Motion counting method and device, electronic equipment and computer readable storage medium

Abstract

The invention discloses a motion counting method and device, electronic equipment and a computer readable storage medium. The method comprises the following steps: acquiring a current motion video frame of a user; identifying body posture information in the current motion video frame; detecting whether each piece of body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame complete the first to Nth standard process postures corresponding to a motion period of the target motion, and according to the detection result, determining whether to add 1 to the count of the target motion completed by the user, the uncounted video frames being video frames of which the count of the target motion is not added with 1. According to the scheme provided by the invention, the standardization of the motion actions of the user can be better considered during motion counting, so that the non-standard actions are excluded in the motion counting process, only the standard actions are counted, and the motion counting accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of video monitoring technology, and in particular to a motion counting method, device, electronic device and computer-readable storage medium. Background Art

[0002] Currently, many people use sports applications on mobile phones or other electronic devices to count exercises, such as counting jumping jacks, sit-ups, burpees, etc., to record exercise data in more detail and improve exercise effects. Among them, electronic devices usually shoot exercise videos of users and analyze exercise counts from the exercise videos.

[0003] When counting sports, the related art usually determines whether the user has completed the target action based on the posture changes of a certain part of the user's sports video, thereby realizing sports counting. However, since many sports are whole-body coordinated sports, the related art's method of counting by posture changes of a certain part can easily ignore the standardization of the user's sports movements, resulting in many non-standard movements being counted during the sports counting process, affecting the accuracy of sports counting and not conducive to improving the exercise effect. Summary of the invention

[0004] The present application provides an exercise counting method, device, electronic device and computer-readable storage medium, which can better consider the standard of user's exercise movements when counting exercises, so that non-standard movements are excluded during exercise counting and only standard movements are counted, thereby improving the accuracy of exercise counting and being more conducive to improving exercise effects. The specific scheme is as follows:

[0005] In a first aspect, the present application provides a motion counting method, the method comprising:

[0006] Get the user's current motion video frame;

[0007] Identifying body posture information in the current motion video frame;

[0008] Detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the target motion and completed one action cycle corresponding to the 1st to Nth standard process postures, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein N is a positive integer, and the uncounted video frame is a video frame for which the target motion count is not added by 1.

[0009] In a second aspect, the present application provides a motion counting device, the device comprising:

[0010] A video acquisition unit, used to acquire a current motion video frame of a user;

[0011] A recognition unit, used to recognize body posture information in the current motion video frame;

[0012] A counting unit is used to detect whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the target motion and completed one action cycle corresponding to the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein the uncounted video frame is a video frame for which the target motion count is not added 1.

[0013] In a third aspect, the present application provides an electronic device comprising: a processor, a memory, and computer program instructions stored in the memory and executable on the processor; when the processor executes the computer program instructions, it implements the method described in any one of the first aspects.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement any of the methods described in the first aspect.

[0015] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the method as described in any one of the first aspects.

[0016] Compared with the prior art, this application has the following advantages:

[0017] The motion counting method provided in the embodiment of the present application obtains the current motion video frame of the user and identifies the body posture information in the current motion video frame, where the body posture information can show the current action of the user. Since the uncounted video frames before the current motion video frame are video frames for which the target motion count has not been increased by 1, that is, the motion posture information corresponding to the uncounted video frames has not been counted into the motion quantity, that is, the motion posture information corresponding to the uncounted video frames has not yet completed the target motion to complete an action cycle corresponding to the standard process posture from the 1st to the Nth. Therefore, the present application detects the body posture information corresponding to the uncounted motion video frames before the current motion video frame. And whether the body posture information corresponding to the current motion video frame has completed the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result. If it is detected that the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth, it means that from the last completion of the count to the current moment, the user has completed N standard process postures corresponding to an action cycle in sequence. Therefore, the count of the user completing the target motion can be added by 1, thereby realizing the counting of the target motion.

[0018] It can be seen that the solution provided by the present application uses the N standard process postures corresponding to an action cycle of completing the target movement as a reference to determine whether the user has completed the action according to each standard process posture within the action cycle, so that when counting, the count is added by 1 after the user has completed each standard process posture within an action cycle, so that the movement count better takes into account the standardization of the user's movement movements, and can exclude non-standard movements during the movement counting process, and only count standard movements, thereby improving the accuracy of movement counting and being more conducive to improving the exercise effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the application scenario of the motion counting solution provided in this application;

[0020] Figure 2 is a flow chart of an example of a motion counting method provided in an embodiment of the present application;

[0021] Figure 3 is a schematic diagram of the process of extracting key points of the human body in an embodiment of the present application;

[0022] Figure 4 These are N standard process postures corresponding to one action cycle of the opening and closing jump in the embodiment of the present application;

[0023] Figure 5is a flowchart of another example of the motion counting method provided in an embodiment of the present application;

[0024] Figure 6 It is a structural block diagram of the electronic device provided in this application. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solution of the present application, the present application is described clearly and completely below in conjunction with the drawings in the embodiments of the present application. However, the present application can be implemented in many other ways different from the following description. Therefore, based on the embodiments provided by the present application, all other embodiments obtained by those of ordinary skill in the art without creative work should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "source domain", "third", etc. in the claims, description and drawings of the present application are used to distinguish similar objects and are not used to describe a specific order or sequence. The data used in this way are interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including", "having" and their variants are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] In order to facilitate understanding of the various embodiments of the present application, the application background of the embodiments is described.

