Running training method and device based on large model and storage medium

Abstract

The invention relates to a running training method and device based on a large model, and belongs to the technical field of artificial intelligence, and the method comprises the steps: carrying out pre-training on a preset first large model, recording multi-modal data of an athlete, inputting the multi-modal data into the pre-trained large model, and obtaining a running training scheme of the athlete. The invention further provides a running training device based on the large model and a storage medium.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a large model-based running training method, device and storage medium. Background Art

[0002] With the increasing attention paid to health education, the analysis of sports training data is becoming increasingly important. Current data applications mainly focus on data collection and basic physical fitness tests, and lack guidance for after-school training. It is usually impossible to dynamically adjust according to students' personal physique, training status, and training results. This leads to the homogeneity and inefficiency of training guidance, which cannot fully meet the needs of students for personalized training. In the training of professional athletes, it is also necessary to formulate personalized training methods based on the athletes' personal physique, previous training performance, etc., so as to provide a more scientific training plan and improve sports performance.

[0003] With the development of artificial intelligence, big models have powerful capabilities in natural language processing, image analysis and multimodal data fusion. How to use big models to provide personalized and real-time training plans based on multimodal data such as athletes' physical test data, training images, and training evaluation results is a technical problem that needs to be solved urgently. Summary of the invention

[0004] In view of the above deficiencies in the prior art, the purpose of the invention is to provide a running training method and device based on a large model, which improves the efficiency of running training, provides a personalized and real-time running training method, and achieves scientific and efficient physical fitness improvement.

[0005] In a first aspect of the present invention, a running training method based on a large model is proposed, comprising:

[0006] S1, pre-training the preset first model;

[0007] S2, records multimodal data of athletes;

[0008] S3, inputting the multimodal data into the pre-trained first large model to obtain a running training program for the athlete;

[0009] The pre-training of the preset first large model includes:

[0010] Record historical multimodal data of athletes’ running;

[0011] Recording the training scheme corresponding to each piece of data in the historical multimodal data;

[0012] Importing the historical multimodal data and the training program into a preset first large model to construct a fine-tuning data set for the running training program;

[0013] The multimodal data includes one or a combination of the following: running ability evaluation results, height, weight, latest running performance score, gender, and age.

[0014] Furthermore, the running ability evaluation result M1 is determined according to the following expression:

[0015] M1 = "Running ability evaluation:" + g1 + g2 + g3 + g4 + g5;

[0016] Among them, g1 is the standard evaluation result of posture;

[0017] g2 is the explosive power evaluation result;

[0018] g3 is the endurance evaluation result;

[0019] g4 is the running skill evaluation result;

[0020] g5 is the training quality evaluation result;

[0021] Among them, the running ability evaluation result M1 is a string data, the + sign in the expression represents the concatenation of the string, and g1, g2, g3, g4 and g5 are string data.

[0022] Furthermore, the posture standard evaluation result g1 is determined according to the following method:

[0023] Pre-training a preset second largest model using the athlete's historical running posture data;

[0024] Recording the running posture images of the athlete for N1 consecutive days, recognizing each running posture image through the pre-trained second largest model, outputting whether the runner's posture meets the standard, and recording the number of images n1 that meet the standard and the number of images n2 that do not meet the standard;

[0025] The posture standard evaluation result g1 is determined according to the following expression:

[0026]

[0027] Wherein, ts1 is a set first judgment threshold, ts2 is a set second judgment threshold, ts3 is a set third judgment threshold, and N1 is an integer greater than or equal to 20.

[0028] Furthermore, the explosive power evaluation result g2 is determined according to the following method:

[0029] Record the athlete's sprint scores for N2 consecutive days g2 i , where i is the serial number of the sprint event score, i=1,2,3,...,n3, and n3 is the number of sprint event scores of the athlete for N2 consecutive days;

[0030] Determine the explosive power evaluation result g2 according to the following expression:

[0031]

[0032] Among them, g 21 is the explosive power evaluation score, and g 21 Determined by the following expression:

[0033]

[0034] Among them, g21 i Score for the partition, and g21 i Determined by the following expression:

[0035]

[0036] Wherein N2 is an integer greater than or equal to 20.

[0037] Furthermore, the endurance evaluation result g3 is determined according to the following method:

[0038] Record the athlete's long-distance running scores for N3 consecutive days g3 j , where j is the serial number of the long-distance running event score, j = 1, 2, 3, ..., n4, and n4 is the number of long-distance running event scores of the athlete for N3 consecutive days;

[0039] The endurance evaluation result g3 is determined according to the following expression:

[0040]

[0041] Among them, g 31 is the endurance rating score and is determined by the following expression:

[0042]

[0043] Wherein, N3 is an integer greater than or equal to 20.

