Real-time swimming stroke identification method and system

Abstract

The invention relates to the technical field of swimming training assistance, in particular to a real-time swimming stroke recognition method and system, which can utilize a head-mounted six-axis sensor to obtain speed data in a swimming process in real time; performing feature recognition on the real-time speed data by using the trained deep neural network to obtain a recognition result carrying the swimming stroke type and the swimming stroke data; and wirelessly receiving the identification result by using a client, and displaying the swimming stroke type and the swimming stroke data carried by the identification result in real time. It can be understood that according to the technical scheme shown by the invention, a coach can monitor the swimming stroke type and the swimming stroke data of the swimmer in real time, then the swimmer can be guided to adjust and optimize actions in time in the training process, the provided swimming stroke data can enable the coach to accurately control the swimming actions of the swimmer, and a better training effect is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of swimming training assistance, and in particular to a real-time swimming posture recognition method and system. Background Art

[0002] In current swimming training practice, the recognition and analysis of swimming styles mainly rely on the manual observation and experience guidance of coaches. The coaches observe the swimmers' movements with their naked eyes and use their professional knowledge and experience to judge whether the swimming style is correct and whether it needs to be improved.

[0003] However, this traditional training method has many limitations. First, manual observation cannot achieve real-time monitoring, and coaches can only give feedback after training, which undoubtedly reduces the efficiency of training. Secondly, due to the limitations of human factors, it is difficult for coaches to make completely accurate observations, especially when observing multiple athletes continuously or training for a long time. Fatigue and distraction may lead to deviations in observation results. In addition, manual observation cannot provide quantitative data support, and lacks a grasp of the subtle differences in swimming styles and the accuracy of movements, which limits the trainees' in-depth understanding and improvement of their own techniques. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a real-time swimming style recognition method and system to solve the problems in the prior art that manual observation methods cannot achieve real-time monitoring, cannot timely adjust and optimize movements during training, and cannot provide quantitative data support.

[0005] According to a first aspect of an embodiment of the present invention, a real-time swimming posture recognition method is provided, comprising:

[0006] Use a head-mounted six-axis sensor to obtain real-time speed data during swimming;

[0007] Use the trained deep neural network to perform feature recognition on real-time speed data, and obtain recognition results that carry swimming style types and swimming style data;

[0008] The recognition result is wirelessly received by the client, and the swimming style type and swimming style data carried by the recognition result are displayed in real time.

[0009] Preferably, the method of using a head-mounted six-axis sensor to obtain speed data in real time during swimming includes:

[0010] Use a head-mounted six-axis sensor to obtain three-dimensional acceleration and three-dimensional angular velocity in real time during swimming;

[0011] The three-dimensional acceleration and the three-dimensional angular velocity are converted into a one-dimensional time series signal, and the one-dimensional time series signal is used as velocity data.

[0012] Preferably, the method of using a trained deep neural network to perform feature recognition on real-time speed data includes:

[0013] Using a trained deep neural network, extract time domain features from the one-dimensional time series signal, perform fast Fourier transform on the one-dimensional time series signal, and extract frequency domain features;

[0014] According to the time domain features and the frequency domain features, a pre-constructed recognition model is used to obtain a recognition result carrying the swimming style type and swimming style data.

[0015] Preferably, the method further comprises:

[0016] deriving the swimmer's movement frequency and movement amplitude according to the swimming stroke data;

[0017] The action frequency and the action amplitude are displayed on the client.

[0018] Preferably, the method further comprises:

[0019] Using the client to time the holding time of the swimming posture type;

[0020] The holding time of the described swimming stroke type is recorded and displayed.

[0021] Preferably, the method further comprises:

[0022] The client is used to analyze the swimming style type and the swimming style data, and to generate and display training suggestions.

