Counting device, counting method, and computer program for counting

Abstract

To provide a counting device capable of accurately counting the number of persons in a predetermined area.SOLUTION: A counting device includes: a detection unit 21 configured to detect, from each of a plurality of time-series images generated by an imaging unit 2 configured to capture images of a predetermined area 1c, one or more persons in the predetermined area; a tracking unit 22 configured to track, for each of the one or more detected persons, the person in one or more images representing the person among the plurality of images; and a counting unit 23 configured to determine, based on tracking results, whether the position of each of the detected persons in the image is within a hiding determination range in an area on the image corresponding to the predetermined area, and, if the duration during which the position in the images of any of the one or more detected persons is outside the hiding determination range is not less than a predetermined time threshold, count the number of persons depicted in any of the images during that duration as the number of persons present in the predetermined area.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a counting device, a counting method, and a computer program for counting the number of people within a predetermined area. [Background technology]

[0002] A technique has been proposed for counting the number of people getting on and off a vehicle based on a moving image including the entrance and exit of the vehicle (see Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-149027 Summary of the Invention [Problem to be solved by the invention]

[0004] When a vehicle is crowded, multiple people may appear to overlap from the camera. In such cases, one person may be hidden by another, making it difficult to accurately count the number of people getting on or off the vehicle or the number of people staying in the vehicle.

[0005] SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide a counting device that can accurately count the number of people within a predetermined area. [Means for solving the problem]

[0006] One aspect of the present invention provides a counting device that includes a detection unit that detects one or more people within a predetermined area from each of a plurality of time-series images generated by an imaging unit configured to capture the predetermined area, a tracking unit that tracks each of the detected people in one or more images among the plurality of images in which the person is depicted, and a counting unit that determines, based on the tracking results, whether the position of each of the detected people in the image is within an occlusion determination range in an area on the image corresponding to the predetermined area, and, if a duration during which none of the detected people in the image is within the occlusion determination range is equal to or longer than a predetermined time threshold, counts the number of people depicted in any of the images during that duration as the number of people present in the predetermined area.

[0007] In one embodiment, if, after counting the number of visitors, the duration is less than a time threshold, the counting device counts, based on the results of tracking, the number of people who crossed an area adjacent to the entrance / exit of the specified area from the entrance / exit side to the inside of the specified area as the number of people entering, and counts the number of people who crossed the adjacent area from the inside of the specified area to the entrance / exit side as the number of people exiting, and further has a correction unit that corrects the number of visitors by adding to the number of people entering minus the number of people exiting.

[0008] In one embodiment, the detection unit detects, for each of one or more persons, a head region representing the person's head and a person region representing the person's torso from the image, the tracking unit tracks the head region and the person region for each of the one or more persons, the counting unit counts the number of person regions represented in any image during the continuous period as the number of visitors, and the correction unit counts the number of people entering and exiting based on the results of tracking the head regions.

[0009] According to another embodiment, there is provided a counting method including: detecting one or more persons within a predetermined area from each of a plurality of time-series images generated by an imaging unit configured to capture the predetermined area; tracking each of the detected one or more persons in one or more images in which the person is depicted among the plurality of images; determining, based on a result of the tracking, whether the position of each of the detected one or more persons in the image is within an occlusion determination range in an area on the image corresponding to the predetermined area; and, if a duration during which the position of any of the detected one or more persons in the image is not within the occlusion determination range is equal to or greater than a predetermined time threshold, counting the number of the one or more persons depicted in any of the images during the duration as the number of visitors present in the predetermined area.

