Personalized intelligent evaluation method and system for user physical ability

Abstract

According to the user physical ability personalized intelligent evaluation method and system provided by the invention, the target user physical ability item type is used for representing the user physical ability item type corresponding to the target evaluation core, and the target credible coefficient is used for representing the association degree between the target user physical ability item type and the target human body form data; the target evaluation thread is used for performing splicing processing on the target user physical ability data queue and the target human body form data through the artificial intelligence evaluation thread to obtain target spliced knowledge characters; and performing splicing processing on the target spliced knowledge characters, the target user physical ability data queue and the target human body form data through an artificial intelligence evaluation thread to obtain target multi-dimensional knowledge characters, and determining a target evaluation kernel, a target user physical ability item type and a target credible coefficient mode according to the target multi-dimensional knowledge characters. Therefore, independent physical fitness evaluation can be carried out for the user, and a personalized training mode can be specified for different users.

Description

Technical Field

[0001] The present application relates to the technical field of physical fitness data evaluation, and more specifically, to a method and system for personalized intelligent evaluation of user physical fitness. Background Art

[0002] Physical fitness is the basic athletic ability of the human body, which is manifested through athletic qualities such as strength, speed, endurance, coordination, flexibility, and agility. It is an important component of an athlete's competitive ability. The level of physical fitness is closely related to the morphological and functional characteristics of the human body. The morphological characteristics of the human body are the qualitative basis of its physical fitness, and the functional characteristics of the human body are the biological functional basis of its physical fitness. Physical fitness aims to improve health and basic activity ability, while competitive sports physical fitness aims to pursue the physical fitness required to create excellent athletic results in competitive competitions.

[0003] At present, the evaluation of human physical fitness requires various tests to make a comprehensive judgment, and it also requires people to exercise to a certain extent, which wastes time and may cause various reasons to lead to inaccurate tests. Therefore, a technical solution is urgently needed to improve the above technical problems. Summary of the invention

[0004] In order to improve the technical problems existing in the related technologies, the present application provides a method and system for personalized intelligent evaluation of user physical fitness.

[0005] In a first aspect, a method for personalized intelligent evaluation of user physical fitness is provided, comprising: Obtaining a target user physical fitness data queue and target human body shape data, wherein the target user physical fitness data queue includes user physical fitness data to be personalized evaluated, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; The target user physical fitness data queue and the target human body morphology data are loaded into a pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, the target user physical fitness item type is used to represent the user physical fitness item type corresponding to the target evaluation core, and the target credibility coefficient is used to represent the degree of association between the target user physical fitness item type and the target human body morphology data, the target evaluation thread is used to splice the target user physical fitness data queue and the target human body morphology data through an artificial intelligence evaluation thread to obtain a target spliced ​​knowledge character, and splice the target spliced ​​knowledge character and the target user physical fitness data queue and the target human body morphology data respectively through the artificial intelligence evaluation thread to obtain a target multi-dimensional knowledge character, and determine the target evaluation core, the target user physical fitness item type and the target credibility coefficient according to the target multi-dimensional knowledge character.

[0006] It can be understood that the target user physical fitness data queue and the target human body shape data are obtained, wherein the target user physical fitness data queue includes the user physical fitness data to be personalized evaluated, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; the target user physical fitness data queue and the target human body shape data are loaded into the pre-trained target evaluation thread to obtain the target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, and the target user physical fitness item type is used to represent the user physical fitness corresponding to the target evaluation core. Item type, the target credibility coefficient is used to indicate the degree of association between the target user's physical fitness item type and the target human body morphological data, the target evaluation thread is used to splice the target user's physical fitness data queue and the target human body morphological data through the artificial intelligence evaluation thread to obtain the target spliced ​​knowledge character, and the target spliced ​​knowledge character and the target user's physical fitness data queue and the target human body morphological data are spliced ​​separately through the artificial intelligence evaluation thread to obtain the target multi-dimensional knowledge character, and the target evaluation core, the target user's physical fitness item type and the target credibility coefficient are determined according to the target multi-dimensional knowledge character, thereby enabling independent physical fitness evaluation to be performed on the user, and personalized training methods can be specified for each user.

[0007] In the present application, the artificial intelligence evaluation thread performs knowledge character splicing on the original user physical fitness data queue knowledge character and the original human body shape knowledge character to obtain the target personalized human body shape knowledge character and the target user physical fitness data queue knowledge character, including: Processing the original user physical fitness data queue knowledge characters based on an artificial intelligence analysis thread to obtain important knowledge characters of the user physical fitness data queue; Adding the values ​​of the important knowledge characters of the user physical fitness data queue and the original human body shape knowledge characters in each dimension to obtain the original personalized human body shape knowledge characters; Based on the artificial intelligence evaluation thread, the original user physical fitness data queue knowledge characters and the original personalized human body morphology knowledge characters are concatenated to obtain the target personalized human body morphology knowledge characters and the target user physical fitness data queue knowledge characters.

[0008] It can be understood that when the original user physical data queue knowledge characters and the original human body shape knowledge characters are spliced ​​based on the artificial intelligence evaluation thread, the splicing abnormality problem is improved, so that the target personalized human body shape knowledge characters and the target user physical data queue knowledge characters can be accurately obtained. In the present application, the artificial intelligence evaluation thread performs knowledge character splicing on the original user physical fitness data queue knowledge character and the original personalized human body shape knowledge character to obtain the target personalized human body shape knowledge character and the target user physical fitness data queue knowledge character, including: Based on the artificial intelligence analysis thread, the original user physical energy data queue knowledge characters and the original personalized human body morphology knowledge characters are processed respectively to obtain a first user physical energy data queue knowledge character and a first personalized human body morphology knowledge character; Using the first user physical energy data queue knowledge character as a first search vector and the first personalized human body shape knowledge character as a first important vector and a first value vector, and determining a second user physical energy data queue knowledge character and a second personalized human body shape knowledge character based on the artificial intelligence evaluation thread; Using the second personalized human body shape knowledge character as a second search vector, and the second user physical fitness data queue knowledge character as a second important vector and a second value vector, and determining a third user physical fitness data queue knowledge character and a third personalized human body shape knowledge character based on the artificial intelligence evaluation thread; The third user physical fitness data queue knowledge character and the third personalized human body morphology knowledge character are input into a decision model module to obtain the target personalized human body morphology knowledge character and the target user physical fitness data queue knowledge character.

[0009] It can be understood that when the original user physical data queue knowledge characters and the original personalized human body morphology knowledge characters are spliced ​​based on the artificial intelligence evaluation thread, the problem of inaccurate first important vector and first value vector is improved, so that the target personalized human body morphology knowledge characters and the target user physical data queue knowledge characters can be accurately obtained.

[0010] In the present application, the target search knowledge characters are sequentially concatenated with the target user physical fitness data queue knowledge characters and the target personalized human body shape knowledge characters based on the artificial intelligence evaluation thread to obtain the target multi-dimensional knowledge characters, including: Processing the target search knowledge characters based on an artificial intelligence analysis thread to obtain important vector search knowledge characters; Using the important vector search knowledge character as a third search vector, the target user physical fitness data queue knowledge character as a third important vector and a third value vector, and determining a first significant search knowledge character based on the artificial intelligence evaluation thread; Using the first significant search knowledge character as a fourth search vector, the target personalized human body shape knowledge character as a fourth importance vector and a fourth value vector, and determining a second significant search knowledge character based on the artificial intelligence evaluation thread; The second significant search knowledge character is input into a decision model module to obtain the target multi-dimensional knowledge character.

