Intelligent stadium health data acquisition equipment

Abstract

The invention discloses intelligent stadium health data acquisition equipment. The method is used for judging positions of gathered crowds. And finding the gathering temperature which is increased due to the increase of carbon dioxide in the area where the gathered crowd is located. And the physical sign value of the human body at the next time point is found through the motion state of the human body at multiple time points, the change of the surrounding gathering temperature and the change of the gathering position. Diseases existing in the human body at different positions are found, and the influence degree of the gathering temperature and the gathering position of the human body on the diseases and the influence degree of the exercise amount generated by the exercise at multiple time points on the diseases are judged. The aggregation temperature, the motion state of the human body and the aggregation state of the human body influence one another and jointly influence diseases of the human body.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an intelligent sports stadium health data collection device. Background Art

[0002] At present, wearable devices are often used to collect health data of the human body. However, since only wearable devices can detect the current human body's vital signs data, it is impossible to predict the human body's vital signs data at the next time point, making it impossible to avoid physical discomfort in advance. In addition, because there will be crowds in stadiums, the gathered crowds or exercising people will cause the temperature in the area where the crowds are located to rise, and will cause the human body to have chest tightness and shortness of breath. If the human body suffers from diseases, such as asthma, it will cause the disease to relapse. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent sports stadium health data collection device to solve the above-mentioned problems existing in the prior art.

[0004] In a first aspect, an embodiment of the present invention provides an intelligent sports stadium health data collection device, including a sports stadium monitoring device, a human body monitoring device, a camera device and a processor: The stadium monitoring device is used to obtain the stadium temperature at multiple time points; the stadium temperature represents the temperature inside the stadium; The camera device is used to capture crowd images at multiple time points; the crowd images represent images containing multiple gathered human bodies; The human body monitoring device is used to obtain n user numbers, human body information corresponding to the user numbers, human body symptoms corresponding to the user numbers and human body positions at multiple time points corresponding to the user numbers; the human body positions represent the positions of the human body in the stadium; n is a positive integer; The processor execution method comprises: Obtaining a clustered human body labeled image based on the crowd image through a clustered position network; the clustered human body labeled image represents an image that marks the positions and boundaries of the clustered human bodies; Matching the position of the human body in the aggregated human body mark image with the position of the human body corresponding to the user number, and associating the aggregated human body mark image with the human body information and human body disease corresponding to the user number; Based on the aggregated human body marker images and human body diseases, a disease feature matrix is ​​obtained; the disease feature matrix is ​​a three-dimensional matrix; Based on the stadium temperature, the disease feature matrix, and the aggregated human body marker images at multiple time points, health data is obtained; the health data includes first predicted human body change data, disease impact value, and aggregated temperature.

[0005] Optionally, the health data is obtained based on the stadium temperature, the disease feature matrix, and the aggregated human body marker images at multiple time points, including: Based on the gathered human body mark image and the stadium temperature, a gathered temperature feature and a gathered temperature are obtained; the gathered temperature represents the temperature of an area affected by the exhaled carbon dioxide of the human body in a gathered state; Multiple aggregation temperature characteristics and aggregation temperatures are obtained corresponding to multiple time points; Based on the human body information and the aggregated temperature characteristics corresponding to the aggregated human body mark images at multiple time points, the motion state of the human body is judged to obtain the first predicted human body change data; the first predicted human body change data represents the vital sign data of the human body at the next time point; Based on the disease feature matrix, the gathering temperature and the first predicted human body change data, a disease impact value is obtained through a disease discrimination network; the disease impact value represents the impact of human body movement and gathering in a sports stadium on the disease.

[0006] Optionally, obtaining a disease feature matrix based on the aggregated human body marker images and human diseases includes: Based on a plurality of human diseases, removing the features of the arrangement order of the human diseases through a first disease neural network, to obtain disease features; the disease features represent features that only include the types of diseases existing in the human body; Multiple human diseases correspond to multiple disease characteristics; Acquire a three-dimensional matrix; the number of rows of the three-dimensional matrix is ​​equal to the width of the aggregated human body marker image; the number of columns of the three-dimensional matrix is ​​equal to the length of the aggregated human body marker image; the number of pages of the three-dimensional matrix is ​​equal to the length of the disease feature; According to the positions corresponding to the gathered human body marker images, the disease characteristics are sequentially filled into the corresponding rows and columns of the three-dimensional matrix to obtain a disease characteristic matrix.