[0010] According to yet another embodiment, there is provided a counting computer program, the counting computer program including instructions for causing a computer to execute the following: detect one or more persons within a predetermined area from each of a plurality of time-series images generated by an imaging unit configured to capture the predetermined area; track each of the detected one or more persons in one or more images among the plurality of images in which the person is depicted; determine, based on the tracking results, whether the position of each of the detected one or more persons in the image is within an occlusion determination range in an area on the image corresponding to the predetermined area; and, if a duration during which the position of any of the detected one or more persons in the image is not within the occlusion determination range is equal to or greater than a predetermined time threshold, count the number of the one or more persons depicted in any of the images during the duration as the number of visitors present in the predetermined area. [Effects of the Invention]

[0011] The counting device according to the present disclosure has the effect of being able to accurately count the number of people within a predetermined area. [Brief explanation of the drawings]

[0012] [Figure 1]1 is a schematic diagram of a vehicle in which a counting device according to an embodiment is mounted; [Figure 2] FIG. [Figure 3] FIG. 2 is a schematic diagram of a counting device. [Figure 4] FIG. 2 is a functional block diagram of a processor related to counting processing. [Figure 5] 1(a) to 1(c) are diagrams each illustrating an outline of the counting process. [Figure 6] 10 is an operational flowchart of a counting process. DETAILED DESCRIPTION OF THE INVENTION

[0013] A counting device, a counting method, and a counting computer program executed on the counting device will be described below with reference to the drawings. The counting device detects one or more people within a predetermined area from each of a plurality of time-series images generated by an imaging unit, and tracks each detected person. Based on the tracking results, the counting device determines whether the position of each detected person on the image is within an occlusion determination range within an area on the image corresponding to the predetermined area. If the duration during which no person's position on the image is within the occlusion determination range is equal to or greater than a predetermined time threshold, the counting device counts the number of people appearing in any of the images during that duration as the number of people present in the predetermined area (hereinafter referred to as the number of visitors).

[0014] In the following, an example will be described in which the counting device is used to count the number of passengers on board a vehicle that can accommodate multiple passengers. Passengers are an example of people to be counted. However, the counting device is not limited to this example and may be used to count the number of people in a predetermined area provided inside a moving object that can accommodate passengers or crew members, such as a railway vehicle, or to count the number of people in a predetermined area provided in some kind of building or facility.

[0015] Fig. 1 is a schematic diagram of a vehicle equipped with a counting device according to one embodiment. Fig. 2 is a diagram of the interior of the vehicle, seen from above, equipped with the counting device. The vehicle 1 equipped with the counting device is a vehicle, such as a bus, that can accommodate multiple passengers and has enough interior space for passengers to stand and move around. The vehicle 1 has a camera 2, a notification device 3, and a counting device 4.

[0016] Additionally, inside vehicle 1, entrance area 1b is set around boarding / alighting door 1a, which is an example of an entrance / exit for the interior area of ​​vehicle 1. Entrance area 1b is an example of an adjacent area adjacent to the entrance / exit, and is set as an area adjacent to boarding / alighting door 1a and through which passengers must pass when boarding vehicle 1 through boarding / alighting door 1a or when disembarking from vehicle 1 through boarding / alighting door 1a.

[0017] Camera 2 is an example of an imaging unit, and is mounted facing downward, for example, near the ceiling on the interior side of vehicle 1, at entrance / exit 1a, so that its imaging range includes the entire interior area 1c where passengers can stay inside vehicle 1. Interior area 1c is an example of a predetermined area to be imaged by the imaging unit. Camera 2 generates an image representing interior area 1c at predetermined imaging intervals (e.g., 1 / 30 to 1 / 10 seconds). Each time it generates an image, camera 2 outputs the generated image to counting device 4 via the in-vehicle network.

[0018] The notification device 3 is a device that can issue predetermined notifications to passengers staying inside the vehicle 1, and has, for example, a speaker, buzzer, or display device, and is attached near the boarding / alighting door 1a or near the ceiling inside the vehicle 1. In response to a notification signal from the counting device 4, the notification device 3 outputs a predetermined notification, for example, a sound that indicates a notification that warns that the number of passengers inside the vehicle 1 has exceeded capacity, or displays a message corresponding to that notification.

[0019] The counting device 4 performs a counting process based on the image generated by the camera 2 .

[0020] Fig. 3 is a hardware configuration diagram of the counting device 4. As shown in Fig. 3, the counting device 4 has a communication interface 11, a memory 12, and a processor 13. The communication interface 11, the memory 12, and the processor 13 may each be configured as separate circuits, or may be integrated into a single integrated circuit.

