display device 115. Referring to Fig. 10, screen 116 displays current installation locations (P1 to P4) in the upper half, and recommended installation locations (P3 to P6) in the lower half.

[0045] In commercial facilities such as Facility 1, the flow of people may change after the start of operations due to the closure or opening of stores. For example, if a popular store appears, many customers will gather at that store's location, making it easier for people to linger there. As a result, if a temperature sensor 3 is not installed at that location, there may be many complaints about air conditioning at that location. In this case, if people are not likely to linger at a particular location, it may be preferable to remove the temperature sensor 3 installed at that location and change its location to a place where people are more likely to linger, as described above.

[0046] By displaying the screen 116 on the display device 115, users are motivated to appropriately change the installation locations of temperature sensors 3-1 to 3-n. As a result, temperature sensors 3 are installed in locations where foot traffic tends to stagnate due to the closure or opening of stores (for example, P5). Therefore, the measured temperature MV at this location is fed back from the temperature sensor 3 to the air conditioning system 5, and the temperature is adjusted to an appropriate level. Thus, even if the foot traffic trend changes, complaints regarding air conditioning can be appropriately suppressed, and the temperature comfort of customers can be appropriately ensured.

[0047] Furthermore, even if facility 1 has just started operations and temperature sensors 3-1 to 3-n have not yet been installed in facility 1, and the storage device 120 does not currently store installation location information 122, the function of the determination unit 250 can still be performed. In this case, the notification unit 260 may display only the lower half of screen 116 on the display device 115 after determining the recommended installation location.

[0048] Referring again to Figure 9, a specific example of how the determination unit 250 determines the recommended installation location will be explained. The determination unit 250 includes a calculation unit 252 and a recommended installation location determination unit 253. The calculation unit 252 calculates the suitability of each target location RX as a recommended installation location. As explained below, the suitability corresponds to the sum of the product of the number of pedestrian traffic dwells and the first weighting coefficient, and the product of the number of temperature complaints and the second weighting coefficient.

[0049] FIG. 11 is a diagram illustrating an example of suitability data 128. Referring to FIG. 11, the suitability data 128 is generated based on the stay history data 124 (FIG. 7) and the complaint history data 126 (FIG. 8). The suitability data 128 is the same as the stay history data 124 and the complaint history data 126 in that it includes a camera number 372, target location information 374, stay count information 125, and complaint count information 127. In contrast, the suitability data 128 differs from these history data in that it further includes suitability information 129. The suitability information 129 represents the suitability (y) of each target location RX as a recommended installation location, calculated for that location.

[0050] The calculation unit 252 generates suitability data 128 based on the stay history data 124 and the complaint history data 126 by weighting the number of stays in the stay count information 125 and the number of complaints in the complaint count information 127 for each target location RX. The first weighting coefficient and the second weighting coefficient are set in advance as appropriate. For example, the second weighting coefficient is larger than the first weighting coefficient (cb>ca).

[0051] The recommended installation location determination unit 253 reads out the suitability data 128 and determines recommended installation locations for the temperature sensors 3-1 to 3-n in accordance with each suitability. Specifically, the recommended installation location determination unit 253 preferentially determines the target location RX with the highest suitability as the recommended installation location. For example, the recommended installation location determination unit 253 selects the first to nth highest suitabilities from the N suitabilities (y1 to yN), and determines the target location RX corresponding to the selected suitability as the recommended installation location for the temperature sensors 3-1 to 3-n. As described above, n is the number of temperature sensors 3.

[0052] The calculation unit 252 and the recommended installation location determination unit 253 can appropriately determine a recommended installation location for the temperature sensor 3 by appropriately setting each weighting coefficient. For example, even if the number of complaints about a certain target location is not that high (for example, even if there are no complaints at the moment), if the number of times people are stagnating is significantly high, the location may be determined as a recommended installation location. In this case, the temperature sensor 3 is installed in advance in a location where complaints are likely to occur, thereby appropriately suppressing complaints about temperature.