[0044] Furthermore, the running skill evaluation result g4 is determined according to the following method:

[0045] Record the time t of the athlete completing each lap in the N4-day long-distance running event m , the corresponding total number of laps n5, the distance of each lap l1, m = 1, ..., n5, is the sequence number of the completed laps;

[0046]

[0047] g4 = steady pace, good technique;

[0048] when and and hour:

[0049] g4 = very unstable pace control, large fluctuations, lack of running skills;

[0050] In other cases:

[0051] g4 = Pace fluctuates, running technique needs to be further improved;

[0052] Wherein ts4 is the set fourth judgment threshold, ts5 is the set fifth judgment threshold, and ts6 is the set sixth judgment threshold;

[0053] Wherein, N4 is an integer greater than or equal to 20; p=1,...,n5 is the sequence number of completed circles.

[0054] Furthermore, the training quality evaluation result g5 is determined according to the following method:

[0055] Record the athlete's running performance for N5 consecutive days p , p=1,...,n3+n4..., is the running sequence number, the smaller the sequence number, the earlier the corresponding training record time, n3 is the number of points scored by the athlete in the sprint event for N2 consecutive days, n4 is the number of points scored by the athlete in the long-distance event for N3 consecutive days, where N2 is an integer greater than or equal to 20, and N3 is an integer greater than or equal to 20;

[0056] The training quality evaluation result g5 is determined according to the following expression:

[0057] g5=g 51 +g 52 +g 53 , where the + sign indicates string concatenation;

[0058] Among them, g 51 For quantitative evaluation, g 52 For quality evaluation, g 53 is the effect evaluation and satisfies the following expression:

[0059]

[0060] Among them, ts7 is the set seventh judgment threshold, ts8 is the set eighth judgment threshold, ts9 is the set ninth judgment threshold, and ts 10 is the tenth judgment threshold, ts 11 is the eleventh judgment threshold, ts 12 is the twelfth judgment threshold, ts 13 is the thirteenth judgment threshold, ts 14 is the fourteenth judgment threshold, ts15 is the fifteenth judgment threshold, ts 16 is the set sixteenth judgment threshold, ts 17 is the set seventeenth judgment threshold;

[0061] Among them, N5 is an integer greater than or equal to 20; q=1,...,n3+n4 is the running sequence number.

[0062] A second aspect of the present invention provides a device for implementing a large model-based running training method, comprising:

[0063] A model pre-training module is configured to pre-train a preset first large model;

[0064] a data recording module configured to record multimodal data of the athlete;

[0065] A calculation module is configured to input the multimodal data into the pre-trained first large model to obtain a running training program for the athlete;

[0066] The pre-training of the preset first large model includes:

[0067] Record historical multimodal data of athletes’ running;

[0068] Recording the training scheme corresponding to each piece of data in the historical multimodal data;

[0069] Importing the historical multimodal data and the training program into a preset first large model to construct a fine-tuning data set for the running training program;

[0070] The multimodal data includes one or a combination of the following: running ability evaluation results, height, weight, latest running performance score, gender, and age.

[0071] In a third aspect, the present application proposes a large model-based running training device, characterized in that it includes a memory, a processor, and a user interface;

[0072] The memory is used to store computer programs;

[0073] The user interface is used to interact with the user;

[0074] The processor is used to read the computer program in the memory, and when the processor executes the computer program, the above-mentioned running training method is implemented.

[0075] In a fourth aspect, the present application proposes a processor-readable storage medium, characterized in that the processor-readable storage medium stores a computer program, and the processor implements the above-mentioned large model-based running training method when executing the computer program.

[0076] The beneficial effects of the present invention are as follows:

[0077] The running training method and device of the present invention records and collects the historical multimodal data of the athlete's running, records the training plan corresponding to each multimodal data, loads the pre-trained large model, and constructs a fine-tuning data set for running training guidance, which includes multimodal data in various running situations and corresponding training guidance labels. Then, the athlete's multimodal data is recorded and input into the pre-trained large model to obtain the corresponding running training plan, thereby providing personalized and real-time running training guidance and achieving scientific and efficient physical fitness improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The accompanying drawings are only used to illustrate specific embodiments and are not considered to limit the present invention. In the entire drawings, the same reference symbols represent the same components. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0079] Figure 1 This is a flow chart of a running training method based on a large model according to an embodiment of the present invention;

[0080] Figure 2 A flow chart of a method for constructing a fine-tuning data set according to an embodiment of the present invention;

[0081] Figure 3 Schematic diagram of a running training device based on a large model according to an embodiment of the present invention;

[0082] Figure 4 Schematic diagram of another large model-based running training device according to an embodiment of the present invention.

[0083] Reference numerals:

[0084] S1, step 1; S2, step 2; S201, step 201; S202, step 202; S203, step 203; 301, pre-training module; 302, data recording module; 303, operation module; 401 processor; 402, memory; 403, user interface. DETAILED DESCRIPTION

[0085] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work should fall within the scope of protection of the present invention.

[0086] Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts disclosed in the present invention.

[0087] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the orientation or position relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. The terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0088] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of methods and systems consistent with some aspects of the present invention as detailed in the appended claims.