[0028] Currently, many people use sports applications on mobile phones or other electronic devices to count exercises, such as counting jumping jacks, sit-ups, burpees, etc., to record exercise data in more detail and improve exercise effects. Among them, electronic devices usually shoot exercise videos of users and analyze exercise counts from the exercise videos.

[0029] When counting movements, the related technology usually determines whether the user has completed the target action based on the posture change of a certain part in the user's movement video, thereby realizing movement counting. For example, when counting sit-ups, counting is performed based on the change in head height.

[0030] However, since many sports are coordinated whole-body movements, the related technology of counting by posture changes of a certain part can easily ignore the standardization of the user's movement movements, resulting in many non-standard movements being counted during the movement counting process, affecting the accuracy of the movement counting and not conducive to improving the exercise effect.

[0031] In order to solve the above problems, the embodiments of the present application provide a method, device, electronic device and computer-readable storage medium for counting exercise, which are intended to better consider the standardization of the user's exercise movements when counting exercise, so that non-standard movements are excluded during the exercise counting process and only standard movements are counted, thereby improving the accuracy of exercise counting and being more conducive to improving the exercise effect.

[0032] The motion counting method provided in the present application can be used in sports applications, and can also be used in sports modules in live broadcast applications, such as live sports counting PK, live sports interaction, etc. It can also be used in other scenarios that require motion counting. The present application does not specifically limit the specific application scenarios of the motion counting solution.

[0033] In order to facilitate understanding of the method embodiments of the present application, the application scenarios are introduced. Figure 1 , Figure 1 The following is a schematic diagram of an application scenario of the solution provided in the embodiment of the present application. The application scenario is a schematic illustration and is not intended to be a specific description of the application scenario. Figure 1 As shown, in this application scenario, a server 102 and a client 101 are provided. In this embodiment, a connection is established between the client 101 and the server 102 via network communication.

[0034] Client 101 can be an electronic device with shooting function, display function and data processing function such as mobile phone, tablet computer (pad), smart watch, desktop computer, smart TV, VR equipment, vehicle-mounted equipment, wearable device, notebook computer, etc., and client 101 is used to shoot the motion video of the user, and can also count the motion according to the motion video, upload the counting result to the server 102, and obtain and display the counting results of other users from the server 102, so that the user can know the counting information of other users, so as to achieve the PK effect of sports competition and improve the enthusiasm of sports. Or, after client 101 shoots the motion video of the user, the motion video can be sent to the server 102, so that the server 102 counts according to the motion video, and sends the counting result to client 101, so that client 101 displays the counting result calculated by the server 102. Wherein, the counting is realized by client 101 without the need to transmit the video to the server 102, but the counting is realized locally in client 101, and the motion counting can be realized more quickly. Client 101 can also be used to obtain sports introduction information from the server 102, upload user information to the server 102, etc., but is not limited to this.

[0035] A specific communication connection needs to be established between the client 101 and the server 102 to perform data transmission.

[0036] The server 102 has high computing power. The server 102 can be a server, and the server 102 has high-speed processor (central processing unit, CPU) computing power, long-term reliable operation, powerful input / output (input / output, I / O) external data throughput capacity and better scalability. The server 102 can be a single server or a server cluster. The server 102 can be used to receive counting results from the client 101, and synchronize the technical results to other clients 101 for other clients 101 to display and realize sports PK. The server 102 can also receive sports videos from the client 101, count according to the sports videos, and send the counting results to the client 101, so that the client 101 displays the counting results calculated by the server 102. The server 102 can also provide other specific services for the client 101, such as user information access, website access, application access, etc., which are not specifically limited in this application.

[0037] The client 101 and the server 102 may communicate with each other using various communication systems, such as a wired communication system or a wireless communication system. The wireless communication system may be, for example, a global system for mobile communications (GSM) system, a code division multiple access (CDMA) system, a wideband code division multiple access (WCDMA) system, a general packet radio service (GPRS), a long term evolution (LTE) system, a LTE frequency division duplex (FDD) system, a LTE time division duplex (TDD) system, a universal mobile telecommunication system (UMTS), a worldwide interoperability for microwave access (WiMAX) communication system, a future fifth generation (5G) system or a new radio (NR), a satellite communication system, etc.

[0038] Embodiment 1

[0039] The first embodiment of the present application provides a motion counting method, which is applied to an electronic device, which may be a mobile phone, a laptop computer, a tablet computer, a desktop computer, a smart watch, a smart TV, a VR device, a client, a server, or other electronic device with a data processing function. The electronic device may also have a shooting function to simultaneously realize video recording and motion counting.

[0040] like Figure 2 As shown, the motion counting method provided in the first embodiment of the present application includes the following steps S110 to S130.

[0041] For the sake of convenience, the following description will take the electronic device as a terminal device such as a mobile phone, a tablet computer, etc. as an example to illustrate the solution of the present application. The execution method of the electronic device as a server is similar to that of the terminal device, and specific reference can be made to each other.

[0042] Step S110: Acquire the user's current motion video frame.

[0043] Specifically, taking the electronic device as a user terminal such as a mobile phone as an example, the electronic device can capture a user motion video in real time, so that the user's current motion video frame can be obtained from the captured user motion video.

[0044] Step S120: identifying body posture information in the current motion video frame.