[0011] It can be understood that when the target search knowledge characters are spliced ​​with the target user physical data queue knowledge characters and the target personalized human body shape knowledge characters in turn based on the artificial intelligence evaluation thread, the accuracy of the important vector search knowledge characters is improved, so that the target multi-dimensional knowledge characters can be obtained more accurately.

[0012] In the present application, the distribution operation is performed on the value of each dimension in the target splicing knowledge character, and the target search knowledge character whose value of each dimension meets the preset requirements is determined according to the distribution result, including: Obtaining the value of the target splicing knowledge character in each dimension, wherein the value of the target splicing knowledge character in each dimension represents the commonality coefficient between the original user physical energy data queue knowledge character and the original human body shape knowledge character in each dimension; The distribution operation is performed on the values ​​on each dimension, and the dimensions where a maximum values ​​are located are determined according to the distribution results, and the target search knowledge characters are determined according to the knowledge character values ​​of the dimensions where the a maximum values ​​are located, wherein a is a positive integer.

[0013] It can be understood that when the distribution operation is performed on the value of each dimension in the target splicing knowledge character, the accuracy of the value is improved, so that the target search knowledge character whose value of each dimension meets the preset requirements can be accurately determined based on the distribution result.

[0014] In the present application, the method further comprises: Obtaining a sample user physical fitness data queue, sample human body shape data, and evaluation core mark information, wherein the sample user physical fitness data queue includes sample user physical fitness data to be subjected to regression analysis, the sample human body shape data is used to specify the type of user physical fitness items to be regressed and analyzed from the sample user physical fitness data queue, and the evaluation core mark information is used to indicate whether there is a user physical fitness data range of a preset user physical fitness item type in the sample user physical fitness data queue, and whether the sample user physical fitness data queue is associated with the sample human body shape data; The example user physical fitness data queue and the example human body shape data are loaded into the original evaluation thread to obtain an example evaluation result, wherein the example evaluation result includes an example evaluation core, an example user physical fitness item type, and an example credibility coefficient, wherein the example evaluation core is used to indicate the user physical fitness data range evaluated from the example user physical fitness data queue, the example user physical fitness item type is used to indicate the user physical fitness item type corresponding to the example evaluation core, and the example credibility coefficient is used to indicate the degree of association between the example user physical fitness item type and the example human body shape data; A target evaluation index value is calculated according to the evaluation core label information and the example evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread.

[0015] It can be understood that by accurately obtaining the example evaluation results, the target evaluation thread is trained to improve the performance of the target evaluation thread.

[0016] In the present application, the target evaluation index value is calculated according to the evaluation core marking information and the example evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread, including: A first evaluation index value, a second evaluation index value and a third evaluation index value are calculated according to the evaluation core marking information and the example evaluation result, wherein the first evaluation index value is determined by the regression analysis probability of the example user physical fitness item type and the marked user physical fitness item type indicated by the evaluation core marking information, the second evaluation index value is determined by the data scale positioning information of the example evaluation core and the data scale positioning information of the marked evaluation core indicated by the evaluation core marking information, and the third evaluation index value is determined by the example credibility coefficient and whether the example user physical fitness data queue indicated by the evaluation core marking information is associated with the example human body shape data; The first evaluation index value, the second evaluation index value and the third evaluation index value are superimposed according to preset weights to obtain the target evaluation index value, and the network parameters of the original evaluation thread are optimized by optimizing the network iteration according to the target evaluation index value to obtain the target evaluation thread.

[0017] It can be understood that when the target evaluation index value is calculated based on the evaluation core marking information and the example evaluation result, the accuracy of the first evaluation index value, the second evaluation index value and the third evaluation index value is improved, so that the original evaluation thread can be accurately trained according to the target evaluation index value to obtain the target evaluation thread, thereby improving the performance of the target evaluation thread.

[0018] In the present application, the target evaluation index value is calculated according to the evaluation core marking information and the example evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread, including: Extracting knowledge characters from the example human body shape data and the example user physical fitness data queue respectively to obtain original example user physical fitness data queue knowledge characters and original example human body shape knowledge characters; Based on the artificial intelligence evaluation thread, the original example user physical data queue knowledge character and the original example human body shape knowledge character are concatenated to obtain an example personalized human body shape knowledge character and an example user physical data queue knowledge character; Performing a list multiplication operation on the example personalized human body shape knowledge character and the example user physical fitness data queue knowledge character to obtain an example splicing knowledge character; Performing a distribution operation on the value of each dimension in the example splicing knowledge character, and determining the example search knowledge character whose value of each dimension meets the preset requirements according to the distribution result; Based on the artificial intelligence evaluation thread, the example search knowledge character is sequentially combined with the example user physical fitness data queue knowledge character and the example personalized human body shape knowledge character to obtain the example multi-dimensional knowledge character; The example evaluation core, the example user physical fitness item type and the example credibility coefficient are determined according to the example multi-dimensional knowledge character to generate the example evaluation result.

[0019] It can be understood that by calculating the target evaluation index value based on the evaluation core marking information and the example evaluation result, and training the original evaluation thread based on the target evaluation index value, the problem of inaccurate training is improved, so that the target evaluation thread can be reliably obtained.

[0020] In the present application, the target user physical fitness data queue and the target human body shape data are loaded into the pre-trained target evaluation thread to obtain the target evaluation result, including: extracting knowledge characters from the target human body shape data and the target user physical fitness data queue respectively to obtain original user physical fitness data queue knowledge characters and original human body shape knowledge characters; performing knowledge character splicing on the original user physical fitness data queue knowledge characters and the original human body shape knowledge characters based on the artificial intelligence evaluation thread to obtain target personalized human body shape knowledge characters and target user physical fitness data queue knowledge characters; performing knowledge character splicing on the target personalized human body shape knowledge characters and the target user physical fitness data queue knowledge characters. The queue knowledge characters perform a list multiplication operation to obtain the target splicing knowledge characters; a distribution operation is performed on the values ​​of each dimension in the target splicing knowledge characters, and the target search knowledge characters whose values ​​of each dimension meet the preset requirements are determined according to the distribution results; based on the artificial intelligence evaluation thread, the target search knowledge characters are sequentially spliced ​​with the target user physical fitness data queue knowledge characters and the target personalized human body morphology knowledge characters to obtain the target multi-dimensional knowledge characters; the target evaluation core, the target user physical fitness item type and the target credibility coefficient are determined according to the target multi-dimensional knowledge characters to generate the target evaluation result.

[0021] It can be understood that when the target user physical fitness data queue and the target human body shape data are loaded into the pre-trained target evaluation thread, the problem of inaccurate thread evaluation is improved, so that the target evaluation result can be accurately obtained.

[0022] In a second aspect, a personalized intelligent evaluation system for user physical fitness is provided, comprising a processor and a memory communicating with each other, wherein the processor is used to read a computer program from the memory and execute it to implement the above method.

