Teenager concentration training system and method

Abstract

The invention provides a teenager concentration training system and method, and the method comprises the steps: carrying out the gaze cluster clustering of eye movement data when a target teenager is subjected to a visual search task, and obtaining a search mode feature; determining a concentration attenuation trend based on the ability portraits of the target teenagers for concentration training and the search mode features; performing similarity matching on the action trajectory of the target teenager in the concentration training process and a standard action trajectory of concentration training to obtain a behavior cooperation coefficient; and generating an adaptive regulation factor of concentration training based on the concentration attenuation trend and the behavior cooperation coefficient, and adjusting the density and moving speed of interference elements in the visual search task in the next concentration training stage according to the adaptive regulation factor. By adopting the scheme of the invention, dynamic regulation and control of training task parameters can be realized by constructing a collaborative perception mechanism of eye movement-limb multi-mode man-machine interaction data so as to improve the training efficiency of teenager concentration.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of human-computer interaction, and more particularly, to a teenager concentration training system and method. BACKGROUND

[0002] With the deepening of digital education and the universalization of the problem of insufficient concentration of teenagers, concentration training has become one of the core needs in the field of quality education and development of teenagers. At present, human-computer interaction technology is gradually integrated into the concentration training scene. By collecting physiological or behavioral data of teenagers during training, the state of concentration is assisted to be judged, and training is carried out combined with visual search and other interactive tasks, which provides technical support for improving the concentration of teenagers.

[0003] However, the existing teenager concentration training method has obvious limitations in implementation: on the one hand, the existing teenager concentration training method relies on single modal data (such as only collecting eye movement data) for state evaluation, and fails to combine the cognitive level of concentration decay law with the body level of action coordination, resulting in one-sided perception of training state and inability to accurately reflect the real concentration level; on the other hand, the core parameters of the training task in the existing teenager concentration training method are mostly fixed preset values, or are only adjusted according to a single index (such as task completion rate), which cannot dynamically adapt to the real-time concentration state and action coordination degree of individuals, resulting in poor training targeting and low efficiency, and it is difficult to meet the individual training needs of different teenagers. Therefore, how to build a collaborative perception mechanism of eye movement-limb multi-modal human-computer interaction data to realize dynamic regulation of training task parameters and improve the training efficiency of teenagers' concentration has become a difficult problem in the industry. SUMMARY

[0004] The present application provides a teenager concentration training system and method, which can realize dynamic regulation of training task parameters by building a collaborative perception mechanism of eye movement-limb multi-modal human-computer interaction data to improve the training efficiency of teenagers' concentration.

[0005] In a first aspect, the present application provides a teenager concentration training method based on human-computer interaction, which adaptively trains the concentration of a target teenager according to the behavior performance data in the concentration training process. The method comprises: Collecting eye movement data of the target teenager when completing a visual search task in the concentration training process; Based on the eye movement data, clustering the fixation clusters to obtain the search pattern features of the target teenager when performing visual search in the concentration training process; construct an ability portrait of the target teenager for concentration training based on historical training data of the target teenager, and then determine a concentration decay trend of the target teenager for concentration training based on the ability portrait and the search pattern feature; obtain body movement data of the target teenager in the process of concentration training, perform similarity matching on action trajectories in the body movement data and standard action trajectories of concentration training, and obtain a behavior coordination coefficient of the target teenager for concentration training; perform state perception fusion on the concentration decay trend and the behavior coordination coefficient to generate an adaptive regulation factor of the target teenager in the process of concentration training, and then adjust the density and moving speed of interference elements in the visual search task of the target teenager in the next concentration training stage according to the adaptive regulation factor.

[0006] In some embodiments, the search pattern feature of the target teenager for visual search in the process of concentration training obtained based on the eye movement data specifically includes: perform spatial clustering on the eye movement data to generate a plurality of fixation clusters; determine an average fixation duration, an intra-cluster fixation density, and an inter-cluster jump rate of the target teenager in the process of concentration training according to all fixation clusters; determine the search pattern feature of the target teenager for visual search in the process of concentration training based on the average fixation duration, the intra-cluster fixation density, and the inter-cluster jump rate.

[0007] In some embodiments, the ability portrait of the target teenager for concentration training constructed based on the historical training data of the target teenager specifically includes: obtain historical training data of the target teenager for concentration training; align the historical training data in the time sequence feature to obtain a training sample set that is spatio-temporally anchored and unified in feature dimension; generate the ability portrait of the target teenager for concentration training based on the training sample set.

[0008] In some embodiments, the determination of the concentration decay trend of the target teenager for concentration training based on the ability portrait and the search pattern feature specifically includes: align the search pattern feature to the feature space of the ability portrait through cross-modal feature mapping, and then obtain a spatio-temporally coupled fusion feature vector; perform iterative deduction based on the fusion feature vector to obtain a concentration decay curve of the target teenager's concentration dynamic evolution with training time; determine the concentration decay trend of the target teenager for concentration training according to the concentration decay curve.

[0009] In some embodiments, the action trajectory in the body movement data is similar matched with a standard action trajectory of the concentration training, to obtain a behavior coordination coefficient of the target teenager when performing the concentration training, specifically comprising: determining an action trajectory of the target teenager when performing the concentration training according to the time sequence coordinate sequence of the skeletal nodes in the body movement data; determining a standard action trajectory of the concentration training; spatiotemporal alignment is performed on the action trajectory and the standard action trajectory, to obtain a behavior coordination coefficient of the target teenager when performing the concentration training.