[0021] The communication interface 11 has an interface circuit for connecting the counting device 4 to the in-vehicle network. The communication interface 11 passes the image received from the camera 2 to the processor 13. The communication interface 11 also outputs the notification signal received from the processor 13 to the notification device 3.

[0022] The memory 12 is an example of a storage unit and includes, for example, a volatile semiconductor memory and a non-volatile semiconductor memory. The memory 12 stores various programs and various data used in the counting process executed by the processor 13 of the counting device 4. For example, the memory 12 stores parameters for identifying a classifier used to detect occupants, the positions and ranges of various regions on an image, etc. The memory 12 also temporarily stores images received from the camera 2 and various data generated during the counting process.

[0023] The processor 13 includes one or more central processing units (CPUs) and their peripheral circuits. The processor 13 may further include other arithmetic circuits such as a logic unit, a numerical calculation unit, or a graphics processing unit. The processor 13 performs counting processing.

[0024] 4 is a functional block diagram of processor 13 related to counting processing. Processor 13 has a detection unit 21, a tracking unit 22, a counting unit 23, a correction unit 24, and a notification processing unit 25. Each of these units in processor 13 is, for example, a functional module realized by a computer program running on processor 13. Alternatively, each of these units in processor 13 may be a dedicated arithmetic circuit provided in processor 13.

[0025] The detection unit 21 detects passengers located within the vehicle interior area from each of a plurality of time-series images generated by the camera 2. In this embodiment, the detection unit 21 detects passengers from the latest image obtained by the camera 2 at predetermined intervals. The detection unit 21 only needs to perform the same processing on each image, so the processing on one image will be described below.

[0026] In this embodiment, the detection unit 21 individually detects, for each passenger, a person region representing at least the passenger's trunk and a head region representing the passenger's head from the image. Note that the person region may include not only the trunk but also other parts of the passenger, such as any or all of the head, arms, and legs. In the following description, the person region is assumed to include the passenger's trunk.

[0027] The detection unit 21 detects the trunk and head of a passenger by inputting the images received by the counting device 4 from the camera 2 into a classifier that has been trained in advance to detect these parts. Such a classifier is based on a so-called deep neural network (DNN). For example, a DNN with a convolutional neural network (CNN)-type architecture, such as Single Shot MultiBox Detector or YOLO, or a DNN with an attention mechanism, such as Vision Transformer, may be used. Alternatively, a classifier based on another machine learning method, such as AdaBoost, may be used. The classifier is trained in advance according to a predetermined learning method, such as backpropagation, using a large number of training images including images depicting the head and trunk.

[0028] The classifier outputs confidence levels for the head and trunk, each representing the likelihood that each of these regions on the input image represents a specific body part. The detection unit 21 then detects, as a head region, a region where the confidence level for the head is equal to or greater than a specific detection threshold, and detects, as a person region, a region where the confidence level for the trunk is equal to or greater than a specific detection threshold. Furthermore, when multiple person regions overlap, the detection unit 21 performs non-maximum suppression (NMS) or soft NMS to prevent multiple detections of a single passenger. Specifically, the detection unit 21 calculates the intersection over union (IoU) of multiple overlapping person regions, and if the IoU is equal to or greater than a specific threshold, removes person regions other than the person region with the highest confidence level. Alternatively, the detection unit 21 reduces the confidence level as the IoU increases, and removes person regions where the reduced confidence level falls below the specific detection threshold. The detection unit 21 may perform the same process on a plurality of overlapping head regions to prevent the head of one passenger from being detected multiple times.

[0029] The detection unit 21 notifies the tracking unit 22, the counting unit 23, and the correction unit 24 of the positions and ranges of the person region and head region for each passenger detected from the image.

[0030] The tracking unit 22 tracks a detected passenger in one or more images in which the passenger is depicted, among a plurality of time-series images generated by the camera 2. In this embodiment, the trunk and head of each passenger are detected separately, and therefore the tracking unit 22 performs tracking processing on each of the detected trunk and head of the passenger. That is, for each passenger detected across a plurality of images, the tracking unit 22 associates person regions and head regions of the same passenger across a plurality of images. The tracking processing for person regions will be described below, but the tracking unit 22 may also perform the same processing on head regions.