[0023] The embodiment of the present application provides a method and system for personalized intelligent evaluation of user physical fitness, which adopts the method of obtaining a target user physical fitness data queue and target human body shape data, wherein the target user physical fitness data queue includes user physical fitness data to be personalized evaluated, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; the target user physical fitness data queue and the target human body shape data are loaded into a pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, and the target user physical fitness item type is used to represent the target The target evaluation core corresponds to the type of user physical fitness item, the target credibility coefficient is used to indicate the degree of association between the type of target user physical fitness item and the target human body morphological data, the target evaluation thread is used to splice the target user physical fitness data queue and the target human body morphological data through the artificial intelligence evaluation thread to obtain the target spliced ​​knowledge character, and the target spliced ​​knowledge character and the target user physical fitness data queue and the target human body morphological data are spliced ​​respectively through the artificial intelligence evaluation thread to obtain the target multi-dimensional knowledge character, and the target evaluation core, the type of target user physical fitness item and the target credibility coefficient are determined according to the target multi-dimensional knowledge character, thereby enabling independent physical fitness evaluation to be performed on the user, and enabling personalized training methods to be specified for each user. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0025] Figure 1 A flowchart of a method for personalized intelligent evaluation of user physical fitness provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0027] See also Figure 1, shows a method for personalized intelligent evaluation of user physical fitness, which may include the technical solutions described in the following steps 101 to S204.

[0028] S202, obtaining a target user physical fitness data queue and target human body shape data, wherein the target user physical fitness data queue includes user physical fitness data to be evaluated individually, and the target human body shape data is used to specify the type of user physical fitness items to be evaluated from the target user physical fitness data queue; For example, the user's physical fitness data queue represents body weight, skeletal muscle content, body water content, body fat, body fat percentage, etc. The target human body shape data includes human body shape measurement, which is one of the sports measurement contents. It is a measurement of human body shape, posture and structure using standardized measuring tools and instruments. The main contents include: (1) human body measurement. Also called physical measurement; (2) human body measurement, (3) body composition measurement, and (4) bone age measurement.

[0029] S204, loading the target user physical fitness data queue and the target human body morphology data into the pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to indicate the range of user physical fitness data evaluated from the target user physical fitness data queue, the target user physical fitness item type is used to indicate the user physical fitness item type corresponding to the target evaluation core, the target credibility coefficient is used to indicate the degree of association between the target user physical fitness item type and the target human body morphology data, the target evaluation thread is used to splice the target user physical fitness data queue and the target human body morphology data through the artificial intelligence evaluation thread to obtain a target spliced ​​knowledge character, and splice the target spliced ​​knowledge character and the target user physical fitness data queue and the target human body morphology data respectively through the artificial intelligence evaluation thread to obtain a target multi-dimensional knowledge character, and determine the target evaluation core, the target user physical fitness item type and the target credibility coefficient according to the target multi-dimensional knowledge character.

[0030] Exemplarily, the target assessment thread is used to assess the user's physical fitness and determine the user's ecological status.

[0031] It can be understood that based on the above-mentioned target evaluation thread proposed in the present application, several types of user physical fitness items can be evaluated from the above-mentioned target user physical fitness data queue. The user physical fitness item type can be a user physical fitness item type that has appeared in the training set during the training process of the above-mentioned target evaluation thread, or a user physical fitness item type that has not appeared in the training set. That is, in an embodiment of the present application, only the above-mentioned target human body shape data can be used to evaluate the user physical fitness item types that do not appear in the training set, so as to improve the scalability, robustness and portability of personalized evaluation.

[0032] Exemplarily, after the target user physical fitness data queue and the target human body morphology data are input into the above-mentioned target evaluation thread, the target user physical fitness data queue and the target human body morphology data are first compressed respectively, and knowledge characters are extracted. Then, the compressed human body morphology knowledge characters and the user physical fitness data queue knowledge characters are used to generate personalized human body morphology knowledge characters. Furthermore, the user physical fitness data queue knowledge characters and the personalized human body morphology knowledge characters are spliced ​​together as mentioned above to obtain more comprehensive and detailed target spliced ​​knowledge characters. That is, the target spliced ​​knowledge characters cover the information of the above-mentioned target user physical fitness data queue and the above-mentioned target human body morphology data, and after the knowledge character extraction and splicing processing, they can better express the user's physical fitness information.

[0033] In a possible implementation, according to another optional user physical fitness personalized intelligent evaluation method of the embodiment of the present application, the user physical fitness personalized intelligent evaluation method proposed in the present application can be applied to the field of physical fitness data evaluation and processing technology, specifically, it can include but is not limited to multi-dimensional knowledge character splicing, personalized evaluation, etc. The user physical fitness personalized intelligent evaluation method proposed in the present application includes but is not limited to: S402, start; S404, obtaining a target user's physical fitness data queue and target human body shape data; S406, loading the target user's physical fitness data queue and the target human body shape data into the pre-trained target evaluation thread for processing, specifically: S408-1, splicing the target user's physical fitness data queue and the target human body shape data through the artificial intelligence evaluation thread to obtain the target splicing knowledge character; S408-2, through the artificial intelligence evaluation thread, the target splicing knowledge character, the target user physical fitness data queue and the target human body shape data are spliced ​​and processed respectively to obtain the target multi-dimensional knowledge character; S408-3, after the target multi-dimensional knowledge character is processed by the regression classification module, a target evaluation result is obtained, including but not limited to a target evaluation core, a target user physical fitness item type, and a target credibility coefficient; S410, end.

[0034] Through the embodiment of the present application, a target user physical fitness data queue and target human body shape data are obtained, wherein the target user physical fitness data queue includes user physical fitness data to be personalized evaluated, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; the target user physical fitness data queue and the target human body shape data are loaded into a pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, and the target user physical fitness item type is used to represent the user physical fitness corresponding to the target evaluation core The target credibility coefficient is used to indicate the degree of association between the target user's physical fitness item type and the target human body morphological data; the target evaluation thread is used to splice the target user's physical fitness data queue and the target human body morphological data through the artificial intelligence evaluation thread to obtain the target spliced ​​knowledge characters; and the target spliced ​​knowledge characters and the target user's physical fitness data queue and the target human body morphological data are spliced ​​respectively through the artificial intelligence evaluation thread to obtain the target multi-dimensional knowledge characters; the target evaluation core, the target user's physical fitness item type and the target credibility coefficient are determined according to the target multi-dimensional knowledge characters, thereby enabling independent physical fitness evaluation to be performed on the user, and personalized training methods can be specified for each user.

[0035] In an alternative embodiment, the target user physical fitness data queue and the target human body shape data are loaded into a pre-trained target evaluation thread to obtain a target evaluation result, including: extracting knowledge characters from the target human body shape data and the target user physical fitness data queue respectively to obtain original user physical fitness data queue knowledge characters and original human body shape knowledge characters; performing knowledge character splicing on the original user physical fitness data queue knowledge characters and the original human body shape knowledge characters based on the artificial intelligence evaluation thread to obtain target personalized human body shape knowledge characters and target user physical fitness data queue knowledge characters; performing knowledge character splicing on the target personalized human body shape knowledge characters and the target user physical fitness data queue knowledge characters. The user physical fitness data queue knowledge characters perform a list multiplication operation to obtain the above-mentioned target splicing knowledge characters; a distribution operation is performed on the values ​​of each dimension in the above-mentioned target splicing knowledge characters, and the target search knowledge characters whose values ​​meet the preset requirements are determined according to the distribution results; based on the above-mentioned artificial intelligence evaluation thread, the above-mentioned target search knowledge characters are respectively and sequentially spliced ​​with the above-mentioned target user physical fitness data queue knowledge characters and the above-mentioned target personalized human body morphology knowledge characters to obtain the above-mentioned target multi-dimensional knowledge characters; the above-mentioned target evaluation core, the above-mentioned target user physical fitness item type and the above-mentioned target credibility coefficient are determined according to the above-mentioned target multi-dimensional knowledge characters to generate the above-mentioned target evaluation result.