[0010] In some embodiments, the state-aware fusion is performed on the concentration decay trend and the behavior coordination coefficient, to generate an adaptive regulation factor of the target teenager in the concentration training process, specifically comprising: extracting a state-aware parameter of the target teenager when performing the concentration training in the concentration decay trend; state fusion is performed based on the state-aware parameter and the behavior coordination coefficient, to obtain a dimension-unified bimodal fusion feature; the density adjustment coefficient and the rate correction coefficient of the interference element in the concentration training process of the target teenager are generated through the bimodal fusion feature; an adaptive regulation factor of the target teenager in the concentration training process is determined according to the density adjustment coefficient and the rate correction coefficient.

[0011] In some embodiments, the eye movement data of the target teenager when performing the visual search task in the concentration training process is collected through a non-invasive infrared eye movement tracking device.

[0012] In a second aspect, the present application provides a teenager concentration training system, comprising: a collection module configured to collect eye movement data of a target teenager when performing a visual search task in a concentration training process; a processing module configured to perform gaze clustering based on the eye movement data, to obtain search pattern features of the target teenager when performing the visual search in the concentration training process; the processing module is configured to construct an ability portrait of the target teenager performing the concentration training through historical training data of the target teenager, and then determine a concentration decay trend of the target teenager when performing the concentration training based on the ability portrait and the search pattern features; the processing module is configured to obtain body movement data of the target teenager in the concentration training process, and perform similar matching on an action trajectory in the body movement data and a standard action trajectory of the concentration training, to obtain a behavior coordination coefficient of the target teenager when performing the concentration training; The execution module is configured to perform state-aware fusion on the attention decay trend and the behavior synergy coefficient to generate an adaptive regulation factor of the target teenager in the attention training process, and then adjust the density and moving speed of the interference elements in the visual search task in the next attention training stage according to the adaptive regulation factor.

[0013] In a third aspect, the present application provides a computer device, comprising a memory and a processor, the memory stores a code, and the processor is configured to acquire the code and execute the above-mentioned human-computer interaction-based attention training method for teenagers.

[0014] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the above-mentioned human-computer interaction-based attention training method for teenagers.

[0015] The technical scheme provided by the embodiments of the present application has the following beneficial effects: In the attention training system and method for teenagers provided by the present application, first, eye movement data of a target teenager in a visual search task in an attention training process is collected; fixation clustering is performed based on the eye movement data to obtain search pattern features of the target teenager in the visual search in the attention training process; an ability portrait of the target teenager in the attention training is constructed through historical training data of the target teenager, and then an attention decay trend of the target teenager in the attention training is determined based on the ability portrait and the search pattern features; limb action data of the target teenager in the attention training process is obtained, and action trajectories in the limb action data are similar matched with standard action trajectories of the attention training to obtain a behavior synergy coefficient of the target teenager in the attention training; state-aware fusion is performed on the attention decay trend and the behavior synergy coefficient to generate an adaptive regulation factor of the target teenager in the attention training process, and then the density and moving speed of the interference elements in the visual search task in the next attention training stage are adjusted according to the adaptive regulation factor.

[0016] It can be seen that, according to the adaptive adjustment factor, the density and moving speed of the interference elements in the visual search task of the target teenager in the next attention training stage are adjusted; first, the determination of the attention decay trend can obtain a feature vector representing the development trend type and trend credibility of the attention of the target teenager in the training process with the change of the training time, the determination of the attention decay trend fuses the real-time eye movement data and the ability portrait constructed based on the historical training data through cross-modal feature mapping, and then the cognitive dimension core parameter is obtained through time sequence deduction, which can completely depict the dynamic evolution law of the attention of the target teenager in the training, and fills the gap of the existing scheme for the perception of the cognitive level decay state, so that the state evaluation is upgraded from a single data segment to a cognitive dynamic panorama; then, the determination of the behavior coordination coefficient can obtain an index reflecting the coordination degree between the body action of the target teenager and the standard action, the determination of the behavior coordination coefficient can quantitatively reflect the coordination degree between the body action of the target teenager and the standard action, so as to supplement the body execution state information not covered by the cognitive modal attention decay trend, so that the overall state evaluation is upgraded from cognitive to two dimensions of cognitive and body, completely solving the problem of single modal perception, avoiding the problem of mismatch between body burden and cognitive difficulty caused by adjusting parameters only according to cognitive data, and changing the parameter regulation from single cognitive orientation to cognitive-body dual adaptation orientation, breaking the limitation of static preset, providing complete data support for subsequent accurate judgment of training state; finally, the determination of the adaptive adjustment factor can obtain structured data integrating the training parameter adjustment coefficient and its confidence in the next stage of attention training process, the determination of the adaptive adjustment factor can provide a direct basis for the adjustment of the next stage training parameter, so as to provide a clear regulation direction and amplitude for the training system, and at the same time, by integrating the attention decay law in the cognitive level and the action coordination in the body level into the structured regulation basis, the deviation of single modal data is avoided, the regulation basis is upgraded from single dimension to cognitive-body comprehensive dimension, the problem of single regulation of the existing scheme is completely solved, the dynamic self-adaptation of attention training is realized, and the individual training efficiency is maximized; in summary, based on the above scheme, the dynamic regulation of the training task parameter can be realized through the construction of the collaborative perception mechanism of eye-limb multi-modal human-computer interaction data to improve the training efficiency of the teenager's attention. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is an exemplary flowchart of a human-computer interaction-based adolescent attention training method according to some embodiments of the present application; Figure 2 is an operation flowchart for determining the attention decay trend according to some embodiments of the present application; Figure 3 is an exemplary flowchart of determining the adaptive adjustment factor according to some embodiments of the present application; Figure 4is a structural schematic diagram of a teenager attention training system according to some embodiments of the present application; Figure 5 is an internal structure diagram of a computer device for implementing a teenager attention training method based on human-computer interaction according to some embodiments of the present application. DETAILED DESCRIPTION

[0018] In order to better understand the technical solutions of the present application, the technical solutions of the present application will be described in detail below in combination with the drawings of the specification and specific embodiments.

