CN-121997247-A - Skiing level evaluation method and system based on multisource information fusion

CN121997247ACN 121997247 ACN121997247 ACN 121997247ACN-121997247-A

Abstract

The invention belongs to the technical field of behavior recognition, and provides a skiing level evaluation method and system based on multi-source information fusion, wherein the skiing level evaluation method and system comprises the steps of synchronously acquiring multi-source time sequence data and biological information and carrying out standardized processing when a skier finishes a preset key action; the method comprises the steps of constructing a data set, constructing a multitasking neural network model, synchronously completing a behavior recognition task and an identity recognition task, training the multitasking neural network model by using a training set containing excellent athlete sample information, extracting high-level fusion characteristics of skiers to be evaluated by using the trained multitasking neural network model, calculating similarity between the skiers to be evaluated and preset high-level fusion characteristics of excellent skiers, and evaluating the skiing level according to the similarity. The invention solves the problem that an evaluation system in the prior art cannot fully consider individual differences among skiers.

Inventors

MA LELE
Sun Canhang
LIU HAO
WANG WENRUI
PENG CHENG
CHEN WEIKANG
WANG ZHICHUANG

Assignees

观云(山东)智能科技有限公司

Dates

Publication Date: 20260508
Application Date: 20251222

Claims (10)

1. A skiing level evaluation method based on multi-source information fusion is characterized by comprising the following steps: When a skier finishes a preset key action, synchronously acquiring multi-source time sequence data and biological information, respectively carrying out standardized processing on the multi-source time sequence data and the biological information, dividing the multi-source time sequence data into samples and correlating the samples with corresponding biological information, constructing a data set comprising a time sequence data set formed by the multi-source time sequence data and a biological data set formed by the biological information, and dividing the data set into a training set and a test set according to a proportion; The method comprises the steps of constructing a multi-task neural network model, synchronously completing a behavior recognition task and an identity recognition task, training the multi-task neural network model by using a training set containing excellent athlete sample information, enabling high-level fusion characteristics learned by the multi-task neural network model to reflect skier level differences, extracting the high-level fusion characteristics of a skier to be evaluated by using the trained multi-task neural network model, calculating the similarity between the skier to be evaluated and the preset high-level fusion characteristics of the excellent skier, and evaluating the skiing level according to the similarity.
2. The skiing level assessment method based on multi-source information fusion according to claim 1, wherein the key actions comprise turning, sliding and jumping, the corresponding key action types are used as supervision information for model training, the multi-source time sequence data are acquired by an inertial measurement unit, a magnetometer and a GPS and comprise acceleration, angular velocity, magnetic field, longitude and latitude, altitude and speed information, and the biological information comprises height, weight, age and leg-body ratio.
3. The method for evaluating skiing level based on multi-source information fusion according to claim 1, wherein the standardized multi-source time series data is divided into samples using a sliding window with a fixed size, the window size is 100, and the step size is 50.
4. The skiing level evaluation method based on multi-source information fusion according to claim 3, wherein the segmented samples are marked, specifically, a behavior recognition task is used for recognizing key behaviors in the skiing process and marking the key behaviors by using key behavior categories, and an identity recognition task is used for reflecting the relation between the completion level of the skiing behaviors and individual differences and marking the key behaviors by using personnel numbers.
5. The skiing level evaluation method based on multi-source information fusion according to claim 1, wherein the multi-task neural network model comprises a multi-source time sequence data processing branch and a biological information processing branch, the multi-source time sequence data processing branch comprises a time sequence feature extraction layer, an attention layer, a bidirectional circulating neural network layer and a full-connection layer which are sequentially connected, the full-connection layer comprises a first full-connection layer and a second full-connection layer, the biological information processing branch is an independent full-connection network and comprises a third full-connection layer and a fourth full-connection layer, and the biological information processing branch is used for high-level semantic extraction of biological information.
6. The skiing level evaluation method based on multi-source information fusion according to claim 5, wherein the output of the first full-connection layer is fused with the high-level semantic features output by the third full-connection layer, the fused features are processed by the fourth full-connection layer, the high-level fusion features are finally output as the input of an identification task, and the time sequence features output by the second full-connection layer are used as the input of a behavior identification task.
7. The skiing level evaluation method based on multi-source information fusion according to claim 1, wherein in the training process of the multi-task neural network model, loss functions of different tasks are calculated, and feedback learning is performed, so that the feature similarity between high-level fusion features extracted by the multi-task neural network model satisfies the condition that feature similarity between skiers at the same level is higher than feature similarity between skiers at different levels.
8. A skiing level assessment system based on multi-source information fusion, comprising: The data set construction module is configured to synchronously acquire multi-source time sequence data and biological information when a skier finishes a preset key action, respectively perform standardized processing on the multi-source time sequence data and the biological information, divide the multi-source time sequence data into samples and correlate the samples with corresponding biological information, construct a data set comprising a time sequence data set formed by the multi-source time sequence data and a biological data set formed by the biological information, and divide the data set into a training set and a test set according to proportion; The skiing level evaluation module is configured to construct a multi-task neural network model, synchronously complete a behavior recognition task and an identity recognition task, train the multi-task neural network model by using a training set containing excellent athlete sample information, enable high-level fusion characteristics learned by the multi-task neural network model to reflect skier level differences, extract high-level fusion characteristics of skiers to be evaluated by using the trained multi-task neural network model, calculate similarity between the skiers to be evaluated and preset high-level fusion characteristics of excellent skiers, and evaluate skiing level according to the similarity.
9. A computer readable storage medium having stored thereon a program, which when executed by a processor, implements the steps of a method of assessing skiing level based on multi-source information fusion according to any one of claims 1 to 7.
10. An electronic device comprising a memory, a processor and a program stored on the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of a skiing level assessment method based on multi-source information fusion according to any one of claims 1-7.

