Campus competition scoring robot assisted by audience emotion

Abstract

The application relates to the technical field of artificial intelligence equipment, in particular to a campus competition scoring robot based on audience emotion auxiliary judgment, which acquires competition action data of a participating object through a motion capture and posture sensing module, generates action rule confidence data based on rule matching, and introduces an audience emotion auxiliary judgment mechanism when the rule confidence is in a preset rule boundary interval; crowd emotion modeling and differential analysis are carried out on audience reaction data to obtain audience emotion difference data, and the audience emotion difference data are weighted and fused with the action rule confidence data to form a final scoring judgment result. The scheme effectively alleviates the uncertainty of judgment of rule boundary actions, improves the fairness, rationality and consistency of the scoring result, reduces the risk of controversial judgment, and enhances the intellectualization and credibility of the campus competition scoring system under the premise of maintaining the dominant position of the competition rules.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence devices, specifically a campus sports scoring robot based on audience emotion-assisted judgment. Background Technology

[0002] With the continuous improvement of informatization and intelligence in campus sports activities and competitive events, automatic scoring equipment based on image recognition, motion sensing interaction, and sensor technology is gradually being applied in campus competitive scenarios to replace or assist human referees in completing action recognition and scoring during the competition. These technologies typically collect motion images, location information, or movement trajectory data of participants, and combine this data with a pre-set scoring rule model to analyze and process the competition events, thereby achieving automatic recording and display of the competition results.

[0003] Existing campus competition scoring systems or scoring robots typically employ deterministic rule-based decision mechanisms, transforming competition rules into quantifiable threshold conditions. Scoring is determined by judging whether a moving target meets preset position, movement amplitude, or time conditions. To improve recognition accuracy, these systems often assume a relatively controllable competition environment and mitigate the impact of environmental interference on image recognition and sensor data processing through noise reduction, shielding spectator areas, or ignoring the actions of non-participants.

[0004] The inventors of this application have discovered that existing campus sports scoring systems are mostly based on deterministic rules. However, campus sports rules are usually verbally agreed upon or pedagogically based, resulting in incomplete and vague rules with considerable flexibility during implementation. This makes it difficult for existing scoring methods to make reasonable judgments when competitive actions fall within the rule boundaries or gray areas. Furthermore, campus competition venues are often non-enclosed or poorly isolated environments, with spectators in close proximity to the competition area. Their voices, facial expressions, and body movements are present throughout the competition, and existing technologies generally treat this information as interference signals and suppress it, leading to insufficient adaptability of the scoring system in complex environments. When competitive actions cannot be clearly distinguished between scoring and non-scoring states based solely on a single image or sensory data, existing technologies also lack effective means to introduce external auxiliary judgment information, making it difficult to achieve reliable and interpretable automatic scoring for boundary actions. Summary of the Invention

[0005] This application provides a campus sports scoring robot based on audience emotion-assisted judgment, which solves the technical problems in the existing sports scoring process that rely solely on fixed rule thresholds for judgment, leading to misjudgments of actions at rule boundaries, lack of objective auxiliary evidence for controversial actions, and insufficient fairness and consistency of scoring results.

[0006] To achieve the above objectives, the embodiments of this application disclose the following technical solutions:

[0007] This solution discloses a campus sports scoring robot based on audience emotion-assisted judgment, including a processor and a storage unit. The processor is configured to execute the following computer program steps:

[0008] The competition motion data of the participants is acquired by a motion capture and posture perception module set up in the competition area. The competition motion data includes at least joint spatial position data, posture change sequence data and motion continuity parameters.

[0009] The competitive action data is processed by rule matching to generate action rule confidence data, which is used to characterize the degree to which the competitive action meets the preset competitive rule conditions;

[0010] When the confidence data of the action rule is within the preset rule boundary range, the audience emotion auxiliary judgment process is triggered;

[0011] Collect audience spatial location data and historical support behavior data, segment the audience groups based on the audience spatial location data and historical support behavior data, and collect audience reaction data for different audience groups separately;

[0012] The audience reaction data is processed by group emotion modeling to generate group emotion characteristic data corresponding to each audience group;

[0013] Perform differential analysis on group emotional characteristic data to generate audience emotional difference data;

[0014] The confidence data of the action rules and the difference data of audience emotions are fused together to generate the final scoring data.

[0015] The final scoring data will be used to output the competition scoring results.