[0036] Optionally, in an embodiment of the present application, the above-mentioned knowledge character extraction refers to extracting representative and distinctive knowledge characters from the target human body morphology data and the target user physical fitness data queue respectively, which are used to describe the respective attributes and characteristics of the target human body morphology data and the target user physical fitness data queue.

[0037] Specifically, it may include but is not limited to: S1, a human body shape compressor may be used to extract knowledge characters from the target human body shape data to generate the original human body shape knowledge characters; S2, using a user physical fitness data compressor to extract knowledge characters from the target user physical fitness data queue to generate knowledge characters from the original user physical fitness data queue.

[0038] Furthermore, after obtaining the above-mentioned original human body morphology knowledge characters and the above-mentioned original user physical fitness data knowledge characters, the above-mentioned artificial intelligence evaluation thread can be further used to perform knowledge character splicing operations on the two to generate the above-mentioned target personalized human body morphology knowledge characters and target user physical fitness data queue knowledge characters.

[0039] Exemplarily, a multi-head attention network is used to further splice the original user physical data queue knowledge characters and the above-mentioned original human body morphology knowledge characters, and then the human body morphology self-attention network and the user physical data self-attention network are respectively input to generate the above-mentioned target personalized human body morphology knowledge characters and target user physical data queue knowledge characters.

[0040] In a possible implementation embodiment, the above-mentioned list multiplication operation is performed on the above-mentioned target personalized human body morphology knowledge character and the above-mentioned target user physical fitness data queue knowledge character to obtain the above-mentioned target splicing knowledge character. The present application does not limit the data scale of the target splicing knowledge character, which can be expressed in the form of a list, wherein the number of rows and columns of the list can be flexibly set, and an element in the list represents the commonality coefficient between a user physical fitness data range and the corresponding human body morphology description.

[0041] In another exemplary embodiment, the above-mentioned distribution operation is performed on the values ​​of each dimension in the above-mentioned target splicing knowledge character in turn to obtain the distributed target splicing knowledge character, wherein the dimension of the target splicing knowledge character refers to the different attributes of the user's physical fitness item types or the classification or grouping of the knowledge character values, which is used to describe or distinguish the user's physical fitness item types.

[0042] Furthermore, any knowledge character values ​​with earlier serial numbers among the target splicing knowledge characters after distribution can be used as the target search knowledge characters. For example, the knowledge character values ​​corresponding to the serial numbers of a maximum values ​​are taken out to generate the target search knowledge characters. The value of a is not limited.

[0043] Exemplarily, the artificial intelligence evaluation thread may then be used to perform knowledge character splicing on the target search knowledge character and the target user physical fitness data queue knowledge character, and then perform knowledge character splicing on the target personalized human body shape knowledge character to generate the target multi-dimensional knowledge character, for example: S1, input the target search knowledge character into the self-attention network in the knowledge character derivation module to generate knowledge characters; S2, taking the knowledge character as H, the knowledge character of the target user physical fitness data queue as a, and P as input into the user physical fitness data cross attention network in the knowledge character derivation module to generate the knowledge character; S3, taking the knowledge character as H, the above-mentioned target personalized human body shape knowledge character as a, and P as input into the human body shape cross attention network in the knowledge character derivation module to generate the knowledge character; S4, inputting the knowledge character into the decision model in the knowledge character derivation module to generate the above-mentioned target multi-dimensional knowledge character.

[0044] In a replaceable embodiment, the above-mentioned artificial intelligence evaluation thread performs knowledge character splicing on the above-mentioned original user physical fitness data queue knowledge characters and the above-mentioned original human body morphology knowledge characters to obtain target personalized human body morphology knowledge characters and target user physical fitness data queue knowledge characters, including: processing the above-mentioned original user physical fitness data queue knowledge characters based on the artificial intelligence analysis thread to obtain important knowledge characters of the user physical fitness data queue; adding the values ​​of the above-mentioned important knowledge characters of the user physical fitness data queue and the above-mentioned original human body morphology knowledge characters in each dimension to obtain original personalized human body morphology knowledge characters; based on the above-mentioned artificial intelligence evaluation thread, the above-mentioned original user physical fitness data queue knowledge characters and the above-mentioned original personalized human body morphology knowledge characters are performed knowledge character splicing to obtain the above-mentioned target personalized human body morphology knowledge characters and the above-mentioned target user physical fitness data queue knowledge characters.

[0045] Optionally, in an embodiment of the present application, the above-mentioned user physical fitness data queue important knowledge characters refer to the user physical fitness data queue knowledge characters after the above-mentioned original user physical fitness data queue knowledge characters are processed by an artificial intelligence analysis thread, and the above-mentioned original personalized human body morphology knowledge characters refer to the human body morphology knowledge characters obtained by adding the values ​​of each dimensional knowledge characters of the above-mentioned user physical fitness data queue important knowledge characters and the above-mentioned original human body morphology knowledge characters, and the above-mentioned user physical fitness data queue important knowledge characters may include but are not limited to vectors, lists, etc.

[0046] In an alternative embodiment, the above-mentioned artificial intelligence evaluation thread performs knowledge character splicing on the above-mentioned original user physical fitness data queue knowledge characters and the above-mentioned original personalized human body morphology knowledge characters to obtain the above-mentioned target personalized human body morphology knowledge characters and the above-mentioned target user physical fitness data queue knowledge characters, including: based on the above-mentioned artificial intelligence analysis thread, respectively processing the above-mentioned original user physical fitness data queue knowledge characters and the above-mentioned original personalized human body morphology knowledge characters to obtain the first user physical fitness data queue knowledge characters and the first personalized human body morphology knowledge characters; using the above-mentioned first user physical fitness data queue knowledge characters as the first search vector, and the above-mentioned first personalized human body morphology knowledge characters as The first important vector and the first value vector are used as the first important vector and the first value vector, and based on the above-mentioned artificial intelligence evaluation thread, the second user physical data queue knowledge character and the second personalized human body morphology knowledge character are determined; the above-mentioned second personalized human body morphology knowledge character is used as the second search vector, and the above-mentioned second user physical data queue knowledge character is used as the second important vector and the second value vector, and based on the above-mentioned artificial intelligence evaluation thread, the third user physical data queue knowledge character and the third personalized human body morphology knowledge character are determined; the above-mentioned third user physical data queue knowledge character and the third personalized human body morphology knowledge character are input into the decision model module to obtain the above-mentioned target personalized human body morphology knowledge character and the above-mentioned target user physical data queue knowledge character.

[0047] Optionally, in an embodiment of the present application, the above-mentioned first user physical data queue knowledge character, first personalized human body morphology knowledge character, second user physical data queue knowledge character, second personalized human body morphology knowledge character, third user physical data queue knowledge character, third personalized human body morphology knowledge character, target personalized human body morphology knowledge character, target user physical data queue knowledge character may all include but are not limited to vectors, lists and the like.

