A method and system for determining crosstalk in a continuous action of a rhythm boxing

Abstract

This invention discloses a method and system for preventing crosstalk in continuous movements of rhythm boxing, relating to the field of motion-sensing interaction technology. The method includes: constructing a movement timeline corresponding to a rhythm boxing course; configuring a corresponding movement determination window for each preset boxing movement on the movement timeline; receiving movement perception events from a motion-sensing device in real time; mapping the timestamp of the motion-sensing device, the receiving time from the application, and the current course audio playback time to the same global time reference axis; for each movement perception event, retrieving several time-adjacent movement determination windows on the aligned global time reference axis to form a candidate window set; determining the final assigned window for the movement perception event from the candidate window set according to preset hierarchical rules; and outputting the final movement result based on the final assigned window. This method helps solve the problem that existing technologies cannot achieve crosstalk prevention in continuous movements of rhythm boxing.

Description

Technical Field

[0001] This invention relates to the field of motion-sensing interaction technology, and in particular to a method and system for preventing crosstalk in rhythmic boxing continuous movements. Background Technology

[0002] Rhythm boxing courses and games generally consist of continuous punching movements with short intervals between adjacent movements, and often involve alternating between ordinary striking actions and defensive actions such as dodging and blocking. Existing rhythm boxing judgment techniques have several significant shortcomings in practical application: First, the recognition time of the motion sensing device, the receiving time of the application, and the playback time of the course audio are independent of each other and the time base is not unified. The time sequence offset of the multi-source timestamps is very easy to cause errors in the determination of action attribution. Second, when the time interval between the previous and subsequent target actions is close, the recognition result of a single action is easily incorrectly assigned to the previous action window or the subsequent action window, resulting in action mismatch problem.

[0003] Furthermore, in scenarios where ordinary striking actions are arranged adjacent to special actions such as dodging and blocking, the existing judgment mechanism lacks action priority control logic. The recognition result of ordinary punches can easily crowd out the judgment window of special actions, causing the judgment of defensive actions to fail.

[0004] Meanwhile, under conditions of continuous punches or continuous combinations of punches with the same hand, stray jitter signals are prone to causing interference from repeated punches, duplicate judgments, and incorrect backfilling of delayed recognition results, resulting in poor judgment stability.

[0005] Especially in high-BPM, high-density motion courses, the motion arrangement is more compact. Traditional methods that use a single judgment window or simple nearest neighbor matching cannot distinguish adjacent motions or filter redundant interference signals, resulting in problems such as delayed user motion feedback, large judgment deviations, and inconsistent somatosensory interactions, which seriously reduce the training experience and scoring accuracy.

[0006] In summary, existing rhythm boxing judgment technology has several technical defects, including inconsistent time references from multiple sources, misallocation of adjacent movements, easy obstruction of special movements, severe crosstalk in continuous movements, and poor adaptability to high beats. These defects make it impossible to meet the accurate judgment requirements of high-density, continuous alternating boxing movements.

[0007] Therefore, there is an urgent need to design a method for preventing crosstalk in rhythmic boxing continuous movements. Summary of the Invention

[0008] In view of this, the present invention proposes a method and system for anti-crosstalk determination of continuous movements in rhythmic boxing, which can realize anti-crosstalk determination of continuous movements in rhythmic boxing.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A method for preventing crosstalk in rhythmic boxing continuous movements, comprising: Read the beats per minute, course segments, level sections, action timeline, and course events from the rhythm boxing course configuration file, and construct the action timeline corresponding to the rhythm boxing course based on the beats per minute, course segments, level sections, action timeline, and course events; Configure a corresponding action determination window for each preset boxing action on the action timeline; Real-time reception of motion sensing events from motion sensing devices; The timestamps of the motion-sensing device, the receiving time of the application, and the current course audio playback time are uniformly mapped to the same global time reference axis. For each motion-aware event, several motion determination windows with similar times are retrieved on the aligned global time reference axis to form a candidate window set; The final assigned window corresponding to the action-aware event is determined from the candidate window set according to a preset hierarchical rule; The final action result is output based on the final attribution window.