[0045] Above-mentioned body posture information is used to represent the body posture of people, and body posture information can include leg posture, arm posture, trunk posture, posture of each joint, neck posture etc., and can also include other postures.In the embodiment of the present application, the body posture information required for identification can be determined according to the action posture involved in the target motion.Due to the different moving parts of different target motions, for example, the moving parts of jumping jacks are mainly the coordinated motion of arms and legs, and push-ups are mainly the coordinated motion of arms, legs, waist etc., therefore, the body posture information identified in step S120 can be matched with N standard process postures corresponding to the target motion, so that it can be judged from the identified body posture information whether the standard process posture is met.For example, it is necessary to match the arm angle and the leg opening angle in the standard process posture, then the identified body posture information should include the information that can reflect the arm angle and the leg opening angle.

[0046] Specifically, the above-mentioned body posture information may include: arm angle, leg angle, torso angle, the arm angle may include the shoulder to elbow angle, the elbow to wrist angle, the elbow joint angle, etc., the leg angle may include the hip to knee angle, the knee to ankle angle, the angle between the legs, etc., the torso angle may include the neck to waist angle, the waist to hip angle, etc., but is not limited to this, and technical personnel in this field can set it according to actual conditions.

[0047] In the embodiment of the present application, the body posture information in the current motion video frame can be identified by a pre-trained posture recognition model, wherein the posture recognition model can directly output body posture information such as arm angles and leg angles for easy use. The posture recognition model can be trained by training samples based on a supervised learning algorithm, a semi-supervised learning algorithm, a contrastive learning algorithm, and other training algorithms in the relevant technology. The specific training method can refer to the relevant technology and will not be described in detail in this application.

[0048] In a specific embodiment, step S120 may identify body posture information in the current motion video frame according to the following steps S121 to S122.

[0049] Step S121: Detecting key points of a human body in the current motion video frame.

[0050] The above-mentioned human body key points may include key points such as the head, shoulders, elbows, wrists, hips, knees, ankles, etc., and may also include other key points according to the body posture information that needs to be identified. For example, hand movements may include finger joints, and movements involving footsteps may include toe joints, etc., but are not limited to this.

[0051] In a specific embodiment, step S121 can detect key points of the human body in the current motion video frame according to the following steps S121a to S121b.

[0052] Step S121a: Determine a key point confidence map and a key point offset error map corresponding to the current motion video frame.

[0053] The key point confidence map is used to represent the position of each human key point in the image. Specifically, the key point confidence map is used to represent the confidence or probability distribution of each key point (such as a joint) in the image at different positions. The key point confidence map can be a two-dimensional matrix or image, and each pixel thereof represents the probability or confidence of the existence of a specific key point at that position. The higher the value of each pixel, the more likely the position is to be the actual position of the key point. The confidence map can reflect the probability distribution near the key point, thereby helping to find the most likely position, such as Figure 3 As shown, the key point confidence map can be a heat map.

[0054] The key point offset error map is used to represent the error compensation of each human key point position. The key point offset error map can guide the adjustment of the key point position predicted in the key point confidence map to the real and accurate key point position. Specifically, the key point offset error map can be a two-dimensional matrix or image, which provides a vector for each predicted key point, which indicates the direction and distance from the predicted position to the real position. In other words, each pixel represents the offset of the key point at that position, that is, the movement vector required to reach the real key point position from that position.

[0055] In an embodiment of the present application, a key point confidence map and a key point offset error map corresponding to the current motion video frame can be generated by a pre-trained machine model. The training process of the machine model can refer to the supervised learning algorithm, semi-supervised learning algorithm, etc. in the relevant technology, which will not be described in detail in this application. The machine model can be a neural network model or other machine models, which is not specifically limited in this application.

[0056] In a specific embodiment, the key point confidence map and the key point offset error map corresponding to the current motion video frame can be determined according to the following steps: based on a pre-trained image feature extraction model, feature extraction is performed on the current motion video frame to obtain image features corresponding to the current motion video frame; based on a pre-trained key point information output model, the key point confidence map and the key point offset error map corresponding to the current motion video frame are determined from the image features.

[0057] The above-mentioned image feature extraction model can extract image features from the current video frame, and the image feature extraction model can be a convolutional neural network model. The image feature extraction model can effectively capture multi-level and different-scale spatial information from the current video frame, including low-level features such as edges and textures, middle-level features of combined patterns such as lines and corners, and high-level features of abstract patterns such as faces and limbs. These multi-level features help to more accurately identify and locate key points. And provide more meaningful input for subsequent key point detection. In addition, the image feature extraction model can also well perceive and process local areas in the image, so that local features can be better extracted for subsequent key point extraction, thereby improving the accuracy of key point extraction.

[0058] The above key point information output model can output the corresponding key point confidence map and key point offset error map when inputting image features.

[0059] In the embodiments of the present application, the training process of the image feature extraction model, the key point information output model and other models can refer to the training process in the relevant technology. The specific training process is not the focus of this application and will not be described in detail.

[0060] Step S121b: determining the human key points in the current motion video frame according to the key point confidence map and the key point offset error map.

[0061] Specifically, the local maximum point can be found from the key point confidence map as the candidate key point position. For each candidate key point position, the offset vector of the position is read from the key point offset error map, and the candidate position is adjusted according to this vector to obtain a more accurate key point position as the human body key point in the current motion video frame.

