WO2022019001A1 - Evaluation device, evaluation method, and program - Google Patents

Abstract

An evaluation device (1) comprises: an image readout unit (12) for reading out an image; a position detection unit (13) for periodically detecting the position coordinate of a human image in the read out image; a distance calculation unit (14) for periodically calculating, for each person, the distance between one person and another person out of multiple persons; a timer (21); determination units (15,17) for periodically determining, for each person, whether the distance between persons is 2 m or less continuously for 3 seconds or not; a totalization unit (19) for successively adding, for each person, a period of at least 3 seconds in the case where the determination by the determination units is continuously affirmed to the previous totalized time and obtaining a new totalized time; and an evaluation unit (20) for performing, for each person, an evaluation on close contact according to the length of the time period totalized by the totalization unit. Accordingly, high accuracy close contact determination is performed on a person in motion.

Description

Evaluation device, evaluation method and program

The present invention mainly relates to an evaluation technique during exercise for social distance for prevention of infection with infectious diseases.

Today, for example, in preventing droplet infection of the new coronavirus, it is said that it is effective to prevent the occurrence of the three dense, closed, close and dense. Non-Patent Document 1 describes a short-range wireless communication technology that uses Bluetooth installed in a mobile terminal, and whether or not it is in close contact within 2 m (social distance) for 15 minutes continuously for close contact. An evaluation device for determining the presence or absence has been proposed. On the other hand, in order to avoid close contact, physical activity may decrease due to self-restraint in social activities (Stayhome), resulting in a state of lack of exercise. Prolonged lack of exercise poses a risk of NCD (non-communicable disease) such as heart disease and mental health. 3 It is important to maintain sufficient physical activity both qualitatively and quantitatively while avoiding stagnation.

In addition, there are various techniques for recognizing a person in an image by deep learning and grasping the position of the person image by inputting an image (moving image) captured by a camera. A system that applies such technology as a countermeasure against the new coronavirus and issues a warning when the distance between people is within a certain range is currently being developed all over the world.

"About the new coronavirus contact confirmation app", released on June 19, 2020, Ministry of Health, Labor and Welfare New Coronavirus Infectious Disease Control Promotion Headquarters, Cabinet Office New Coronavirus Infectious Disease Control Tech Team Secretariat, [Rewa 2 July 13 Day search], Internet <https://www.mhlw.go.jp/content/10900000/000641655.pdf>

By the way, in addition to physical activities that make up for lack of daily exercise, there are also professional sports, school sports, club activities, private / public sports, and exercises for children / elderly people, both inside and outside the building. 3 It is desirable to avoid a dense environment. On the other hand, since it is not possible to carry a mobile terminal while exercising, it is difficult to apply a social distance determination technique using a mobile terminal such as the conventional technique. In addition, the system that applies the technology to grasp the position of the human image captured by the camera to the new coronavirus countermeasures currently only issues a warning when it is located within a certain distance, and the time condition is also for people in sports. No consideration is given to the characteristics specific to the movement of.

The present invention has been made in view of the above, and is an evaluation that enables highly accurate dense contact determination or evaluation by adopting a method of applying a time condition in consideration of the peculiarity of being under the movement of a person during exercise. It provides equipment, evaluation methods and programs.

The evaluation device and method according to the present invention include an image reading means, a position detecting means, a distance calculating means, a time measuring means, a determination means, an aggregation means, and an evaluation means. The image reading means reads out the captured images of a plurality of people in motion captured by the imaging means. The position detecting means periodically detects the position coordinates of each of the plurality of people from the captured image read out. The distance calculation means periodically calculates the distance between one person and another person among the plurality of people for each person. The timekeeping means measures the passage of time. The determination means periodically determines for each person whether or not the distance between the persons is within the close range for the reference time. The aggregation means sequentially aggregates at least the reference time when the determination by the determination means is continuously affirmed for each person into the aggregation time up to the previous time, and sets it as a new aggregation time. Then, the evaluation means evaluates the close contact for each person according to the magnitude of the time totaled by the totaling means.

Further, the present invention is a program for making a computer function as the evaluation device.