[0089] Current data applications mainly focus on data collection and basic physical fitness tests, and lack guidance for after-school training. It is usually impossible to dynamically adjust according to students' personal physique, training status, and training results. This leads to the homogeneity and inefficiency of training guidance, which cannot fully meet students' needs for personalized training. In the training of professional athletes, it is also necessary to formulate personalized training methods based on the athletes' personal physique, previous training performance, etc., so as to provide more scientific training plans and improve sports performance. How to use large models to provide personalized and real-time training plans based on multimodal data such as athletes' physical test data, training images, and training evaluation results is a technical problem that needs to be solved urgently.

[0090] In response to the problems in the prior art, the present invention proposes a running training method and device based on a large model, which solves the problems of low automation, inaccurate target positioning and labeling, low work efficiency, high technical requirements for testers, complicated and time-consuming operations, etc. in the existing methods.

[0091] By utilizing the analytical reasoning capabilities of large models and the processing capabilities of massive data, the efficiency of training program production can be greatly improved compared to manual analysis and processing, and the scientificity and rationality of training programs can be improved. For example, manual analysis and processing of multimodal data requires recording multimodal data on a computer disk, making a large number of intermediate analysis tables, sorting and organizing intermediate information and saving them as files, and matching them with the athlete's height, weight, latest running score, gender, age, etc., and then finding patterns in the data by writing software programs for analysis and manual analysis. This analysis method requires the storage of a large number of intermediate files and consumes more computer storage resources. If a special program is written for analysis, it will consume more storage space and computing resources. Using the analysis and processing capabilities of large models for multimodal data can not only save a lot of storage space for intermediate files, but also greatly improve processing efficiency.

[0092] Method Embodiment

[0093] In a first aspect, the present invention provides a running training method based on a large model, such as Figure 1 As shown, specifically, the method includes steps S1 to S3:

[0094] S1, pre-training the preset first model;

[0095] S2, records multimodal data of athletes;

[0096] S3, inputting the multimodal data into the pre-trained first large model to obtain a running training program for the athlete;

[0097] Among them, the process of pre-training the preset first large model is as follows Figure 2As shown, including:

[0098] S201, recording historical multimodal data of the athlete's running;

[0099] S202, recording the training scheme corresponding to each piece of data in the historical multimodal data;

[0100] S203, importing the historical multimodal data and the training plan into the preset first large model to construct a fine-tuning data set for the running training plan;

[0101] The multimodal data includes one or a combination of the following: running ability evaluation results, height, weight, latest running performance score, gender, and age.

[0102] As an optional example, the first large model is the InternVL2 large model. Among them, the first large model is fine-tuned through historical multimodal data to make it more accurate when calculating the running training plan. Specifically, it can be to record and collect multimodal data of athletes' running through an intelligent running device, including running ability evaluation in historical training, as well as height, weight, latest running performance score, gender, age, and collect the training guidance corresponding to each data, load the pre-trained InternVL2 model, and build a fine-tuning data set for running training guidance. The data set includes multimodal data in various running situations and corresponding training guidance program labels. The InternVL2 model is fine-tuned using the fine-tuning data set so that the InternVL2 model learns how to generate a personalized running training plan based on the multimodal data during the running process.

[0103] It should be noted that the historical multimodal data in the present embodiment refers to the multimodal data recorded before S2 in the above-mentioned embodiment, and the preset first large model is pre-trained using these historical multimodal data. The multimodal data in S2 refers to the multimodal data of the current athlete's movement, and the multimodal data of the current athlete's movement is input into the preset first large model after pre-training and fine-tuning to obtain the corresponding subsequent training program under the current motion state. For example, for a certain student, the historical multimodal data of N days before today has been collected, and the after-school training program corresponding to the multimodal data of each day is collected, and these historical multimodal data and the corresponding after-school training program are input into the InternVL2 model for training to obtain the pre-trained InternVL2 model. For this student, today's multimodal data is collected and input into the pre-trained InternVL2 model to obtain today's after-school running training program. For example, a student A's 100-meter sprint result today is 14 seconds. The student A's running ability evaluation M1, height, weight, latest running score, gender, and age are input into the pre-trained InternVL2 model, and the training plan after today's class is to raise the legs high on the spot to increase the running stride frequency.

[0104] It should be noted that the training plan corresponding to each piece of data in the historical multimodal data can be a training plan labeled by a professional training coach. That is to say, for a piece of multimodal data, its corresponding training plan is the best training plan for the athlete's state corresponding to the multimodal data. It is a targeted training plan designed for the problems in running or the directions for improvement reflected in the multimodal data.

[0105] It should be noted that the running training method of the present invention can be used to train athletes of all ages, such as students, professional athletes, etc.

[0106] As an optional example, in an embodiment of the present invention, the running ability evaluation result is determined according to the following method:

[0107] The running ability evaluation result M1 is determined according to the following expression 1:

[0108] M1 = "Running ability evaluation: " + g1 + g2 + g3 + g4 + g5; (Expression 1)

[0109] Among them, g1 is the posture standard evaluation result, which is used to indicate whether the runner's posture meets the standard;

[0110] g2 is the explosive power evaluation result, which is used to indicate the running explosive power level of the athlete;

[0111] g3 is the endurance evaluation result, which is used to indicate the running endurance level of the athlete;

[0112] g4 is a running skill evaluation result, which is used to indicate the running skill level of the athlete;

[0113] g5 is a training quality evaluation result, which is used to indicate the running training level of the athlete.