[0062] Exemplarily, for head key points, the electronic device can find the maximum value point of the head in the key point confidence map as a candidate head key point, and then find the offset vector corresponding to the candidate head key point from the key point offset error map, and adjust the position of the head according to the offset vector to obtain the head key point in the current motion video frame.

[0063] This embodiment can accurately determine the key points of the human body in the current motion video frame through the key point confidence map and the key point offset error map.

[0064] The following is an introduction to the process of detecting the key points of the human body in the current motion video frame in step 121 of the present application through a specific example. Figure 3 As shown, in this example, detecting the key points of the human body in the current motion video frame includes the following steps S1 to S4.

[0065] Step S1: down-sample the current motion video frame at least once to obtain a down-sampled image.

[0066] The above-mentioned down sampling is to reduce the resolution of the current motion video frame to reduce the amount of calculation for subsequent key point extraction and improve the calculation speed.

[0067] Specifically, the current motion video frame may be downsampled once, twice, or three times, or more times. The number of downsampling times may be determined based on the actual situation of computing resources, the complexity of key point extraction, etc., so as to reduce computing resource consumption as much as possible while ensuring the accuracy of key point extraction. Figure 3 As shown, the current motion video frame can be downsampled three times.

[0068] Step S2: extract image features from the downsampled image using an image feature extraction model to obtain corresponding image features.

[0069] Specifically, the image feature extraction model may include a plurality of mobilenetblock modules arranged in sequence, each mobilenetblock module extracting and converting features layer by layer to gradually form high-level abstract image features.

[0070] Step S3: Based on the key point information output model and the above-mentioned image features, a key point confidence map and a key point offset error map corresponding to the current motion video frame are obtained.

[0071] like Figure 3 As shown, the above-mentioned key point information output model may include a first optimization module and a second optimization module. The first optimization module and the second optimization module may include multiple sliced ​​convolution modules. The first optimization module is used to refine the above-mentioned image features to obtain refined features. The second optimization module is used to process the refined features to obtain the key point confidence map heatmap and the key point offset error map offset corresponding to the current motion video frame.

[0072] Step S4: The post-processing module postprocess calculates the accurate coordinates of each human key point based on the key point confidence map and the key point offset error map.

[0073] The key points of the human body are as follows Figure 3 As shown in the last figure of . In the embodiment of the present application, each model used to extract key points can be a miniaturized model obtained by performing network miniaturization acceleration on the basis of mobilenetV2 structure or other model structure, so as to be more suitable for deployment on mobile terminals and quickly perform key point recognition. Network miniaturization acceleration can be model distillation compression, parameter quantization processing, model pruning and acceleration and other processing based on the original model structure.

[0074] Step S122: Determine body posture information in the current motion video frame according to the human body key points.

[0075] Specifically, the detected key points can be paired and connected according to the human body structure to form a human skeleton structure. For example, the head key point is connected to the neck key point, and the neck key point is connected to the left shoulder and right key points respectively. By analogy, each key point is connected to form a human skeleton. Then the body posture information is determined from the human skeleton structure. Since the human skeleton structure has well reflected the position and posture of each bone, the body posture information can be easily determined from the human skeleton structure.

[0076] This embodiment can accurately and efficiently determine the body posture information in the current motion video frame by detecting key points of the human body.

[0077] Step S130: Detect whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the target motion and completed one action cycle corresponding to the 1st to Nth standard process postures, and determine whether to add 1 to the count of the user completing the target motion based on the detection result.

[0078] Specifically, when it is detected that the body posture information corresponding to the uncounted motion video frames before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth, the count of the user completing the target motion can be increased by 1.

[0079] The above-mentioned target motion is the motion to be counted. The target motion can be a preset motion or a motion identified based on the current motion video. That is, the motion corresponding to the current motion video can be determined as the target motion. Different motions correspond to different standard process postures. The electronic device can obtain N standard process postures corresponding to the target motion to be counted to complete an action cycle.

[0080] Among them, N is a positive integer, and the N standard process postures corresponding to the target motion completing an action cycle can be manually pre-set, and the electronic device can obtain the N manually set standard process postures. The N standard process postures corresponding to different target motions completing an action cycle are different, and the number N of standard process postures contained in an action cycle corresponding to different target motions is usually different, but may also be the same. Technical personnel in this field can flexibly set it according to the specific motion.

[0081] Specifically, the electronic device can obtain N standard process postures corresponding to a target motion to be counted that is pre-set by the user to complete an action cycle. In this case, the user can independently set N standard process postures to flexibly move according to his or her own movement habits, rather than being limited to the standard movement postures preset by the system, thereby improving the user's movement flexibility.

[0082] Alternatively, the electronic device may also obtain N standard process postures corresponding to a target motion to be counted that are pre-configured in the sports application program to complete an action cycle. In this case, the user may be required to complete the action according to the usual standard postures, which can more standardize the user's motion and improve the exercise effect.

[0083] Those skilled in the art may select a flexible standard process posture setting method, which is not specifically limited in this application.

[0084] The target exercise can be any one of jumping jacks, sit-ups, push-ups, squats, and burpees, or other exercises that need to be counted, which is not specifically limited in this application. Those skilled in the art can set the N standard process postures of a corresponding cycle according to the specific type of the target exercise.

[0085] In an embodiment of the present application, when the above-mentioned electronic device is a terminal device, if the electronic device obtains the N standard process postures corresponding to the target movement to be counted completing an action cycle pre-configured by the motion application, the terminal device can obtain the N standard process postures corresponding to the target movement to be counted completing an action cycle from the server; when the electronic device is a server, the server can directly obtain the N standard process postures corresponding to the target movement to be counted completing an action cycle.