According to the present invention, the position coordinates are periodically detected for the captured human image, and the distance between the people is calculated for each person. Then, it is periodically determined for each person whether or not the distance between the people is within the close distance continuously for the reference time. When the determination by the determination means is continuously affirmed, the reference time or the duration exceeding the reference time is sequentially aggregated for each person by the aggregation means. Then, the evaluation means evaluates the close contact according to the magnitude of the time totaled for each person by the totaling means.

In sports, people often approach or separate from each other for a very short time during play (exercise). Considering the droplet infection of the new coronavirus, it is important whether or not people were breathing when they were approaching each other, and it is not possible to evaluate the risk of close contact and thus infection simply by distance. On average, a person's breathing rate is said to be more than 10 to 20 times per minute. Focusing on the one with the highest number of times, it means that one breath is performed in 3 seconds. Therefore, it is preferable to use whether or not the proximity state has continued for 3 seconds as a criterion for determining whether or not the person has breathed at the time of proximity. Therefore, in the present invention, the time required for one breath (3 seconds) is set as the reference time, and when the time is close to the reference time exceeding the reference time, the time is totaled as evaluation information. The reference time is not limited to the breathing time, and may be shorter as a probability, or may be longer, for example, two breaths (6 seconds). In addition, the exercise environment (temperature, sunshine, indoors and outdoors) is measured with a sensor, and the physical condition and movement state (body temperature, heart rate, acceleration) of a person are measured with a wearable terminal, and the result is used as a parameter for changes in the respiratory cycle. It can be reflected in the time adjustment. Further, an appropriate reference time may be set depending on the type of exercise. In addition to the reference time, these can also be applied to the adjustment of the close distance. Since the accumulation time of each person is acquired with each other person, risk information for each other person can be obtained.

According to the present invention, it is possible to make a highly accurate rich contact determination for a person who is exercising.

It is a block diagram which shows one Embodiment of the evaluation apparatus which concerns on this invention. It is a figure explaining the position of a person in a captured image and the distance between people, (A) is a diagram showing an imaging environment from the side, (B) is a diagram of an image captured by a camera, and (C) is a diagram showing a camera. It is a figure explaining that the 2D plane imaged in 1 was converted into a horizontal plane, and a person was re-plotted in it. (A) is a diagram showing an example of a skeleton image created from a captured human image, and (B) is a diagram showing that the height of the horizontal plane on which a person is plotted is set to the height of a person's waist. C) is a diagram for explaining the outline of the homography transformation. It is a flowchart which shows an example of the procedure of the monitoring process which a control part executes. It is a flowchart which shows an example of the procedure of the evaluation process which a control part executes.

FIG. 1 is a block diagram showing an embodiment of the evaluation device according to the present invention. The evaluation device 1 includes a control unit 10 including a computer, a storage unit 101, and a touch panel 102. The storage unit 101 includes a control program for executing the evaluation process according to the present invention, a machine learning model (program and learned parameters) for operating AI described later, and a processing program for converting the imaging surface of the camera 30 into a horizontal plane. I remember. Further, as will be described later, the storage unit 101 stores various processing programs such as detecting a human body from a human image, labeling, detecting a position, and creating a skeleton image from a human image. .. In the touch panel 102, for example, a transparent pressure-sensitive sheet is laminated on a liquid crystal display panel (display unit 1021), and the display coordinates of buttons and the like are associated with the coordinates on the pressure-sensitive sheet (input unit 1022) in advance. It enables selection of buttons and the like at the pressing position. Further, the input unit 1022 is provided with a keyboard and a mouse (not shown) so that necessary information can be input.

At least one camera 30 and, if necessary, a sensor unit 31 are connected to the control unit 10. The control unit 10 is configured to be able to receive the image captured by the camera 30 by wire or wirelessly. The sensor unit 31 obtains information on the exercise environment, for example, receives information from a thermometer by wire or wirelessly. In addition, the sensor unit 31 receives sensing information from a wearable small thermometer, heart rate monitor, accelerometer, etc. that can be carried around each person, for example, by wire or wirelessly. In a mode other than real-time evaluation, it may be temporarily stored in a storage unit in the sensor so that it can be read out later.