[0114] Among them, the running ability evaluation result M1 is a string data, and the + sign in the expression represents the splicing of the string, and g1, g2, g3, g4 and g5 are string data. When splicing different strings, a separator can be added between the two spliced ​​strings after splicing, such as a comma "," or a semicolon ";", or a space "", or other special characters such as an asterisk "*", a pound sign "#", an exclamation mark "!", etc., determined as needed. For example, if g1 = "perfect posture, efficient and labor-saving", g2 = "explosive and powerful, with advantages in short-distance running", g3 = "lack of endurance, long-distance running ability needs to be improved", g4 = "stable pace, good skills", g5 = "very frequent training, high training quality and high training effect", the separator added in the middle is a semicolon ";", then the result after the above expression 1 is processed is: M1 = "running ability evaluation: perfect posture, efficient and labor-saving; explosive and powerful, with advantages in short-distance running; lack of endurance, long-distance running ability needs to be improved; stable pace, good skills; very frequent training, high training quality and high training effect".

[0115] In the embodiment of the present invention, the running ability evaluation result is comprehensively determined by the runner's posture, explosive power, endurance, skills and training quality.

[0116] As an optional example, the posture standard evaluation result g1 is determined according to the following method:

[0117] Pre-training a preset second largest model using the athlete's historical running posture data;

[0118] Recording the running posture images of the athlete for N1 consecutive days, recognizing each running posture image through the pre-trained second largest model, outputting whether the runner's posture meets the standard, and recording the number of images n1 that meet the standard and the number of images n2 that do not meet the standard;

[0119] The posture standard evaluation result g1 is determined according to the following expression 3:

[0120]

[0121] Wherein ts1 is the set first judgment threshold, ts2 is the set second judgment threshold, ts3 is the set third judgment threshold, and N1 is an integer greater than or equal to 20, for example, it can also be 30 days, 50 days, 60 days, etc.;

[0122] Pre-training the preset second largest model using the historical running posture data of the athlete includes:

[0123] Record the athlete's historical running posture data, annotate the historical running posture data, and annotate the athlete's key posture elements. The key posture elements include one or a combination of the following: torso angle, arm swing angle, stride length, and landing method;

[0124] Label the historical running posture data and divide them into standard and non-standard categories;

[0125] The preset second largest model is pre-trained using the annotated historical running posture data, so that the preset second largest model can recognize key posture features and associate the key posture features with "standard" or "non-standard" labels.

[0126] As an optional example, the preset second largest model is the InternVL2 large model.

[0127] For example, the student's posture images during running are recorded by the intelligent running device and the image acquisition date is marked. The student's running posture images for 30 consecutive days are obtained (the evaluation defaults to the highest option when there is no record). Each running posture image is identified by the fine-tuned InternVL2 large model to output whether the runner's posture meets the standard. The number of images that meet the standard n1 and the number of images that do not meet the standard n2 are recorded, and the posture standard evaluation is calculated according to the above expression 3. For example, for a certain athlete, the posture standard evaluation result g1 = "Standard posture, slight room for improvement" is calculated.

[0128] The fine-tuning process of the InternVL2 large model is as follows:

[0129] Collect a large amount of running posture data (for example, more than 10,000 pieces), annotate the collected data, and annotate the key posture elements of the runner, including torso angle, arm swing angle, stride, landing method, etc. At the same time, divide the data into "standard" and "non-standard" categories. Use the collected standard and non-standard running posture data to fine-tune the multimodal model so that it can recognize and distinguish the subtle differences in running posture. During the fine-tuning process, ensure that the model can recognize key posture features and associate these features with "standard" or "non-standard" labels.

[0130] As an optional example, the explosive power evaluation result g2 is determined according to the following method:

[0131] Record the athlete's sprint scores for N2 consecutive days g2 i , where i is the serial number of the sprint event score, i = 1, 2, 3, ..., n3, and n3 is the number of sprint event scores of the athlete for N2 consecutive days;

[0132] Determine the explosive force evaluation result g2 according to the following expression 4:

[0133]

[0134] Among them, g 21 is the explosive power evaluation score, and g 21 Determined by the following expression 5:

[0135]

[0136] Among them, g21 i Score for the partition, and g21 i Determined by the following expression six:

[0137]

[0138] Wherein N2 is an integer greater than or equal to 20.

[0139] For example, a student's scores for 50m, 50m*8, 60m, 100m, 200m and other sprint events are recorded through a smart running device, and the student's sprint results for 30 consecutive days are obtained (the evaluation defaults to the highest option when there is no record), and the explosive power evaluation result is determined according to the above expression 4. Assume that for a certain athlete, the calculated explosive power evaluation result is g2 = "strong explosive power, very good at sprinting".