[0023] According to a second aspect of an embodiment of the present invention, a real-time swimming style recognition system is provided, comprising:

[0024] A data acquisition and recognition module is used to use a head-mounted six-axis sensor to obtain speed data in real time during swimming, and use a trained deep neural network to perform feature recognition on the real-time speed data to obtain an identification result containing the swimming style type and swimming style data;

[0025] The intelligent terminal is wirelessly connected to the data acquisition and identification module, receives the identification result by using the client running on the intelligent terminal, and displays the swimming style type and swimming style data carried by the identification result in real time.

[0026] Preferably, the data acquisition and identification module includes:

[0027] Head-mounted six-axis sensor, head-mounted microcontroller and wireless connection module;

[0028] The head-mounted six-axis sensor transmits the acquired three-dimensional acceleration and three-dimensional angular velocity to the head-mounted single-chip microcomputer;

[0029] The head-mounted single-chip computer converts the three-dimensional acceleration and the three-dimensional angular velocity into a one-dimensional time series signal, and uses a trained deep neural network to perform feature recognition on the one-dimensional time series signal to obtain a recognition result;

[0030] The wireless connection module sends the recognition result to the smart terminal.

[0031] Preferably, the wireless connection module is a WIFI module or a Bluetooth module.

[0032] Preferably, the data acquisition and identification module further includes:

[0033] A rechargeable battery is used to provide power to the head-mounted six-axis sensor, the head-mounted single-chip microcomputer and the wireless connection module.

[0034] The technical solution provided by the embodiments of the present invention may have the following beneficial effects:

[0035] It is understandable that the technical solution shown in the present invention can use a head-mounted six-axis sensor to obtain speed data in real time during swimming; use a trained deep neural network to perform feature recognition on the real-time speed data to obtain an identification result that carries the swimming style type and swimming style data; use a client to wirelessly receive the identification result, and display the swimming style type and swimming style data carried by the identification result in real time. It is understandable that the technical solution shown in the present invention enables the coach to monitor the swimmer's swimming style type and swimming style data in real time, and then can guide the swimmer to adjust and optimize the movements in time during training, and the provided swimming style data can enable the coach to accurately control the swimmer's swimming movements, achieving better training effects.

[0036] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0038] Figure 1 is a schematic diagram of steps of a real-time swimming posture recognition method according to an exemplary embodiment;

[0039] Figure 2 The figure is a schematic diagram of a real-time swimming posture recognition application according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0041] In one embodiment, Figure 1 is a schematic diagram of the steps of a real-time swimming posture recognition method according to an exemplary embodiment. Figure 1 , a real-time swimming posture recognition method is provided, comprising:

[0042] Step S11: using a head-mounted six-axis sensor to obtain speed data during swimming in real time.

[0043] Step S12: Use the trained deep neural network to perform feature recognition on the real-time speed data to obtain a recognition result that carries the swimming style type and swimming style data.

[0044] Step S13: using the client to wirelessly receive the recognition result, and displaying the swimming style type and swimming style data carried in the recognition result in real time.

[0045] In specific practice, a data acquisition and identification module can be fixed on the swimmer's head, and the data acquisition and identification module includes a head-mounted six-axis sensor, such as Figure 2 As shown, the data acquisition and identification module can be connected to the client via wireless communication. Preferably, the data acquisition and identification module can be configured to be light and strong to ensure that it remains stable during swimming and does not hinder the athlete's vision and movements.

[0046] A single-chip microcomputer can be set in the data acquisition and recognition module, and the single-chip microcomputer is used to perform data analysis and feature recognition, and the recognition results are sent to the client. After the client receives the recognition results, the data is displayed, thereby realizing real-time monitoring and analysis of the data. In addition, the single-chip microcomputer can be powered by a lithium battery to ensure its independent operation and long-term use requirements.

[0047] It should be noted that the use of a head-mounted six-axis sensor to obtain real-time speed data during swimming includes: using a head-mounted six-axis sensor to obtain real-time three-dimensional acceleration and three-dimensional angular velocity during swimming; converting the three-dimensional acceleration and the three-dimensional angular velocity into a one-dimensional time series signal, and using the one-dimensional time series signal as speed data.