The camera 30 is a digital video camera capable of acquiring time-series images (moving images). The camera 30 may be a camera mounted on a smartphone or a personal computer in addition to a normal digital camera. Further, the image output from the camera 30 to the control unit 10 may be a recorded image temporarily stored in the internal storage unit (not shown) of the camera 30 in addition to the real-time data. The image may be data stored in a server on the network. In the present embodiment, the camera 30 is supported by a support tool at an appropriate position on the side wall of the gymnasium, for example, in a posture of maintaining a predetermined depression / elevation angle so as to overlook the entire exercise area. In the aspect of adopting a plurality of cameras, images are taken from different directions, for example, from the front-back direction and the left-right direction, or from the lateral direction and the ceiling, and both images are combined to identify a person in the image. Position detection processing becomes possible with high accuracy.

By executing each program of the storage unit 101, the control unit 10 executes the initial setting unit 11, the captured image reading unit 12, the position detection unit 13, the distance calculation unit 14, the proximity distance determination unit 15, the time integration unit 16, and the proximity. It functions as a time determination unit 17, an alarm processing unit 18, a proximity time totaling unit 19, an evaluation unit 20, a timer 21, an image display processing unit 22, and a search unit 23. In the present embodiment, the position detection unit 13 includes a human body detection unit 131, a labeling processing unit 132, a skeleton image creation unit 133, and a coordinate conversion unit 134.

The initial setting unit 11 sets various initial conditions according to the situation of the observation target. As the setting items, for example, the viewpoint and line of sight of the camera 30, a sign, and a close contact condition are assumed.

The captured image reading unit 12 reads the captured image from the camera 30 to the position detection unit 13 in real time, or reads the recorded image stored in the memory. Specifically, the captured image (moving image) to be read out is output to an running control program or the like and used for image processing. The read image may be output (displayed) to the display unit 1021 as needed.

Subsequently, the processing of the position detection unit 13 will be described with reference to FIGS. 2 and 3. 2A and 2B are views for explaining the position of a person and the distance between people in a captured image, FIG. 2A is a diagram showing an imaging environment from the side, and FIG. 2B is a diagram of an image captured by a camera. C) is a diagram for explaining that a two-dimensional surface imaged by a camera is converted into a horizontal plane and a person is plotted in the horizontal plane. In FIG. 3, (A) is a diagram showing an example of a skeletal image created from a captured human image, and (B) shows that the height of the horizontal plane on which a person is plotted is set to the height of a person's waist. The figure shown, (C) is a diagram for explaining the outline of the homography conversion.

In FIG. 2, it is assumed that five players A to E are exercising in the imaging space in the gymnasium, for example. Actually, each player A to E is exercising, but in FIG. 2, for convenience of drawing, it is shown by a similar figure of the standing posture. The human body detection unit 131 detects the players A to E, and in the present embodiment, the human body detection unit 131 acquires time-series two-dimensional position information at a plurality of specific parts of the moving human body from the image captured from the camera 30. .. The human body detection unit 131 repeatedly detects the position information of each joint portion from the acquired two-dimensional image of the human body at a predetermined cycle, for example, 20 Hz. Various methods for detecting the two-dimensional position information associated with the joint site have been proposed, and for example, machine learning, deep learning, image recognition processing, or the like may be used. The human body detection unit 131 may detect the human body by performing tracking processing at the predetermined cycle once the human body is detected.

The labeling processing unit 132 labels the human image detected by the human body detection unit 131 like players A, B, C, D, and E. Labeling may be a personal name or a pseudonym (for example, A, B ...). In addition, in the case of personal identification, it is possible by, for example, face recognition, uniform number recognition, wearing different color wear, and identifying those colors.

The skeleton image creation unit 133 is based on the labeled moving images of the players A, B, C, D, and E, and here, as shown in FIG. 3A, the feature point (joint position: node) and the feature direction (skeleton). A known skeleton detection algorithm (OpenPose) for extracting Sk: skeleton) is exemplified.