[0053] The determination unit 250 does not necessarily need to calculate the suitability as long as it determines the recommended installation location based on the stay history data 124 and the complaint history data 126. For example, the determination unit 250 determines i target locations RX for which the number of complaints is equal to or greater than the threshold number as recommended installation locations for the temperature sensor 3 (i≧1). On the other hand, when the remaining number of temperature sensors 3 that can be installed in the facility 1 (j=ni) is 1 or greater and there are multiple target locations RX for which the number of complaints is less than the threshold number and the number of complaints is closest to the threshold number, the determination unit 250 may preferentially determine the location with the highest number of stays among these target locations as the recommended installation location. For example, when the number of target locations RX with the same number of complaints is k (k>j), the determination unit 250 determines the locations with the first to j-th highest number of stays among the k target locations RX as recommended installation locations.

[0054] 12 is a flowchart illustrating the processing executed by the control device 130 in the first embodiment. Hereinafter, step will be abbreviated as "S".

[0055] 12, the control device 130 reads out the staying history data 124 and the complaint history data 126 from the storage device 120 (S105). The control device 130 calculates the aforementioned suitability (y1 to yN in FIG. 11) for each target location RX based on the staying history data 124 and the complaint history data 126 (S115). The control device 130 determines recommended installation locations for the temperature sensors 3-1 to 3-n in accordance with these suitabilities (S120).

[0056] The control device 130 determines whether the recommended installation location is different from the current installation location (S125). Specifically, it determines whether at least one of the recommended installation locations is different from any of the current installation locations. If the recommended installation location is different from the current installation location (YES in S125), the control device 130 notifies the user of information urging them to change the installation location of the temperature sensor 3 (S130). The control device 130 controls the display device 115 to display, for example, screen 116. On the other hand, if the recommended installation location is exactly the same as the current installation location (NO in S125), the control device 130 controls the display device 115 to display a screen notifying the user that there is no need to change the installation location of the temperature sensor 3 (S135). After S130 and S135, the processing ends.

[0057] As described above, according to the first embodiment, both the history of whether or not there is a stagnation of people at each target location RX and the history of complaints about temperature are reflected in the recommended installation locations. This allows the installation locations of the temperature sensors 3-1 to 3-n to be determined in a way that is suitable for the tendency of people to flow within the facility 1. As a result, it is possible to prevent complaints about air conditioning in the facility 1 and appropriately ensure the thermal comfort of the customers CS.

[0058] [Embodiment 2] In Embodiment 2, the control device 130 uses a trained model generated by machine learning to predict the number of temperature complaints for each target location RX. Then, the control device 130 determines the recommended installation locations for temperature sensors 3-1 to 3-n according to the prediction results.

[0059] Fig. 13 is a diagram illustrating the overall configuration of an information processing system according to the second embodiment. Referring to Fig. 13, the information processing system 10A differs from the information processing system 10 of the first embodiment (Fig. 4) in that the storage device 120 of the terminal device 100 stores a training dataset 305 and a trained model 320, and the terminal device 100 includes a model generation unit 310. In other respects, the information processing system 10A is basically the same as the information processing system 10 unless otherwise specified. Therefore, detailed description will not be repeated.

[0060] This example shows a configuration for generating the trained model 320. The information processing system 10A also has a configuration for utilizing the trained model 320. This will be described in detail later.

[0061] The training dataset 305 is used to generate the trained model 320. The trained model 320 is generated using a machine learning algorithm, and in this example, it is a neural network model generated by supervised learning. The model generation unit 310 is a function performed by the CPU built into the control unit 130 or by the dedicated processing circuit described above.

[0062] Figure 14 is a diagram showing an example of each training data included in the training dataset 305. Referring to Figure 14, each training data 307 is generated by the control device 130 based on databases 36 and 44, with, for example, a past month as the information collection period. Each training data 307 includes camera number 372, target location information 374, dwell time information 125a, and complaint time information 127a. Camera number 372 and target location information 374 are the same as those shown in Figure 7.

[0063] The stay count information 125a differs from the stay count information 125 in that it indicates the number of stays per month for each target location RX. The complaint count information 127a differs from the complaint count information 127b in that it indicates the number of complaints about temperature per month for each target location RX.