[0016] This invention addresses the problem that existing competitive scoring processes rely solely on fixed rule thresholds, leading to misjudgments of rule-boundary actions and a lack of objective supporting evidence for controversial actions. It proposes a campus competitive scoring robot based on audience emotion-assisted judgment. By performing rule matching calculations on competitive action data, it generates continuously valued action rule confidence data and presets rule boundary intervals to identify boundary situations where rule judgment is uncertain. Only in these uncertain situations is audience emotion difference data, after group segmentation and differential analysis, introduced as auxiliary information to ensure that the dominant position of rule judgment is not disturbed at the system level. Through comprehensive analysis of the differences in emotional intensity and the synchronicity of emotional changes among different audience groups, it effectively characterizes the degree of divergence in audience reactions to the same competitive action. Combined with time-series correlation data generated by timestamp alignment, it avoids irrelevant emotional fluctuations from affecting the judgment results. Furthermore, by weighted fusion of action rule confidence data and audience emotion difference data, continuous correction of rule boundary judgment results can be achieved, thereby improving the fairness, consistency and rationality of scoring results in rule-critical scenarios, while enhancing the interpretability of the scoring process and the credibility of the results. This is suitable for automatic scoring scenarios in campus competitions with highly controversial actions. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall method of an embodiment of the present invention;

[0018] Figure 2 This is a flowchart of the competitive action data acquisition and processing according to an embodiment of the present invention;

[0019] Figure 3 This is a diagram illustrating the audience segmentation and emotion modeling structure of an embodiment of the present invention.

[0020] Figure 4 This is a logic diagram for rule boundary triggering and fusion determination in an embodiment of the present invention. Detailed Implementation

[0021] Specific embodiments of the invention will now be described in detail. Although the invention is described in conjunction with these specific embodiments, it should be understood that the invention is not intended to be limited to these specific embodiments. Rather, these embodiments are intended to cover alternative, modified, or equivalent embodiments that may be included within the spirit and scope of the invention as defined by the claims. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. The invention may be practiced without some or all of these specific details. In other instances, well-known processes have not been described in detail so as not to unnecessarily obscure the invention.

[0022] When used in conjunction with the terms "comprising," "method comprising," or similar language in this specification and appended claims, the singular forms "a," "some," and "the" include plural references unless the context clearly indicates otherwise. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] Application Overview:

[0024] Campus sports competitions often rely on verbal agreements or instructional rules, which are often incomplete and vaguely defined. This makes it difficult for existing scoring systems to make stable and consistent judgments when competitive actions fall into a gray area of ​​the rules. For example, when an action's completion is close to the scoring threshold or when there is a dispute, relying solely on action or location information can easily lead to misjudgments or inconsistent assessments. Furthermore, campus competitions typically take place in non-closed or low-isolation environments, with spectators close to the competition area. Their continuous sounds, facial expressions, and body movements are treated as interference signals and suppressed by existing technology, further weakening the system's adaptability in complex environments. If these problems are not addressed, the scoring system will ultimately fail to closely resemble the judgment logic of human referees in real-world campus sports scenarios, resulting in insufficient stability and reliability of automatic scoring results.

[0025] In response to the aforementioned problems, this application first considers the natural temporal correlation between the audience's immediate reaction and the progress of the competition in a campus competition setting. This application attempts to introduce the audience's reaction as auxiliary judgment information and conduct correlation analysis with the competition action data to provide additional reference for scoring judgment under the rule boundary conditions. Thus, without changing the existing competition rules, the scoring robot can make more reasonable and interpretable automatic scoring judgments on boundary actions in complex environments.

[0026] Example

[0027] The process involves acquiring the contestant's athletic motion data, which is collected by a motion capture and posture perception module located in the competition area. This data includes at least joint spatial location data, posture change sequence data, and motion continuity parameters. The motion capture and posture perception module continuously captures frames of the contestant, extracting the joint spatial location data and generating posture change sequence data based on joint changes between adjacent frames. The posture change sequence data undergoes temporal smoothing to obtain motion continuity parameters. The joint spatial location data, posture change sequence data, and motion continuity parameters are then stored as components of the athletic motion data.