[0140] As an optional example, the endurance evaluation result g3 is determined according to the following method:

[0141] Record the athlete's long-distance running scores for N3 consecutive days g3 j , where j is the serial number of the long-distance running event score, j = 1, 2, 3, ..., n4, and n4 is the number of long-distance running event scores of the athlete for N3 consecutive days;

[0142] The endurance evaluation result g3 is determined according to the following expression 7:

[0143]

[0144] Among them, g 31 is the endurance evaluation score and is determined by the following expression:

[0145]

[0146] Wherein, N3 is an integer greater than or equal to 20.

[0147] For example, obtain a student's long-distance running results for 30 consecutive days (the evaluation defaults to the highest option when there is no record), and then use the above expression 7 to determine the endurance evaluation result g3. Assume that a student's endurance evaluation result g3 = "poor endurance, fatigue in the latter part of long-distance running".

[0148] As an optional example, the running skill evaluation result g4 is determined according to the following method:

[0149] Record the time t of the athlete completing each lap in the N4-day long-distance running event m , the corresponding total number of laps n5, the distance of each lap l1, m = 1, ..., n5, is the sequence number of the completed laps;

[0150] when hour:

[0151] g4 = steady pace, good technique;

[0152] when and and hour:

[0153] g4 = very unstable pace control, large fluctuations, lack of running skills;

[0154] In other cases:

[0155] g4 = Pace fluctuates, running technique needs to be further improved;

[0156] Among them, ts4 is the set fourth judgment threshold, ts5 is the set fifth judgment threshold, and ts6 is the set sixth judgment threshold.

[0157] Wherein, N4 is an integer greater than or equal to 20; p=1,...,n5 is the sequence number of completed circles.

[0158] For example, a smart running device can be used to record the time it takes an athlete to complete each lap of a 400-meter, 800-meter, 1000-meter, or 3000-meter long-distance running event. m , and the corresponding total number of laps n5, the distance of each lap l1, m = 1, ..., n5 is the serial number of the completed laps (the evaluation defaults to the highest option when there is no record), and the running skill evaluation result is calculated according to the above steps. Assume that the calculated running skill result of the athlete is g4 = "The pace fluctuates, and the running skill needs to be further improved."

[0159] As an optional example, the training quality evaluation result g5 is determined according to the following method:

[0160] Record the athlete's running performance for N5 consecutive days p, p=1,...,n3+n4..., is the running sequence number, the smaller the sequence number, the earlier the corresponding training record time, n3 is the number of points scored by the athlete in the sprint event for N2 consecutive days, n4 is the number of points scored by the athlete in the long-distance event for N3 consecutive days, where N2 is an integer greater than or equal to 20, and N3 is an integer greater than or equal to 20;

[0161] The training quality evaluation result g5 is determined according to the following expression 9:

[0162] g5=g 51 +g 52 +g 53 ; (Expression 9)

[0163] Among them, expression nine represents three strings g 51 , g 52 and g 53 The concatenation processing is the same as that of expression 1 and will not be repeated here.

[0164] Among them, g 51 For quantitative evaluation, g 52 For quality evaluation, g 53 is an effect evaluation, and satisfies the following Expression 10, Expression 11, Expression 12 and Expression 13:

[0165]

[0166] Among them, ts7 is the set seventh judgment threshold, ts8 is the set eighth judgment threshold, ts9 is the set ninth judgment threshold, and ts 10 is the tenth judgment threshold, ts 11 is the eleventh judgment threshold, ts 12 is the twelfth judgment threshold, ts 13 is the thirteenth judgment threshold, ts 14 is the fourteenth judgment threshold, ts 15 is the fifteenth judgment threshold, ts 16 is the set sixteenth judgment threshold, ts 17 is the set seventeenth judgment threshold.

[0167] Among them, N5 is an integer greater than or equal to 20; q=1,...,n3+n4 is the running sequence number.

[0168] For example, the student's running training results are recorded through the intelligent running device, and the student's running results for 60 consecutive days are taken, including 20 days of short-distance running results and 40 days of long-distance running results, and then the student's training quality evaluation result g5 is calculated according to the above expression 9. Assume that the student's training quality evaluation result is g5 = "high training volume # high training quality # high training effect".

[0169] In the embodiment of the present invention, the historical multimodal data of the athlete's running is recorded and collected, and the training plan corresponding to each multimodal data is recorded, and the pre-trained large model is loaded to construct a fine-tuning data set for running training guidance, which includes multimodal data in various running situations and corresponding training guidance labels. Then the multimodal data of the athlete is recorded and input into the pre-trained large model to obtain the corresponding running training plan, thereby providing personalized and real-time running training guidance and achieving scientific and efficient physical fitness improvement.