[0019] Reference Figure 1 The figure is an exemplary flowchart of a teenager attention training method based on human-computer interaction according to some embodiments of the present application, which mainly includes the following steps: In step 101, eye movement data of the target teenager during the visual search task in the attention training process is collected.

[0020] It should be noted that in the present application, the eye movement data is a set of physiological data reflecting the cognitive state changes of the target teenager during the visual search process in the attention training, which can capture the visual focusing features of the target teenager during the attention training process and provide objective basis for analyzing the search pattern and inferring the changes of attention, so that the adaptive adjustment of attention training is more accurate and scientific; when specifically implemented, the collection of eye movement data of the target teenager during the visual search task in the attention training process can be realized in the following manner, that is, the existing non-invasive infrared eye tracking device can be fixed below the training screen to ensure that the lens is directly opposite the eye area of the target teenager, the eyeball is illuminated by infrared illumination technology, and the movement trajectory of the eyeball is recorded by using the corneal reflection principle, so as to collect the pupil diameter change, the time sequence coordinates of the fixation point, the saccade amplitude and the saccade frequency of the target teenager during the visual search process in the attention training with a sampling frequency of 30 frames per second and a sampling area of 1024x768 pixels, and collect all the pupil diameter change, the time sequence coordinates of the fixation point, the saccade amplitude and the saccade frequency of the target teenager during the visual search process in the attention training. The data set is used as eye movement data; wherein, the time sequence coordinates are the positions of the eyeball on the screen at different times; the saccade amplitude is the distance of the rapid movement of the eyeball; and the saccade frequency is the number of saccades per unit time.

[0021] In step 102, gaze clustering is performed based on the eye movement data to obtain the search pattern features of the target teenager during the visual search in the attention training process.

[0022] In some embodiments, the gaze clustering based on the eye movement data to obtain the search pattern features of the target teenager during the visual search in the attention training process can be realized in the following steps: spatially clustering the eye movement data to generate a plurality of gaze clusters; determining, according to all the gaze clusters, an average gaze duration, an intra-cluster gaze density and an inter-cluster jump rate of the target teenager in the concentration training process; determining, based on the average gaze duration, the intra-cluster gaze density and the inter-cluster jump rate, a search pattern feature of the target teenager when performing visual search in the concentration training process.

[0023] In a specific implementation, the spatial clustering of the eye movement data to generate a plurality of gaze clusters can be implemented in the following manner: a density-based spatial clustering algorithm with noise can be used to calculate the Euclidean distance between the time-series coordinates of each gaze point in the eye movement data, and gaze points that are less than the neighborhood radius in spatial distance and continuously appear are divided into the same cluster, and finally a plurality of independent gaze clusters are output, each of which contains the time-series coordinates, occurrence time and duration of all associated gaze points; wherein the neighborhood radius of the clustering algorithm can be dynamically adjusted to 8-12 pixels according to the resolution of the training screen and the accuracy of the eye tracker, and the minimum sample number is set to 5; the gaze cluster is a set formed by clustering gaze points that are adjacent in space and continuous in time, which can reflect the area that the target teenager focuses on in visual search, thereby providing a basic analysis unit for visual focus, so that the spatial feature of the search pattern can be quantified.

[0024] In a specific implementation, the average fixation duration, the intra-cluster fixation density and the inter-cluster jump rate of the target teenager in the concentration training process can be determined according to all fixation clusters in the following manner: first, for each fixation cluster, the sum of the fixation point duration of all fixation points in the fixation cluster is divided by the total number of fixation points in the fixation cluster to obtain the average fixation duration of the fixation cluster. The average fixation duration of each fixation cluster is obtained by the above steps, and the average value of the average fixation duration of all fixation clusters is taken as the average fixation duration of the target teenager in the concentration training process; then, for each fixation cluster, the total number of fixation points in the fixation cluster is divided by the spatial coverage area of the fixation cluster, that is, the area of the minimum circumscribed rectangle formed by the coordinates of all fixation points in the fixation cluster, to obtain the fixation density of each fixation cluster. The fixation density of each fixation cluster is obtained by the above steps, and the average value of the fixation density of all fixation clusters is taken as the intra-cluster fixation density of the target teenager in the concentration training process; finally, the effective saccade number from one fixation cluster to another fixation cluster in a unit time is counted as the inter-cluster jump rate of the target teenager in the concentration training process; wherein the movement distance exceeding 2 times the neighborhood radius is regarded as an effective saccade; the average fixation duration is an index reflecting the fixation duration of the target teenager in the attention area in the visual search of the concentration training, which can quantify the attention depth by duration and assist in judging the concentration degree; the intra-cluster fixation density is an index reflecting the visual focus density of the target teenager in the attention area in the visual search of the concentration training, which can quantify the focus degree of the visual search and provide a basis for evaluating the stability of concentration; the inter-cluster jump rate is an index reflecting the frequency of attention transfer of the target teenager in the visual search process of the concentration training, which reflects the flexibility of attention allocation of the target teenager and assists in inferring the dispersion trend of concentration.

[0025] In a specific implementation, the search mode feature of the target teenager in the visual search process of the concentration training can be determined based on the average fixation duration, the intra-cluster fixation density and the inter-cluster jump rate in the following manner: the average fixation duration, the intra-cluster fixation density and the inter-cluster jump rate can be standardized and normalized to the range of 0-1, respectively, and then a three-dimensional feature vector constructed by the normalized average fixation duration, the intra-cluster fixation density and the inter-cluster jump rate is taken as the search mode feature of the target teenager in the visual search process of the concentration training; wherein when the average fixation duration is greater than 0.5, the intra-cluster fixation density is greater than 0.6 and the inter-cluster jump rate is less than 0.4, it is determined that the target teenager is in a concentrated search mode; when the average fixation duration is less than 0.3, the intra-cluster fixation density is less than 0.4 and the inter-cluster jump rate is greater than 0.6, it is determined that the target teenager is in a dispersed search mode; and the remaining combinations correspond to a transitional search mode.