[0086] The N standard process postures may be key process postures among the standard process postures for the target motion to complete one action cycle, that is, the N standard process postures may be key process postures among all the standard process postures for the target motion to complete one action cycle. Figure 4 As shown, if the opening and closing jump includes completing five standard process postures in the order of state one → state two → state three → state two → state one, then the three key process postures of state one → state three → state one can be the N standard process postures corresponding to the opening and closing jump.

[0087] The N standard process postures may also be the standard process postures of the target motion to complete one action cycle. Figure 4 As shown, if the opening and closing jump includes completing five standard process postures in the order of state one → state two → state three → state two → state one, then the five process postures of state one → state two → state three → state two → state one can be the N standard process postures corresponding to the opening and closing jump.

[0088] In an embodiment of the present application, after determining the body posture information in the current motion video frame, the electronic device can detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the standard process postures from the 1st to the Nth.

[0089] The uncounted video frames are video frames for which the target motion count is not increased by 1, that is, the body posture recorded in each uncounted video frame has not yet completed the standard process posture from the 1st to the Nth, so the motion count has not yet been increased by 1 according to the uncounted video frames. For example, the current motion video frame corresponds to the Nth labeled process posture, and there are N-1 uncounted video frames before the current motion video frame, corresponding to the 1st to the N-1th standard process postures, respectively. Since these N-1 uncounted video frames have not completed the standard process postures from the 1st to the Nth, the count has not yet been increased by 1. If the current motion video frame corresponds to the Nth standard process posture, the current motion video frame and these N-1 uncounted video frames can complete the standard process postures from the 1st to the Nth, indicating that a motion cycle of the target motion is completed, and the count can be increased by 1.

[0090] It is understandable that the motion rhythm of the user's target motion is usually different from the frequency of the video frames. Therefore, more than N motion video frames may be required to complete the 1st to Nth standard process postures, that is, there may be more than N-2 motion video frames between the motion video frame corresponding to the completion of the 1st standard process posture and the motion video frame corresponding to the completion of the Nth standard process posture, among which there may be motion video frames that do not meet the standard process postures.

[0091] In an embodiment of the present application, if it is detected that the body posture information corresponding to the uncounted motion video frames before the current motion video frame and the body posture information corresponding to the current motion video frame have not completed the standard process postures from the 1st to the Nth, then return to execute steps S110 to S130 to detect the next frame, thereby continuing to complete the counting periodically.

[0092] Step S130 can count the standard process postures satisfied by the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the standard process postures satisfied by the body posture information corresponding to the current motion video frame. If the counted satisfied standard process postures satisfy the standard process postures from the 1st to the Nth, it is determined that the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth.

[0093] In one embodiment, step S130 can detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the standard process postures from the 1st to the Nth according to the following steps S131 to S132.

[0094] Step S131: adding 1 to the posture state number according to whether the body posture information corresponding to the current motion video frame satisfies one of the 1st to Nth standard process postures corresponding to the target motion completing an action cycle.

[0095] The posture state number indicates the number of postures completed so far in the current motion action cycle. If the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures, it means that the current user has completed a standard process posture, so the posture state number can be increased by 1.

[0096] Specifically, when the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures, the posture state number may be increased by 1.

[0097] It is understandable that if the body posture information corresponding to the current motion video frame does not satisfy any one of the 1st to Nth standard process postures, the process returns to step S110 to reacquire the motion video frame for status determination and to continue counting.

[0098] Step S132: According to whether the number of posture states is N, it is determined whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth.

[0099] Specifically, when the number of posture states is N, it can be determined that the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth. When the number of posture states is less than N, it is determined that the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have not completed the standard process postures from the 1st to the Nth.

[0100] The above method may further include the following step S140.

[0101] Step S140: when it is detected that the count of the user completing the target movement is increased by 1, the posture state number is cleared.

[0102] When the target motion count is completed and 1 is added to the count, it means that the motion count of one motion cycle is completed. Then, it is necessary to restart the count of another motion action cycle. Therefore, it is necessary to clear the posture state number to perform the posture completion number statistics of the new motion cycle.

[0103] In some possible cases, in order to record the user's motions in a standard motion sequence, improve the counting accuracy, and prevent the user from omitting a certain standard posture, step S132 can be implemented according to the following steps: when the number of posture states is N, and the body posture information corresponding to the current motion video frame satisfies the Nth standard process posture, it is determined that the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth. This embodiment determines that the standard process postures from the 1st to the Nth are completed when the current motion video frame corresponds to the Nth standard process posture, which can better ensure that the user completes the N standard process postures in sequence from beginning to end.

[0104] In a specific embodiment, step S131 can also be implemented by adding 1 to the posture state number according to the following step S131a.

[0105] Step S131a: When the body posture information corresponding to the current motion video frame satisfies one of the 1st to Nth standard process postures, and the relationship between the currently detected standard posture and the last detected standard posture satisfies the adjacent previous standard process posture and the next standard process posture among the N standard process postures, the posture state number is increased by 1.

[0106] Among them, the currently detected standard posture is the standard process posture satisfied by the current motion video frame, the last detected standard posture is the standard process posture satisfied by the most recent standard motion video frame before the current motion video frame, and the standard motion video frame is a motion video frame that satisfies the standard process posture.