[0048] The three-dimensional acceleration and the three-dimensional angular velocity reflect the dynamic behavior of the swimmer in the water and are key information for identifying swimming styles. For example, the breaststroke is characterized by the arms extending forward into the water at the same time and the legs making frog-leg-like movements; the butterfly stroke is characterized by its continuous whole-body wave movements and the simultaneous swinging of the arms; and the freestyle is usually characterized by alternating swinging arms and kicking legs. These different swimming styles will show different patterns in the acceleration and angular velocity data.

[0049] In practice, a head-mounted six-axis sensor is fixed on the swimmer's head to obtain accurate motion posture data. The six-axis sensor is a composite sensor that integrates an accelerometer and a gyroscope, which can simultaneously measure three-axis acceleration and three-axis angular velocity. When the swimmer swims, the sensor will record the movement trajectory of his head in real time, including the acceleration changes in the front and back, left and right, and up and down, as well as the rotation speed around these three axes.

[0050] Preferably, after acquiring the three-dimensional acceleration and the three-dimensional angular velocity, preprocessing operations such as denoising and normalization may be performed on the collected raw data to improve the accuracy of subsequent feature extraction and classification.

[0051] Converting the three-dimensional acceleration and the three-dimensional angular velocity into a one-dimensional time series signal can facilitate subsequent deep neural network analysis and feature extraction of the data.

[0052] It should be noted that, with regard to swimming style recognition, the use of a trained deep neural network to perform feature recognition on real-time speed data includes: using the trained deep neural network to extract time domain features of the one-dimensional time series signal, performing fast Fourier transform on the one-dimensional time series signal, and extracting frequency domain features; based on the time domain features and the frequency domain features, a pre-constructed recognition model is used to derive a recognition result that carries the swimming style type and swimming style data.

[0053] A deep neural network is a multi-layer neural network structure that can automatically learn and extract features from raw data and then classify based on these features. In the present invention, the deep neural network is specially trained in advance to recognize six different swimming styles. During the training process, the network adjusts its internal parameters to better match the relationship between the input data and the swimming style labels.

[0054] In practice, deep neural networks are specially designed and trained to efficiently process and identify one-dimensional signals. Deep neural networks have the ability to self-learn and extract features from large amounts of data, which enables them to accurately classify swimmers' swimming styles and maintain a high recognition rate even in complex underwater environments.

[0055] Deep neural networks can extract time domain features from one-dimensional time series signals, such as mean, variance, maximum, minimum, etc. The frequency domain features of the signal are extracted through methods such as Fast Fourier Transform (FFT). Fast Fourier Transform (FFT) is an efficient algorithm for calculating Discrete Fourier Transform (DFT) and its inverse transform. The main formula of DFT is as follows:

[0056] Given a sequence of length N (x[n]), its DFT is defined as:

[0057]

[0058] Where X[k] is the transformed complex sequence, representing the signal in the frequency domain, (x[n]) is the signal sequence in the original time domain, N is the length of the sequence, k is the frequency index, ranging from 0 to N-1, and i is the imaginary unit, satisfying i 2 =-1. It is the core of Fourier transform and represents the rotation factor. The FFT algorithm uses properties such as symmetry and periodicity to reduce the computational complexity of DFT from (O(N 2 )) is reduced to O(N\logN).

[0059] The recognition model uses a lightweight ResNet model that can specifically process one-dimensional signals. The model uses a residual module to improve the training efficiency of the model, allowing the training of deeper network structures. It uses a one-dimensional convolution layer to process one-dimensional signals, reducing the number of parameters and improving computational efficiency. It uses deep separable convolution to further reduce model complexity and computational complexity. In terms of model training and optimization, the training data is enhanced through time expansion, window slicing and other methods to improve the generalization ability of the model. Select an appropriate loss function, such as the cross entropy loss function, to optimize the classification performance of the model.