[0026] It should be noted that in the present application, the search mode feature is a feature reflecting the visual search strategy of the target teenager in the visual search process of concentration training, which fully characterizes the cognitive mode of visual search and can provide a core basis for determining the decay trend.

[0027] In step 103, the ability portrait of the target teenager for concentration training is constructed through the historical training data of the target teenager, and then the concentration decay trend of the target teenager for concentration training is determined based on the ability portrait and the search mode feature.

[0028] In some embodiments, the ability portrait of the target teenager for concentration training constructed through the historical training data of the target teenager can be implemented by the following steps: Obtain the historical training data of the target teenager for concentration training; Align the time sequence features of the historical training data to obtain a training sample set with spatiotemporal anchoring and unified feature dimensions; Generate the ability portrait of the target teenager for concentration training based on the training sample set.

[0029] It should be noted that in the present application, the historical training data is a multi-modal data set generated by the past concentration training of the target teenager, including the search mode feature, the task completion accuracy, the training duration and the number of interruptions of each training, which provides a historical reference for the concentration training of the target teenager and provides an original basis for subsequent data processing and model training, ensuring the individualization of the ability portrait construction and the integrity of the data support.

[0030] In specific implementation, the historical training data of the target teenager for concentration training can be implemented in the following way, that is, the historical data of the target teenager for concentration training can be extracted from the background database of the concentration training system, including the search mode feature sequence, the task completion accuracy, and the training duration and the number of interruptions recorded automatically by the system for each training, and then all the extracted historical data are sorted by training date to form the historical training data of the target teenager for concentration training.

[0031] In a specific implementation, the time sequence feature alignment of the historical training data can be achieved in the following manner: a dynamic time warping algorithm can be used to stretch or compress the time axis to make the lengths of the historical data of different training durations consistent, and then a sliding window with a step of 1 minute is used to slice the aligned historical data to extract the search pattern features and the task completion rates in each window, and the extracted search pattern features and the task completion rates are standardized and uniformly mapped to the interval [-1, 1] by Z standardization, thereby generating a training sample set that is time and space anchored and has uniform feature dimensions and contains timestamp labels. The training sample set is a structured data set that is time and space anchored and has uniform dimensions after the historical training data is aligned by dynamic time warping, feature extraction, and standardization. The training sample set can eliminate the time sequence differences and dimension inconsistencies in the historical training data, improve the accuracy of model training, and provide a qualified data basis for the generation of the ability portrait.

[0032] In a specific implementation, the ability portrait of the target teenager for concentration training can be generated based on the training sample set in the following manner: an attention-enhanced long short-term memory network can be initialized, the feature sequence in the training sample set can be taken as input, and the ability portrait of the target teenager for concentration training can be generated. The hidden layer of the long short-term memory network can calculate the weights of each feature in the feature sequence through a self-attention mechanism, for example, the search pattern feature weight is 0.5-0.6, and the task completion feature weight is 0.3-0.4. The output layer can be mapped to the core parameters of concentration, such as the concentration base value, the fatigue threshold, and the interference sensitivity, through a fully connected layer. The model can be trained using 70% of the data in the training sample set and validated using 30% of the data. The model can be optimized by adjusting the learning rate and the number of iterations, and finally the concentration trait map containing the dynamic change curve of the core parameters can be output as the ability portrait of the target teenager for concentration training.

[0033] It should be noted that in this application, the ability portrait is a dynamic trait map that characterizes the individual traits of the target teenager's concentration. The ability portrait is automatically updated by retraining with the latest data after each new training, and can accurately reflect the individual's concentration characteristics, providing personalized basis for subsequent determination of concentration decay trend and development of adaptive training strategies, and ensuring the pertinence of training adjustment.

[0034] In some embodiments, with reference to Figure 2 The figure is an operation flow chart for determining the concentration decay trend according to some embodiments of the present application. The concentration decay trend of the target teenager for concentration training can be determined based on the ability portrait and the search pattern feature in the following steps: align the search pattern features to a feature space of the capability portrait through cross-modal feature mapping, and then obtain a spatio-temporal coupling fusion feature vector; perform iterative deduction based on the fusion feature vector to obtain a concentration decay curve of the target adolescent concentration over time; determine a concentration decay trend of the target adolescent during concentration training according to the concentration decay curve.

[0035] In a specific implementation, the aligning of the search pattern features to a feature space of the capability portrait through cross-modal feature mapping, and then obtaining a spatio-temporal coupling fusion feature vector can be implemented in the following manner: a dynamic feature adaptation matrix algorithm can be used to adapt the feature values in the search pattern features to the core parameters in the capability portrait, to generate a spatio-temporal coupling fusion feature vector; wherein the coupling coefficient of the feature values in the search pattern features and the core parameters in the capability portrait can be set to 0.65, and a nonlinear activation function (such as a ReLU function) can be used to weight and fuse the two types of features, to eliminate the dimension difference between the features, and thus output a heterogeneous fusion feature vector containing a training time stamp; wherein the fusion feature vector is a structured vector containing spatio-temporal correlation information between the visual search behavior of the target adolescent and the individual concentration trait, which can eliminate the dimension difference and information island problem of features from different sources, provide a unified and comprehensive feature input for the concentration decay trend prediction, and improve the initial accuracy of the decay trend prediction.