[0010] Based on the above technical solution, the present invention can be further improved as follows: Optionally, the step of reading the beats per minute, course segments, level sections, action timelines, and course events from the rhythm boxing course configuration file, and constructing the action timeline corresponding to the rhythm boxing course based on the beats per minute, course segments, level sections, action timelines, and course events, includes: Calculate the duration of a single beat based on the number of beats per minute; Read the start time, end time, or start beat number of each course segment or level segment to form a time reference for each course segment or level segment in order to determine the segment to which the action belongs; Each action item in the action timeline is analyzed, and the action category, left and right hands and related events of the action item are extracted. The action sequence number is automatically generated in order, and the original time or beat of the action is converted into a unified audio time to obtain the target hit time. By merging the action category, left and right hands, action sequence number, action segment, target hit time and related events, an action timeline is obtained, arranged in ascending order of time.

[0011] Optionally, configuring a corresponding action determination window for each preset boxing action on the action timeline includes: Read the preset window parameters for each preset boxing move, the preset window parameters including advance tolerance time and delay tolerance time; Calculate the start time, best hit time, and judgment end time for each preset boxing action entering the recognition area to obtain the window start and end time for each preset boxing action; Determine the action category, left and right hands, window start and end times, window state, and whether it is a special action for each preset boxing action, and configure a corresponding action judgment window for each preset boxing action on the action timeline.

[0012] Optionally, mapping the timestamp recognized by the motion-sensing device, the receiving time of the application, and the current course audio playback time to the same global time reference axis includes: Determine whether the timestamp identified by the motion sensing device is valid. If valid, use the timestamp identified by the device for time-series mapping. If the timestamp identified by the motion-sensing device is invalid, then the current course audio playback time or the application receiving time will be used for time-series mapping.

[0013] Optionally, determining the final belonging window corresponding to the action-aware event from the candidate window set according to a preset hierarchical rule includes: Perform action type compatibility checks and left / right hand compatibility checks on the candidate windows in the candidate window set, and compare the time distance between the action perception event and the time center of each candidate window. If a conflict still exists, the final belonging window corresponding to the action-aware event is determined based on the preset action priority rules. When both a normal punch window and a special action window exist in the candidate window set, the special action window is protected first.

[0014] Optionally, after determining the final home window corresponding to the action-aware event from the candidate window set according to a preset hierarchical rule, the method further includes: Calculate the dynamic cooldown time of the current action based on the type of the next target action, left and right hands, beat interval, and target hit time.

[0015] Optionally, after determining the final home window corresponding to the action-aware event from the candidate window set according to a preset hierarchical rule, the method further includes: Mark action-aware events that cannot match any candidate window as isolated results; Determine whether the isolated result meets the remedy conditions. If yes, temporarily store the isolated result and wait for a short time for remedy matching. If no, discard the isolated result directly.

[0016] A crosstalk detection system for continuous movements in rhythmic boxing includes: The timeline construction module is used to read the number of beats per minute, course segments, level segments, action timelines, and course events from the rhythm boxing course configuration file, and construct the action timeline corresponding to the rhythm boxing course based on the number of beats per minute, course segments, level segments, action timelines, and course events. The judgment window configuration module is used to configure a corresponding action judgment window for each preset boxing action on the action timeline. The event receiving module is used to receive motion sensing events from the motion sensing device in real time. The alignment module is used to map the timestamp of the motion sensing device, the time received by the application, and the current course audio playback time to the same global time reference axis. The window filtering module is used to search for several action judgment windows that are close in time on the aligned global time reference axis for each action perception event, and form a candidate window set. The attribution determination module is used to determine the final attribution window corresponding to the action-aware event from the candidate window set according to preset hierarchical rules. The result output module is used to output the final action result based on the final attribution window.

[0017] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the method described herein.

[0018] A non-transitory computer-readable storage medium having a computer program stored thereon, the computer program implementing the steps of the method when executed by a processor.