For the position information in the human body detection, it is necessary to replace the two-dimensional position information of the players A to E on the image pickup screen by the camera 30 with the position information corresponding to the floor surface. The positions Pa to Pe of the players A to E indicate the absolute or relative positions with respect to the floor surface FL in the gymnasium, which is an exercise space. Since various methods are known for detecting the position of the player, a brief description thereof will be given. In the competition space captured by the camera 30, the displayed image is distorted into a trapezoidal shape along the depth direction by the perspective method. Therefore, the known line mark 40 or the separately set known marks 411 and 412 are imaged, and coordinate conversion is performed so that the image pickup surface is projected onto another image pickup surface as described later. In addition to the method of attaching the mark to the floor surface, scale bars of a predetermined height or the like are erected at at least four places, and the coordinate conversion of the image pickup surface is performed from the length of each scale bar imaged by the camera 30. The method of doing it can also be adopted. In a mode in which a plurality of cameras 30, for example, two cameras are used, a method of identifying each player by referring to two two-dimensional images and obtaining position coordinates may be used.

As shown in FIG. 3C, the coordinate conversion unit 134 converts the coordinates of the trapezoidal image pickup plane PL1 viewed from the camera 30 at a predetermined depression / elevation angle θ into a rectangular bird's-eye view plane PL2 viewed from directly above. As the coordinate transformation in this case, a known bird's-eye view transformation or transformation using a homography matrix is applied by utilizing the mark information. This makes it possible to convert each coordinate position on the plane PL1 into a coordinate position on the plane PL2.

As shown in FIG. 3B, the coordinate conversion unit 134 is based on the assumption that the height of the waist does not change significantly even when the player performs various movements, and the main part of the skeleton image Hu, for example, The height h of the waist portion Wa is converted into a bird's-eye view when cut by a horizontal surface PL. The coordinate conversion unit 134 uses the coordinates in the camera image of the waist portion Wa detected by the skeleton detection algorithm (OpenPose) as the coordinates when converted into a bird's-eye view, and uses these coordinates as the position information of the player.

The distance calculation unit 14 calculates the distance between the players for all the players A to E based on the converted bird's-eye view. In this case, the lengths of the line mark 40 or the separately set marks 411 and 412 are known, and the distance between the players is calculated based on the known lengths. FIG. 2C describes a case where the distance between the player B and another player on the horizontal plane PL is calculated. In this figure, the distance between the player BAs is the distance Dba, the distance between the player BCs is the distance Dbc, the distance between the player BDs is the distance Dbd, and the distance between the player BEs is the distance Dbe. The distance between the players may be determined by the number of combinations. For example, if the distance Dba between the player BAs is measured, the distance Dab between the players AB does not need to be measured.

Further, the distance calculation unit 14 may calculate the distance in the three-dimensional space in consideration of the height, instead of the method of performing the distance between the players after the conversion to the bird's-eye view surface. In this way, the distance can be measured in three dimensions. For example, the distance between the heads may be the distance between people.

The proximity distance determination unit 15 is one of the conditions for close contact, and determines whether or not the distance between the players is within a predetermined proximity distance. The close range is set to 2 m here from the viewpoint of preventing droplet infection. It should be noted that it may be farther outdoors, for example, 3 m.

The time integration unit 16 integrates the duration for each other player while the proximity distance determination unit 15 determines that the distance between the players is within the proximity distance. In this example, the time integration unit 16 repeatedly integrates the time having a cycle corresponding to 20 Hz.

The proximity time determination unit 17 is one of the conditions for close contact, and determines whether or not each player has continued for a predetermined time (proximity time or longer duration) within the proximity distance. The proximity time is set to, for example, about 3 seconds in consideration of the respiratory cycle.

The alarm processing unit 18 is provided as needed, and issues an alarm when the determination results of the proximity distance determination unit 15 and the proximity time determination unit 17 are both affirmative. Alarms include images, sounds, and sounds.

When the determination results of the proximity distance determination unit 15 and the proximity time determination unit 17 are both affirmative, the proximity time aggregation unit 19 adds the duration of this time to the aggregation time up to the previous time and stores it as a new aggregation time. And make it a cumulative target.

The evaluation unit 20 determines for each player whether or not the total time exceeds the time corresponding to the close contact, and if it exceeds, evaluates as a close contact. The evaluation unit 20 can also give points such as a rich contact risk point according to the magnitude of the totaled time during the series of exercise time and the practice time monitored.