[0048] Exemplarily, another optional method for personalized intelligent evaluation of user physical fitness according to an embodiment of the present application includes but is not limited to: S1, inputting the original user physical data queue knowledge character into the user physical data self-attention network module to generate a first personalized human body shape knowledge character, and inputting the original personalized human body shape knowledge character into the human body shape self-attention network module to generate a first personalized human body shape knowledge character; S2, taking the first personalized human body shape knowledge character as the first search vector H1 input of the user physical data-human body shape cross attention network module, taking the first personalized human body shape knowledge character as the first important vector a1 and the first value vector P1 input of the user physical data-human body shape cross attention network module, generating a second user physical data queue knowledge character and a second personalized human body shape knowledge character, wherein, in the user physical data-human body shape cross attention network module, the first search vector is used to guide the degree of attention to the input, the first important vector represents the importance of the input, and the first value vector represents the knowledge character of the input; S3, taking the second personalized human body shape knowledge character as the second search vector H2 input of the human body shape-user physical data cross attention network module, taking the second user physical data queue knowledge character as the second important vector a2 and the second value vector P2 input of the user physical data-human body shape cross attention network module, generating a third user physical data queue knowledge character and a third personalized human body shape knowledge character, wherein, in the human body shape-user physical data cross attention network module, the second search vector is used to guide the degree of attention to the input, the second important vector represents the importance of the input, and the second value vector represents the knowledge character of the input; S4, the third user physical fitness data queue knowledge character and the third personalized human body shape knowledge character are respectively passed through a decision model module to generate the above-mentioned target personalized human body shape knowledge character and the above-mentioned target user physical fitness data queue knowledge character.

[0049] In an alternative embodiment, the above-mentioned target search knowledge characters are spliced ​​with the target user physical fitness data queue knowledge characters and the target personalized human body morphology knowledge characters in sequence based on the above-mentioned artificial intelligence evaluation thread to obtain the above-mentioned target multi-dimensional knowledge characters, including: processing the above-mentioned target search knowledge characters based on the artificial intelligence analysis thread to obtain important vector search knowledge characters; using the above-mentioned important vector search knowledge characters as the third search vector, the above-mentioned target user physical fitness data queue knowledge characters as the third important vector and the third value vector, and determining the first significant search knowledge characters based on the above-mentioned artificial intelligence evaluation thread; using the above-mentioned first significant search knowledge characters as the fourth search vector, the above-mentioned target personalized human body morphology knowledge characters as the fourth important vector and the fourth value vector, and determining the second significant search knowledge characters based on the above-mentioned artificial intelligence evaluation thread; inputting the above-mentioned second significant search knowledge characters into the decision model module to obtain the above-mentioned target multi-dimensional knowledge characters.

[0050] Optionally, in the embodiment of the present application, the first significant search knowledge character and the second significant search knowledge character may both be in the form of, but not limited to, vectors, lists, and the like.

[0051] In a possible implementation example, it may include but is not limited to: S1, passing the target search knowledge characters through the spatial self-attention network module to generate the important vector search knowledge characters; S2, using the important vector search knowledge character as the third search vector, the target user physical fitness data queue knowledge character as the third important vector and the third value vector, inputting them into the user physical fitness data cross-attention network module to generate a first significant search knowledge character, wherein in the user physical fitness data cross-attention network module, the third search vector is used to guide the attention degree of the input, the third important vector represents the importance of the input, and the third value vector represents the knowledge character of the input; S3, taking the first salient search knowledge character as the fourth search vector, the target personalized human body shape knowledge character as the fourth importance vector and the fourth value vector, inputting them into the human body shape cross attention network module, and generating a second salient search knowledge character, wherein in the human body shape cross attention network module, the fourth search vector is used to guide the degree of attention to the input, the fourth importance vector represents the importance of the input, and the fourth value vector represents the knowledge character of the input; S4, inputting the second significant search knowledge character into the decision model module for processing to obtain the above-mentioned target multi-dimensional knowledge character.

[0052] In a replaceable embodiment, the above-mentioned distribution operation is performed on the values ​​of each dimension in the above-mentioned target splicing knowledge character, and the target search knowledge character whose value meets the preset requirements is determined according to the distribution result, including: obtaining the value of the above-mentioned target splicing knowledge character in each dimension, wherein the value of the above-mentioned target splicing knowledge character in each dimension represents the commonality coefficient between the above-mentioned original user physical fitness data queue knowledge character and the above-mentioned original human body shape knowledge character in each dimension; performing the above-mentioned distribution operation on the values ​​in each dimension, and determining the dimensions where a maximum values ​​are located according to the above-mentioned distribution result, and determining the above-mentioned target search knowledge character according to the knowledge character values ​​of the dimensions where the above-mentioned a maximum values ​​are located, wherein a is a positive integer.

[0053] Optionally, in an embodiment of the present application, the above distribution operation can be understood as a descending distribution according to the commonality coefficient value on each dimension, that is, the dimension with the largest commonality coefficient value is arranged in front, and the dimension with the smallest commonality coefficient value is arranged in the back. According to the above distribution results, the dimensions where a maximum values ​​are located can be determined, and then, the knowledge characters of these a dimensions are taken to determine the above target search knowledge characters.

[0054] Exemplarily, the value of the target splicing knowledge character in each dimension is first obtained, and the value represents the commonality coefficient between the original user physical data queue knowledge character and the original human body shape knowledge character in each dimension. Then, the values ​​in each dimension are distributed to determine the dimensions where a maximum values ​​are located, and the target search knowledge character is determined according to the knowledge character values ​​of these dimensions.

[0055] In an alternative embodiment, the method further includes: obtaining a sample user physical fitness data queue, sample human body shape data and evaluation core marking information, wherein the sample user physical fitness data queue includes sample user physical fitness data to be regressed and analyzed, the sample human body shape data is used to specify the type of user physical fitness items that need to be regressed and analyzed from the sample user physical fitness data queue, and the evaluation core marking information is used to indicate whether there is a user physical fitness data range of a preset user physical fitness item type in the sample user physical fitness data queue, and whether the sample user physical fitness data queue is associated with the sample human body shape data; loading the sample user physical fitness data queue and the sample human body shape data into the original An evaluation thread is started to obtain a sample evaluation result, wherein the sample evaluation result includes a sample evaluation core, a sample user physical fitness item type and a sample credibility coefficient, the sample evaluation core is used to represent the user physical fitness data range evaluated from the sample user physical fitness data queue, the sample user physical fitness item type is used to represent the user physical fitness item type corresponding to the sample evaluation core, and the sample credibility coefficient is used to represent the degree of association between the sample user physical fitness item type and the sample human body shape data; the target evaluation index value is calculated according to the evaluation core marking information and the sample evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread.

[0056] It is understandable that the present application does not limit the time for obtaining the above-mentioned example user physical fitness data queue and example human body shape data.

[0057] It can be understood that in the above-mentioned original evaluation thread, it may include but is not limited to human body shape-user physical data cross-attention network module, human body shape cross-attention network module, user physical data-human body shape cross-attention network module, user physical data cross-attention network module, etc.

[0058] Furthermore, after obtaining the above-mentioned example evaluation results, the above-mentioned target evaluation index value can be calculated based on the example evaluation results and the row number evaluation core marking information, and then, the above-mentioned original evaluation thread can be trained using the target evaluation index value to obtain the above-mentioned target evaluation thread. For example, the target evaluation index value is calculated using the cross entropy evaluation index function or the mean square error evaluation index function, and the thread parameters of the original evaluation thread are optimized according to the evaluation index value through optimization algorithms such as gradient descent, so that the original evaluation thread can better regress and analyze the positioning and category of the target evaluation core to obtain the above-mentioned target evaluation thread.