[0019] The present invention has the following advantages: The anti-crosstalk determination method for continuous movements in rhythmic boxing in this invention achieves multi-source timestamp time sequence alignment by constructing a unified global time reference axis, effectively eliminating determination offset errors caused by different time dimensions. This invention forms a candidate set by searching neighboring determination windows and, in conjunction with preset hierarchical attribution determination rules, accurately distinguishes neighboring movements, avoiding misassignment of movements. Simultaneously, relying on standardized matching logic, it avoids the defect of ordinary punches occupying special movement windows, ensuring the normal determination of dodging and blocking movements. Attached Figure Description

[0020] For illustrative and not limiting purposes, the present invention will now be described in conjunction with embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the method for preventing crosstalk in continuous rhythmic boxing movements according to an embodiment of the present invention. Figure 2A schematic diagram of the main components of the anti-crosstalk determination system for continuous rhythmic boxing movements in an embodiment of the present invention. Figure 3 This is a schematic diagram of the physical structure of the electronic device provided by the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be used interchangeably where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] It should be noted that, where there is no conflict, the embodiments and features of the present invention can be combined with each other. The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0024] Figure 1 This is a flowchart illustrating the anti-crosstalk determination method for continuous rhythmic boxing movements in an embodiment of the present invention, as shown below. Figure 1 As shown, the method for preventing crosstalk in rhythmic boxing continuous movements provided in this embodiment of the invention includes the following steps S101 to S107.

[0025] S101 reads the number of beats per minute, course segments, level segments, action timeline, and course events from the rhythm boxing course configuration file, and constructs the action timeline corresponding to the rhythm boxing course based on the number of beats per minute, course segments, level segments, action timeline, and course events.

[0026] Read the rhythm boxing course and construct a complete action timeline based on its BPM, section, timeline, and events. BPM (Beats Per Minute) represents the number of beats per minute, used to determine the course rhythm and beat duration; section represents a segment or level segment, used to determine the start and end times, action groups, difficulty, or stance; timeline represents the action timeline, used to record target actions in chronological order; events represent course events, used to record musical cues, commands, start / end signals, vibrations, visual cues, or other auxiliary control information.

[0027] When constructing the timeline, first calculate the duration of a single beat based on the number of beats per minute, for example, beatMs=60000 / BPM; then read the start time, end time, or start beat number of each section to form the time base for each course segment or level segment to determine the segment to which the action belongs; then parse the action items in the timeline in sequence, extract the action category, left and right hands, and related events of the action item, and automatically generate action sequence numbers in order, converting the time or beat in the action item into a unified audio time audioTimeMs; finally, merge the action category, left and right hands, action sequence number, segment to which the action belongs, target hit time, and related events to obtain an action timeline arranged in ascending order of time.

[0028] A preferred embodiment is as follows: the course configuration file is set with 150 BPM and 356 boxing moves, and the timeline is configured with moves such as straight punch, slip, and hook, along with their specific time points. Since 150 BPM corresponds to a beat duration of approximately 400 ms, if a section starts at audio 10000 ms, and the 12th move in the timeline is configured with time=12800ms, action=hook, hand=right, and index=12, then the established timeline record can be represented as: index=12, section=0, action=hook, hand=right, targetTimeMs=12800ms, beat=32, and can be associated with sound effects, commands, or vibration cues near that time.

[0029] Preset boxing moves: The rhythm boxing course pre-sets all interactive actions that users need to complete in sync with the music. It is divided into two categories: ① Normal punching moves: offensive punches such as straight punch, hook punch, and uppercut; ② Special defensive moves: defensive moves such as dodging, blocking, and diving. (2) Course segments: The complete boxing course is divided into independent sections according to difficulty / style. Each segment has an independent start and end time and beat interval; (3) Level segment: A complete training level consisting of several consecutive course segments, corresponding to a complete training process; (4) Course events: Auxiliary interactive events bound to the course timeline, including music prompts, voice commands, device vibration, visual effects, and score trigger markers; (5) Related events: Exclusive course events tied to a single boxing move, such as sound effects, vibration feedback, and animation prompts when the punch hits; (6) Optimal hit time: The theoretical optimal hitting time for each preset boxing action. The calculation formula is: Optimal hit time = target hit time (targetTimeMs); the center of the window is the optimal hit time. (7) Window Status: Used to mark the lifecycle of the judgment window, including four states: inactive, matchable, settled, and locked (cooldown). Not activated: The event time has not been included in the window start time and is not included in the matching. Matchable: The event falls within the start and end range of the window, allowing for matching of motion-aware events; Settled: The window has successfully matched the action and will no longer receive new recognition events; Cooling lock: After matching is completed, a dynamic cooling cycle is entered to suppress repeated matching of stray signals.