[0028] In this embodiment, the motion capture and posture perception module includes an RGB camera, a depth camera, and a binocular or multi-view camera array. The perception method employed is a posture estimation algorithm based on human keypoint recognition or a 3D joint point reconstruction based on multi-view fusion. Continuous frame acquisition is performed on the participant to obtain joint point spatial position data. The motion capture and posture perception module is positioned within the competition area, enabling a global perception perspective. If positioned on the robot itself, only a local perspective following the robot would be obtained. This embodiment further proposes that the motion capture and posture perception module be positioned at a fixed or adjustable location within the competition area to achieve non-contact acquisition of competitive motion data.

[0029] By continuously acquiring and calculating the displacement changes of joints between adjacent frames, posture change sequence data is formed, reflecting the evolution of the action over time. Simultaneously, temporal smoothing of the posture change sequence data reduces the impact of instantaneous jitter or acquisition noise on action judgment, thus obtaining action continuity parameters characterizing the smoothness of the action. Unifying and storing joint spatial location data, posture change sequence data, and action continuity parameters provides multi-dimensional joint features for subsequent rule confidence calculations. For example, in a campus taekwondo competition, the athlete's kicking motion is captured at a frequency of 30 frames per second, recording the spatial coordinates of the knee and ankle joints. Posture change sequence data is generated through displacement changes between adjacent frames, allowing subsequent rule judgments to simultaneously consider both action amplitude and action continuity.

[0030] In motion capture-based sports motion acquisition, motion continuity reflects the stability and consistency of joint motion trajectories over time. Since sensor jitter, occlusion errors, or momentary false triggers can cause anomalies in local frames, it is necessary to characterize the overall continuity of the motion by analyzing the statistical properties of joint displacement changes between adjacent frames.

[0031] Based on the above analysis, this embodiment defines the statistical stability of joint displacement changes in consecutive frames as the motion continuity parameter. By normalizing the joint displacement amplitude in consecutive frames, an engineering quantitative calculation model is constructed to characterize the smoothness of the motion trajectory. The calculation method of the motion continuity parameter is as follows:

[0032] ;

[0033] in, ; The number of frames captured consecutively. For the first The spatial position vector of the preset joint point in the frame, Euclidean distance is used to measure the displacement amplitude of joints. The value ranges from 0 to 1, with a value closer to 1 indicating smoother and more consistent movement. This calculation method allows for a unified representation of the overall stability of competitive movements without depending on specific movement types, thus providing continuously comparable input parameters for subsequent rule matching.

[0034] The competitive action data is processed by rule matching to generate action rule confidence data, which is used to characterize the degree to which the competitive action meets the preset competitive rule conditions. The competitive action data is matched and calculated with the action threshold conditions corresponding to the preset competitive rules to generate action satisfaction parameters. The action rule confidence data is calculated based on the action satisfaction parameters, and the action rule confidence data is used as the basis for determining whether to trigger the audience emotion auxiliary judgment process.

[0035] Competition rules are typically composed of multiple quantifiable action conditions, and the importance of each condition to the rule determination is not entirely the same. Therefore, mapping each action feature directly corresponding to the rule to an action satisfaction parameter and weighting and summing them according to their relative importance in the rule system can form a continuous quantitative expression of the overall rule compliance degree. Based on the above analysis, this embodiment defines the weighted fusion result of the action satisfaction parameter as action rule confidence data, used to characterize the degree to which the overall competitive action satisfies the preset competitive rule conditions, and is obtained through the following weighted calculation model:

[0036]

[0037] in, Action rule confidence data is used to characterize the degree to which the overall competitive action meets the preset competitive rule conditions; The number of action features involved in the rule determination; The parameter representing the satisfaction of the i-th action feature, such as the satisfaction of kicking height or the satisfaction of action duration, is calculated as follows: based on the joint spatial location data, posture change sequence data or action continuity parameters, the corresponding action feature measured value is extracted, and the measured value is normalized or mapped to the action threshold condition in the corresponding competition rules. The value range is [0,1]. This represents the weight coefficient of the corresponding feature in the rule, and =1. To illustrate with a specific numerical example, if the kick height motion satisfaction parameter is 0.9 and the motion continuity motion satisfaction parameter is 0.8, with weights of 0.6 and 0.4 respectively, then the motion rule confidence data C = 0.9 × 0.6 + 0.8 × 0.4 = 0.86. This transforms the traditional binary judgment of compliance into a continuously comparable quantified result of rule confidence, providing input conditions for subsequent boundary judgments.

[0038] Based on the aforementioned action rule confidence data calculation model, in order to further consider the trajectory stability during the acquisition of competitive actions, the aforementioned action continuity parameter Ac is introduced as a global constraint factor into the calculation process of action rule confidence data.