[0064] Referring again to Figure 13, the model generation unit 310 generates a trained model 320 using a machine learning algorithm according to the training dataset 305. The trained model 320 learns the relationship between input parameters, which include multiple stay counts for multiple target locations RX (x1 to xN in Figure 14), and output parameters, which include multiple complaint counts for multiple target locations RX (z1 to zN in Figure 14), for a large number of patterns.

[0065] Figure 15 shows the configuration of the neural network used by the model generation unit 310. The neural network includes an input layer, an intermediate layer (hidden layer), and an output layer. Each of these layers consists of multiple neurons.

[0066] Referring to Figure 15, when the input parameters (x1, x2, ... xN) of the training data 307 are input to the input layer (X1-XN), the values ​​obtained by multiplying the input parameters by weights W1 (w11, w12, ...) are input to the hidden layer (Y1-Yk). The values ​​obtained by further multiplying the input values ​​to the hidden layer by weights W2 (w21, w22, ...) are output from the output layer (Z1-ZN) as output parameters (z1, z2, ... zN). The output result of the neural network changes depending on the values ​​of weights W1 and W2. The neural network performs supervised learning by adjusting the weights W1 and W2 so that the result output from the output layer approaches the output parameters of the training data 307 when the input parameters of the training data 307 are input to the input layer. The neural network may have two or more hidden layers.

[0067] Figure 16 shows the configuration of the information processing system 10A for utilizing the trained model 320. Referring to Figure 16, the storage device 120 of the information processing system 10A contains current installation location information 122, the trained model 320, and recent dwell time data 420.

[0068] The recent dwell time count data 420 represents the number of times a customer dwelled at each target location RX during the most recent period. The most recent period is the period from a predetermined time (e.g., one week) prior to the current time to the current time, which in this example is after the time the trained model 320 was generated. The recent dwell time count data 420 is generated by the control device 130 based on a plurality of dwell time determination data 37 associated with times during the most recent period.

[0069] The information processing system 10A includes, as its functional configuration, a prediction unit 410, a decision unit 415, and a notification unit 260. The functions of the prediction unit 410 and the decision unit 415 are performed by a CPU built into the control device 130 or by the dedicated processing circuit described above. The notification unit 260 is the same as the one shown in Figure 9.

[0070] The prediction unit 410 uses the trained model 320 to execute a prediction process for predicting the future number of complaints about temperature at multiple target locations RX in accordance with the most recent stay count data 420. Specifically, the prediction unit 410 first provides the number of times people have stayed at multiple target locations RX during the most recent period as input parameters to a neural network ( FIG. 15 ). Then, the prediction unit 410 acquires the future number of complaints about temperature at multiple target locations RX as output parameters of the neural network.

[0071] The determining unit 415 determines recommended installation locations for the temperature sensors 3-1 to 3-n according to the results of the prediction process (specifically, the above-mentioned output parameters). The determining unit 415 determines, for example, the target locations RX with the first to nth highest predicted number of complaints as the recommended installation locations. The notifying unit 260 notifies the user of information indicating the recommended installation locations determined by the determining unit 415.

[0072] The floor map of Facility 1 is updated due to the closure or opening of stores within Facility 1. In this case, the trends in pedestrian flow within Facility 1 change. As a result, for each target location RX, the likelihood of pedestrian congestion at that location changes. Therefore, the locations where temperature complaints are likely to occur also change.

[0073] According to Embodiment 2, a pre-trained model 320 can be created in advance that has learned the relationship between the number of pedestrian dwell times and the number of temperature complaints during the period before the floor map is updated (for example, the relationship for one month, the relationship for another month, and the relationship for yet another month, etc.). Then, using the pre-trained model 320, the number of temperature complaints during the period after the floor map is updated can be predicted immediately after the floor map is updated, according to the most recent dwell time data 420. This makes it possible to appropriately determine recommended installation locations while appropriately addressing changes in locations where temperature complaints are likely to occur due to floor map updates.

[0074] [Modification 1 of Embodiment 2] The tendency of people flowing inside the facility 1 may also change depending on the temperature (outside temperature) outside the facility 1. For example, the tendency of people flowing in seasons when the outside temperature is very high or low, such as summer or winter, differs from the tendency of people flowing in seasons when the outside temperature is moderate, such as spring or autumn.