[0059] In an alternative embodiment, the target evaluation index value is calculated according to the evaluation core marking information and the example evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread, including: calculating the first evaluation index value, the second evaluation index value and the third evaluation index value according to the evaluation core marking information and the example evaluation result, wherein the first evaluation index value is determined by the regression analysis probability of the example user physical fitness event type and the marked user physical fitness event type indicated by the evaluation core marking information, and the second evaluation index value is determined by the example user physical fitness event type. The data scale positioning information of the example evaluation core and the data scale positioning information of the marked evaluation core indicated by the above-mentioned evaluation core marking information are determined, and the above-mentioned third evaluation index value is determined by whether the above-mentioned example user physical fitness data queue indicated by the above-mentioned example credibility coefficient and the above-mentioned evaluation core marking information is associated with the above-mentioned example human body shape data; the above-mentioned first evaluation index value, the above-mentioned second evaluation index value and the above-mentioned third evaluation index value are superimposed according to preset weights to obtain the above-mentioned target evaluation index value, and according to the above-mentioned target evaluation index value, the network parameters of the above-mentioned original evaluation thread are optimized by optimizing the network iteration to obtain the above-mentioned target evaluation thread.

[0060] Optionally, in the embodiment of the present application, there is no limitation on the specific values ​​of the first evaluation index value, the second evaluation index value and the third evaluation index value, and the evaluation core marking information may include but is not limited to the types of user physical fitness items marked by the evaluation core, the data scale positioning information of the marked evaluation core and the corresponding example user physical fitness data queue, etc. The example evaluation result may include but is not limited to the regression analysis probability of the example user physical fitness item type, the data scale positioning information of the evaluation core, the example credibility coefficient, etc.

[0061] Furthermore, the above-mentioned preset weights can be set flexibly, and the present application does not impose any limitation on this. Thus, the above-mentioned first evaluation index value, the second evaluation index value and the third evaluation index value are superimposed according to the preset weights to obtain the above-mentioned target evaluation index value, and then, the network parameters of the above-mentioned original evaluation thread are optimized by optimizing the network iteration to obtain the target evaluation thread.

[0062] In an alternative embodiment, the target evaluation index value is calculated according to the evaluation core marking information and the example evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread, including: extracting knowledge characters from the example human body shape data and the example user physical fitness data queue respectively to obtain original example user physical fitness data queue knowledge characters and original example human body shape knowledge characters; performing knowledge character splicing on the original example user physical fitness data queue knowledge characters and the original example human body shape knowledge characters based on the artificial intelligence evaluation thread to obtain example personalized human body shape knowledge characters and example user physical fitness data queue knowledge characters; and performing knowledge character splicing on the above-mentioned original example user physical fitness data queue knowledge characters and the above-mentioned original example human body shape knowledge characters based on the artificial intelligence evaluation thread to obtain example personalized human body shape knowledge characters and example user physical fitness data queue knowledge characters; and performing knowledge character splicing on the above-mentioned example personalized human body shape knowledge characters and example user physical fitness data queue knowledge characters. The human body morphology knowledge characters and the above-mentioned example user physical fitness data queue knowledge characters are list multiplied to obtain the above-mentioned example spliced ​​knowledge characters; a distribution operation is performed on the values ​​of each dimension in the above-mentioned example spliced ​​knowledge characters, and the example search knowledge characters whose values ​​meet the above-mentioned preset requirements are determined according to the distribution results; based on the above-mentioned artificial intelligence evaluation thread, the above-mentioned example search knowledge characters are respectively and sequentially spliced ​​with the above-mentioned example user physical fitness data queue knowledge characters and the above-mentioned example personalized human body morphology knowledge characters to obtain the above-mentioned example multi-dimensional knowledge characters; the above-mentioned example evaluation core, the above-mentioned example user physical fitness item type and the above-mentioned example credibility coefficient are determined according to the above-mentioned example multi-dimensional knowledge characters to generate the above-mentioned example evaluation results.

[0063] Optionally, in an embodiment of the present application, the above-mentioned original example user physical data queue knowledge characters, original example human body shape knowledge characters, example personalized human body shape knowledge characters, example user physical data queue knowledge characters, example splicing knowledge characters, example search knowledge characters, example multi-dimensional knowledge characters can all include but are not limited to vectors, lists and the like.

[0064] Exemplarily, assuming that there is an example data set covering an example user physical fitness data queue and example human body morphology data, first, knowledge characters are extracted from the example user physical fitness data queue and the example human body morphology data, respectively, to obtain the corresponding original example user physical fitness data queue knowledge characters and original example human body morphology knowledge characters, and then, the original example user physical fitness data queue knowledge characters and the original example human body morphology knowledge characters are spliced ​​using an artificial intelligence evaluation thread to obtain the above-mentioned example personalized human body morphology knowledge characters and the above-mentioned example user physical fitness data queue knowledge characters.

[0065] Furthermore, a list multiplication operation is performed on the example personalized human morphology knowledge characters and the example user physical fitness data queue knowledge characters to obtain example splicing knowledge characters, and each dimensional value of the example splicing knowledge characters is distributed, and the example search knowledge characters that meet the preset requirements are determined according to the distribution results; then, the artificial intelligence evaluation thread is used to perform knowledge character splicing on the example search knowledge characters with the example user physical fitness data queue knowledge characters and the example personalized human morphology knowledge characters respectively to obtain the example multi-dimensional knowledge characters; finally, the example evaluation core, the example user physical fitness item type and the example credibility coefficient are determined according to the example multi-dimensional knowledge characters to generate the above-mentioned example evaluation results.

[0066] For example, the embodiments of the present application may include but are not limited to: S1, generating a data set: selecting physical data action queues in different time periods, converting them into user physical data for storage and preprocessing, and using the preprocessed data as training and testing examples, wherein each user physical data queue (the above-mentioned example user physical data queue) in the training and testing examples generates a corresponding human body morphology description (the above-mentioned example human body morphology data) to train the above-mentioned original evaluation thread to obtain the above-mentioned target evaluation thread; S2, knowledge character extraction module: human body shape description uses human body shape compressor to extract knowledge characters, generating human body shape knowledge characters (the above original example human body shape knowledge characters); user physical fitness data queue uses user physical fitness data compressor to extract knowledge characters, generating user physical fitness data queue knowledge characters (the above original example user physical fitness data queue knowledge characters); S3, personalized physical data description generation module: the user physical data queue knowledge character first passes through the multi-head attention network module to generate the self-attention example user physical data queue knowledge character, where the human body shape knowledge character and the self-attention example user physical data queue knowledge character are added together, and then pass through two layers of fully connected layers to obtain the original example personalized human body shape knowledge character; S4, a knowledge character splicing module, inputs the original example personalized human body shape knowledge characters and the user physical data queue knowledge characters into the human body shape self-attention network module and the user physical data self-attention network module respectively; Next, a first personalized example human body shape knowledge character and a first example user physical data queue knowledge character are generated respectively; a second personalized example human body shape knowledge character and a second example user physical data queue knowledge character are generated through a user physical data-human body shape cross attention network module; S5, search knowledge character generation module: list multiplication of the example personalized human body shape knowledge character and the example user physical data queue knowledge character to obtain the above-mentioned example splicing knowledge character, distribute the values ​​of each dimension of the example splicing knowledge character, take out the knowledge character values ​​where a maximum values ​​are located, and generate the above-mentioned example search knowledge character; S6, knowledge character derivation module: the example search knowledge character first passes through the self-attention network module to generate the self-attention example search knowledge character, and the self-attention example search knowledge character is used as H, the example user physical data queue knowledge character is used as a, P is input into the user physical data cross-attention network module to generate the first cross-attention example search knowledge character, and the first cross-attention example search knowledge character is used as H, the example personalized human body shape knowledge character is used as a, P is input into the human body shape cross-attention network module to generate the second cross-attention example search knowledge character, and finally the second cross-attention example search knowledge character is passed through the decision model module (FFN) to generate the example multi-dimensional knowledge character; S7, regression classification module: load the example multi-dimensional knowledge character into an 11 convolution layer, pass through the ReLu layer (activation layer), and then pass through an 11 convolution layer to obtain the example evaluation kernel, the example user physical fitness item type and the example credibility coefficient; S8, thread training: CrossEntropy Loss (cross entropy evaluation index function), Focal Loss (focus evaluation index function) and L1 Loss (absolute value evaluation index function) are used to supervise the intermediate knowledge character graph, user physical fitness item type, credibility coefficient and evaluation kernel coordinates of the original evaluation thread respectively. Among them, the three evaluation index functions are superimposed according to the set ratio, and the SGD optimizer (stochastic gradient descent optimizer) is used to optimize the network parameters by optimizing the network iteration. The results on the test example select the best thread. Specifically: In summary, the embodiments of the present application realize end-to-end multi-dimensional personalized evaluation of physical fitness data. By splicing human body shape data and user physical fitness data information, it is possible to locate the action behavior in the physical fitness data through human body shape. It also includes personalized human body shape description prompt generation: it can generate human body shape description prompts corresponding to the user physical fitness data queue in a personalized manner according to the global information of the user's physical fitness data and the type of user's physical fitness items; multi-dimensional knowledge character splicing strategy: splicing human body shape data with user physical fitness data queue information, so that the target evaluation thread pays more attention to the user physical fitness data queue range corresponding to the human body shape data; personalized evaluation of physical data: the types of user physical fitness items that do not appear in the training set can be evaluated only through human body shape data, which has stronger scalability and robustness.