[0060] In terms of swimming posture recognition, a multi-classifier that can recognize multiple swimming postures (such as freestyle, breaststroke, butterfly stroke, etc.) can be designed in the deep neural network, and the model fusion technology is used to combine the prediction results of multiple prediction models to improve the recognition accuracy. The model is compressed and accelerated through model pruning, quantization and other technologies to meet the needs of real-time monitoring.

[0061] It is understandable that in the real-time processing process, the deep neural network will quickly and accurately analyze the data collected by the six-axis sensor, and maintain a high recognition rate even during continuous swimming. The application of this technology not only improves the accuracy of swimming style recognition, but also provides swimmers with a real-time feedback system to help them understand and adjust their swimming style instantly, thereby improving training effects and competition performance.

[0062] It should be noted that the method further includes: obtaining the swimmer's movement frequency and movement amplitude according to the swimming posture data; and displaying the movement frequency and the movement amplitude on the client.

[0063] During swimming training, client monitoring plays a vital role, which enables the coach to accurately monitor and analyze the swimmer's performance in a digital way. The client of this embodiment enables the coach to monitor multiple key indicators of the swimmer in real time, including the movement frequency and the movement amplitude.

[0064] In practice, the coach can monitor the swimmer's swimming style in real time through the client, which is one of the core functions of the client. Through the recognition results received by wireless communication technology, the client can display the type of swimming style currently adopted by the swimmer, such as breaststroke, butterfly stroke or freestyle. In addition, the client can also derive the swimmer's movement frequency based on the swimming style data. The movement frequency shown, that is, the number of strokes or kicks completed per unit time, is crucial for evaluating the swimmer's rhythm and endurance.

[0065] In addition to swimming style and movement frequency, the client can also monitor the swimmer's range of motion. The size of the range of motion is directly related to swimming efficiency. Too small a range of motion may lead to insufficient propulsion, while too large a range of motion may increase unnecessary energy consumption. Through the client, the coach can observe the details of each movement of the swimmer and provide precise guidance.

[0066] Preferably, the interface design of the client should also consider ease of use and intuitiveness of information, which needs to be specifically set according to specific circumstances and is not specifically limited here. Preferably, the swimmer's real-time data can generate corresponding graphics, which are clearly displayed in the form of graphics and numbers, so that the coach can see all key indicators at a glance.

[0067] It should be noted that the method further includes: using the client to time the holding time of the swimming style type; recording and displaying the holding time of the swimming style type.

[0068] In practice, when the client detects that the swimmer starts to perform a swimming stroke, it starts real-time timing, and stops timing when the swimmer changes to the next swimming stroke, and records and displays the time the swimming stroke type is maintained. This is of great significance for analyzing the timing of the swimmer's swimming stroke transition and the overall training plan.

[0069] Preferably, the client can also continuously store historical data to facilitate coaches and swimmers to conduct long-term tracking and analysis of training effects.

[0070] It should be noted that the method further includes: using the client to analyze the swimming style type and the swimming style data, and generate and display training suggestions.

[0071] In specific practice, training suggestions corresponding to the swimming stroke type and the swimming stroke data can be stored in advance on the server. When the swimmer's swimming stroke type and swimming stroke data are monitored to meet the corresponding template, the client is used to generate and display training suggestions to help coaches and swimmers analyze the training effect, thereby optimizing swimming skills and improving training efficiency.

[0072] Understandably, this comprehensive monitoring system has greatly improved the scientific and systematic nature of training, enabling coaches to develop training plans based on data rather than experience alone, thereby helping swimmers improve their skills and physical fitness in a more efficient way.