[0007] Optionally, the method of removing the features of the arrangement order of human diseases through a first disease neural network based on multiple human diseases to obtain disease features includes: Obtaining a random number; the random number represents the number of times a plurality of human diseases are arranged; Randomly arranging the plurality of human disease symptoms multiple times according to a random number of times to obtain a plurality of human disease symptom sequences; Inputting the human disease sequence into a first disease neural network to obtain a first disease feature; obtaining a plurality of first disease features corresponding to a plurality of human disease sequences; Calculate the variance of the values ​​of the subscripts corresponding to the plurality of first disease features to obtain a feature change set; the values ​​in the feature change set represent changes in the features when the order is changed; the number of elements in the feature change set is equal to the length of the disease feature; If the value in the feature change set is less than the change threshold, the average value of the subscripts corresponding to the multiple first disease features is calculated; if the value in the feature change set is greater than or equal to the change threshold, the value of the subscript corresponding to the first disease feature is set to 0 to obtain the disease feature.

[0008] Optionally, obtaining a clustered human body marked image based on the crowd image by clustering the location network includes: Based on the crowd image, mark the position of the human body and the frame containing the human body to obtain a second human body marked image; the second human body marked image includes multiple first detection positions and a second human body frame; Obtaining a clustering radius; the clustering radius represents a radius of a clustered human body that will cause a change in health data; A circle is drawn with the first detection position as the center and the aggregation radius as the radius to obtain an aggregation circle; multiple aggregation circles are obtained corresponding to multiple first detection positions; Based on k clustering circles, an n*uth second clustering distance is obtained; one first detection position corresponds to u second clustering distances; the second clustering distance represents the distance of the human body clustered around the first detection position; If the number of the second clustering distances corresponding to the first detection position is less than the clustering number threshold, the first detection position is deleted from the second human body mark image to obtain a clustered human body mark image.

[0009] Optionally, obtaining the n*uth second clustering distance based on the k clustering circles includes: The number of first detection positions contained in the gathering circle is used as the number of people gathered; If the number of people gathered is greater than the threshold number of people gathered, the gathering circle corresponding to the number of people gathered is used as the first gathering circle; If the center of a first gathering circle is contained in another first gathering circle, the first gathering circle with a smaller number of people is deleted; Clustering the multiple first detection positions in the first clustering circle, determining the cluster center, and obtaining a cluster position set; Calculating the distances between the plurality of first detection positions in the cluster position set to obtain a plurality of first cluster distances; The first clustering distance that is smaller than the clustering position threshold is used as the second clustering distance.

[0010] Optionally, the step of determining the motion state of a human body based on the human body information and the aggregated temperature characteristics corresponding to the aggregated human body marker images at multiple time points to obtain the first predicted human body change data includes: Inputting the human body information corresponding to the human body marker images gathered at multiple time points into the first time convolution network in sequence to obtain a first human body change feature; the first human body change feature represents the change feature of the vital signs of the gathered human body during the movement at multiple time points; Inputting the aggregated temperature features of multiple time points into a second time convolutional network to obtain a second human body change feature; the second human body change feature represents the change of the ambient temperature of the human body during the movement of the multiple time points; The first human body change feature and the second human body change feature are input into a human body change prediction neural network to determine the changes in the aggregation and physical signs of the human body in motion, and obtain first predicted human body change data.

[0011] Optionally, obtaining the aggregation temperature feature and the aggregation temperature based on the aggregated human body marker image and the stadium temperature includes: Extract the area corresponding to a clustering circle in the clustered human body mark image to obtain a clustering circle image; obtain multiple clustering circle images corresponding to multiple clustering circles in the clustered human body mark image; The regions within the second human body borders in the clustered circle image are set to black, and the regions outside the second human body borders are set to white, to obtain a binary clustered circle image; The binary cluster circle image and the stadium temperature are input into the cluster space detection network to obtain the cluster temperature feature and the cluster temperature; the cluster temperature feature represents the feature before the cluster temperature is classified.