[0067] Based on the above, a user physical fitness personalized intelligent evaluation device is provided, the device comprising: A data acquisition module, used to obtain a target user physical fitness data queue and target human body shape data, wherein the target user physical fitness data queue includes user physical fitness data to be evaluated individually, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; The result evaluation module is used to load the target user physical fitness data queue and the target human body morphology data into the pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, the target user physical fitness item type is used to represent the user physical fitness item type corresponding to the target evaluation core, and the target credibility coefficient is used to represent the degree of association between the target user physical fitness item type and the target human body morphology data, the target evaluation thread is used to splice the target user physical fitness data queue and the target human body morphology data through an artificial intelligence evaluation thread to obtain a target spliced ​​knowledge character, and splice the target spliced ​​knowledge character and the target user physical fitness data queue and the target human body morphology data respectively through the artificial intelligence evaluation thread to obtain a target multi-dimensional knowledge character, and determine the target evaluation core, the target user physical fitness item type and the target credibility coefficient according to the target multi-dimensional knowledge character.

[0068] On the basis of the above, a user physical fitness personalized intelligent evaluation system is shown, including a processor and a memory that communicate with each other, and the processor is used to read a computer program from the memory and execute it to implement the above method.

[0069] Based on the above, a computer-readable storage medium is also provided, on which a computer program stored implements the above method when running.

[0070] In summary, based on the above scheme, a target user physical fitness data queue and target human body shape data are obtained, wherein the target user physical fitness data queue includes the user physical fitness data to be personalized evaluated, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; the target user physical fitness data queue and the target human body shape data are loaded into the pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation kernel, a target user physical fitness item type and a target credibility coefficient, the target evaluation kernel is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, and the target user physical fitness item type is used to represent the user corresponding to the target evaluation kernel. The type of physical fitness items and the target credibility coefficient are used to indicate the degree of association between the type of physical fitness items of the target user and the target human body morphological data. The target evaluation thread is used to splice the target user's physical fitness data queue and the target human body morphological data through the artificial intelligence evaluation thread to obtain the target spliced ​​knowledge characters, and the target spliced ​​knowledge characters and the target user's physical fitness data queue and the target human body morphological data are spliced ​​respectively through the artificial intelligence evaluation thread to obtain the target multi-dimensional knowledge characters. The target evaluation core, the type of physical fitness items of the target user and the target credibility coefficient are determined according to the target multi-dimensional knowledge characters. Thus, independent physical fitness evaluation can be performed for the user, and personalized training methods can be specified for each user.

[0071] It should be understood that the system and its modules shown above can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or a dedicated design hardware. Those skilled in the art will understand that the above methods and systems can be implemented using computer executable instructions and / or included in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The system and its modules of the present application can not only be implemented by hardware circuits such as ultra-large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., but can also be implemented by software such as executed by various types of processors, and can also be implemented by a combination of the above hardware circuits and software (e.g., firmware).

[0072] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or any other beneficial effects that may be obtained.

Claims

1. A personalized intelligent evaluation method for user physical fitness, characterized in that: include: Obtaining a target user physical fitness data queue and target human body shape data, wherein the target user physical fitness data queue includes user physical fitness data to be personalized evaluated, and the target human body shape data is used to specify the type of user physical fitness items that need to be evaluated from the target user physical fitness data queue; The target user physical fitness data queue and the target human body morphology data are loaded into a pre-trained target evaluation thread to obtain a target evaluation result, wherein the target evaluation result includes a target evaluation core, a target user physical fitness item type and a target credibility coefficient, the target evaluation core is used to represent the user physical fitness data range evaluated from the target user physical fitness data queue, the target user physical fitness item type is used to represent the user physical fitness item type corresponding to the target evaluation core, and the target credibility coefficient is used to represent the degree of association between the target user physical fitness item type and the target human body morphology data, the target evaluation thread is used to splice the target user physical fitness data queue and the target human body morphology data through an artificial intelligence evaluation thread to obtain a target spliced ​​knowledge character, and splice the target spliced ​​knowledge character and the target user physical fitness data queue and the target human body morphology data respectively through the artificial intelligence evaluation thread to obtain a target multi-dimensional knowledge character, and determine the target evaluation core, the target user physical fitness item type and the target credibility coefficient according to the target multi-dimensional knowledge character.

2. The method according to claim 1, characterized in that The step of performing knowledge character splicing on the original user physical fitness data queue knowledge character and the original human body shape knowledge character based on the artificial intelligence evaluation thread to obtain target personalized human body shape knowledge character and target user physical fitness data queue knowledge character includes: Processing the original user physical fitness data queue knowledge characters based on an artificial intelligence analysis thread to obtain important knowledge characters of the user physical fitness data queue; Adding the values ​​of the important knowledge characters of the user physical fitness data queue and the original human body shape knowledge characters in each dimension to obtain the original personalized human body shape knowledge characters; Based on the artificial intelligence evaluation thread, the original user physical fitness data queue knowledge characters and the original personalized human body morphology knowledge characters are concatenated to obtain the target personalized human body morphology knowledge characters and the target user physical fitness data queue knowledge characters.