[0031] The tracking unit 22 applies a predetermined tracking method, such as KLT tracking or ByteTrack, to each person region in the latest image. As a result, the tracking unit 22 associates each person region in the latest image with a person region of the same passenger detected and currently being tracked in a previously acquired image (hereinafter referred to as a past image). Each time the tracking unit 22 receives a detection result for the latest image from the detection unit 21, the tracking unit 22 repeats the above process to track the torso of each passenger, assigns a unique identification number (hereinafter referred to as a passenger ID) to the torso of each passenger currently being tracked, and defines a line connecting the center of gravity of each person region currently being tracked in chronological order as the passenger's trajectory. For a person region detected in the latest image that cannot be associated with any person region representing a passenger currently being tracked in a past image, the tracking unit 22 considers the passenger represented in that person region to have newly entered the vehicle interior and begins new tracking. Conversely, if the person area for any of the passengers being tracked in the past images is not associated with any of the person areas in the latest images, the tracking unit 22 assumes that the passenger being tracked has exited the interior of the vehicle and terminates tracking.

[0032] As described above, in this embodiment, the trunk and head of the same passenger are detected and tracked separately, and therefore separate passenger IDs are assigned to the trunk and head.

[0033] Counting unit 23 determines whether the position on the image of each of the one or more detected passengers is included within an occlusion determination range in the area on the image corresponding to the vehicle interior area, based on the tracking result by tracking unit 22. If the duration during which the position on the image of any of the one or more detected passengers is not included within the occlusion determination range is equal to or longer than a predetermined time threshold (for example, several seconds to 10-odd seconds), counting unit 23 counts the number of one or more people included in the vehicle interior area depicted in any of the images during that duration as the number of passengers present in the vehicle interior area.

[0034] The counting unit 23 may determine the number of person regions present in the vehicle interior area as the number of visitors, or may determine the number of head regions present in the vehicle interior area as the number of visitors. However, it is preferable to use, for counting the number of visitors, either the person region or the head region, whichever is more accurately detected by the detection unit 21 in the vehicle interior area. For example, if the detection accuracy of the person region is higher than the detection accuracy of the head region, the counting unit 23 counts the number of person regions present in the vehicle interior area as the number of visitors.

[0035] Note that the counting unit 23 may count each passenger present in the vehicle interior area as one passenger if either a person area or a head area is detected. In this case, the counting unit 23 determines that the person area and the head area represent the same passenger if the center of gravity of a certain person area and the center of gravity of a certain head area are within a predetermined distance. Alternatively, the counting unit 23 may determine that the person area and the head area represent the same passenger if the average value of the difference between the positions of the trajectories of a certain person area and the trajectory of a certain head area in each image being tracked is within a predetermined distance.

[0036] When counting the number of visitors based on person regions, counting unit 23 may count, as included in the vehicle interior region, any person region whose overlapping ratio with the vehicle interior region is equal to or greater than a predetermined ratio (e.g., 50% to 80%). The same can be done when counting the number of visitors based on head regions. Furthermore, when the overlapping ratio of the person region (or head region) with the occlusion determination region is equal to or greater than a predetermined ratio for any passenger, counting unit 23 determines that the position of the passenger on the image is included in the occlusion determination range. Alternatively, counting unit 23 may estimate the position of the passenger's feet shown in the person region to determine whether the person region is included in the occlusion determination range. If the estimated position of the passenger's feet is included in the occlusion determination range, counting unit 23 determines that the position of the passenger shown in the person region is included in the occlusion determination range. In this embodiment, camera 2 is mounted so as to face downward from the ceiling inside the vehicle. Therefore, the counting unit 23 estimates the position of the intersection between a line extending from a reference point in the person area toward the vanishing point of the image and one of the sides of the person area as the position of the feet of the passenger depicted in that person area. Note that the reference point may be set to, for example, the center of gravity of the person area.

[0037] The counting unit 23 notifies the correction unit 24 and the notification processing unit 25 of the number of visitors.