[0107] For example, the current detection standard posture satisfied by the current motion video frame is the Nth standard process posture. If the standard motion video frame most immediately before the current motion video frame satisfies the N-1th standard process posture, the posture state number is increased by 1. If the standard motion video frame most immediately before the current motion video frame satisfies the N-2th standard process posture, it means that the user did not complete the action in sequence, and the posture state number is not increased by 1.

[0108] This embodiment can detect whether the standard postures completed twice by the user are completed in the order of the 1st to Nth standard process postures, so that the number of postures can be counted only when the user completes the action in the standard order. This can better ensure that the user completes the exercise in accordance with the standard action and standard order before counting, and the counting is more accurate.

[0109] In one embodiment, when the target motion is jumping jacks, it is possible to detect whether the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures in the following manner: based on whether the up and down swinging amplitude of the arms of the current motion video frame satisfies the swinging threshold and whether the opening amplitude of the two legs satisfies the opening threshold, it is detected whether the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures.

[0110] Specifically, when the target motion is jumping jacks, the N standard process postures may include the 1st to 5th standard process postures. The 1st standard process posture is: the up and down swinging amplitude of the arms is less than the first swinging threshold, and the opening amplitude of the two legs is less than the first opening threshold. The 2nd standard process posture is: the up and down swinging amplitude of the arms is greater than the second swinging threshold and less than the third swinging threshold, and the opening amplitude of the two legs is greater than the first opening threshold, wherein the second swinging threshold is greater than the first swinging threshold, and the third swinging threshold is greater than the second swinging threshold. The 3rd standard process posture is: the up and down swinging amplitude of the arms is greater than the fourth swinging threshold, and the opening amplitude of the two legs is greater than the second opening threshold, wherein the second opening threshold is greater than the first opening threshold, and the fourth swinging threshold is greater than the third swinging threshold. The 4th standard process posture is the same as the 2nd standard process posture, and the 5th standard process posture is the same as the 1st standard process posture. That is to say, if Figure 4 As shown, for the jumping jack target exercise, follow Figure 4 When the five standard process postures are completed in the order of state 1 → state 2 → state 3 → state 2 → state 1, it is considered that one opening and closing jump cycle is completed, and the motion count is increased by 1.

[0111] The swing threshold and the opening threshold can be represented by the swing angle and the opening angle, and the specific value of the swing threshold can be determined according to the jumping jack posture and is not specifically limited in this application. The above arm swing range refers to the swing range of the two arms.

[0112] Specifically, when the target movement is jumping jacks, the first standard process posture is specifically: the elbow bending angle is greater than 120°, the hand and upper arm swinging angle is less than 45°, and the leg angle is less than 15°, wherein the hand and upper arm swinging angle is the up and down swinging amplitude of the above-mentioned arms, and the leg angle is the opening amplitude of the above-mentioned legs. 45° is the above-mentioned first swinging threshold, and 15° is the above-mentioned first opening threshold. The second standard process posture is specifically: the elbow bending angle is greater than 120°, the hand and upper arm swinging angle is greater than 90° and less than 120°, and the leg angle is greater than 15°, wherein 90° is the above-mentioned second swinging threshold, 120° is the above-mentioned third swinging threshold, and 15° is the above-mentioned second opening threshold. The third standard process posture is specifically: the elbow bending angle is greater than 120°, the hand and upper arm swinging angle is greater than 150°, and the leg angle is greater than 30°, wherein 150° is the above-mentioned fourth swinging threshold, and 30° is the above-mentioned third opening threshold. Those skilled in the art may also set other standard process postures as the N standard process postures corresponding to jumping jacks. The N standard process postures can accurately determine whether the user has completed the standard action, and the specific standard postures are not limited.

[0113] This embodiment determines the first to fifth standard process postures according to the movement characteristics of jumping jacks. It can accurately judge whether the user has completed a standard jumping jack based on these five standard process postures and complete accurate jumping jack movement counting.

[0114] The motion counting method provided by the embodiment of the present application obtains N standard process postures corresponding to a motion cycle of the target motion to be counted, and the N standard process postures are subsequently used to determine whether the user has completed the motion according to the standard process posture of a motion cycle, and then obtains the user's current motion video frame, identifies the body posture information in the current motion video frame, and the body posture information can show the user's current motion. Since the uncounted video frames before the current motion video frame are video frames for which the target motion count is not increased by 1, that is, the motion posture information corresponding to the uncounted video frames has not been counted in the number of motions, that is, the motion posture information corresponding to the uncounted video frames has not completed the standard process postures from the 1st to the Nth. Therefore, if it is detected that the body posture information corresponding to the uncounted motion video frames before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth, it means that from the last completion of the count to the current moment, the user has completed the N standard process postures corresponding to a motion cycle in sequence. Therefore, the count of the user completing the target motion can be increased by 1, thereby realizing the counting of the target motion.

[0115] It can be seen that the solution provided by the present application uses the N standard process postures corresponding to one action cycle of the target movement as a reference to determine whether the user has completed the action according to each standard process posture within the action cycle. As a result, when counting, the count is incremented by 1 only after the user has completed each standard process posture within one action cycle. This enables the movement count to better consider the standardization of the user's movement actions, exclude non-standard actions during the movement counting process, only count standard actions, improve the accuracy of the movement count, and is more conducive to improving the exercise effect.