[0036] In a specific implementation, the performing iterative deduction based on the fusion feature vector to obtain a concentration decay curve of the target adolescent concentration over time can be implemented in the following manner: a pre-trained time-series decay deduction model (such as a deduction model constructed based on a gated recurrent unit and a self-attention mechanism) can be loaded, and the fusion feature vector can be used as the input of the time-series decay deduction model, with the fatigue threshold in the capability portrait of the target adolescent as the model constraint boundary; the update gate and the reset gate of the gated recurrent unit can be used to control the forgetting and retention of feature information, and the self-attention mechanism can be used to strengthen the weight proportion of key features in the feature information, such as the inter-cluster jump rate in the search pattern features; after 100 iterations, a time-series curve with the training time on the horizontal axis and the concentration value on the vertical axis is output as the concentration decay curve; wherein the concentration value ranges from 0 to 10; the concentration decay curve is a time-series curve reflecting the dynamic change of the concentration value of the target adolescent over time during the concentration training, which can visualize the abstract concentration, intuitively present the evolution process of the concentration of the target adolescent, and facilitate the extraction of key parameters to provide visual basis for subsequent trend judgment.

[0037] In a specific implementation, the determination of the concentration attenuation trend of the target teenager during the concentration training according to the concentration attenuation curve can be implemented in the following manner: first, the attenuation acceleration can be obtained by calculating the difference between the concentration reduction rates within adjacent 5 minutes in the concentration attenuation curve, and the longest duration of the concentration value maintained above 6.0 in the concentration attenuation curve is calculated as the tolerance limit duration; then, three threshold intervals of the concentration attenuation curve attenuation acceleration can be set, for example, the attenuation acceleration less than 0.2 is stable attenuation, the attenuation acceleration between 0.2-0.5 is moderate attenuation, and the attenuation acceleration greater than 0.5 is accelerated attenuation, and the type corresponding to the interval in which the attenuation acceleration falls is taken as the development trend type of the target teenager's concentration training, then the fitting residual of the concentration attenuation curve is calculated by the mean square error, the statistical significance coefficient of the attenuation acceleration is verified by the t-test, and the trend confidence is calculated according to the formula "trend confidence=(1-fitting residual) x 100 x significance coefficient"; finally, the vector composed of the development trend type and the trend confidence is taken as the concentration attenuation trend of the target teenager during the concentration training; wherein the development trend type of the concentration attenuation curve includes stable attenuation or accelerated attenuation, and the value range of the trend confidence is 0-100.

[0038] It should be noted that in the present application, the concentration attenuation trend is a feature vector representing the development trend type and the trend confidence of the target teenager's concentration during the training process, which quantifies the dynamic change direction and degree of concentration, provides core cognitive dimension data for the state perception parameter, effectively supports the accuracy of the bimodal fusion feature, and further provides a reliable basis for the generation of the density adjustment coefficient and the rate correction coefficient, ensuring that the adaptive regulation of the concentration training can accurately match the individual attenuation state, and improving the training pertinence and efficiency.

[0039] In step 104, the body movement data of the target teenager during the concentration training is obtained, the action trajectory in the body movement data is matched with the standard action trajectory of the concentration training, and the behavior coordination coefficient of the target teenager during the concentration training is obtained.

[0040] It should be noted that in the present application, the body action data is a time sequence set reflecting the body behavior state of the target teenager during the concentration training process. The body action data can supplement the cognitive level information reflected by the eye movement data. By quantifying the coordination degree of the body action and the standard trajectory, the one-sidedness of relying only on cognitive data evaluation is avoided, the state perception of the concentration training is more comprehensive, and the scientificity and accuracy of the adaptive regulation strategy are further improved. In specific implementation, the body action data of the target teenager during the concentration training process can be obtained in the following manner, that is, a depth camera can be fixed at 1.5-2 meters in front of the training area, the lens is horizontally aligned with the upper body of the target teenager, and the shoulder, elbow, wrist and other body parts are ensured to be covered. The time sequence coordinate sequence of 21 skeletal nodes including the head, neck, shoulder, elbow, wrist and other parts of the upper body of the target teenager is captured in real time at a sampling frequency of 30 frames per second. The missing values of node coordinates caused by body occlusion in each time sequence coordinate sequence are removed by ensuring that the interval between adjacent effective frames is not more than 3 frames. The abnormal jump values caused by rapid shaking in each time sequence coordinate sequence are filtered by regarding the coordinate change greater than 50 centimeters between adjacent frames as abnormal. Finally, the set of time sequence coordinate sequences of all processed skeletal nodes is taken as the body action data of the target teenager during the concentration training process.

[0041] In some embodiments, the behavior coordination coefficient of the target teenager during the concentration training can be obtained by performing similarity matching between the action trajectory in the body action data and the standard action trajectory of the concentration training in the following steps: determining the action trajectory of the target teenager during the concentration training according to the time sequence coordinate sequence of the skeletal node in the body action data; determining the standard action trajectory of the concentration training; spatiotemporal alignment of the action trajectory and the standard action trajectory to obtain the behavior coordination coefficient of the target teenager during the concentration training.

[0042] In a specific implementation, the action trajectory of the target teenager during the concentration training can be determined according to the time sequence coordinate sequence of the bone nodes in the limb action data in the following manner: the time sequence coordinate sequence of the training-related bone points, such as the shoulder, elbow, wrist, and hip, can be extracted from the limb action data, the missing coordinate values in each time sequence coordinate sequence caused by temporary limb occlusion can be supplemented by using a linear interpolation method, the jump abnormal values in each time sequence coordinate sequence whose coordinate change in adjacent frames is greater than 50 cm can be corrected by using a moving average filter, and the time sequence coordinate sequence of each related bone point after correction is sequentially connected according to the training action time sequence to form a three-dimensional action trajectory containing timestamp information as the action trajectory of the target teenager during the concentration training; wherein the action trajectory is a path representing the actual limb behavior state of the target teenager during the concentration training, the action trajectory is the basis for quantifying the limb behavior state of the target teenager, can provide original data support for subsequent similar matching, and ensures that the behavior coordination coefficient calculation has a reliable data source.