[0012] Optionally, the step of marking the position of a human body and a frame containing the human body based on the crowd image to obtain a second human body marked image includes: The clustered location network includes a first target detection network and a second target detection network; Inputting the crowd image into a first target detection network to detect the position of a human body to obtain a first detection position; Acquire a binary image; the size of the binary image is the same as the size of the crowd image; Marking the location of the first detection position in the binary image to obtain a first human body marked image; Inputting the first human body marked image into a super-resolution network to perform image enhancement to obtain a first enhanced image; Inputting the first enhanced image into a second object detection network, detecting the position of a human body boundary, and obtaining a second human body border; The second human body frame is marked in the first human body marked image to obtain a second human body marked image.

[0013] Optionally, obtaining the disease impact value through a disease discrimination network based on the disease feature matrix, the aggregate temperature and the first predicted human body change data includes: Inputting the first predicted human body change data into a first prediction neural network to obtain a first disease impact feature; the first disease impact feature indicates the probability of a disease recurrence caused by a physical sign of the human body at the next time point; Inputting the human body information and the aggregate temperature into a second prediction neural network to obtain a second disease impact feature; the second disease impact feature indicates the probability of disease recurrence caused by the human body state and the external temperature; The second symptom impact feature and the second symptom impact feature are input into the symptom discrimination network to obtain the symptom impact value.

[0014] Compared with the prior art, the embodiments of the present invention achieve the following beneficial effects: The embodiment of the present invention also provides an intelligent sports stadium health data collection device: In the present invention, the position of the human body in the stadium is judged, so as to judge the position of the gathered crowd. The gathering temperature of the area where the gathered crowd is located, which is increased due to the increase of carbon dioxide, is found. And the value of the human body's vital signs at the next time point is found through the movement state of the human body at multiple time points, the change of the corresponding surrounding gathering temperature and the change of the gathering position. It is also found by finding the symptoms of the human body at different positions, and judging the influence of the gathering temperature and the gathering position of the human body on the symptoms and the influence of the amount of exercise generated by the movement at multiple time points on the symptoms. The gathering temperature, the movement state of the human body, and the gathering state of the human body affect each other and jointly affect the symptoms of the human body. The technical effect of being able to obtain the value of the human body's vital signs in the stadium at the next time point in advance, being able to cool down according to the gathering temperature and judging the symptoms caused by gathering and movement. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of a method executed by a processor in an intelligent sports stadium health data collection device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The present invention will be described in detail below in conjunction with the accompanying drawings. Example

[0017] like Figure 1 As shown, an embodiment of the present invention provides an intelligent sports stadium health data collection device, including a sports stadium monitoring device, a human body monitoring device, a camera device and a processor: The stadium monitoring device is used to obtain the stadium temperature at multiple time points; the stadium temperature represents the temperature inside the stadium; The camera device is used to capture crowd images at multiple time points; the crowd images represent images containing multiple gathered human bodies.

[0018] The camera device is placed obliquely above the sports stadium so that the top of the human head and a part of the torso can be photographed.

[0019] In this embodiment, the size of the crowd image is 1024*1024.

[0020] The human body monitoring device is used to obtain n user numbers, human body information corresponding to the user numbers, human body symptoms corresponding to the user numbers and human body positions at multiple time points corresponding to the user numbers; the human body positions represent the positions of the human bodies in the sports stadium.

[0021] The human body information includes human body invariant information and human body variable information. In this embodiment, the human body invariant information includes age, height and weight, and the human body variable information represents heart rate, blood pressure and body temperature.

[0022] Wherein, one user number corresponding to the human body information corresponds to m human body diseases, and n user numbers correspond to n*m human body diseases. n and m are natural integers greater than 0.

[0023] The processor execution method comprises: S101: obtaining a clustered human body labeled image based on the crowd image through a clustered position network; the clustered human body labeled image represents an image that marks the positions and boundaries of the clustered human bodies.

[0024] S102: Matching the position of the human body in the aggregated human body mark image with the position of the human body corresponding to the user number, and associating the aggregated human body mark image with the human body information and human body disease corresponding to the user number.

[0025] The position of the human body in the human body marker image is used, that is, the first detected position of the marker, and the distance is calculated with multiple human body positions. The human body position within the boundary of the human body corresponding to the first detected position, that is, within the second human body frame and with the smallest distance is taken as the associated human body position. The first detected position is associated with the associated human body position.