[0075] For these reasons, if the outside temperature during the period in which the number of complaints is predicted (hereinafter also referred to as the "prediction period") is close to the average outside temperature to the extent that it falls within the same temperature range as the outside temperature during past information collection periods (for example, if the season of the prediction period is the same as the season of the information collection period), then the trend in human movement during the prediction period is likely to be similar to the trend in human movement during the information collection period. The temperature range is, for example, one of the following: 5°C to 15°C (winter temperature range), 15°C to 25°C (spring / autumn temperature range), or 25°C to 30°C (summer temperature range). On the other hand, if the outside temperature during the prediction period falls within a different temperature range than the outside temperature during past information collection periods (for example, if the season of the prediction period is different from the season of the information collection period), then the trend in human movement during the prediction period may be significantly different from the trend in human movement during the information collection period. Note that the prediction period is the period after the information collection period.

[0076] Therefore, in the information processing system according to this first modification, the control device 130 generates multiple trained models for predicting the number of complaints, each associated with one of the temperature ranges described above. Then, from the multiple trained models, the control device 130 selects one trained model associated with the temperature range to which the temperature index value outside the facility 1 during the prediction period belongs. In the following description, the temperature index value is assumed to be a value indicating the average daytime temperature outside the facility 1, but it may also be a value indicating the maximum or minimum daytime temperature. The temperature index value is determined, for example, for each month. The control device 130 uses the selected trained model to predict the number of complaints for each target location RX.

[0077] Fig. 17 is a diagram showing the overall configuration of an information processing system according to this first modification. Referring to Fig. 17, information processing system 10B differs from information processing system 10A of embodiment 2 (Fig. 13) in that storage device 120 stores training data sets 305-1, 305-2... and trained models 320-1, 320-2... In other respects, unless otherwise specified, information processing system 10B is basically the same as information processing system 10A. Therefore, detailed description will not be repeated.

[0078] Each training dataset indicates the number of stays and the number of complaints at multiple target locations RX during a corresponding information collection period (e.g., one month). Each training dataset corresponds to training dataset 305 (Figure 13). Each trained model learns the relationship between the number of stays and the number of complaints at multiple target locations RX during a corresponding information collection period. For each trained model, the information collection period corresponding to the trained model is the period during which information on the training data used to generate the trained model was collected. Each trained model corresponds to trained model 320.

[0079] This example shows a configuration for generating trained models 320-1, 320-2, .... The information processing system 10B also has a configuration for utilizing these trained models. This will be explained in detail later.

[0080] FIG. 18 is a diagram showing an example of each piece of training data included in the training data sets 305-1, 305-2, .... Referring to FIG. 18, each piece of training data 308 is generated by the control device 130 based on the databases 36, 44. Each piece of training data 308 differs from the training data 307 (FIG. 14) in that it further includes temperature range information 309. The temperature range information 309 indicates the temperature range to which the temperature index value during the information collection period (in the month in which the number of stays and the number of complaints were determined) corresponding to the training data 308 belongs. The temperature index value is determined based on the measurement value of an outside air temperature sensor (not shown) installed outside the facility 1. The measurement value of this sensor is provided to the control device 130.

[0081] For example, the temperature index value associated with the training data 308 of the training data set 305-1 belongs to the temperature range r1. The temperature index value associated with the training data 308 of the training data set 305-2 belongs to the temperature range r2. In one example, the temperature range r1 is 15°C to 25°C, and the temperature range r2 is 25°C to 30°C.

[0082] In this example, each piece of training data 308 in the training data set 305-1 represents the number of stays and the number of complaints at each target location RX in the spring, and each piece of training data 308 in the training data set 305-2 represents the number of stays and the number of complaints at each target location RX in the summer.

[0083] The model generation unit 310 generates a trained model for each training dataset by, for example, supervised learning based on a neural network model. In one example, the model generation unit 310 generates trained model 320-1 using training dataset 305-1. Similarly, the model generation unit 310 generates trained model 320-2 using training dataset 305-2. Each trained model is generated by the model generation unit 310 in association with the temperature range to which the temperature index value belongs during the information collection period corresponding to the trained model (for example, in the case of trained model 320-1, the period during which information on the training data included in training dataset 305-1 was collected).