[0043] In a specific implementation, the standard action trajectory of the concentration training can be determined in the following manner: first, three professional trainers in the field of concentration training can be invited to demonstrate standard actions, and the same depth camera as that of the target teenager can be used to collect bone data of three complete training actions of each person at a sampling frequency of 30 frames per second in the same training scene; then, the missing values of the collected nine groups of original data can be supplemented by using a linear interpolation method, the jump abnormal values of the nine groups of original data can be corrected by using a moving average filter, the average value of the coordinates of the same bone points in each frame of the nine groups of original data can be calculated and sequentially connected according to the training action time sequence to generate a single-time sequence standard trajectory template, and the standard trajectory template is taken as the standard action trajectory of the concentration training; wherein the standard action trajectory is a benchmark path of the concentration training generated by the professional training instructors demonstrating the standard action of the concentration training, and the standard action trajectory can be used as a standard reference for whether the action of the target teenager is standard, provides a unified standard reference for similar matching, avoids errors caused by inconsistent references, and ensures the objectivity and accuracy of the behavior coordination coefficient calculation.

[0044] In a specific implementation, the spatiotemporal alignment of the action trajectory and the standard action trajectory to obtain the behavior coordination coefficient of the target teenager during the concentration training can be implemented in the following manner: the dynamic time warping algorithm can be used to perform the spatiotemporal alignment of the action trajectory and the standard action trajectory to obtain the behavior coordination coefficient of the target teenager during the concentration training; wherein the neighborhood search radius of the dynamic time warping algorithm can be set to 5 frames to limit the search range and reduce the calculation amount, and the path penalty coefficient can be set to 0.2 to avoid excessive stretching or compression of the trajectory, the dynamic time warping algorithm calculates the sum of the Euclidean distances between all skeletal point time coordinates in each pair of aligned frames of the action trajectory and the standard action trajectory, and finds the aligned path that minimizes the sum of the Euclidean distances through dynamic time warping, and the minimum total distance in the aligned path is taken as the dissimilarity between the two action trajectories, the greater the dissimilarity, the greater the action difference between the two action trajectories; and then the dissimilarity is normalized, and the value obtained by subtracting the normalized dissimilarity from 1 is taken as the behavior coordination coefficient of the target teenager during the concentration training.

[0045] It should be noted that in the present application, the behavior coordination coefficient is an index reflecting the coordination degree between the target teenager's body action and the standard action, which can complement the body level information reflected by the action data to make the state perception of the concentration training more comprehensive, and thus improve the scientificity and accuracy of the adaptive regulation factor generation.

[0046] In step 105, the state perception fusion of the concentration decay trend and the behavior coordination coefficient is performed to generate an adaptive regulation factor of the target teenager during the concentration training process, and then the density and moving speed of the interference elements in the next concentration training stage of the target teenager are adjusted according to the adaptive regulation factor.

[0047] In some embodiments, with reference to Figure 3 The figure is an exemplary flow chart for determining an adaptive regulation factor according to some embodiments of the present application, and the state perception fusion of the concentration decay trend and the behavior coordination coefficient to generate an adaptive regulation factor of the target teenager during the concentration training process can be implemented in the following steps: In step 1051, a state perception parameter of the target teenager during the concentration training is extracted from the concentration decay trend; In step 1052, state fusion is performed based on the state perception parameter and the behavior coordination coefficient to obtain a dimensionally unified bimodal fusion feature; In step 1053, the density adjustment coefficient and the rate correction coefficient of the interference elements in the concentration training process of the target teenager are generated through the bimodal fusion feature; In step 1054, the adaptive regulation factor of the target teenager in the concentration training process is determined according to the density adjustment coefficient and the rate correction coefficient.

[0048] In a specific implementation, the state perception parameter of the target teenager during the concentration training can be extracted from the concentration decay trend in the following manner: the decay acceleration of concentration of the target teenager during the concentration training can be extracted from the concentration decay trend as the state perception parameter of the target teenager during the concentration training; wherein the state perception parameter is a parameter that characterizes the change trend of the concentration cognitive state of the target teenager during the concentration training. This state perception parameter can provide core data at the cognitive level for subsequent dual-modal fusion, avoiding the one-sidedness of state perception caused by relying only on body data.

[0049] In a specific implementation, the state fusion based on the state perception parameter and the behavior coordination coefficient to obtain the dual-modal fusion feature with unified dimensions can be implemented in the following manner: the maximum-minimum standardization method can be used to standardize the state perception parameter and the behavior coordination coefficient and map them to the 0-1 interval, and then the standardized state perception parameter and behavior coordination coefficient are input into the dual-modal attention fusion network. The state perception parameter and the behavior coordination coefficient are expanded to equal-length sequences through the network input layer, and the cosine similarity between the state perception parameter sequence and the behavior coordination coefficient sequence is calculated through the attention mechanism to dynamically allocate weights. The state perception parameter sequence and the behavior coordination coefficient sequence are fused and compressed into a 64-dimensional vector through the full connection layer, thereby outputting the dual-modal fusion feature with unified dimensions; wherein the dual-modal fusion feature is a unified vector formed by integrating the cognitive modal information and the body modal information of the target teenager during the concentration training. This dual-modal fusion feature eliminates information islands by fusing two types of modal information, provides comprehensive feature support for the generation of the density adjustment coefficient and the rate correction coefficient, and improves the accuracy and adaptability of the coefficient calculation.