[0026] Among them, by S103: Based on the clustered human body marked images and human body diseases, a disease feature matrix is ​​obtained; the disease feature matrix is ​​a three-dimensional matrix.

[0027] The human diseases corresponding to the multiple user numbers are mapped to the positions of the human body in the stadium. The disease feature matrix is ​​used to determine the influence of the temperature around the human body, the human body gathering state and the human body movement state on different diseases.

[0028] S104: Obtain health data based on the stadium temperature, the disease feature matrix, and the aggregated human body marker images at multiple time points; the health data includes first predicted human body change data, disease impact value, and aggregate temperature.

[0029] Optionally, the health data is obtained based on the stadium temperature, the disease feature matrix, and the aggregated human body marker images at multiple time points, including: Based on the aggregated human body marker image and the stadium temperature, an aggregated temperature feature and an aggregated temperature are obtained; the aggregated temperature represents the temperature of an area affected by the exhaled carbon dioxide of the human body in an aggregated state.

[0030] Multiple aggregation temperature characteristics and aggregation temperatures are obtained corresponding to multiple time points; Based on the human body information and aggregated temperature characteristics corresponding to the aggregated human body marker images at multiple time points, the movement state of the human body is judged to obtain the first predicted human body change data; the first predicted human body change data represents the vital sign data of the human body at the next time point.

[0031] Among them, in this embodiment, the human body's physical sign data are heart rate, blood pressure and body temperature.

[0032] Based on the disease feature matrix, human body information, gathering temperature and the first predicted human body change data, a disease impact value is obtained through a disease discrimination network; the disease impact value represents the impact of human body movement and gathering in a sports stadium on the disease.

[0033] The value range of the symptom impact value is 0-1, and the larger the value, the greater the impact on the symptom, and the smaller the value, the smaller the impact on the symptom.

[0034] Optionally, obtaining a disease feature matrix based on the aggregated human body marker images and human diseases includes: Based on a plurality of human diseases, the features of the arrangement order of human diseases are removed through a first disease neural network to obtain disease features; the disease features represent features that only include the types of diseases existing in the human body.

[0035] The lengths of the disease features corresponding to all user numbers are the same. The disease features are one-dimensional vectors.

[0036] Multiple human diseases correspond to multiple disease characteristics; A three-dimensional matrix is ​​obtained; the number of rows of the three-dimensional matrix is ​​equal to the width of the aggregated human body marker image; the number of columns of the three-dimensional matrix is ​​equal to the length of the aggregated human body marker image; and the number of pages of the three-dimensional matrix is ​​equal to the length of the disease feature.

[0037] The initial value of the three-dimensional matrix is ​​0.

[0038] According to the positions corresponding to the gathered human body marker images, the disease characteristics are sequentially filled into the corresponding rows and columns of the three-dimensional matrix to obtain a disease characteristic matrix.

[0039] Optionally, the method of removing the features of the arrangement order of human diseases through a first disease neural network based on multiple human diseases to obtain disease features includes: Obtain a random number; the random number represents the number of times a plurality of human diseases are arranged.

[0040] In this embodiment, the random number is 4, which means that the multiple human diseases are randomly arranged 4 times.

[0041] The multiple human disease symptoms are randomly arranged multiple times according to a random number of times to obtain multiple human disease disease sequences.

[0042] For example, a user number corresponds to three human diseases, namely 112, 110, and 090. 112 represents anemia, 110 represents bronchial asthma, and 090 represents calcium deficiency. The human disease sequences arranged four times are [112, 110, 090], [112, 090, 110], [110, 090, 112], and [090, 112, 110].

[0043] The human disease sequence is input into a first disease neural network to obtain a first disease feature; and a plurality of human disease sequences are used to obtain a plurality of first disease features.

[0044] Among them, the first disease neural network is a fully connected neural network (Fully Connected Neural Network, FCN).

[0045] The variance of the values ​​of the subscripts corresponding to the multiple first disease features is calculated to obtain a feature change set; the values ​​in the feature change set represent the changes of the features when the order is changed; the number of elements in the feature change set is equal to the length of the disease feature.

[0046] If the value in the feature change set is less than the change threshold, the average value of the subscripts corresponding to the multiple first disease features is calculated; if the value in the feature change set is greater than or equal to the change threshold, the value of the subscript corresponding to the first disease feature is set to 0 to obtain the disease feature.