[0073] According to a second aspect of an embodiment of the present invention, a real-time swimming style recognition system is provided, comprising:

[0074] A data acquisition and recognition module is used to use a head-mounted six-axis sensor to obtain speed data in real time during swimming, and use a trained deep neural network to perform feature recognition on the real-time speed data to obtain an identification result containing the swimming style type and swimming style data;

[0075] The intelligent terminal is wirelessly connected to the data acquisition and identification module, receives the identification result by using the client running on the intelligent terminal, and displays the swimming style type and swimming style data carried by the identification result in real time.

[0076] It should be noted that the data acquisition and identification module includes: a head-mounted six-axis sensor, a head-mounted single-chip microcomputer and a wireless connection module.

[0077] The head-mounted six-axis sensor transmits the acquired three-dimensional acceleration and three-dimensional angular velocity to the head-mounted single-chip microcomputer.

[0078] The head-mounted single-chip microcomputer converts the three-dimensional acceleration and the three-dimensional angular velocity into a one-dimensional time series signal, and uses a trained deep neural network to perform feature recognition on the one-dimensional time series signal to obtain a recognition result.

[0079] The wireless connection module sends the recognition result to the smart terminal.

[0080] In practice, the six-axis sensor is closely connected to the microcontroller. The six-axis sensor can collect the swimmer's motion data in three-dimensional space in real time, including acceleration and angular velocity. These data are the key to analyzing swimming styles, because different swimming styles will show different characteristics in these data. The accuracy and response speed of the six-axis sensor ensure the reliability and real-time nature of the data.

[0081] After receiving the data collected by the six-axis sensor, the MCU will immediately process it. Through the built-in deep neural network model, the MCU can quickly identify the swimmer's swimming style, such as breaststroke, butterfly stroke or freestyle. The recognition results will be sent to the client in real time through the wireless connection module, and the client software will visualize the data, so that coaches and swimmers can instantly understand the current swimming style status and make corresponding adjustments and optimizations. This real-time feedback mechanism greatly improves the efficiency and effectiveness of training.

[0082] The data acquisition and identification module is made into a head-mounted module, which includes a head-mounted six-axis sensor and a head-mounted single-chip microcomputer. The head-mounted single-chip microcomputer is a lightweight and sturdy electronic device designed to fit snugly on the swimmer's head to capture and transmit motion data with minimal interference. The six-axis sensor connected to the single-chip microcomputer is used to monitor the swimmer's three-dimensional motion state in the water in real time, including acceleration and angular velocity. At the same time, the head-mounted module is equipped with a rechargeable battery to ensure long-term operation without the need for an external power supply. In addition, the built-in wireless connection module enables the single-chip microcomputer to wirelessly transmit the collected data to external devices, such as clients. This design allows swimmers to train in a natural swimming state without worrying about the limitations of the equipment.

[0083] The client is a software application that can run on a smart terminal, such as a smartphone, tablet or computer. Its main function is to receive real-time data sent by the head-mounted microcontroller through the wireless connection module and convert the data into easy-to-understand graphics and text information. The interface design of the client allows users to easily view key indicators such as the swimmer's stroke type, swimming frequency, and range of motion. In addition, the client can also provide real-time feedback and training suggestions to help coaches and swimmers analyze training effects, thereby optimizing swimming skills and improving training efficiency. The user-friendly design of the client makes data monitoring and analysis simple and intuitive, and is a valuable tool in swimming training.

[0084] The wireless connection module is a WIFI module or a Bluetooth module.

[0085] It is understandable that the real-time swimming style recognition system shown in this embodiment can bring better changes to swimming training. The application of this technology can significantly improve the efficiency of swimming training, because it can instantly capture the swimmer's movements and accurately determine the type of swimming style, thereby providing accurate data support for training. This real-time feedback mechanism allows swimmers to immediately understand whether their movements are correct, and then quickly adjust, avoiding the inefficiency caused by feedback delays in traditional training.

[0086] In addition, this system makes it easier for coaches to provide personalized guidance to swimmers. Through in-depth analysis of swimmers' swimming style, frequency, range of motion and other data, coaches can develop personalized training plans based on the characteristics of each swimmer, thereby improving training results. This personalized approach can better tap the potential of each swimmer and improve their technical level and competitive state.