[0050] In a specific implementation, the density adjustment coefficient and the speed correction coefficient of the interference element in the target adolescent concentration training process can be obtained by inputting the bimodal fusion feature into a pre-trained three-layer fully connected network for coefficient prediction. The first layer of the three-layer fully connected network has 32 neurons, and the second layer has 16 neurons, both using ReLU activation function. The output layer has 2 neurons, using Sigmoid activation function combined with linear transformation formula to map the output to the range of 0.5-1.5. The first output value of the three-layer fully connected network is defined as the density adjustment coefficient of the interference element, and the second output value is defined as the speed correction coefficient of the interference element. The bimodal fusion feature, the optimal density adjustment coefficient, and the optimal speed correction coefficient are inputted into the three-layer fully connected network, and the mean square error loss function is used to train the network to obtain the pre-trained three-layer fully connected network. The density adjustment coefficient is a coefficient that determines the adjustment range of the interference element density in the next stage of concentration training. If the density adjustment coefficient is greater than 1, the density increases, and if it is less than 1, the density decreases. The density adjustment coefficient provides a quantitative basis for adjusting the density of the interference element, ensuring that the density adapts to the current concentration state and avoiding excessive difficulty or low difficulty affecting the training effect. The speed correction coefficient is a coefficient that determines the adjustment range of the interference element speed in the next stage of concentration training. If the speed correction coefficient is greater than 1, the speed increases, and if it is less than 1, the speed decreases. The speed correction coefficient provides a standardized basis for adjusting the speed of the interference element, ensuring that the speed of the interference element matches the concentration level and improving the scientific nature of the training.

[0051] It should be noted that in this application, the adaptive control factor is a structured data that integrates the training parameter adjustment coefficient and its confidence in the next stage of concentration training process. The adaptive control factor provides a direct basis for adjusting the training parameters in the next stage, thereby providing a clear control direction and amplitude for the training system, achieving dynamic adaptation of concentration training, and maximizing individual training efficiency.

[0052] In a specific implementation, the adaptive control factor of the target adolescent in the concentration training process can be determined according to the density adjustment coefficient and the speed correction coefficient. First, the effectiveness of the density adjustment coefficient and the speed correction coefficient is verified. If the coefficient exceeds the range of 0.5-1.5, it is considered abnormal and replaced by the effective coefficient at the previous time. Then, the confidence of the density adjustment coefficient and the speed correction coefficient is calculated using the confidence formula, for example: confidence = (1-bimodal fusion feature fitting residual) x 100. Finally, the density adjustment coefficient, the speed correction coefficient, and the confidence are packaged as structured data and labeled with the generation timestamp and the corresponding training stage to form the adaptive control factor of the target adolescent in the concentration training process.

[0053] In a specific implementation, adjusting the density and moving speed of the interference elements in the visual search task in the next attention training phase of the target teenager according to the adaptive regulation factor can be implemented in the following manner: the initial interference element density and the initial moving speed of the current training phase are determined first, and then the interference element density adjustment coefficient and the dynamic speed correction coefficient in the adaptive regulation factor are used as the core adjustment basis to adjust the density and moving speed of the interference elements in the visual search task in the next attention training phase of the target teenager by using a training task control component (such as a scene parameter control component of the Unity engine) according to adjusted density = initial interference element density x density adjustment coefficient and adjusted speed = initial moving speed x speed correction coefficient; wherein the initial interference element density can be set based on the basic difficulty of the visual search task, and the initial moving speed can be set according to the average visual tracking ability of the teenager, which is not limited here.

[0054] In addition, another aspect of the present application, in some embodiments, the present application provides a teenager attention training system, referring to Figure 4 The figure is a structural schematic diagram of a teenager attention training system according to some embodiments of the present application, which includes a collection module 401, a processing module 402 and an execution module 403, which are described as follows: The collection module 401 is mainly used for collecting the eye movement data of the target teenager when completing the visual search task in the attention training process in the present application; The processing module 402 is mainly used for clustering the fixation clusters based on the eye movement data to obtain the search pattern features of the target teenager when performing visual search in the attention training process in the present application; It should be noted that the processing module 402 is also used to construct the ability portrait of the target teenager for attention training through the historical training data of the target teenager, and then determine the attention decay trend of the target teenager when performing attention training based on the ability portrait and the search pattern features; In addition, it should be noted that the processing module 402 is also used to obtain the body movement data of the target teenager in the attention training process, and perform similarity matching between the action trajectory in the body movement data and the standard action trajectory of the attention training to obtain the behavior coordination coefficient of the target teenager when performing attention training; The execution module 403 is mainly used for state perception fusion of the concentration attenuation trend and the behavior coordination coefficient, to generate an adaptive regulation factor of the target teenager in the concentration training process, and then adjust the density and moving speed of the interference elements in the visual search task of the target teenager in the next concentration training stage according to the adaptive regulation factor.

[0055] The modules in the teenager concentration training system can be implemented by software, hardware, or a combination thereof. The modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory in the computer device in software form, so as to be called and executed by the processor.

[0056] In addition, in an embodiment, the present application provides a computer device, which can be a server, and the internal structure diagram thereof can be as shown in Figure 5 The computer device includes a processor, a memory, and a network interface connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data of the human-computer interaction-based teenager concentration training method. The network interface of the computer device is used to communicate with the external terminal through the network connection. The computer program is executed by the processor to implement a human-computer interaction-based teenager concentration training method.

[0057] Those skilled in the art can understand that Figure 5 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0058] In an embodiment, a computer device is also provided, which includes a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above-mentioned human-computer interaction-based teenager concentration training method embodiments when executing the computer program.