The timer 21 is a timekeeping means composed of software for measuring the passage of time. The image display processing unit 22 causes the display unit 1021 to display the state in which the evaluation process is being executed or the result at the end in a predetermined format. For example, the total time for each player and the total time for each player are displayed. In these displays, it is preferable to display the cumulative status of the total time over time in two dimensions by plotting time on the horizontal axis and aggregation time on the vertical axis.

When it is later found that any of the players A to E is infected with the infectious disease, the search unit 23 activates the search program to evaluate the evaluation result of the player who exercised with the infected person, particularly. Check the total time. The search unit 23 searches (extracts) the player who has the longest close gathering time with the player confirmed to be infected, and the player whose time corresponds to or longer than the close contact.

FIG. 4 is a flowchart showing an example of the procedure of the monitoring process executed by the control unit 10. First, when the monitoring process is executed, the initial setting process for setting various initial conditions including the input setting of the shape information of the mark which is the reference of the position detection is performed (step S1). Next, when monitoring is started, it is determined whether or not the predetermined time (the cycle of 20 Hz described above) timed by the timer 21 has elapsed (step S3). When the predetermined time has elapsed, the two-dimensional image captured by the camera 30 including the image of the moving human body to be measured is captured (step S5). Next, a human body image is detected from the acquired two-dimensional image of the human body, and labeling processing and skeleton image creation processing are performed on the image (step S7).

Next, the position of each player image in the imaging space is calculated (step S9). Further, coordinate conversion to the bird's-eye view plane is performed (step S11). After that, the evaluation process (FIG. 5) is executed every predetermined time elapse (Yes in step S3) (step S13). Then, until the monitoring process is completed (step S15), the process returns to step S3 and the same process is repeated.

FIG. 5 is a flowchart showing an example of the procedure of the evaluation process executed by the control unit 10. The evaluation process will be described with reference to the example of FIG. First, the control unit 10 sequentially selects each pair that is a combination of the players A to E, and calculates the distance between the selected pair of people (step S21). Then, it is determined whether or not the calculated distance is within the close range (step S23), and if it is within the close range, a predetermined time (time corresponding to 20 Hz in this example) is integrated with the integrated time of the pair (step). S25). Next, it is determined whether or not the accumulation time of the pair exceeds the reference time (typically 3 seconds) (step S27). If the reference time is exceeded, an alarm is instructed as necessary (step S29), and then it is determined whether or not there is a remaining pair (step S31). If there is a remaining pair, step S21 is performed. Go back and do the same for the next pair. If there are no remaining pairs and the processing for all the pairs is completed, the process proceeds to step S15.

If it is determined in step S23 that the distance is not within the close range, it is determined whether or not the integrated time of the pair is equal to or longer than the reference time (step S33). If it is equal to or longer than the reference time, the totalization time of this time is added to each totalization time of the pair, and the totalization time is reset (step S35). The time added this time is temporarily stored as a new aggregation time of the pair, and is subject to addition in a later evaluation (step S23, step S35, and then step S35).

On the other hand, if the integrated time of the pair does not exceed the reference time in step S33, it is considered that there is no close contact, and the integrated time of the pair is reset without totaling (step S37).

In this way, by acquiring the reference time or duration between each pair, it is possible to make adjustments that further reduce the reference time or duration by changing the exercise or practice method even if it does not lead to close contact, and it is more secure. It will be possible to propose and adopt the exercise and practice method that you have done.

Further, in the present embodiment, it is possible to total the reference time to the total time between the players, evaluate the exercise and practice according to the magnitude of the time, and change the exercise method as appropriate. In this case, the integrated time to be aggregated may be aggregated so as to include the reference time itself at the time when the reference time is reached or the duration exceeding the reference time.

The present invention is also applicable to diseases that are more infectious by close contact, including droplet infection.

As described above, it is preferable that the evaluation device and method according to the present invention include image reading means, position detection means, distance calculation means, timekeeping means, determination means, tabulation means, and evaluation means. The image reading means reads out the captured images of a plurality of people in motion captured by the imaging means. The position detecting means periodically detects the position coordinates of each of the plurality of people from the captured image read out. The distance calculation means periodically calculates the distance between one person and another person among the plurality of people for each person. The timekeeping means measures the passage of time. The determination means periodically determines for each person whether or not the distance between the persons is within the close range for the reference time. The aggregation means sequentially aggregates at least the reference time when the determination by the determination means is continuously affirmed for each person into the aggregation time up to the previous time, and sets it as a new aggregation time. Then, the evaluation means evaluates the close contact for each person according to the magnitude of the time totaled by the totaling means.