[0039] Specifically, by applying overall modulation of the action continuity parameter to the weighted fusion result of each action satisfaction parameter, the abnormal amplification of rule confidence under unstable action trajectories is suppressed, thereby constructing a data calculation model for action rule confidence that incorporates action continuity constraints. The calculation method is expressed as follows:

[0040]

[0041] in, For action rule confidence data; The number of action features involved in rule matching; is the weight coefficient of the i-th action feature; For action satisfaction parameters; This is a parameter for the continuity of action.

[0042] In practice, competition rules typically constrain competitive actions using thresholds, such as setting minimum or maximum threshold conditions for features like action height and duration. Since measurement uncertainties inevitably exist near these thresholds during sensor acquisition, using a hard threshold method can easily introduce instability in the critical threshold region.

[0043] Based on the above considerations, this embodiment does not directly perform a binary determination of whether the action feature meets the threshold. Instead, it continuously quantifies and models the action satisfaction parameter based on the degree of deviation of the action feature from the rule threshold. Furthermore, to avoid a single action feature from excessively affecting the overall action rule confidence data under abnormal conditions, a nonlinear adjustment mechanism is introduced for the action satisfaction parameter.

[0044] Specifically, after normalizing each movement feature in the competitive movement data, a nonlinear mapping model is constructed between the movement features and the movement satisfaction parameter. The calculation method is defined as follows:

[0045] ;

[0046] in, Let be the action satisfaction parameter of the i-th action feature, with a value range of [0,1]. For the first The measured value of a motion feature, which is a derived feature quantity directly corresponding to a specific competition rule, obtained by rule mapping from at least one of the following: joint spatial location data, posture change sequence data, and motion continuity parameters. The threshold parameters for the corresponding rules are derived from a preset competition rule library; This is the rule sensitivity coefficient, set by the project type, used to control the smoothness of the transition near the threshold. This method ensures that when action features approach the threshold boundary, their impact on confidence is smoother, avoiding abrupt changes in rule judgments at boundary regions, thereby improving the stability and interpretability of rule confidence calculation.

[0047] When the confidence data of the action rule is within the preset rule boundary range, the audience emotion auxiliary judgment process is triggered.

[0048] This step establishes the triggering relationship between action rule determination and audience emotion-assisted determination. Its core function is to avoid redundant calculations when rules are clear and to introduce auxiliary information only when there is uncertainty in the determination. Specifically, the system pre-sets a preset rule boundary range, such as 0.45 to 0.55. When the action rule confidence data falls within this preset boundary range, it indicates that the competitive action neither clearly violates the rules nor fully meets the rule conditions, thus triggering the audience emotion-assisted determination process. The technical effect of this dynamic adjustment logic is to improve the overall determination efficiency and fairness, and to avoid interference from audience emotions on the results of clear rules. The threshold range can be adjusted according to different competitive events; for example, the range can be appropriately expanded in highly competitive events. This step is one of the important innovations of this solution. Through the interval control of rule confidence, it achieves conditional introduction of audience emotion information, rather than indiscriminate introduction.

[0049] The preset rule boundary interval is essentially used to characterize the degree of uncertainty of the system's judgment result for the current action rule. In existing implementations, the rule boundary interval is usually set in the form of a fixed threshold range, which makes it difficult to reflect the uncertainty changes of the action rule confidence data in different value ranges.

[0050] To avoid the preset rule boundary interval existing only as a static threshold, this embodiment further models the rule determination uncertainty through continuous quantification. Specifically, using the aforementioned action rule confidence data as the input variable for uncertainty assessment, a continuous uncertainty function related to the action rule confidence data is constructed to elevate the preset rule boundary interval from a discrete threshold to a continuous engineering quantity, which is then used to drive subsequent process control logic. Based on the above modeling approach, the rule determination uncertainty function is defined as:

[0051] ;

[0052] in, This is a rule uncertainty index, with a value range of [0,1]. For action rule confidence data; when action rule confidence data When approaching the rule decision boundary, the rule uncertainty index The value of approaches 1; when the action rule confidence data When approaching the defined decision interval, the rule uncertainty index The value of approaches 0, thereby achieving a continuous characterization of the uncertainty of rule determination within the normalized interval from the preset rule boundary interval 0 to the preset rule boundary interval 1.