[0059] In an embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by the processor to implement the steps in the above-mentioned human-computer interaction-based teenager concentration training method embodiments.

[0060] In one embodiment, a computer program product or computer program is provided, which includes computer instructions stored in a computer readable storage medium. A processor of a computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in the above-mentioned human-computer interaction-based adolescent concentration training method embodiments.

[0061] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0062] The technical features of the above embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the present application.

[0063] The above embodiments only express several implementation manners of the present application, and the description is specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present application, some modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of the patent protection of the present application should be subject to the appended claims.

Claims

1. A human-computer interaction-based adolescent concentration training method, which adaptively trains the concentration of a target adolescent according to behavior performance data in a concentration training process, characterized in that, The method comprises the following steps: Collecting eye movement data of the target teenager during visual search task in the process of concentration training; Based on the eye movement data, the gaze cluster is clustered to obtain the search mode characteristics of the target teenager during visual search in the process of concentration training; Through the historical training data of the target teenager, the ability portrait of the target teenager in concentration training is constructed, and then the concentration decay trend of the target teenager in concentration training is determined based on the ability portrait and the search mode characteristics; Obtain the body movement data of the target teenager in the process of concentration training, and perform similarity matching on the action trajectory in the body movement data and the standard action trajectory of concentration training to obtain the behavior coordination coefficient of the target teenager in concentration training; The state perception fusion of the concentration decay trend and the behavior coordination coefficient is performed to generate an adaptive regulation factor of the target teenager in the process of concentration training, and then the density and moving speed of the interference elements in the visual search task of the target teenager in the next concentration training stage are adjusted according to the adaptive regulation factor.

2. The method of claim 1, wherein, Based on the eye movement data, the gaze cluster is clustered to obtain the search mode characteristics of the target teenager during visual search in the process of concentration training, which specifically comprises: The eye movement data is spatially clustered to generate a plurality of gaze clusters; The average gaze duration, intra-cluster gaze density and inter-cluster jump rate of the target teenager in the process of concentration training are determined according to all gaze clusters; The search mode characteristics of the target teenager during visual search in the process of concentration training are determined based on the average gaze duration, the intra-cluster gaze density and the inter-cluster jump rate.

3. The method of claim 1, wherein, Through the historical training data of the target teenager, the ability portrait of the target teenager in concentration training is constructed, which specifically comprises: Obtain the historical training data of the target teenager in concentration training; The time sequence characteristics of the historical training data are aligned to obtain a training sample set with spatio-temporal anchoring and unified feature dimensions; Based on the training sample set, the ability portrait of the target teenager in concentration training is generated.

4. The method of claim 1, wherein, Based on the ability portrait and the search mode characteristics, the concentration decay trend of the target teenager in concentration training is determined, which specifically comprises: The search mode characteristics are aligned to the feature space of the ability portrait through cross-modal feature mapping, and then a spatio-temporally coupled fusion feature vector is obtained; Based on the fusion feature vector, iterative deduction is performed to obtain a concentration decay curve of the target teenager's concentration dynamic evolution with training time; The concentration decay trend of the target teenager in concentration training is determined according to the concentration decay curve.

5. The method of claim 1, wherein, The action trajectory in the body movement data is matched with the standard action trajectory of concentration training to obtain the behavior coordination coefficient of the target teenager in concentration training, which specifically comprises: The action trajectory of the target teenager in concentration training is determined according to the time sequence coordinate sequence of the skeletal node in the body movement data; The standard action trajectory of concentration training is determined; The action trajectory and the standard action trajectory are spatio-temporally aligned to obtain the behavior coordination coefficient of the target teenager in concentration training.

6. The method of claim 1, wherein, The state awareness fusion of the concentration attenuation trend and the behavior synergy coefficient generates an adaptive regulation factor of the target teenager in the concentration training process, and specifically includes: Extracting a state awareness parameter of the target teenager when performing concentration training from the concentration attenuation trend; Performing state fusion based on the state awareness parameter and the behavior synergy coefficient to obtain a dimension-unified bimodal fusion feature; Generating a density adjustment coefficient and a rate correction coefficient of the interference element in the concentration training process of the target teenager through the bimodal fusion feature; Determining the adaptive regulation factor of the target teenager in the concentration training process according to the density adjustment coefficient and the rate correction coefficient.

7. The method of claim 1, wherein, The eye movement data of the target teenager when performing a visual search task in the concentration training process is collected through a non-invasive infrared eye movement tracking device.

8. A system for training concentration in adolescents, characterized in that, Comprise: A collection module for collecting eye movement data of the target teenager when performing a visual search task in the concentration training process; A processing module for clustering gaze clusters based on the eye movement data to obtain search pattern features of the target teenager when performing a visual search in the concentration training process; The processing module is configured to construct an ability portrait of the target teenager performing concentration training through historical training data of the target teenager, and then determine a concentration attenuation trend of the target teenager performing concentration training based on the ability portrait and the search pattern features; The processing module is configured to obtain body movement data of the target teenager in the concentration training process, perform similarity matching between the movement trajectory in the body movement data and a standard movement trajectory of the concentration training, and obtain a behavior synergy coefficient of the target teenager performing concentration training; An execution module for state awareness fusion of the concentration attenuation trend and the behavior synergy coefficient to generate an adaptive regulation factor of the target teenager in the concentration training process, and then adjusting the density and moving speed of the interference element in the visual search task of the target teenager in the next concentration training stage according to the adaptive regulation factor. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the human-computer interaction-based teenager concentration training method in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to realize the steps of the human-computer interaction-based teenager concentration training method in any one of claims 1 to 7.

Images