[0084] Figure 19 shows the configuration of the information processing system 10B for utilizing the trained models 320-1, 320-2, etc. Referring to Figure 19, the storage device 120 of the information processing system 10A contains current installation location information 122, trained models 320-1, 320-2, etc., and recent dwell time data 420.

[0085] The information processing system 10B according to this first modification differs from the information processing system 10A of the second embodiment in that its functional configuration further includes a selection unit 404. The function of the selection unit 404 is performed by a CPU built into the control device 130 or the dedicated processing circuit described above. The prediction unit 410, the determination unit 415, and the notification unit 260 are the same as those shown in FIG. 16.

[0086] The selection unit 404 selects one trained model from the trained models 320-1, 320-2, ... that is associated with the temperature range to which the temperature index value within the prediction period belongs. Information indicating the temperature index value within the prediction period is input by the user using, for example, the input device 110 (FIG. 4). For example, if the temperature index value within the prediction period belongs to temperature range r1, the selection unit 404 selects trained model 320-1. Similarly, if the temperature index value within the prediction period belongs to temperature range r2, the selection unit 404 selects trained model 320-2.

[0087] The prediction unit 410 executes a prediction process according to the result of the selection of the trained model. This prediction process corresponds to a process of predicting the number of complaints for multiple target locations RX within a prediction period using one trained model selected by the selection unit 404 according to the most recent stay count data 420. The determination unit 415 determines recommended installation locations for the temperature sensors 3-1 to 3-n according to the result of the prediction process.

[0088] According to Modification 1, the selection unit 404 selects a trained model from among the trained models 320-1, 320-2, etc., that is associated with an outside temperature close to the outside temperature during the prediction period. Then, the number of complaints for multiple target locations RX during the prediction period is predicted using the selected trained model. As a result, the trend of pedestrian flow caused by the outside temperature is reflected in the prediction result of the number of complaints. For example, if the season of the prediction period is summer (in one example, June to August), a trained model that has learned the relationship between the number of pedestrian dwells and the number of complaints in these months over the past few years is selected, and the prediction process is performed using this trained model. As a result, the number of complaints for multiple target locations RX during the prediction period can be predicted more appropriately. Therefore, the recommended installation locations for temperature sensors 3-1 to 3-n can be determined more appropriately according to the predicted number of complaints.

[0089] [Modification 2 of Embodiment 2] Whether or not a customer (CS) is likely to complain about temperature may vary depending on the customer's attributes (e.g., age group and gender). Specifically, customers with certain attributes may be relatively more likely to complain about temperature, while customers without those attributes may be relatively less likely to complain. The main customer base for each store in Facility 1 is determined by customer attributes, but the flow of people may change due to store closures or openings. In this case, the locations where customers with certain attributes are likely to be found also change. As a result, the locations where temperature complaints are likely to occur also change. Therefore, in this modified example, the aforementioned prediction process is performed based on the customer's attributes as well.

[0090] Figure 20 is a diagram showing the overall configuration of the information processing system according to this modified example 2. Referring to Figure 20, the information processing system 10C differs from the information processing system 10A of Embodiment 2 in that the processing unit 32 of the server 30 further includes an attribute recognition unit 38 and a customer determination unit (person determination unit) 41. The information processing system 10C also differs from the information processing system 10A in that the storage device 33 of the server 30 stores stagnant customer determination data 39-1, 39-2, 39-3… in the database 36 instead of stagnant determination data 37-1, 37-2, 37-3…. Each of the stagnant customer determination data is also referred to as "stagnant customer determination data 39". The information processing system 10C also differs from the information processing system 10A in that the storage device 120 includes a training dataset 405 and a trained model 330. In other respects, unless otherwise specified, the information processing system 10C is basically the same as the information processing system 10A. Therefore, detailed description will not be repeated.