[0047] In this embodiment, the change threshold is 0.1.

[0048] Optionally, obtaining a clustered human body marked image based on the crowd image by clustering the location network includes: Based on the crowd image, mark the position of the human body and the frame containing the human body to obtain a second human body marked image; the second human body marked image includes multiple first detection positions and a second human body frame; Get the clustering radius; the clustering radius represents the radius of the clustered human body that will cause the health data to change.

[0049] In this embodiment, the gathering radius is 8, which is 1024 / 8=128 times that of the crowd image.

[0050] A circle is drawn with the first detection position as the center and the aggregation radius as the radius to obtain an aggregation circle; multiple aggregation circles are obtained corresponding to multiple first detection positions; Based on k clustering circles, the n*uth second clustering distance is obtained; one first detection position corresponds to u second clustering distances; the second clustering distance represents the distance of the human body clustered around the first detection position.

[0051] Wherein, k and u are natural numbers greater than 0.

[0052] If the number of the second clustering distances corresponding to the first detection position is less than the clustering number threshold, the first detection position is deleted from the second human body mark image to obtain a clustered human body mark image.

[0053] By the above method, a clustering circle is drawn. The first clustering circle selected from the clustering circles indicates that a fixed area contains more than a certain number of people. Therefore, clustering is performed within the first clustering circle, and some non-clustered human bodies are excluded. Then, among the clustered human bodies, the human bodies in the surrounding vacant areas are excluded at the second clustering distance to obtain a clustered human body marker image. The changes in physical signs caused by clustering on the human body can be obtained more accurately. For example, an increase in carbon dioxide and an increase in temperature lead to an increase in heart rate, blood pressure, and body temperature, which manifests as chest tightness and shortness of breath.

[0054] Optionally, obtaining the n*uth second clustering distance based on the k clustering circles includes: The number of first detection positions contained in the gathering circle is used as the number of people gathered; If the number of people gathered is greater than the threshold number of people gathered, the gathering circle corresponding to the number of people gathered is used as the first gathering circle.

[0055] Among them, the threshold for the number of people gathering is 150.

[0056] If the center of a first gathering circle is contained in another first gathering circle, the first gathering circle with a smaller number of people gathered than the other first gathering circle will be deleted.

[0057] By using the above method, the corresponding first detection position is determined in two intersecting first clustering circles, so that the clustered people can be classified more accurately and the misjudgment of clustering of overlapping people can be avoided.

[0058] Cluster the multiple first detection positions in the first clustering circle, determine the cluster center, and obtain a cluster position set.

[0059] In this embodiment, k-means is used for clustering.

[0060] The distances between the plurality of first detection positions are calculated in the cluster position set to obtain a plurality of first cluster distances.

[0061] In this embodiment, the Euclidean distance is used to calculate the distance.

[0062] The first clustering distance that is greater than the clustering position threshold is used as the second clustering distance.

[0063] Among them, in this embodiment, the aggregation position threshold is 2.

[0064] By using the above method, the distances from the multiple first detection positions in the first clustering circle to the cluster center and the distances between the multiple first detection positions are used together to identify the first detection positions, so that the first detection positions that are not clustered can be excluded more accurately.

[0065] Optionally, the step of determining the motion state of a human body based on the human body information and the aggregated temperature characteristics corresponding to the aggregated human body marker images at multiple time points to obtain the first predicted human body change data includes: The human body information corresponding to the human body marker images aggregated at multiple time points is sequentially input into a first time convolutional network to obtain a first human body change feature; the first human body change feature represents the change feature of the vital signs of the aggregated human body during the movement at multiple time points.

[0066] In this embodiment, the first time convolution network (TCN) has a number of input layers equal to the number of human body information, which is input sequentially from early to late according to the time point.

[0067] The aggregated temperature features of multiple time points are input into a second time convolutional network to obtain a second human body change feature; the second human body change feature represents the change in ambient temperature of the human body during movement at multiple time points.

[0068] In this embodiment, the second time convolution network (TCN) has a number of input layers equal to the length of the aggregated temperature feature and is input sequentially from early to late according to the time point.

[0069] The first human body change feature and the second human body change feature are input into a human body change prediction neural network to determine the changes in the aggregation and physical signs of the human body in motion, and obtain first predicted human body change data.