[0030] S102, Configure a corresponding action judgment window for each preset boxing action on the action timeline.

[0031] First, read the preset window parameters for the action, including the early tolerance time earlyMs and the late tolerance time lateMs; Then calculate the start time, best hit time, and decision end time for entering the recognition area; finally, write the action category, left or right hand, window start and end time, window state, and whether it is a special action into the window object.

[0032] For example, if a right-hand hook has a targetTimeMs of 12800ms, and a normal punch is configured with earlyMs=300ms and lateMs=250ms, then the action window is 12500ms to 13050ms, with 12800ms being the optimal hit time. If a slip has a targetTimeMs of 13600ms, and a special action is configured with earlyMs=250ms and lateMs=300ms, then the action window is 13350ms to 13900ms. Only events falling within the corresponding window are allowed to participate in the action's attribution and scoring.

[0033] The action determination window includes the recognition area start time, the best hit time, and the determination end time.

[0034] Different types of boxing moves are configured with different window parameters and judgment strategies; among them, ordinary striking moves use general window parameters, while special defensive moves such as dodging and blocking use exclusive judgment strategies to adapt to the movement characteristics of different moves.

[0035] S103 receives motion sensing events from the motion sensing device in real time.

[0036] Among them, motion perception events are divided into striking motion events and defensive motion events. Ordinary punching motions rely on the motion recognition model to output recognition results, while special motions such as dodging and blocking rely on the sensor signals collected by acceleration sensors and angular velocity sensors to generate trigger results, providing raw data support for subsequent judgment processes.

[0037] Complete steps for building an action recognition model: S01 Sample Collection: Collects a large number of user striking action samples. The collection device is equipped with an accelerometer and an angular velocity sensor. The samples include various common striking actions such as straight punches, hook punches, and uppercuts. Action labels and left and right hand markings are simultaneously added. S02 Sample Preprocessing: The original sensor signal is denoised, sliced ​​by sliding window, and normalized to extract time-domain features such as peak velocity, acceleration extremum, and motion duration. S03 Model Training: A lightweight classification neural network is used as the basic model, with preprocessed sensor features as input and action type and left / right hand as classification labels to complete supervised training. S04 Model Deployment: After training, the lightweight model is deployed to the local terminal, receiving the controller sensor data stream in real time and outputting the corresponding ordinary punch action recognition results; Differentiation Mechanism: The amplitude and acceleration characteristics of dodging and blocking actions are very different from those of striking actions. No neural network model is required. Recognition is triggered directly by the sensor signal threshold, thereby reducing the terminal's computing power consumption.

[0038] S104 maps the timestamp of the motion sensing device, the time received by the application, and the current course audio playback time to the same global time reference axis.

[0039] Determine whether the timestamp identified by the motion sensing device is valid. If valid, use the timestamp identified by the device for time-series mapping. If the timestamp identified by the motion sensing device is invalid, the current course audio playback time or the application receiving time is used for time-series mapping to avoid the judgment offset problem caused by the inconsistency of multiple source time bases.

[0040] In addition to audio time, a unified time base can also be the course master clock, rendering clock, or server synchronization clock.

[0041] S105: For each motion-aware event, retrieve several motion determination windows that are close in time on the aligned global time reference axis to form a candidate window set.

[0042] The candidate window set can be determined as follows: compare the event time eventTimeMs after the action-aware event is mapped with each action judgment window. If windowStartMs≤eventTimeMs≤windowEndMs, or if |eventTimeMs-windowCenterMs| is less than the preset candidate distance threshold, then the action judgment window is added to the candidate window set.