[0053] Collect audience spatial location data and historical support behavior data; divide the audience into groups based on the audience spatial location data and historical support behavior data; and collect audience reaction data for each audience group separately; divide the audience into at least two audience group regions based on the audience spatial location data and historical support behavior data; collect audience reaction data in each audience group region separately; the audience reaction data includes group voice feature data, group body movement feature data, and group facial expression change feature data; and add timestamp information to the audience reaction data.

[0054] After triggering the audience emotion assessment process, this step aims to construct comparable audience group areas and acquire audience reaction data related to the competitive event. By collecting audience spatial location data, the distribution of the audience at the competition venue can be characterized; combined with historical audience support behavior data, such as historical cheering directions or past reaction records, the audience can be divided into at least two audience group areas with different support tendencies. The technical effect of this step is to avoid treating the audience as a single whole, but rather to form a group structure with different stances, thus providing a basis for subsequent emotion difference analysis. The collection of audience reaction data includes group voice feature data, group body movement feature data, and group facial expression change feature data, and uniformly adds timestamp information to ensure alignment with the time of the competitive action. As for alternative solutions, audience spatial location data can be obtained through camera positioning or seat number mapping; historical support behavior data can be obtained from historical competition records or on-site interaction behavior records.

[0055] The audience reaction data is processed to perform group emotion modeling to generate group emotion characteristic data corresponding to each audience group; energy and spectrum analysis is performed on the group voice characteristic data to generate voice emotion parameters; amplitude and frequency analysis is performed on the group body movement characteristic data to generate movement emotion parameters; expression change trend analysis is performed on the group facial expression change characteristic data to generate facial expression emotion parameters; the voice emotion parameters, movement emotion parameters, and facial expression emotion parameters are fused to generate group emotion characteristic data corresponding to the audience group region.

[0056] This step involves energy and spectrum analysis of the group's vocal characteristics to reflect the intensity and magnitude of their emotions; amplitude and frequency analysis of their body movements to reflect their activity level; and trend analysis of their facial expressions to reflect the direction of their emotions. Given that audience emotions are typically expressed across multiple dimensions, including vocal intensity, body movement amplitude, and facial expression trends, and that single-modal emotion parameters are susceptible to local anomalies, it is necessary to model multimodal emotion parameters uniformly. Therefore, this embodiment constructs a group emotion fusion model. By weighted fusion of different modal emotion parameters, stable and representative group emotion characteristic data is formed. The calculation method can be expressed as follows:

[0057]

[0058] in, This refers to data on the collective emotional characteristics of the audience. These are vocal emotion parameters obtained through energy and spectral analysis based on group vocal characteristic data. These are action-emotion parameters obtained from group body movement characteristic data through amplitude and frequency analysis. These are facial expression parameters obtained through facial expression trend analysis based on group facial expression change characteristic data. , , To preset the weighting coefficients for the corresponding modes, and + + =1. For example, when =0.4, =0.3, =0.3, and when the corresponding parameters are 0.8, 0.6, and 0.7, then E = 0.4 × 0.8 + 0.3 × 0.6 + 0.3 × 0.7 = 0.71. This step involves using group-level emotion fusion rather than individual emotion judgment, thereby reducing the impact of individual abnormal viewers on the overall judgment.

[0059] Differential analysis is performed on the group emotion characteristic data to generate audience emotion difference data; comparative calculations are performed on the group emotion characteristic data corresponding to different audience group regions to generate group emotion intensity difference parameters and group emotion change synchronicity parameters, and these parameters are used as audience emotion difference data.

[0060] After completing the group emotion modeling, this step is used to extract the antagonism and differences between different audience groups, aiming to avoid bias in the judgment results caused by a single emotion intensity. By comparing and calculating the group emotion characteristic data of different audience groups, a group emotion intensity difference parameter can be obtained, which is used to characterize the strength of the supporting positions. At the same time, by analyzing the synchronicity of emotion changes over time, a group emotion change synchronicity parameter can be obtained, which is used to determine whether emotional reactions are concentrated on the same competitive event. The technical effect of this step is to enhance the reliability of the audience emotion data judgment and provide differentiated input for subsequent fusion judgment.