[0170] Device Embodiment

[0171] In a second aspect, another specific embodiment of the present invention discloses a running training device based on a large model, such as Figure 3 As shown, including:

[0172] The model pre-training module 301 is configured to pre-train a preset first large model;

[0173] A data recording module 302 configured to record multimodal data of an athlete;

[0174] The operation module 303 is configured to input the multimodal data into the pre-trained first large model to obtain a running training program for the athlete;

[0175] The pre-training of the preset first large model includes:

[0176] Record historical multimodal data of athletes’ running;

[0177] Recording the training scheme corresponding to each piece of data in the historical multimodal data;

[0178] Importing the historical multimodal data and the training program into a preset first large model to construct a fine-tuning data set for the running training program;

[0179] The multimodal data includes one or a combination of the following: running ability evaluation results, height, weight, latest running performance score, gender, and age.

[0180] As an optional example, the operation module 303 is further configured to determine the running ability evaluation result M1 according to the following method:

[0181] The running ability evaluation result M1 is determined according to the following expression:

[0182] M1 = "Running ability evaluation:" + g1 + g2 + g3 + g4 + g5;

[0183] Among them, g1 is the posture standard evaluation result, which is used to indicate whether the runner's posture meets the standard;

[0184] g2 is the explosive power evaluation result, which is used to indicate the running explosive power level of the athlete;

[0185] g3 is the endurance evaluation result, which is used to indicate the running endurance level of the athlete;

[0186] g4 is a running skill evaluation result, which is used to indicate the running skill level of the athlete;

[0187] g5 is the training quality evaluation result, which is used to indicate the running training level of the athlete;

[0188] Among them, the running ability evaluation result M1 is a string data, the + sign in the expression represents the concatenation of the string, and g1, g2, g3, g4 and g5 are string data.

[0189] As an optional example, the operation module 303 is further configured to determine the posture standard evaluation result g1 according to the following method:

[0190] Pre-train the preset second largest model using the athlete’s historical running posture data;

[0191] Record the running posture images of athletes for N1 consecutive days, use the pre-trained second largest model to recognize each running posture image, output whether the runner's posture meets the standard, and record the number of images that meet the standard n1 and the number of images that do not meet the standard n2;

[0192] The attitude standard evaluation result g1 is determined according to the following expression:

[0193]

[0194] Wherein ts1 is a set first judgment threshold, ts2 is a set second judgment threshold, ts3 is a set third judgment threshold, and N1 is an integer greater than or equal to 20;

[0195] Pre-training the second largest model using historical running posture data of athletes includes:

[0196] Record the athlete's historical running posture data, annotate the historical running posture data, and annotate the athlete's key posture elements. The key posture elements include one or a combination of the following: torso angle, arm swing angle, stride length, and landing method;

[0197] Label the historical running posture data and divide them into standard and non-standard categories;

[0198] The preset second largest model is pre-trained using the annotated historical running posture data, so that the preset second largest model can recognize key posture features and associate the key posture features with "standard" or "non-standard" labels.

[0199] As an optional example, the operation module 303 is further configured to determine the explosive force evaluation result g2 according to the following method:

[0200] Record the athlete's sprint scores for N2 consecutive days g2 i , where i is the serial number of the sprint event score, i = 1, 2, 3, ..., n3, and n3 is the number of sprint event scores of the athlete for N2 consecutive days;

[0201] Determine the explosive power evaluation result g2 according to the following expression:

[0202]

[0203] Among them, g 21 is the explosive power evaluation score, and g 21 Determined by the following expression:

[0204]

[0205] Among them, g21 i Score for the partition, and g21 i Determined by the following expression:

[0206]

[0207] Wherein N2 is an integer greater than or equal to 20.

[0208] As an optional example, the operation module 303 is further configured to determine the endurance evaluation result g3 according to the following method:

[0209] Record the athlete's long-distance running scores for N3 consecutive days g3 j , where j is the serial number of the long-distance running event score, j = 1, 2, 3, ..., n4, and n4 is the number of long-distance running event scores of the athlete for N3 consecutive days;

[0210] The endurance evaluation result g3 is determined according to the following expression:

[0211]

[0212] Among them, g 31 is the endurance rating score and is determined by the following expression:

[0213]

[0214] Wherein, N3 is an integer greater than or equal to 20;

[0215] As an optional example, the operation module 303 is further configured to determine the running skill evaluation result g4 according to the following method:

[0216] Record the time t of the athlete completing each lap in the N4-day long-distance running event m , the corresponding total number of laps n5, the distance of each lap l1, m = 1, ..., n5 ..., is the sequence number of the completed laps;

[0217]

[0218] g4 = steady pace, good technique;

[0219] when and and hour:

[0220] g4 = very unstable pace control, large fluctuations, lack of running skills;

[0221] In other cases:

[0222] g4 = Pace fluctuates, running technique needs to be further improved;

[0223] Wherein ts4 is the set fourth judgment threshold, ts5 is the set fifth judgment threshold, and ts6 is the set sixth judgment threshold;

[0224] Wherein, N4 is an integer greater than or equal to 20; p=1,...,n5 is the sequence number of completed circles.