Further, it is preferable that the present invention is a program for making a computer function as the evaluation device.

In sports, people often approach or separate from each other for a very short time during play (exercise). Considering the droplet infection of the new coronavirus, it is important whether or not people were breathing when they were approaching each other, and it is not possible to evaluate the risk of close contact and thus infection simply by distance. On average, a person's breathing rate is said to be more than 10 to 20 times per minute. Focusing on the one with the highest number of times, it means that one breath is performed in 3 seconds. Therefore, it is preferable to use whether or not the proximity state has continued for 3 seconds as a criterion for determining whether or not the person has breathed at the time of proximity. Therefore, in the present invention, the time required for one breath (3 seconds) is set as the reference time, and when the time is close to the reference time exceeding the reference time, the time is totaled as evaluation information. The reference time is not limited to the breathing time, and may be shorter as a probability, or may be longer, for example, two breaths (6 seconds). In addition, the exercise environment (temperature, sunshine, indoors and outdoors) is measured with a sensor, and the physical condition and movement state (body temperature, heart rate, acceleration) of a person are measured with a wearable terminal, and the result is used as a parameter for changes in the respiratory cycle. It can be reflected in the time adjustment. Further, an appropriate reference time may be set depending on the type of exercise. In addition to the reference time, these can also be applied to the adjustment of the close distance. Since the accumulation time of each person is acquired with each other person, risk information for each other person can be obtained.

Further, it is preferable that the number of the imaging means is at least one. According to this configuration, in the case of one unit, a method of constructing a two-dimensional coordinate system on the captured image surface using signs (mark size information and height information) so that the coordinate position of the human image can be calculated can be applied. It becomes. On the other hand, if there are two or more units, it is possible to identify the captured human image by collating a plurality of captured images.

Further, it is preferable that the position detection means and the distance calculation means execute position detection and distance calculation in a short cycle shorter than the reference time, and the determination means performs a determination operation in each short cycle. According to this configuration, by repeating the position detection and the distance calculation in a short cycle corresponding to, for example, 20 Hz, it is possible to accurately monitor whether the determination condition is continuously affirmed within the reference time such as 3 seconds. The short cycle is not limited to the time of 20 Hz, and may be shorter, or at least shorter than the reference time.

Further, in the determination means, the first determination means for determining whether or not the distance between the persons is within the close distance and the affirmative determination by the first determination means continue for the reference time. It is preferable to include a second determination unit for determining whether or not. According to this configuration, determination is made for each determination condition of the proximity distance and the reference time.

Further, it is preferable that the aggregation means sequentially aggregates the reference time when the determination by the determination means is affirmed. According to this configuration, when it is determined that the proximity is so-called, the reference time is totaled. Therefore, in this embodiment, when the integration process continues longer than the reference time, when the integration process reaches the reference time, the aggregation for the reference time is once completed, and then the integration time is reset and the integration time is again calculated. It is possible to repeat the continuous integration process. In this case, the aggregation process for each reference time is possible.

Further, the present invention includes a sensor that measures at least one of the physical condition and the exercise environment of an exercising person, and the determination means adjusts at least one of the reference time and the proximity distance using the measurement result of the sensor as a parameter. Those are preferable. According to this configuration, since the physical condition and the state of the exercise environment affect the respiratory rate, it is possible to evaluate with high accuracy by adjusting the evaluation conditions using the measurement result as a parameter.

Further, the present invention preferably includes a sensor for measuring at least one of the physical condition and the exercise environment of an exercising person, and the evaluation means preferably adjusts the evaluation of the dense contact using the measurement result of the sensor as a parameter. According to this configuration, the parameters can be applied to the evaluation as well as the determination.

Further, the tabulation means further totals the totaling time of each person, and it is preferable that the evaluation means evaluates the close contact with the whole exercise. According to this configuration, it is possible to present a close contact opportunity as a whole separately from each person for a series of exercises, and it is possible to adjust a safer exercise method and practice method.