[0061] Audience emotion difference data is used to characterize the degree of disagreement among different audience groups regarding the same competitive event, rather than to judge the correctness of the competitive result. Since simple differences in emotion intensity cannot distinguish between incidental noise and genuine emotional disagreement regarding the same competitive event, it is necessary to constrain emotion differences by incorporating the temporal synchronicity of emotion changes. Therefore, this embodiment constructs an audience emotion difference calculation model. By simultaneously considering the differences in group emotion intensity and the synchronicity of emotion changes among different audience groups, an audience emotion difference index is formed to characterize the degree of controversy. Its calculation method can be expressed as follows:

[0062]

[0063] in, The Audience Emotion Difference Index is used to quantify the results of audience emotion difference data. , Group emotional characteristics data corresponding to different audience groups; The synchronicity coefficient of emotion changes is obtained from the correlation analysis results of the time series of group emotion characteristic data;

[0064] The action rule confidence data and the audience emotion difference data are fused and judged to generate the final scoring judgment data; the timestamps of the action occurrence time corresponding to the competitive action data and the audience reaction data are aligned to generate time-series correlation data, and the audience emotion difference data is filtered for validity based on the time-series correlation data; when the action rule confidence data is within the preset rule boundary range, the action rule confidence data and the audience emotion difference data are weighted and fused according to the preset weight to generate the corrected confidence data, and the final scoring judgment data is generated based on the corrected confidence data.

[0065] When the confidence data for action rules falls within the preset rule boundary range, relying solely on rule-based judgments may introduce uncertainty. Therefore, audience emotional difference data, after differential analysis, is introduced as supplementary information. By weighted fusion of these two types of information, the boundary judgment results can be reasonably corrected while maintaining the dominance of the rules, thereby improving the overall performance of the scoring results in terms of fairness and consistency. In cases of rule uncertainty, the confidence data for action rules is corrected using audience emotional difference data. First, time-series correlation data is generated through timestamp alignment to ensure a one-to-one correspondence between audience emotional reactions and specific competitive actions. Second, the weighted fusion of the two types of data is only performed when the confidence data for action rules falls within the preset rule boundary range.

[0066] When the confidence data of action rules falls within the preset rule boundary range, relying solely on rule judgment may introduce uncertainty. Therefore, audience emotion difference data after differential analysis is introduced as auxiliary information. To achieve reasonable correction of boundary judgment results while maintaining the dominant position of rule judgment, this embodiment adopts an adaptive fusion mechanism that dynamically adjusts the fusion weights of the two types of information based on the uncertainty of rule judgment. Specifically, based on the aforementioned rule uncertainty index U(C), an adaptive fusion weight related to the rule uncertainty is constructed to reflect the credibility of rule judgment at the current moment. Its calculation method is as follows:

[0067] =1 ; ;

[0068] in, To correct the confidence level data, which is used to generate the final scoring data; C is the confidence level data of the action rules; D is the audience emotion difference data, which is calculated by combining the group emotion intensity difference parameter and the group emotion change synchronicity parameter. The fusion weight coefficient, with a value range of (0.5, 1), is used to ensure the dominant role of rule determination; when The larger the value, the greater the impact of the rule confidence on the final result. For example, when C=0.52 and D=0.70, When the value is 0.7, then R = 0.7 × 0.52 + 0.3 × 0.70 = 0.574. This dynamic adjustment logic ensures that the rule determination always takes precedence, and audience emotional information only participates in the correction when the rule is uncertain.

[0069] The processor outputs the competition scoring result based on the final scoring judgment data; it outputs the scoring result data and the judgment confidence information corresponding to the scoring result data, and stores the scoring result data and judgment confidence information in the memory; during the execution of the computer program steps, the processor performs unified data identification management on the competition action data, action rule confidence data, audience reaction data, group emotional characteristic data, audience emotional difference data, and final scoring judgment data to ensure the continuous access of each data in the computer program flow.