[0038] If the duration is less than the time threshold after the counting unit 23 counts the number of staying passengers, the correction unit 24 counts, based on the results of tracking by the tracking unit 22 after the counting, the number of passengers who crossed the entrance area 1b from the boarding / alighting door 1a side to the interior of the vehicle interior area (hereinafter referred to as "enterers") as the number of entering passengers. Furthermore, the correction unit 24 counts the number of passengers who crossed the entrance area 1b from the interior of the vehicle interior area to the boarding / alighting door 1a side (hereinafter referred to as "exiting passengers") as the number of exiting passengers. The correction unit 24 then corrects the number of staying passengers by adding the number obtained by subtracting the number of exiting passengers from the number of entering passengers to the number of staying passengers. In this way, by counting the number of passengers who cross the entrance area when passengers located within the hiding determination range have not disappeared, the correction unit 24 can accurately count the number of passengers staying in the vehicle.

[0039] The correction unit 24 refers to the trajectory of each passenger being tracked to determine whether the passenger has crossed the entrance area 1b. The correction unit 24 then determines, as an entering passenger, a passenger whose trajectory includes entering the entrance area 1b from the side of the entrance area 1b on the doorway 1a side (hereinafter simply referred to as the doorway side) and exiting the entrance area 1b from the side of the entrance area 1b located inside the interior of the train (hereinafter referred to as the car's inner side). Note that a passenger may be detected for the first time after entering the entrance area 1b. Therefore, the correction unit 24 may also determine, as an entering passenger, a passenger whose first detected position is within the entrance area 1b and whose position is closer to the doorway side edge than the car's inner side edge. Similarly, the correction unit 24 determines, as an exiting passenger, a passenger whose trajectory includes entering the entrance area 1b from the car's inner side edge and exiting the entrance area 1b from the doorway side edge. Note that tracking of a passenger may end before the passenger exits the vehicle through the boarding / alighting door 1a. Therefore, the correction unit 24 may also count a passenger who enters the entrance area 1b from the inside edge of the vehicle and whose last detected position is closer to the boarding / alighting door side edge than the inside edge of the vehicle within the entrance area 1b as an exiting passenger. However, the correction unit 24 does not count a passenger who enters the entrance area from the boarding / alighting door side edge and whose trajectory exits the entrance area from the boarding / alighting door side edge as either an entering or exiting passenger. Similarly, the correction unit 24 does not count a passenger who enters the entrance area from the inside edge of the vehicle and whose trajectory exits the entrance area from the inside edge of the vehicle as either an entering or exiting passenger. Furthermore, the correction unit 24 may not count a passenger who has been in the entrance area for a certain period of time or more as either an entering or exiting passenger.

[0040] The correction unit 24 may count the number of entering and exiting passengers based on the trajectory of the head region of each detected passenger, or may count the number of entering and exiting passengers based on the trajectory of the person region. However, it is preferable that the correction unit 24 use, for counting the number of entering and exiting passengers, either the person region or the head region, whichever can more accurately determine the trajectory crossing the entrance region. For example, if the trajectory of the head region can be determined more accurately than the trajectory of the person region, the correction unit 24 counts the number of entering and exiting passengers based on the trajectory of the head region of each detected passenger. In this way, the counting unit 23 and the correction unit 24 use, respectively, the person region or the head region with higher detection accuracy or tracking accuracy, thereby more accurately counting the number of staying passengers.

[0041] In addition, the difference between the number of entering passengers and the number of exiting passengers counted by the correction unit 24 at a point in time before the counting unit 23 counts the number of passengers staying in the vehicle may be counted as the number of staying passengers.

[0042] Every time the correction unit 24 corrects the number of visitors, it notifies the notification processing unit 25 of the corrected number of visitors.

[0043] 5(a) to 5(c) are diagrams each illustrating an outline of the process of counting the number of visitors. In this example, the entrance / exit door 1a is shown near the bottom of an image 500 showing the interior area 1c of the vehicle 1, as shown in FIG. 5(a) to FIG. 5(c).