Description

Skiing level evaluation method and system based on multisource information fusion Technical Field The invention belongs to the technical field of behavior recognition, and particularly relates to a skiing level evaluation method and system based on multi-source information fusion. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. In recent years, skiing sports have been favored by more and more people, and the number of participants has been significantly increased. Among them, the skiing beginners occupy the vast majority, and although the beginners are enthusiasm for skiing, due to lack of experience and professional guidance, the beginners often have difficulty in showing a good technical level on the snow field, and even may cause sports injury due to improper operation. Under the background, skiing coaches play a key role, however, the current coaching resources have obvious defects in quantity and energy, and the personalized requirements of vast beginners are difficult to meet. With the continuous development and wide application of computer technology, it has become possible to construct such a virtual trainer system. The virtual trainer system can be used to evaluate the skill level of skiers, identify deficiencies in their movements, and provide improvement advice and training programs to help beginners quickly raise skill levels, reducing the risk of athletic injuries. Although many researches are carried out on skiing level evaluation or action analysis at present, the evaluation system has certain limitation that a part of evaluation methods are relatively rough and can only give wide level grades, and the other part only focuses on comparison with standard actions, so that individual differences among skiers cannot be fully considered, and the accuracy and the comprehensiveness of evaluation results are influenced. Disclosure of Invention The invention aims to provide a skiing level evaluation method and system based on multi-source information fusion, which synchronously complete a behavior recognition task and an identity recognition task and extract high-level fusion characteristics based on a multi-task neural network model so as to solve the problem that an evaluation system in the prior art cannot fully consider individual differences among skiers. In order to achieve the above purpose, the invention adopts the following technical scheme: the first aspect of the invention provides a skiing level evaluation method based on multi-source information fusion, which comprises the following steps: When a skier finishes a preset key action, synchronously acquiring multi-source time sequence data and biological information, respectively carrying out standardized processing on the multi-source time sequence data and the biological information, dividing the multi-source time sequence data into samples and correlating the samples with corresponding biological information, constructing a data set comprising a time sequence data set formed by the multi-source time sequence data and a biological data set formed by the biological information, and dividing the data set into a training set and a test set according to a proportion; The method comprises the steps of constructing a multi-task neural network model, synchronously completing a behavior recognition task and an identity recognition task, training the multi-task neural network model by using a training set containing excellent athlete sample information, enabling high-level fusion characteristics learned by the multi-task neural network model to reflect skier level differences, extracting the high-level fusion characteristics of a skier to be evaluated by using the trained multi-task neural network model, calculating the similarity between the skier to be evaluated and the preset high-level fusion characteristics of the excellent skier, and evaluating the skiing level according to the similarity. The multi-source time sequence data are acquired by an inertial measurement unit, a magnetometer and a GPS, and comprise acceleration, angular velocity, a magnetic field, longitude and latitude, altitude and speed information, and the biological information comprises height, weight, age and leg-body ratio. As a further technical solution, a sliding window with a fixed size is used to divide the normalized multi-source time sequence data into samples, the window size is 100, and the step length is 50. The method comprises the steps of dividing a sample into a plurality of samples, marking the samples after division, specifically, identifying a key behavior in the skiing process by a behavior identification task, marking by using a key behavior category, and marking by using a person number by an identity identification task, wherein the person identification task is used for reflecting the relation between the completion level of the skiing behavio