[0070] Wherein, the human body change prediction neural network is a fully connected neural network (Fully Connected Neural Network, FCN). The first human body change data represents the change of physiological information of the human body.

[0071] Optionally, obtaining the aggregation temperature feature and the aggregation temperature based on the aggregated human body marker image and the stadium temperature includes: Extract the area corresponding to a clustering circle in the clustered human body mark image to obtain a clustering circle image; obtain multiple clustering circle images corresponding to multiple clustering circles in the clustered human body mark image; The regions within the second human body borders in the clustered circle image are set to black, and the regions outside the second human body borders are set to white, to obtain a binary clustered circle image; The binary cluster circle image and the stadium temperature are input into the cluster space detection network to obtain the cluster temperature feature and the cluster temperature; the cluster temperature feature represents the feature before the cluster temperature is classified.

[0072] The aggregation temperature refers to the temperature that rises due to the aggregation of human bodies.

[0073] In this embodiment, the gathering space detection network is a convolutional neural network (CNN). The output of the convolutional neural network is stretched from top to bottom and from left to right to obtain the gathering temperature feature. The stadium temperature and the gathering temperature feature are classified by a fully connected neural network (FCN) to obtain the gathering temperature. The gathering space detection network is trained with the distribution of the crowds gathered in multiple stadiums and the rising temperature in the gathering area.

[0074] Optionally, the step of marking the gathered human bodies based on the crowd image through the gathered position network to obtain the gathered human body marked image includes: The clustered location network includes a first object detection network and a second object detection network.

[0075] In this embodiment, the first target detection network is a YOLOv5 network. The second target detection network includes an edge detection algorithm and a YOLOv5 network.

[0076] The crowd image is input into a first target detection network to detect the position of the human body to obtain a first detection position.

[0077] A binary image is obtained; the size of the binary image is the same as the size of the crowd image.

[0078] The initial values ​​of the binary images are all 255.

[0079] The position where the first detection position is located in the binary image is marked to obtain a first human body marked image.

[0080] The center point corresponding to the first detection position is set to 0 in the binary image.

[0081] The first human body marked image is input into a super-resolution network to perform image enhancement to obtain a first enhanced image.

[0082] Among them, in this embodiment, a super-resolution network (SRGAN) is used.

[0083] The first enhanced image is input into a second object detection network to detect the position of a human body boundary and obtain a second human body border.

[0084] Wherein, the second human body frame is a rectangular frame.

[0085] The second human body frame is marked in the first human body marked image to obtain a clustered human body marked image.

[0086] Optionally, the obtaining of the disease impact value through a disease discrimination network based on the disease feature matrix, the stadium temperature, the gathering temperature and the first predicted human body change data includes: The first predicted human body change data is input into a first prediction neural network to obtain a first disease impact feature; the first disease impact feature represents the probability of a disease recurrence caused by a human body's physical signs at the next time point.

[0087] Among them, in this embodiment, the first prediction neural network is a fully connected neural network (Fully Connected Neural Network, FCN).

[0088] The human body information and the aggregate temperature are input into a second prediction neural network to obtain a second disease impact feature; the second disease impact feature represents the probability of disease recurrence caused by the human body's physical condition and the external temperature.

[0089] Among them, in this embodiment, the second prediction neural network is a fully connected neural network (Fully Connected Neural Network, FCN).

[0090] The second symptom impact feature and the second symptom impact feature are input into the symptom discrimination network to obtain the symptom impact value.

[0091] Among them, in this embodiment, the disease discrimination network is a fully connected neural network (Fully Connected Neural Network, FCN).

[0092] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed method should not be interpreted as reflecting the intention that the claimed invention requires more features than those explicitly recited in each claim. More specifically, as reflected in the claims below, inventive aspects lie in less than all the features of the individual embodiments disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, with each claim itself serving as a separate embodiment of the present invention.