3. The method according to claim 2, characterized in that The step of performing knowledge character splicing on the original user physical fitness data queue knowledge character and the original personalized human body morphology knowledge character based on the artificial intelligence evaluation thread to obtain the target personalized human body morphology knowledge character and the target user physical fitness data queue knowledge character includes: Based on the artificial intelligence analysis thread, the original user physical energy data queue knowledge characters and the original personalized human body morphology knowledge characters are processed respectively to obtain a first user physical energy data queue knowledge character and a first personalized human body morphology knowledge character; Using the first user physical energy data queue knowledge character as a first search vector and the first personalized human body shape knowledge character as a first important vector and a first value vector, and determining a second user physical energy data queue knowledge character and a second personalized human body shape knowledge character based on the artificial intelligence evaluation thread; Using the second personalized human body shape knowledge character as a second search vector, and the second user physical fitness data queue knowledge character as a second important vector and a second value vector, and determining a third user physical fitness data queue knowledge character and a third personalized human body shape knowledge character based on the artificial intelligence evaluation thread; The third user physical fitness data queue knowledge character and the third personalized human body morphology knowledge character are input into a decision model module to obtain the target personalized human body morphology knowledge character and the target user physical fitness data queue knowledge character.

4. The method according to claim 1, characterized in that: The artificial intelligence evaluation thread sequentially performs knowledge character splicing on the target search knowledge character, the target user physical fitness data queue knowledge character and the target personalized human body shape knowledge character to obtain the target multi-dimensional knowledge character, including: Processing the target search knowledge characters based on an artificial intelligence analysis thread to obtain important vector search knowledge characters; Using the important vector search knowledge character as a third search vector, the target user physical fitness data queue knowledge character as a third important vector and a third value vector, and determining a first significant search knowledge character based on the artificial intelligence evaluation thread; Using the first significant search knowledge character as a fourth search vector, the target personalized human body shape knowledge character as a fourth importance vector and a fourth value vector, and determining a second significant search knowledge character based on the artificial intelligence evaluation thread; The second significant search knowledge character is input into a decision model module to obtain the target multi-dimensional knowledge character.

5. The method according to claim 1, characterized in that The step of performing a distribution operation on the value of each dimension in the target splicing knowledge character, and determining the target search knowledge character whose value of each dimension meets the preset requirements according to the distribution result, includes: Obtaining the value of the target splicing knowledge character in each dimension, wherein the value of the target splicing knowledge character in each dimension represents the commonality coefficient between the original user physical energy data queue knowledge character and the original human body shape knowledge character in each dimension; The distribution operation is performed on the values ​​on each dimension, and the dimensions where a maximum values ​​are located are determined according to the distribution results, and the target search knowledge characters are determined according to the knowledge character values ​​of the dimensions where the a maximum values ​​are located, wherein a is a positive integer.

6. The method according to claim 1, characterized in that The method further comprises: Obtaining a sample user physical fitness data queue, sample human body shape data, and evaluation core mark information, wherein the sample user physical fitness data queue includes sample user physical fitness data to be subjected to regression analysis, the sample human body shape data is used to specify the type of user physical fitness items to be regressed and analyzed from the sample user physical fitness data queue, and the evaluation core mark information is used to indicate whether there is a user physical fitness data range of a preset user physical fitness item type in the sample user physical fitness data queue, and whether the sample user physical fitness data queue is associated with the sample human body shape data; The example user physical fitness data queue and the example human body shape data are loaded into the original evaluation thread to obtain an example evaluation result, wherein the example evaluation result includes an example evaluation core, an example user physical fitness item type, and an example credibility coefficient, wherein the example evaluation core is used to indicate the user physical fitness data range evaluated from the example user physical fitness data queue, the example user physical fitness item type is used to indicate the user physical fitness item type corresponding to the example evaluation core, and the example credibility coefficient is used to indicate the degree of association between the example user physical fitness item type and the example human body shape data; A target evaluation index value is calculated according to the evaluation core label information and the example evaluation result, and the original evaluation thread is trained according to the target evaluation index value to obtain the target evaluation thread.

7. The method according to claim 6, characterized in that The step of calculating a target evaluation index value according to the evaluation core label information and the example evaluation result, and training the original evaluation thread according to the target evaluation index value to obtain the target evaluation thread includes: A first evaluation index value, a second evaluation index value and a third evaluation index value are calculated according to the evaluation core marking information and the example evaluation result, wherein the first evaluation index value is determined by the regression analysis probability of the example user physical fitness item type and the marked user physical fitness item type indicated by the evaluation core marking information, the second evaluation index value is determined by the data scale positioning information of the example evaluation core and the data scale positioning information of the marked evaluation core indicated by the evaluation core marking information, and the third evaluation index value is determined by the example credibility coefficient and whether the example user physical fitness data queue indicated by the evaluation core marking information is associated with the example human body shape data; The first evaluation index value, the second evaluation index value and the third evaluation index value are superimposed according to preset weights to obtain the target evaluation index value, and the network parameters of the original evaluation thread are optimized by optimizing the network iteration according to the target evaluation index value to obtain the target evaluation thread.

8. The method according to claim 6, characterized in that The step of calculating a target evaluation index value according to the evaluation core label information and the example evaluation result, and training the original evaluation thread according to the target evaluation index value to obtain the target evaluation thread includes: Extracting knowledge characters from the example human body shape data and the example user physical fitness data queue respectively to obtain original example user physical fitness data queue knowledge characters and original example human body shape knowledge characters; Based on the artificial intelligence evaluation thread, the original example user physical data queue knowledge character and the original example human body shape knowledge character are concatenated to obtain an example personalized human body shape knowledge character and an example user physical data queue knowledge character; Performing a list multiplication operation on the example personalized human body shape knowledge character and the example user physical fitness data queue knowledge character to obtain an example splicing knowledge character; Performing a distribution operation on the value of each dimension in the example splicing knowledge character, and determining the example search knowledge character whose value of each dimension meets the preset requirements according to the distribution result; Based on the artificial intelligence evaluation thread, the example search knowledge character is sequentially combined with the example user physical fitness data queue knowledge character and the example personalized human body shape knowledge character to obtain the example multi-dimensional knowledge character; The example evaluation core, the example user physical fitness item type and the example credibility coefficient are determined according to the example multi-dimensional knowledge character to generate the example evaluation result.

9. The method according to claim 1, characterized in that: The method of loading the target user physical fitness data queue and the target human body shape data into the pre-trained target evaluation thread to obtain the target evaluation result includes: extracting knowledge characters from the target human body shape data and the target user physical fitness data queue respectively to obtain original user physical fitness data queue knowledge characters and original human body shape knowledge characters; performing knowledge character splicing on the original user physical fitness data queue knowledge characters and the original human body shape knowledge characters based on the artificial intelligence evaluation thread to obtain target personalized human body shape knowledge characters and target user physical fitness data queue knowledge characters; performing knowledge character splicing on the target personalized human body shape knowledge characters and the target user physical fitness data queue knowledge characters. The target search knowledge character is executed by performing a list multiplication operation on the recognition characters to obtain the target splicing knowledge character; a distribution operation is performed on the value of each dimension in the target splicing knowledge character, and a target search knowledge character whose value of each dimension meets the preset requirements is determined according to the distribution result; based on the artificial intelligence evaluation thread, the target search knowledge character is sequentially spliced ​​with the target user physical fitness data queue knowledge character and the target personalized human body shape knowledge character to obtain the target multi-dimensional knowledge character; the target evaluation core, the target user physical fitness item type and the target credibility coefficient are determined according to the target multi-dimensional knowledge character to generate the target evaluation result.

10. A user physical fitness personalized intelligent evaluation system, characterized in that: The invention comprises a processor and a memory communicating with each other, wherein the processor is used to read a computer program from the memory and execute the computer program to implement the method according to any one of claims 1 to 9.

Images