[0043] The candidate distance threshold must not be less than the single-sided fault tolerance time of the window, ensuring that all action events within the complete decision window can be included in the candidate set. Narrowing the matching search range avoids interference from distant and irrelevant windows in action determination, adapting to high-BPM and high-density compact action layout scenarios.

[0044] S106, determine the final window to which the motion-aware event belongs from the candidate window set according to the preset hierarchical rules.

[0045] Perform action type compatibility checks and left / right hand compatibility checks on the candidate windows in the candidate window set, and compare the time distance between the action perception event and the time center of each candidate window. If a matching conflict still exists after multiple levels of verification, the final belonging window corresponding to the action-aware event is determined based on the preset action priority rules.

[0046] When there are multiple windows in the candidate set, they are assigned in the following order: First, determine if the action types are compatible: Action type compatibility is determined through an action mapping table. For standard punches, straight, hook, and uppercut preferably require the identified category to match the target category; if the target action contains directional prefixes such as left_straight, right_hook, etc., it is first normalized to basic punch types such as straight and hook, and then the left and right hands are considered together. Special actions such as dodging, blocking, and diving are classified as special action categories and are not compatible with standard punches; different names for the same special action, such as slip, left_slip, right_slip, or dodge, can be mapped to the slip class first, and then the direction or triggering hand is considered together.

[0047] Next, determine left-hand compatibility: When the target window specifies hand=left or hand=right, the hand side of the identified event must be consistent with the target hand side; when the target is triggered by both hands or a pair of special actions, both left and right windows are allowed to be calculated together, but the primary triggering hand is used as the basis for attribution. If the identified event has no hand side information, it can only be matched with windows that do not require hand side distinction, or it will be downgraded to a low-priority candidate.

[0048] Next, compare the distance between the identified event and the center time of each window: In candidate windows that are compatible with both action types and left and right hands, calculate the absolute difference between the identified event time eventTimeMs and the center time windowCenter of each window. The smaller the difference, the higher the time matching degree, and the window that is given priority is the final assigned window.

[0049] If a conflict still exists, the final assigned window will be determined according to a preset priority. The preset priority is based on the risk of crosstalk and the determinism of the action. The example order is as follows: settled windows will no longer participate; special action windows take priority over ordinary punch windows; windows with fully compatible action categories and compatibility with both left and right hands take priority; special actions with compatible action categories but requiring direction conversion are secondary; those closer to the center of the window in time take priority; if they are still the same, the window that has not yet been settled and has an earlier sequence number will be selected to avoid a single recognition result hitting multiple target actions repeatedly.

[0050] This method reduces misallocation between adjacent actions.

[0051] A special action priority protection mechanism is set up. When both a normal punch window and a special action window exist in the candidate window set, the special action window is given priority for protection. When both a normal punch window and a special action window exist in the candidate window set, and the action perception event meets the triggering condition or characteristic condition of the special action, the action perception event is given priority to be assigned to the special action window. If the action perception event only meets the normal punch action condition but does not meet the special action condition, it is not forced to be assigned to the special action window.

[0052] The special action priority protection mechanism can be applied to actions such as dodging, blocking, and diving.

[0053] After determining the final home window corresponding to the motion-aware event from the candidate window set according to preset hierarchical rules, the method further includes: The dynamic cooldown time calculation logic is as follows: Based on the type, hand, beat interval, and current action hit time of the next target action, the dynamic cooldown time is calculated using a preset formula. If the next action and the current action are consecutive actions performed with the same hand, the corresponding hand cooldown time is extended. If there are no-beats between adjacent actions, a compensation time corresponding to the number of no-beats is added to the base cooldown time. Invalid recognition results are suppressed, ignored, or delayed during the cooldown period. The isolated result processing logic is as follows: Action perception events that cannot be matched with any candidate window are marked as isolated results, and isolated results are prohibited from being backfilled into already settled historical action windows. It is determined whether the isolated result meets the remediation conditions. If so, the isolated result is temporarily stored and awaits short-term remediation matching; otherwise, the isolated result is directly discarded to reduce stray signal interference.