Claims

1. An intelligent sports stadium health data collection device, characterized in that: Including stadium monitoring equipment, human monitoring equipment, camera equipment and processors: The stadium monitoring device is used to obtain the stadium temperature at multiple time points; the stadium temperature represents the temperature inside the stadium; The camera device is used to capture crowd images at multiple time points; the crowd images represent images containing multiple gathered human bodies; The human body monitoring device is used to obtain n user numbers, human body information corresponding to the user numbers, human body symptoms corresponding to the user numbers, and human body positions at multiple time points corresponding to the user numbers; The human body position represents the position of the human body in the sports stadium; n is a positive integer; The processor execution method comprises: Obtaining a clustered human body labeled image based on the crowd image through a clustered position network; the clustered human body labeled image represents an image that marks the positions and boundaries of the clustered human bodies; Matching the position of the human body in the aggregated human body mark image with the position of the human body corresponding to the user number, and associating the aggregated human body mark image with the human body information and human body disease corresponding to the user number; Based on the aggregated human body marker images and human body diseases, a disease feature matrix is ​​obtained; the disease feature matrix is ​​a three-dimensional matrix; Based on the stadium temperature, the disease feature matrix, and the aggregated human body marker images at multiple time points, health data is obtained; the health data includes first predicted human body change data, disease impact value, and aggregated temperature.

2. The intelligent sports stadium health data collection device according to claim 1 is characterized in that: The health data obtained based on the stadium temperature, the disease feature matrix, and the aggregated human body marker images at multiple time points include: Based on the gathered human body mark image and the stadium temperature, a gathered temperature feature and a gathered temperature are obtained; the gathered temperature represents the temperature of an area affected by the exhaled carbon dioxide of the human body in a gathered state; Multiple aggregation temperature characteristics and aggregation temperatures are obtained corresponding to multiple time points; Based on the human body information and the aggregated temperature characteristics corresponding to the aggregated human body mark images at multiple time points, the motion state of the human body is judged to obtain the first predicted human body change data; the first predicted human body change data represents the vital sign data of the human body at the next time point; Based on the disease feature matrix, the gathering temperature and the first predicted human body change data, a disease impact value is obtained through a disease discrimination network; the disease impact value represents the impact of human body movement and gathering in a sports stadium on the disease.

3. The intelligent sports stadium health data collection device according to claim 1 is characterized in that: The method of obtaining a disease feature matrix based on the aggregated human body marker images and human disease symptoms includes: Based on a plurality of human diseases, removing the features of the arrangement order of the human diseases through a first disease neural network, to obtain disease features; the disease features represent features that only include the types of diseases existing in the human body; Multiple human diseases correspond to multiple disease characteristics; Acquire a three-dimensional matrix; the number of rows of the three-dimensional matrix is ​​equal to the width of the aggregated human body marker image; the number of columns of the three-dimensional matrix is ​​equal to the length of the aggregated human body marker image; the number of pages of the three-dimensional matrix is ​​equal to the length of the disease feature; According to the positions corresponding to the gathered human body marker images, the disease characteristics are sequentially filled into the corresponding rows and columns of the three-dimensional matrix to obtain a disease characteristic matrix.

4. The intelligent sports stadium health data collection device according to claim 3 is characterized in that: Based on the plurality of human diseases, the characteristics of the arrangement order of human diseases are removed through the first disease neural network to obtain disease characteristics, including: Obtaining a random number; the random number represents the number of times a plurality of human diseases are arranged; Randomly arranging the plurality of human disease symptoms multiple times according to a random number of times to obtain a plurality of human disease symptom sequences; Inputting the human disease sequence into a first disease neural network to obtain a first disease feature; obtaining a plurality of first disease features corresponding to a plurality of human disease sequences; Calculate the variance of the values ​​of the subscripts corresponding to the plurality of first disease features to obtain a feature change set; the values ​​in the feature change set represent changes in the features when the order is changed; the number of elements in the feature change set is equal to the length of the disease feature; If the value in the feature change set is less than the change threshold, the average value of the subscripts corresponding to the multiple first disease features is calculated; if the value in the feature change set is greater than or equal to the change threshold, the value of the subscript corresponding to the first disease feature is set to 0 to obtain the disease feature.