[0225] As an optional example, the operation module 303 is further configured to determine the training quality evaluation result g5 according to the following method:

[0226] Record the athlete's running performance for N5 consecutive days p , p=1,...,n3+n4…, is the running sequence number, the smaller the sequence number, the earlier the corresponding training record time, n3 is the number of points scored by the athlete in the sprint event for N2 consecutive days, n4 is the number of points scored by the athlete in the long-distance event for N3 consecutive days, where N2 is an integer greater than or equal to 20, and N3 is an integer greater than or equal to 20;

[0227] The training quality evaluation result g5 is determined according to the following expression:

[0228] g5=g 51 +g 52 +g53 , where the + sign indicates string concatenation;

[0229] Among them, g 51 For quantitative evaluation, g 52 For quality evaluation, g 53 is the effect evaluation and satisfies the following expression:

[0230]

[0231]

[0232] Among them, ts7 is the set seventh judgment threshold, ts8 is the set eighth judgment threshold, ts9 is the set ninth judgment threshold, and ts 10 is the tenth judgment threshold, ts 11 is the eleventh judgment threshold, ts 12 is the twelfth judgment threshold, ts 13 is the thirteenth judgment threshold, ts 14 is the fourteenth judgment threshold, ts 15 is the fifteenth judgment threshold, ts 16 is the set sixteenth judgment threshold, ts 17 is the set seventeenth judgment threshold.

[0233] Among them, N5 is an integer greater than or equal to 20; q=1,...,n3+n4 is the running sequence number.

[0234] It should be noted that the running training device provided in the second aspect and the running training method provided in the first aspect belong to the same inventive concept, solve the same technical problem, and achieve the same technical effect, which will not be described in detail here.

[0235] In a third aspect, the present invention proposes a running training device based on a large model, such as Figure 4 As shown, it includes: a memory and one or more processors.

[0236] One or more application programs are stored in the memory, and the one or more application programs are suitable for being executed by the one or more processors to implement the running training device method described in the first aspect.

[0237] like Figure 4 As shown, the electronic device includes: a processor 401 and a memory 402. The processor 401 and the memory 402 are connected, for example, through a bus interface.

[0238] The structure of the electronic device does not constitute a limitation on the embodiments of the present invention.

[0239] Processor 401 may be a CPU, a general purpose processor, a DSP, an ASIC, an FPGA or other programmable logic device, a transistor logic device, a hardware component or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 401 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0240] The bus interface may include a path to transfer information between the above components. The bus interface may be a PCI bus or an EISA bus, etc. The bus interface may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0241] The memory 402 may be a ROM or other type of static storage device that can store static information and instructions, a RAM or other type of dynamic storage device that can store information and instructions, or an EEPROM, a CD-ROM or other optical disk storage, an optical disk storage (including a compressed optical disk, a laser disk, an optical disk, a digital versatile disk, a Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0242] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program can be loaded and executed by a processor to perform the running training method described in the first aspect.

[0243] The applicant of the present invention has made a detailed explanation and description of the implementation examples of the present invention in conjunction with the drawings in the specification. However, those skilled in the art should understand that the above implementation examples are only preferred implementation schemes of the present invention, and the detailed description is only to help readers better understand the spirit of the present invention, but not to limit the scope of protection of the present invention. On the contrary, any improvements or modifications based on the inventive spirit of the present invention should fall within the scope of protection of the present invention.

[0244] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the above embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

Claims

1. A running training method based on a large model, characterized in that: include: S1, pre-training the preset first model; S2, records multimodal data of athletes; S3, inputting the multimodal data into the pre-trained first large model to obtain a running training program for the athlete; The pre-training of the preset first large model includes: Record historical multimodal data of athletes’ running; Recording the training scheme corresponding to each piece of data in the historical multimodal data; Importing the historical multimodal data and the training program into a preset first large model to construct a fine-tuning data set for the running training program; The multimodal data includes one or a combination of the following: running ability evaluation results, height, weight, latest running performance score, gender, and age.

2. A running training method according to claim 1, characterized in that: The running ability evaluation result is determined according to the following method: The running ability evaluation result M1 is determined according to the following expression: M1 = "Running ability evaluation:" + g1 + g2 + g3 + g4 + g5; Among them, g1 is the standard evaluation result of posture; g2 is the explosive power evaluation result; g3 is the endurance evaluation result; g4 is the running skill evaluation result; g5 is the training quality evaluation result; Among them, the running ability evaluation result M1 is a string data, the + sign in the expression represents the concatenation of the string, and g1, g2, g3, g4 and g5 are string data.