[0091] The attribute recognition unit 38 performs attribute recognition processing for each captured image (IMG) to recognize the attributes of each person (customer CS) appearing in the captured image (IMG) according to the image. The attribute recognition processing is performed at regular time intervals using image processing technology. In this example, the attributes of the customer CS are their age group and gender. Therefore, the attribute recognition processing is a process of recognizing the age group and gender by, for example, determining the facial contours, the relative positions of the eyes, nose, and mouth, and the skin color of the customer CS appearing in the captured image (IMG).

[0092] The customer determination unit 41 performs a customer determination process for each target location RX, determining whether the customers CS present at the location include customers with specific attributes, based on the results of the recognition process, if congestion of people is occurring at the location RX. Specific attributes are predetermined as attributes of customers CS who are considered more likely to complain about temperature than customers CS who do not possess these specific attributes. Specific attributes are determined by prior statistical surveys, such as teenage males, 20s females, 30s males, 40s females, 50s males, or 60s females. The customer determination process, like the attribute recognition process, is performed at regular time intervals. Congestion / customer determination data 39 is generated according to the results of the aforementioned congestion determination process and customer determination process.

[0093] Figure 21 shows an example of dwelling / customer determination data 39. Referring to Figure 21, dwelling / customer determination data 39 differs from dwelling determination data 37 (Figure 5) in that it further includes customer determination information 378. Customer determination information 378 indicates whether or not those customers CS include customers with specific attributes when it is determined that there is dwelling in the flow of people. Since the dwelling determination process and the customer determination process are executed at regular time intervals, dwelling / customer determination data 39 is generated at regular time intervals. For example, dwelling / customer determination data 39-1, 39-2, 39-3… are generated at regular time intervals and are associated with their generation date and generation time.

[0094] Figure 22 shows an example of each training data included in the training dataset 405. Referring to Figure 22, each training data 407 is generated by the control device 130 based on databases 36 and 44, for example, every month. The training data 407 differs from the training data 307 (Figure 14) in that it further includes judgment count information 123a.

[0095] The judgment count information 123a indicates the number of times, for each target location RX, that the customer determination unit 41 determined that customers in the target location RX included customers with specific attributes (specific customer determination count) during the period in which the learning data 407 information was collected.

[0096] 20 again, the model generation unit 310 generates a trained model 330 using the training dataset 405. The trained model 330 learns the relationship between input parameters including multiple numbers of stays and multiple numbers of specific customer determinations for multiple target locations RX (x1 to xN and xa1 to xaN in FIG. 22) and output parameters including multiple numbers of complaints for multiple target locations RX (z1 to zN in FIG. 22) for a large number of patterns.

[0097] Figure 23 illustrates the method by which the model generation unit 310 generates the trained model 330. Referring to Figure 23, the model generation unit 310 learns the relationship between the input and output parameters described above through supervised learning based on the neural network model shown.

[0098] Figure 24 is a diagram showing the configuration of the information processing system 10C for utilizing the trained model 330. Referring to Figure 24, the storage device 120 of the information processing system 10C contains current installation location information 122, the trained model 330, recent dwell time data 420, and recent judgment time data 440.

[0099] The most recent determination count data 440 indicates, for each target location RX, the number of times during the most recent period that the customer determination unit 41 determined that customers in the target location RX include a customer CS with a specific attribute. As described above, the most recent period is the period from a time a predetermined time before the current time to the current time, which in this example is after the generation of the trained model 330.

[0100] The prediction unit 410 uses the trained model 330 to predict the number of complaints for each target location RX according to the most recent dwell time data 420 and the most recent judgment time data 440. Specifically, the prediction unit 410 first provides the number of dwell times and the number of specific customer judgments for multiple target locations RX during the most recent period as input parameters to the neural network (Figure 23). Then, the prediction unit 410 obtains the number of future complaints regarding temperature for multiple target locations RX as an output parameter of the neural network, and thereby predicts the number of complaints for each target location RX.

[0101] According to this modified version 2, the input parameters further include multiple specific customer determination counts (xa1 to xaN) for multiple target locations RX. As a result, these counts are reflected in the predicted number of complaints. For example, if there is congestion of people at one target location RX and those customers CS often include specific customers, the determination unit 415 determines the recommended installation locations for temperature sensors 3-1 to 3-n, assuming that there is a relatively high probability of complaints about temperature being made at that location in the future. Therefore, the user is notified of the recommended installation locations, and temperature sensors 3-1 to 3-n are installed in such locations. As a result, temperature sensors 3 are pre-installed in locations where specific customers are likely to be present. Thus, complaints about temperature within facility 1 can be suppressed more effectively.