[0054] In the formula: T is the preset base cooling time; if the next action is a continuous action of the same hand as the current action, then H>1 is set to extend the cooling time of the corresponding hand; for actions of different hands, H=1 is used; N is the number of idle shots between the current action and the next action; K is the unit idle shot compensation time, and the compensation time is added when there are idle shots. During the cooling period, the system suppresses, ignores, or delays the generation of invalid and spurious identification results.

[0055] Dynamic cooling can be set according to action type, hand position, beat difference, or user's historical performance.

[0056] S107, output the final action result based on the final destination window.

[0057] Simultaneously, the final action result is distributed to the display layer, scoring layer, sound effect layer, vibration layer, and log layer to achieve multi-dimensional feedback, data retention, and interaction optimization, thus completing the action judgment process.

[0058] A preferred embodiment is that the present invention can be further extended to scenarios such as two-player battles, multiplayer online rhythm boxing, or scenarios with visual recognition fusion.

[0059] Figure 2 This is a schematic diagram of the main components of the anti-crosstalk determination system for continuous rhythmic boxing movements in an embodiment of the present invention. Figure 2 As shown, the anti-crosstalk determination system 1 for continuous rhythmic boxing movements provided in this embodiment of the invention includes a reading module 10, a time axis construction module 20, a determination window configuration module 30, an event receiving module 40, an alignment module 50, a window filtering module 60, an attribution determination module 70, and a result output module 80.

[0060] The reading module 10 is used to read the number of beats per minute, course segments, level segments, action timeline and course events from the rhythm boxing course configuration file; Timeline construction module 20 is used to construct the action timeline corresponding to the rhythm boxing course based on the number of beats per minute, course segments, level sections, action timelines and course events; The determination window configuration module 30 is used to configure a corresponding action determination window for each preset boxing action on the action timeline. Event receiving module 40 is used to receive motion sensing events from the motion sensing device in real time; Alignment module 50 is used to map the timestamp of the motion sensing device, the time received by the application, and the current course audio playback time to the same global time reference axis. The window filtering module 60 is used to search for several action judgment windows that are close in time on the aligned global time reference axis for each action perception event, and form a candidate window set. The attribution determination module 70 is used to determine the final attribution window corresponding to the action-aware event from the candidate window set according to a preset hierarchical rule. The result output module 80 is used to output the final action result based on the final attribution window.

[0061] Figure 3 This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as... Figure 3 As shown, the electronic device 90 includes: a processor 901, a memory 902, and a bus 903; The processor 901 and the memory 902 communicate with each other via the bus 903. The processor 901 is used to call program instructions in the memory 902 to execute the methods provided in the above-described method embodiments, and to execute the methods provided in the embodiments of the present invention.

[0062] This embodiment provides a non-transitory computer-readable storage medium that stores computer instructions, which cause a computer to execute the method provided in this embodiment of the invention.

[0063] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various storage media capable of storing program code, such as ROM, RAM, magnetic disk, or optical disk.

[0064] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for preventing crosstalk in continuous movements of rhythmic boxing, characterized in that, include: Read the beats per minute, course segments, level sections, action timeline, and course events from the rhythm boxing course configuration file, and construct the action timeline corresponding to the rhythm boxing course based on the beats per minute, course segments, level sections, action timeline, and course events; Configure a corresponding action determination window for each preset boxing action on the action timeline; Real-time reception of motion sensing events from motion sensing devices; The timestamps of the motion-sensing device, the receiving time of the application, and the current course audio playback time are uniformly mapped to the same global time reference axis. For each motion-aware event, several motion determination windows with similar times are retrieved on the aligned global time reference axis to form a candidate window set; The final assigned window corresponding to the action-aware event is determined from the candidate window set according to a preset hierarchical rule; The final action result is output based on the final attribution window.