Further, in the present invention, the total time between the input unit for inputting the information of the infected person infected with the infectious disease and the input infected person among the plurality of persons exceeds the rich contact standard. Those provided with a search means for searching for a person are preferable. According to this configuration, when an infected person is found, it is possible to quickly search for a close contact person who exceeds a standard (for example, 15 minutes).

1 Evaluation device 10 Control unit 101 Storage unit 102 Touch panel (input unit)
12 Captured image reading unit (image reading means)
13 Position detection unit (position detection means)
131 Human body detection unit 132 Labeling processing unit 133 Skeleton image creation unit 134 Coordinate conversion unit 14 Distance calculation unit (distance detection means)
15 Proximity distance determination unit (determination means, first determination means)
16 Hours integration unit 17 Proximity time determination unit (determination means, second determination means)
18 Alarm processing unit 19 Proximity time totaling unit (aggregation means)
20 Evaluation Department (Evaluation Means)
21 Timer (timekeeping means)
22 Image display processing unit 23 Search unit (search means)
30 Camera 31 Sensor unit (sensor)

Claims (11)

An image reading means for reading out images captured by a plurality of people in motion, which are captured by the imaging means, and
A position detection means for periodically detecting the position coordinates of each person among the plurality of people from the captured image read out.
A distance calculation means for periodically calculating the distance between one person and another person among the plurality of people, and a distance calculation means for periodically calculating the distance between each person.
Timekeeping means to measure the passage of time and
A determination means for periodically determining whether or not the distance between the persons is within the close range for a continuous reference time, and
At least when the determination by the determination means is continuously affirmed, at least the reference time is sequentially aggregated for each person in the aggregation time up to the previous time, and the aggregation means is used as a new aggregation time.
An evaluation device provided with an evaluation means for evaluating each person in close contact according to the amount of time totaled by the totaling means. The evaluation device according to claim 1, wherein the image pickup means is at least one or more. The claim means that the position detection means and the distance calculation means execute position detection and distance calculation in a short cycle shorter than the reference time, and the determination means performs a determination operation in each short cycle. Item 2. The evaluation device according to Item 1 or 2. The determination means is a first determination means for determining whether or not the distance between the persons is within the close distance, and whether or not the affirmative determination by the first determination means continues for the reference time. The evaluation device according to any one of claims 1 to 3, further comprising a second determination unit for determining whether or not. The evaluation device according to any one of claims 1 to 4, wherein the aggregation means sequentially aggregates the reference time when the determination by the determination means is affirmed. Equipped with sensors that measure at least one of the physical condition and exercise environment of the person exercising,
The evaluation device according to any one of claims 1 to 5, wherein the determination means adjusts at least one of the reference time and the proximity distance using the measurement result of the sensor as a parameter. Equipped with sensors that measure at least one of the physical condition and exercise environment of the person exercising,
The evaluation device according to any one of claims 1 to 5, wherein the evaluation means adjusts the evaluation of the dense contact using the measurement result of the sensor as a parameter. The totaling means further totals the totaling time of each person.
The evaluation device according to any one of claims 1 to 7, wherein the evaluation means evaluates the dense contact with respect to the whole movement. An input unit for inputting information on an infected person infected with an infectious disease among the plurality of people.
The evaluation device according to any one of claims 1 to 8, further comprising a search means for searching for a person whose total time with respect to the input infected person exceeds the dense contact criterion. The image reading means reads out the captured images of a plurality of people in motion, which are captured by the imaging means.
The position detecting means periodically detects the position coordinates of each of the plurality of people from the captured image read out.
The distance calculation means periodically calculates the distance between one person and another person among the plurality of people for each person.
Timekeeping means measure the passage of time and
The determination means periodically determines for each person whether or not the distance between the persons is within the close range for the reference time.
At least the reference time when the determination by the determination means is continuously affirmed by the aggregation means is sequentially aggregated for each person in the aggregation time up to the previous time, and is set as a new aggregation time.
An evaluation method in which the evaluation means evaluates each person for close contact according to the magnitude of the time totaled by the totaling means. A program for making a computer function as the evaluation device according to any one of claims 1 to 9.

Images