[0102] The embodiments disclosed herein should be considered to be illustrative in all respects and not restrictive. The scope of the present invention is defined by the claims, not by the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0103] 1 Facility, 3 Temperature sensor, 5 Air conditioning system, 10, 10A, 10B, 10C Information processing system, 20 Camera, 30, 40 Server, 35 Dwelling determination unit, 38 Attribute recognition unit, 41 Customer determination unit, 250, 415 Decision unit, 252 Calculation unit, 253 Recommended installation location determination unit, 260 Notification unit, 310 Model generation unit, 404 Selection unit, 410 Prediction unit.

Claims

1. an imaging unit that captures images of a plurality of target locations in a facility equipped with an air conditioning system, thereby generating a plurality of captured images each showing the plurality of target locations; a stagnation determination unit that determines, for each of the plurality of target locations, whether or not a stagnation of people is occurring at the target location based on a photographed image that shows the target location; a storage unit configured to store first history data indicating a history of the results of the determination by the stay determination unit, and second history data indicating a history of complaints about temperature at each of the plurality of target locations; and a determination unit that determines, based on the first history data and the second history data, a recommended installation location within the facility for a temperature sensor that outputs a measurement value to the air conditioning system.

2. the storage unit further stores information indicating a current installation location, which is a location in the facility where the temperature sensor has already been installed; 2. The information processing system according to claim 1, further comprising a notification unit that, when the recommended installation location is different from the current installation location, notifies information urging a change of installation location of the temperature sensor from the current installation location to the recommended installation location.

3. The first history data includes, for each of the plurality of target locations, a first number of times indicating the number of times the user stayed at the target location within a first period; The information processing system according to claim 1 or 2, wherein the second history data includes, for each of the plurality of target locations, a second number indicating the number of complaints at the target location within the first period.

4. The determination unit a calculation unit that calculates, for each of the plurality of target locations, an added value of a product of the first number of times and a first weighting coefficient and a product of the second number of times and a second weighting coefficient; The information processing system according to claim 3 , further comprising: a recommended installation location determination unit that determines the recommended installation location in accordance with the additional value.

5. the storage unit stores a trained model in which a relationship between input parameters including the first number of times for the plurality of target locations and output parameters representing the second number of times for the plurality of target locations is trained; A prediction unit that predicts the number of complaints for the plurality of target locations after the first period using the trained model according to the number of stays at the plurality of target locations after the first period, The information processing system according to claim 3 , wherein the determination unit determines the recommended installation location in accordance with a prediction result of the prediction unit.

6. The storage unit stores a plurality of the trained models, Each of the plurality of trained models is associated with a temperature range to which the temperature index value outside the facility belongs during the first period, which is a period during which information on training data used to generate the trained model was collected; and The information processing system further includes a selection unit that selects, from the plurality of trained models, a trained model associated with a temperature range to which the outside-facility temperature index value in a second period after the first period belongs, The information processing system according to claim 5 , wherein the prediction unit predicts the number of complaints for the plurality of target locations within the second period using the selected trained model.

7. a recognition unit that recognizes attributes of people appearing in each of the plurality of captured images; a person determination unit that determines, for each of the plurality of target locations, whether or not people present at the target location include a person with a specific attribute when the stay occurs at the target location, according to a recognition result of the recognition unit; the specific attribute is predetermined as an attribute of a person who is more likely to make the complaint than a person who does not have the specific attribute; The input parameters further include a number of times that the person determination unit has determined that people in the target location include people with the specific attribute within the first period, 6. The information processing system according to claim 5, wherein the prediction unit uses the trained model to predict the number of complaints based on the number of times the person has stayed at the plurality of target locations after the first period and the number of times determined by the person determination unit after the first period.

8. The information processing system according to claim 7 , wherein the attributes of the person appearing in the photographed image include an age group and a gender of the person.

Images