[0070] This step is used to complete the output and data closed-loop management of the entire technical solution. Its purpose is to provide clear scoring results to external systems or users and to maintain a complete data link for subsequent review or learning optimization. By outputting the scoring result data and corresponding judgment confidence information, the reliability of the scoring results can be intuitively reflected. At the same time, through a unified data identification management mechanism, the data of each stage is stored in association, which is conducive to subsequent traceability analysis and system iterative optimization. The data management method used in this step is a mature technology, but it forms a complete closed loop with the aforementioned multi-stage judgment process, making this solution interpretable and scalable.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation methods of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A campus sports scoring robot based on audience emotion-assisted judgment, characterized in that, Includes a processor and storage, the processor being configured to execute the following computer program steps: The competition motion data of the participants is acquired by a motion capture and posture perception module set up in the competition area. The competition motion data includes at least joint spatial position data, posture change sequence data and motion continuity parameters. The competitive action data is processed by rule matching to generate action rule confidence data, which is used to characterize the degree to which the competitive action meets the preset competitive rule conditions; When the confidence data of the action rule is within the preset rule boundary range, the audience emotion auxiliary judgment process is triggered; Collect audience spatial location data and historical support behavior data, segment the audience groups based on the audience spatial location data and historical support behavior data, and collect audience reaction data for different audience groups separately; The audience reaction data is processed by group emotion modeling to generate group emotion characteristic data corresponding to each audience group; Perform differential analysis on group emotional characteristic data to generate audience emotional difference data; The confidence data of the action rules and the difference data of audience emotions are fused together to generate the final scoring data. The final scoring data will be used to output the competition scoring results.

2. The campus sports scoring robot based on audience emotion-assisted judgment as described in claim 1, characterized in that, The steps for acquiring competitive motion data include: continuously capturing frames of the participants through a motion capture and posture perception module, extracting the spatial position data of the participants' joints, and generating posture change sequence data based on the joint changes between adjacent frames; performing temporal smoothing on the posture change sequence data to obtain motion continuity parameters; and storing the joint spatial position data, posture change sequence data, and motion continuity parameters as components of the competitive motion data.

3. The campus sports scoring robot based on audience emotion-assisted judgment as described in claim 1, characterized in that, The rule matching process for competitive action data includes: matching and calculating the competitive action data with the action threshold conditions corresponding to the preset competitive rules to generate action satisfaction parameters; calculating action rule confidence data based on the action satisfaction parameters; and using the action rule confidence data as the basis for determining whether to trigger the audience emotion auxiliary judgment process.

4. The campus sports scoring robot based on audience emotion-assisted judgment as described in claim 1, characterized in that, Acquiring audience reaction data includes: dividing the audience into at least two audience group areas based on audience spatial location data and historical support behavior data; collecting audience reaction data in each of the different audience group areas, including group voice feature data, group body movement feature data, and group facial expression change feature data, and adding timestamp information to the audience reaction data.

5. The campus sports scoring robot based on audience emotion-assisted judgment according to claim 1, characterized in that, The process of modeling group emotions based on audience reaction data includes: performing energy and spectrum analysis on group voice feature data to generate voice emotion parameters; performing amplitude and frequency analysis on group body movement feature data to generate movement emotion parameters; performing expression change trend analysis on group facial expression change feature data to generate facial expression emotion parameters; and fusing the voice emotion parameters, movement emotion parameters, and facial expression emotion parameters to generate group emotion feature data for the corresponding audience group region.

6. The campus sports scoring robot based on audience emotion-assisted judgment according to claim 1, characterized in that, Differential analysis of group emotion characteristic data includes: comparing and calculating group emotion characteristic data corresponding to different audience groups in different regions, generating group emotion intensity difference parameters and group emotion change synchronicity parameters, and using the group emotion intensity difference parameters and group emotion change synchronicity parameters as audience emotion difference data.

7. The campus sports scoring robot based on audience emotion-assisted judgment as described in claim 1, characterized in that, The fusion and judgment of action rule confidence data and audience emotion difference data includes: aligning the time of action occurrence corresponding to the competitive action data with the timestamp of audience reaction data to generate time-series correlation data, and filtering the audience emotion difference data for validity based on the time-series correlation data.

8. The campus sports scoring robot based on audience emotion-assisted judgment according to claim 7, characterized in that, The fusion determination includes: when the action rule confidence data is within a preset rule boundary range, the action rule confidence data and the audience emotion difference data are weighted and fused according to preset weights to generate corrected confidence data, and the final scoring determination data is generated based on the corrected confidence data.

9. The campus sports scoring robot based on audience emotion-assisted judgment according to claim 1, characterized in that, The output of the competition scoring results includes: output scoring result data and judgment confidence information corresponding to the scoring result data, and the scoring result data and judgment confidence information are stored in the memory.

10. The campus sports scoring robot based on audience emotion-assisted judgment according to claim 9, characterized in that, During the execution of the computer program steps, the processor performs unified data identification and management on the competitive action data, action rule confidence data, audience reaction data, group emotional characteristic data, audience emotional difference data, and final scoring determination data.

Images