[0044] In the example shown in Fig. 5(a), of the passengers detected in the interior area 510 of the vehicle shown in the image 500, passenger 501 is included in the occlusion determination range 520. The occlusion determination range 520 is set as a range within which, if there is a passenger captured within the range, that passenger may obscure other passengers as viewed from the camera 2. Therefore, in the state of the example shown in Fig. 5(a), the counting unit 23 does not count the number of passengers.

[0045] In the example shown in Fig. 5(b), there are no passengers included in the occlusion determination range 520 in the image 500. Therefore, when the duration of the state shown in the example of Fig. 5(b) exceeds the time threshold, the counting unit 23 counts the number of person areas of individual passengers 502 detected from the interior area 510 as the number of visitors (three in this example).

[0046] In the example shown in FIG. 5(c), when the duration of the state in which there are no passengers in the occlusion determination range is less than the time threshold, the number of entering and exiting passengers is counted based on the trajectories of the passengers' head regions crossing the entrance area 1b adjacent to the boarding / alighting door 1a, as shown in image 500. In this example, passengers 503 and 504 have their head region trajectories 503a and 504a crossing the entrance area 1b from the boarding / alighting door 1a side toward the inside of the train. Therefore, passengers 503 and 504 are counted as entering passengers (number of entering passengers: 2). Furthermore, passenger 505's head region trajectory 505a crosses the entrance area 1b from the inside of the train toward the boarding / alighting door 1a side. Therefore, passenger 505 is counted as exiting passengers (number of exiting passengers: 1). Therefore, if the previous number of staying passengers was three, the corrected number of staying passengers becomes four.

[0047] When the number of visitors notified by the counting unit 23 or the correction unit 24 exceeds the allowable upper limit, the notification processing unit 25 outputs a notification signal indicating a warning that the capacity is exceeded to the notification device 3 via the communication interface 11. Alternatively, the notification processing unit 25 may output a signal indicating that the capacity is exceeded to an electronic control unit (ECU) that controls the doors of the vehicle 1 via the communication interface 11. While receiving the signal indicating that the capacity is exceeded, the ECU controls the door of the entrance 1a to keep the door open.

[0048] In addition, the notification processing unit 25 may transmit the number of passengers when the vehicle 1 arrives at or departs from a specified point (e.g., a specified bus stop) to a device external to the vehicle 1 via a wireless communication terminal (not shown) mounted on the vehicle 1.

[0049] 6 is an operational flowchart of the counting process. The processor 13 executes the counting process in accordance with the operational flowchart shown below.

[0050] The detection unit 21 detects passengers from the image generated by the camera 2 (step S101). The tracking unit 22 tracks the detected passengers (step S102).

[0051] It is determined whether the duration during which any of the detected passengers remains outside the occlusion determination range is equal to or greater than a predetermined time threshold Th (step S103). If the duration is equal to or greater than the time threshold Th (step S103-Yes), the counting unit 23 counts the number of person regions in the vehicle interior region shown in the images obtained during that duration as the number of visitors (step S104).

[0052] On the other hand, if the duration is less than the time threshold Th (step S103-No), the correction unit 24 counts the number of passengers who cross the entrance area 1b shown in the image from the boarding / alighting door 1a side toward the inside of the train as the number of entering passengers, based on the tracking results of the head areas since the counting unit 23 last counted the number of staying passengers (step S105). Similarly, the correction unit 24 counts the number of passengers who cross the entrance area 1b shown in the image from the inside of the train toward the boarding / alighting door 1a side as the number of exiting passengers, based on the tracking results (step S106). The correction unit 24 then corrects the number of staying passengers by adding the number of passengers obtained by subtracting the number of exiting passengers from the number of entering passengers to the latest number of staying passengers (step S107).

[0053] After step S104 or S107, when the number of passengers exceeds the allowable upper limit, notification processing unit 25 notifies passengers in the vehicle of a warning that the vehicle is over capacity via notification device 3 (step S108). Then, processor 13 ends the counting process.

[0054] As described above, this counting device determines whether the position on an image of each person detected and tracked from multiple images in a time series is within an occlusion determination range in an area on the image corresponding to a predetermined area. If the duration during which the position on an image of any person is not within the occlusion determination range is equal to or greater than a predetermined time threshold, the counting device counts the number of people within the predetermined area depicted in any of the images during that duration as the number of visitors. This counts the number of visitors under the assumption that no person is obscured by another person, allowing the counting device to accurately count the number of visitors within the predetermined area.