5. The intelligent sports stadium health data collection device according to claim 1 is characterized in that: The method of obtaining a clustered human body marked image based on the crowd image by clustering the location network includes: Based on the crowd image, mark the position of the human body and the frame containing the human body to obtain a second human body marked image; the second human body marked image includes multiple first detection positions and a second human body frame; Obtaining a clustering radius; the clustering radius represents a radius of a clustered human body that will cause a change in health data; A circle is drawn with the first detection position as the center and the aggregation radius as the radius to obtain an aggregation circle; multiple aggregation circles are obtained corresponding to multiple first detection positions; Based on k clustering circles, an n*uth second clustering distance is obtained; one first detection position corresponds to u second clustering distances; the second clustering distance represents the distance of the human body clustered around the first detection position; If the number of the second clustering distances corresponding to the first detection position is less than the clustering number threshold, the first detection position is deleted from the second human body mark image to obtain a clustered human body mark image.

6. The intelligent sports stadium health data collection device according to claim 5 is characterized in that: The method of obtaining the n*uth second clustering distance based on the k clustering circles includes: The number of first detection positions contained in the gathering circle is used as the number of people gathered; If the number of people gathered is greater than the threshold number of people gathered, the gathering circle corresponding to the number of people gathered is used as the first gathering circle; If the center of a first gathering circle is contained in another first gathering circle, the first gathering circle with a smaller number of people is deleted; Clustering the multiple first detection positions in the first clustering circle, determining the cluster center, and obtaining a cluster position set; Calculating the distances between the plurality of first detection positions in the cluster position set to obtain a plurality of first cluster distances; The first clustering distance that is smaller than the clustering position threshold is used as the second clustering distance.

7. The intelligent sports stadium health data collection device according to claim 2 is characterized in that: The method of determining the motion state of a human body based on the human body information and the aggregated temperature characteristics corresponding to the aggregated human body marked images at multiple time points to obtain the first predicted human body change data includes: Inputting the human body information corresponding to the human body marker images gathered at multiple time points into the first time convolution network in sequence to obtain a first human body change feature; the first human body change feature represents the change feature of the vital signs of the gathered human body during the movement at multiple time points; Inputting the aggregated temperature features of multiple time points into a second time convolutional network to obtain a second human body change feature; the second human body change feature represents the change of the ambient temperature of the human body during the movement of the multiple time points; The first human body change feature and the second human body change feature are input into a human body change prediction neural network to determine the changes in the aggregation and physical signs of the human body in motion, and obtain first predicted human body change data.

8. The intelligent sports stadium health data collection device according to claim 2 is characterized in that: The method of obtaining the aggregation temperature feature and the aggregation temperature based on the aggregated human body mark image and the stadium temperature includes: Extract the area corresponding to a clustering circle in the clustered human body mark image to obtain a clustering circle image; obtain multiple clustering circle images corresponding to multiple clustering circles in the clustered human body mark image; The regions within the second human body borders in the clustered circle image are set to black, and the regions outside the second human body borders are set to white, to obtain a binary clustered circle image; The binary cluster circle image and the stadium temperature are input into the cluster space detection network to obtain the cluster temperature feature and the cluster temperature; the cluster temperature feature represents the feature before the cluster temperature is classified.

9. The intelligent sports stadium health data collection device according to claim 5, characterized in that: The step of marking the position of a human body and a frame containing the human body based on the crowd image to obtain a second human body marked image includes: The clustered location network includes a first target detection network and a second target detection network; Inputting the crowd image into a first target detection network to detect the position of a human body to obtain a first detection position; Acquire a binary image; the size of the binary image is the same as the size of the crowd image; Marking the location of the first detection position in the binary image to obtain a first human body marked image; Inputting the first human body marked image into a super-resolution network to perform image enhancement to obtain a first enhanced image; Inputting the first enhanced image into a second object detection network, detecting the position of a human body boundary, and obtaining a second human body border; The second human body frame is marked in the first human body marked image to obtain a second human body marked image.

10. The intelligent sports stadium health data collection device according to claim 2, characterized in that: The method of obtaining the disease impact value based on the disease feature matrix, the aggregate temperature and the first predicted human body change data through a disease discrimination network includes: Inputting the first predicted human body change data into a first prediction neural network to obtain a first disease impact feature; the first disease impact feature indicates the probability of a disease recurrence caused by a physical sign of the human body at the next time point; Inputting the human body information and the aggregate temperature into a second prediction neural network to obtain a second disease impact feature; the second disease impact feature indicates the probability of disease recurrence caused by the human body state and the external temperature; The second symptom impact feature and the second symptom impact feature are input into the symptom discrimination network to obtain the symptom impact value.

Images