3. A running training method according to claim 2, characterized in that: The posture standard evaluation result g1 is determined according to the following method: Pre-training a preset second largest model using the athlete's historical running posture data; Recording the running posture images of the athlete for N1 consecutive days, recognizing each running posture image through the pre-trained second largest model, outputting whether the runner's posture meets the standard, and recording the number of images n1 that meet the standard and the number of images n2 that do not meet the standard; The posture standard evaluation result g1 is determined according to the following expression: Wherein ts1 is a set first judgment threshold, ts2 is a set second judgment threshold, ts3 is a set third judgment threshold, and N1 is an integer greater than or equal to 20; The pre-training of the preset second largest model using the historical running posture data of the athlete comprises: Recording the historical running posture data of the athlete, marking the historical running posture data, marking the key posture elements of the athlete, wherein the key posture elements include one or a combination of the following: torso angle, arm swing angle, stride and landing method; Labeling the historical running posture data and dividing them into standard and non-standard categories; The preset second largest model is pre-trained using the annotated historical running posture data, so that the preset second largest model can recognize key posture features and associate the key posture features with "standard" or "non-standard" labels.

4. A running training method according to claim 2, characterized in that: The explosive power evaluation result g2 is determined according to the following method: Record the athlete's sprint scores for N2 consecutive days g2 i , where i is the serial number of the sprint event score, i=1,2,3,...,n3, and n3 is the number of sprint event scores of the athlete for N2 consecutive days; The explosive power evaluation result g2 is determined according to the following expression: Among them, g 21 is the explosive power evaluation score, and g 21 Determined by the following expression: Among them, g21 i Score for the partition, and g21 i Determined by the following expression: Wherein N2 is an integer greater than or equal to 20.

5. A running training method according to claim 2, characterized in that: The endurance evaluation result g3 is determined according to the following method: Record the athlete's long-distance running scores for N3 consecutive days g3 j , where j is the serial number of the long-distance running event score, j=1,2,3,...,n4, and n4 is the number of the athlete's long-distance running event scores for N3 consecutive days; The endurance evaluation result g3 is determined according to the following expression: Among them, g 31 The endurance rating score is determined by the following formula: Wherein, N3 is an integer greater than or equal to 20.

6. A running training method according to claim 2, characterized in that: The running skill evaluation result g4 is determined according to the following method: Record the time t of each lap completed by the athlete in the N4-day long-distance running event m , the corresponding total number of laps n5, the distance of each lap l1, m = 1, ..., n5, is the sequence number of the completed laps; when hour: g4 = steady pace, good technique; when and and hour: g4 = very unstable pace control, large fluctuations, lack of running skills; In other cases: g4 = Pace fluctuates, running technique needs to be further improved; Wherein ts4 is the set fourth judgment threshold, ts5 is the set fifth judgment threshold, and ts6 is the set sixth judgment threshold; Wherein, N4 is an integer greater than or equal to 20; p=1,...,n5 is the sequence number of completed circles.

7. A running training method according to claim 2, characterized in that: The training quality evaluation result g5 is determined according to the following method: Record the athlete's running performance for N5 consecutive days g5 p , p=1,...,n3+n4, is the running sequence number, the smaller the sequence number, the earlier the corresponding training record time, n3 is the number of points scored by the athlete in the sprint event for N2 consecutive days, n4 is the number of points scored by the athlete in the long-distance event for N3 consecutive days, where N2 is an integer greater than or equal to 20, and N3 is an integer greater than or equal to 20; The training quality evaluation result g5 is determined according to the following expression: g5=g 51 +g 52 +g 53 , where the + sign indicates string concatenation; Among them, g 51 For quantitative evaluation, g 52 For quality evaluation, g 53 is the effect evaluation and satisfies the following expression: Among them, ts7 is the set seventh judgment threshold, ts8 is the set eighth judgment threshold, ts9 is the set ninth judgment threshold, and ts 10 is the tenth judgment threshold, ts 11 is the eleventh judgment threshold, ts 12 is the twelfth judgment threshold, ts 13 is the thirteenth judgment threshold, ts 14 is the fourteenth judgment threshold, ts 15 is the fifteenth judgment threshold, ts 16 is the set sixteenth judgment threshold, ts 17 is the set seventeenth judgment threshold; Wherein, N5 is an integer greater than or equal to 20; q=1,...,n3+n4 is the running sequence number.

8. A running training device based on a large model, characterized in that: Used to implement the running training method according to any one of claims 1 to 7, the device comprises: A model pre-training module is configured to pre-train a preset first large model; a data recording module configured to record multimodal data of the athlete; A calculation module is configured to input the multimodal data into the pre-trained first large model to obtain a running training program for the athlete; The pre-training of the preset first large model includes: Record historical multimodal data of athletes’ running; Recording the training scheme corresponding to each piece of data in the historical multimodal data; Importing the historical multimodal data and the training program into a preset first large model to construct a fine-tuning data set for the running training program; The multimodal data includes one or a combination of the following: running ability evaluation results, height, weight, latest running performance score, gender, and age.

9. A running training device based on a large model, characterized in that: including a memory, a processor and a user interface; The memory is used to store computer programs; The user interface is used to interact with the user; The processor is used to read the computer program in the memory, and when the processor executes the computer program, the running training method according to any one of claims 1 to 7 is implemented.

10. A processor-readable storage medium, characterized in that: The processor-readable storage medium stores a computer program, and when the processor executes the computer program, the large model-based running training method as described in any one of claims 1 to 7 is implemented.

Images