[0055] According to a modified example, counting unit 23 may increase the hiding determination range as the number of visitors increases prior to the latest counting of the number of visitors. As the number of visitors increases, the range in which any person is hidden by another person becomes wider as viewed from camera 2. Therefore, by increasing the hiding determination range as the number of visitors increases, counting unit 23 can reduce the possibility that any person is hidden by another person when counting the number of visitors.

[0056] A computer program that causes a computer to execute the processing executed by the processor 13 of the counting device 4 according to the above embodiment or modification may be distributed in the form of being recorded on a recording medium such as an optical recording medium or a magnetic recording medium.

[0057] As described above, those skilled in the art can make various modifications to the embodiments within the scope of the present invention. [Explanation of symbols]

[0058] REFERENCE SIGNS LIST 1 vehicle, 1a boarding / exiting door, 1b entrance area, 1c vehicle interior area, 2 camera, 3 notification device, 4 counting device, 11 communication interface, 12 memory, 13 processor, 21 detection unit, 22 tracking unit, 23 counting unit, 24 correction unit, 25 notification processing unit

Claims

1. a detection unit that detects one or more people within a predetermined area from each of a plurality of time-series images generated by an imaging unit that is configured to capture the predetermined area; a tracking unit configured to track, for each of the one or more detected persons, the person in one or more images in which the person is depicted among the plurality of images; a counting unit that determines, based on a result of the tracking, whether or not the position of each of the one or more detected persons on the image is included within an occlusion determination range in an area on the image corresponding to the predetermined area, and when a duration during which the position of any of the detected one or more persons on the image is not included within the occlusion determination range is equal to or longer than a predetermined time threshold, counts the number of the one or more persons shown in any of the images during the duration as the number of people present in the predetermined area; A counting device having:

2. and a correction unit that, when the duration is less than the time threshold after counting the number of visitors, counts, based on the result of the tracking, the number of people who crossed an area adjacent to an entrance of the specified area from the entrance side to the inside of the specified area as the number of people entering, and counts the number of people who crossed the adjacent area from the inside of the specified area to the entrance side as the number of people leaving, and corrects the number of visitors by adding a number obtained by subtracting the number of people leaving from the number of people entering. The counting device according to claim 1 .

3. the detection unit detects, for each of the one or more persons, a head region representing the head of the person and a person region representing the trunk of the person from the image; the tracking unit tracks the head region and the person region for each of the one or more people; the counting unit counts the number of person regions shown in any image during the duration as the number of visitors; The counting device according to claim 2 , wherein the correction unit counts the number of entering persons and the number of exiting persons based on a result of tracking the head region.

4. detecting one or more people within a predetermined area from each of a plurality of time-series images generated by an imaging unit provided to capture the predetermined area; For each of the one or more detected persons, tracking the person in one or more images in which the person is depicted in the plurality of images; Based on the result of the tracking, it is determined whether or not the position of each of the detected one or more persons on the image is included within an occlusion determination range of an area on the image corresponding to the predetermined area; If the duration during which the position of any of the detected one or more people on the image is not included in the occlusion determination range is equal to or greater than a predetermined time threshold, the number of the one or more people shown in any of the images during the duration is counted as the number of people present in the predetermined area. A counting method including:

5. detecting one or more people within a predetermined area from each of a plurality of time-series images generated by an imaging unit provided to capture the predetermined area; For each of the one or more detected persons, tracking the person in one or more images in which the person is depicted in the plurality of images; Based on the result of the tracking, it is determined whether or not the position of each of the detected one or more persons on the image is included within an occlusion determination range of an area on the image corresponding to the predetermined area; If the duration during which the position of any of the detected one or more people on the image is not included in the occlusion determination range is equal to or greater than a predetermined time threshold, the number of the one or more people shown in any of the images during the duration is counted as the number of people present in the predetermined area. A computer program for counting that causes a computer to perform the following.

Images