2. The method for preventing crosstalk in continuous rhythmic boxing movements according to claim 1, characterized in that, The process involves reading the beats per minute, course segments, level sections, action timelines, and course events from the rhythm boxing course configuration file, and constructing the corresponding action timeline for the rhythm boxing course based on these data. This includes: Calculate the duration of a single beat based on the number of beats per minute; Read the start time, end time, or start beat number of each course segment or level segment to form a time reference for each course segment or level segment in order to determine the segment to which the action belongs; Each action item in the action timeline is analyzed, and the action category, left and right hands and related events of the action item are extracted. The action sequence number is automatically generated in order, and the original time or beat of the action is converted into a unified audio time to obtain the target hit time. By merging the action category, left and right hands, action sequence number, action segment, target hit time and related events, an action timeline is obtained, arranged in ascending order of time.

3. The method for preventing crosstalk in continuous rhythmic boxing movements according to claim 1, characterized in that, The step of configuring a corresponding action determination window for each preset boxing action on the action timeline includes: Read the preset window parameters for each preset boxing move, the preset window parameters including advance tolerance time and delay tolerance time; Calculate the start time, best hit time, and judgment end time for each preset boxing action entering the recognition area to obtain the window start and end time for each preset boxing action; Determine the action category, left and right hands, window start and end times, window state, and whether it is a special action for each preset boxing action, and configure a corresponding action judgment window for each preset boxing action on the action timeline.

4. The method for preventing crosstalk in continuous rhythmic boxing movements according to claim 1, characterized in that, The process of mapping the timestamp of the motion-sensing device, the receiving time of the application, and the current course audio playback time to the same global time reference axis includes: Determine whether the timestamp identified by the motion sensing device is valid. If it is valid, use the timestamp identified by the motion sensing device for time-series mapping. If the timestamp identified by the motion-sensing device is invalid, then the current course audio playback time or the application receiving time will be used for time-series mapping.

5. The method for preventing crosstalk in continuous rhythmic boxing movements according to claim 1, characterized in that, The step of determining the final belonging window corresponding to the action-aware event from the candidate window set according to a preset hierarchical rule includes: Perform action type compatibility checks and left / right hand compatibility checks on the candidate windows in the candidate window set, and compare the time distance between the action perception event and the time center of each candidate window. If a conflict still exists, the final belonging window corresponding to the action-aware event is determined based on the preset action priority rules. When both a normal punch window and a special action window exist in the candidate window set, the special action window is protected first.

6. The method for preventing crosstalk in continuous rhythmic boxing movements according to claim 1, characterized in that, After determining the final home window corresponding to the motion-aware event from the candidate window set according to preset hierarchical rules, the method further includes: Calculate the dynamic cooldown time of the current action based on the type of the next target action, left and right hands, beat interval, and target hit time.

7. The method for preventing crosstalk in continuous rhythmic boxing movements according to claim 1, characterized in that, After determining the final home window corresponding to the motion-aware event from the candidate window set according to preset hierarchical rules, the method further includes: Mark action-aware events that cannot match any candidate window as isolated results; Determine whether the isolated result meets the remedy conditions. If yes, temporarily store the isolated result and wait for a short time for remedy matching. If no, discard the isolated result directly.

8. A crosstalk detection system for continuous movements in rhythmic boxing, characterized in that, include: The read module is used to read the beats per minute, course segments, level sections, action timeline, and course events from the rhythm boxing course configuration file; The timeline construction module is used to construct the action timeline corresponding to the rhythm boxing course based on the number of beats per minute, course segments, level segments, action timelines, and course events. The judgment window configuration module is used to configure a corresponding action judgment window for each preset boxing action on the action timeline. The event receiving module is used to receive motion sensing events from the motion sensing device in real time. The alignment module is used to map the timestamp of the motion sensing device, the time received by the application, and the current course audio playback time to the same global time reference axis. The window filtering module is used to search for several action judgment windows that are close in time on the aligned global time reference axis for each action perception event, and form a candidate window set. The attribution determination module is used to determine the final attribution window corresponding to the action-aware event from the candidate window set according to preset hierarchical rules. The result output module is used to output the final action result based on the final attribution window.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images