[0116] The following introduces the movement counting method provided by the present application through a specific example. In this example, the target movement is jumping jacks. As Figure 5 shown, the movement counting method provided in this example includes the following steps S10 to S80.

[0117] Step S10: Obtain the current movement video frame of the user.

[0118] Step S20: Detect the human key points in the current movement video frame.

[0119] Step S30: Determine whether the human body is entirely within the frame.

[0120] That is, determine whether the human body in the frame is complete. If step S30 determines that the human body is entirely within the frame, then execute step S40; otherwise, return to execute step S10.

[0121] Step S40: Calculate five angles A, B, C, D, and E.

[0122] The above angles A and D are the bending angles of the two elbows respectively, the angles B and C are the swinging angles of the two upper arms respectively, and the angle E is the angle between the two legs. Step S40 is the process of identifying the body posture information in the current movement video frame in the above step S130, and the five angles A, B, C, D, and E are the above body posture information.

[0123] Step S50: Determine whether it is in state one. State one is: A > 120 degrees, B < 45 degrees, C < 45 degrees, D > 120 degrees, E < 15 degrees.

[0124] The above state one is the first standard process posture in the above text.

[0125] If step S50 determines that it is in state one, then execute step S80; if step S50 determines that it is not in state one, then execute step S60.

[0126] Step S60: Determine whether it is in state two. State two is: A > 120 degrees, 90 degrees < B < 120 degrees, 90 degrees < C < 120 degrees, D > 120 degrees, E > 15 degrees.

[0127] If step S60 determines that the state is in state 2, step S80 is executed. If step S60 determines that the state is not in state 2, step S70 is executed.

[0128] The above-mentioned state 2 is the second standard process posture mentioned above.

[0129] Step S70: Determine whether it is in state three, state three is: A>120 degrees, B>150 degrees, C>150 degrees, D>120 degrees, E>30 degrees.

[0130] If step S70 determines that the state is in state three, then step S80 is executed. If step S70 determines that the state is not in state three, then the process returns to step S10.

[0131] The above-mentioned state three is the third standard process posture in the above text. Since the fourth standard process posture is the same as the second standard process posture, and the fifth standard process posture is the same as the first standard process posture, that is, state four is the same as state two, and state five is the same as state one, therefore only state one, state two, and state three can be judged.

[0132] Step S80: Posture state number +1.

[0133] Step S90: Determine whether the posture state number is equal to 5.

[0134] If the value in step S90 is equal to 5, the process proceeds to step S100 . If the value in step S90 is not equal to 5, the process returns to step S10 .

[0135] Step S100: Clear the posture state number and increase the movement count by 1.

[0136] This embodiment uses the specific example of jumping jacks to exemplify the motion counting method in the present application, which does not constitute a specific limitation on the present application, and the same idea can also be used for counting other sports.

[0137] The second embodiment of the present application also provides a motion counting device corresponding to the motion counting method embodiment provided in the first embodiment. Since the device embodiment is basically similar to the method embodiment, the description is relatively simple. For the details of the relevant technical features and the effects achieved, please refer to the corresponding description of the motion counting method embodiment provided above. The motion counting device provided in this embodiment includes:

[0138] A video acquisition unit, used to acquire a current motion video frame of a user;

[0139] A recognition unit, used to recognize body posture information in the current motion video frame;

[0140] A counting unit is used to detect whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the target motion and completed one action cycle corresponding to the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein the uncounted video frame is a video frame for which the target motion count is not added 1.

[0141] Optionally, the device further comprises:

[0142] The detection unit is used to: increase the posture state number by 1 according to whether the body posture information corresponding to the current motion video frame satisfies one of the 1st to Nth standard process postures; determine whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the 1st to Nth standard process postures according to whether the posture state number is N; and clear the posture state number when it is detected that the count of the user completing the target motion is increased by 1.

[0143] Optionally, the detection unit is specifically used for:

[0144] When the number of posture states is N, and the body posture information corresponding to the current motion video frame satisfies the Nth standard process posture, it is determined that the body posture information corresponding to the uncounted motion video frames before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth.

[0145] Optionally, the detection unit is specifically used to: when the body posture information corresponding to the current motion video frame satisfies one of the 1st to Nth standard process postures, and the relationship between the currently detected standard posture and the last detected standard posture satisfies the adjacent previous standard process posture and the next standard process posture among the N standard process postures, add 1 to the posture state number; wherein the currently detected standard posture is the standard process posture satisfied by the current motion video frame, the last detected standard posture is the standard process posture satisfied by the most recent standard motion video frame before the current motion video frame, and the standard motion video frame is a motion video frame that satisfies the standard process posture.

[0146] Optionally, when the target motion is jumping jacks, the detection unit detects whether the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures in the following manner: based on whether the up and down swinging amplitude of the arms of the current motion video frame satisfies the swinging threshold and whether the opening amplitude of the two legs satisfies the opening threshold, it is detected whether the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures.

[0147] Optionally, the recognition unit is specifically used to: detect human key points in the current motion video frame; and determine body posture information in the current motion video frame according to the human key points.

[0148] Optionally, the recognition unit is specifically used to: determine a key point confidence map and a key point offset error map corresponding to the current motion video frame, wherein the key point confidence map is used to represent the position of each human key point in the image, and the key point offset error map is used to represent the error compensation of the position of each human key point; determine the human key points in the current motion video frame based on the key point confidence map and the key point offset error map.