[0087] At the same time, the system has the advantages of simple structure, easy use and high precision. The head-mounted microcontroller is compact and does not affect the normal movement of the swimmer, while the intuitive interface of the client software makes it easy to interpret and analyze the data. During use, there is no need for complicated settings, and users can quickly get started, which makes the system suitable not only for the training of professional athletes, but also for amateur swimming enthusiasts. High-precision sensors and data processing algorithms ensure the reliability of the collected data, providing swimmers with a solid foundation for scientific training.

[0088] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0089] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0090] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0091] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0092] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0093] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0094] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0095] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0096] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A real-time swimming style recognition method, characterized in that: include: Use a head-mounted six-axis sensor to obtain real-time speed data during swimming; Use the trained deep neural network to perform feature recognition on real-time speed data, and obtain recognition results that carry swimming style types and swimming style data; The recognition result is wirelessly received by the client, and the swimming style type and swimming style data carried by the recognition result are displayed in real time.

2. The method according to claim 1, characterized in that The method of using a head-mounted six-axis sensor to obtain speed data in real time during swimming includes: Use a head-mounted six-axis sensor to obtain three-dimensional acceleration and three-dimensional angular velocity in real time during swimming; The three-dimensional acceleration and the three-dimensional angular velocity are converted into a one-dimensional time series signal, and the one-dimensional time series signal is used as velocity data.

3. The method according to claim 2, characterized in that The method of using the trained deep neural network to perform feature recognition on real-time speed data includes: Using a trained deep neural network, extract time domain features from the one-dimensional time series signal, perform fast Fourier transform on the one-dimensional time series signal, and extract frequency domain features; According to the time domain features and the frequency domain features, a pre-constructed recognition model is used to obtain a recognition result carrying the swimming style type and swimming style data.

4. The method according to claim 1, characterized in that: Also includes: deriving the swimmer's movement frequency and movement amplitude according to the swimming stroke data; The action frequency and the action amplitude are displayed on the client.

5. The method according to claim 1, characterized in that Also includes: Using the client to time the holding time of the swimming posture type; The holding time of the described swimming stroke type is recorded and displayed.

6. The method according to claim 1, characterized in that Also includes: The client is used to analyze the swimming style type and the swimming style data, and to generate and display training suggestions.

7. A real-time swimming style recognition system, characterized in that: include: A data acquisition and recognition module is used to use a head-mounted six-axis sensor to obtain speed data in real time during swimming, and use a trained deep neural network to perform feature recognition on the real-time speed data to obtain an identification result containing the swimming style type and swimming style data; The intelligent terminal is wirelessly connected to the data acquisition and identification module, receives the identification result by using the client running on the intelligent terminal, and displays the swimming style type and swimming style data carried by the identification result in real time.

8. The real-time swimming style recognition system according to claim 7, characterized in that: The data acquisition and identification module comprises: Head-mounted six-axis sensor, head-mounted microcontroller and wireless connection module; The head-mounted six-axis sensor transmits the acquired three-dimensional acceleration and three-dimensional angular velocity to the head-mounted single-chip microcomputer; The head-mounted single-chip computer converts the three-dimensional acceleration and the three-dimensional angular velocity into a one-dimensional time series signal, and uses a trained deep neural network to perform feature recognition on the one-dimensional time series signal to obtain a recognition result; The wireless connection module sends the recognition result to the smart terminal.

9. The real-time swimming style recognition system according to claim 8, characterized in that: The wireless connection module is a WIFI module or a Bluetooth module.

10. The real-time swimming style recognition system according to claim 8, characterized in that: The data acquisition and identification module further includes: A rechargeable battery is used to provide power to the head-mounted six-axis sensor, the head-mounted single-chip microcomputer and the wireless connection module.

Images