[0149] Optionally, the recognition unit is specifically used to: perform feature extraction on the current motion video frame based on a pre-trained image feature extraction model to obtain image features corresponding to the current motion video frame; and determine the key point confidence map and key point offset error map corresponding to the current motion video frame from the image features based on a pre-trained key point information output model.

[0150] The third embodiment of the present application also provides an electronic device embodiment corresponding to the motion counting method provided in the first embodiment. The following description of the electronic device embodiment is only illustrative. The electronic device embodiment is as follows:

[0151] Please refer to Figure 6 Understand the above electronic devices, Figure 6 The electronic device provided in this embodiment includes: a processor 1001, a memory 1002, a communication bus 1003, and a communication interface 1004;

[0152] The memory 1002 is used to store computer instructions for data processing. When the computer instructions are read and executed by the processor 1001, the following steps are performed:

[0153] Get the user's current motion video frame;

[0154] Identifying body posture information in the current motion video frame;

[0155] Detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the target motion and completed one action cycle corresponding to the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein the uncounted video frame is a video frame for which the target motion count is not added 1.

[0156] The fourth embodiment of the present application also provides a computer-readable storage medium for implementing the method described in the first embodiment. The computer-readable storage medium embodiment provided in the present application is described relatively simply. For the relevant parts, please refer to the corresponding description of the above method embodiment. The embodiment described below is only illustrative.

[0157] The computer-readable storage medium provided in this embodiment stores computer instructions, and the computer-readable storage medium can be applied to electronic devices; when the instructions are executed by a processor, the following steps are implemented:

[0158] Get the user's current motion video frame;

[0159] Identifying body posture information in the current motion video frame;

[0160] Detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the target motion and completed one action cycle corresponding to the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein the uncounted video frame is a video frame for which the target motion count is not added 1.

[0161] The fifth embodiment of the present application also provides a computer program product for implementing the method described in the first embodiment. The computer program product embodiment provided in this application is described relatively simply, and the relevant parts can refer to the corresponding description of the above method embodiment. The embodiment described below is only illustrative.

[0162] The computer program product provided in this embodiment includes a computer program, which, when executed by a processor, implements the following steps based on the estimation model:

[0163] Get the user's current motion video frame;

[0164] Identifying body posture information in the current motion video frame;

[0165] Detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the target motion and completed one action cycle corresponding to the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein the uncounted video frame is a video frame for which the target motion count is not added 1.

[0166] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0167] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0168] 1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0169] 2. Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0170] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A motion counting method, characterized in that: The method comprises: Get the user's current motion video frame; Identifying body posture information in the current motion video frame; Detect the body posture information corresponding to each uncounted motion video frame before the current motion video frame and whether the body posture information corresponding to the current motion video frame has completed the target motion and completed one action cycle corresponding to the 1st to Nth standard process postures, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein N is a positive integer, and the uncounted video frame is a video frame for which the target motion count is not added by 1.

2. The motion counting method according to claim 1, characterized in that: The following steps are used to detect whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth corresponding to one action cycle of the target motion: According to whether the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures corresponding to the target motion completing one action cycle, the posture state number is increased by 1; According to whether the number of posture states is N, determining whether each body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame complete the standard process postures from the 1st to the Nth; The method further comprises: When it is detected that the count of the user completing the target movement is increased by 1, the posture state number is cleared.

3. The motion counting method according to claim 2, characterized in that: The step of determining, based on whether the number of posture states is N, whether each body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame complete the standard process postures from the 1st to the Nth, comprises: When the number of posture states is N, and the body posture information corresponding to the current motion video frame satisfies the Nth standard process posture, it is determined that the body posture information corresponding to the uncounted motion video frames before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the standard process postures from the 1st to the Nth.

4. The motion counting method according to claim 2, characterized in that: The step of adding 1 to the posture state number according to whether the body posture information corresponding to the current motion video frame satisfies one of the first to Nth standard process postures corresponding to the target motion completing one action cycle comprises: When the body posture information corresponding to the current motion video frame satisfies one of the 1st to Nth standard process postures corresponding to the target motion completing an action cycle, and the relationship between the currently detected standard posture and the last detected standard posture satisfies the adjacent previous standard process posture and the next standard process posture among the N standard process postures, the posture state number is increased by 1; wherein the currently detected standard posture is the standard process posture satisfied by the current motion video frame, the last detected standard posture is the standard process posture satisfied by the most recent standard motion video frame before the current motion video frame, and the standard motion video frame is a motion video frame that satisfies the standard process posture.

5. The motion counting method according to claim 2, characterized in that: When the target motion is jumping jacks, the body posture information corresponding to the current motion video frame is detected in the following manner to determine whether it satisfies one of the first to Nth standard process postures corresponding to the target motion completing one action cycle: According to whether the up and down swinging amplitude of the arms in the current motion video frame meets the swinging threshold and whether the opening amplitude of the two legs meets the opening threshold, it is detected whether the body posture information corresponding to the current motion video frame meets one of the 1st to Nth standard process postures corresponding to the target motion completing an action cycle.

6. The motion counting method according to any one of claims 1 to 5, characterized in that: The identifying body posture information in the current motion video frame includes: Detecting key points of a human body in the current motion video frame; The body posture information in the current motion video frame is determined according to the human body key points.

7. The motion counting method according to claim 6, characterized in that: The detecting of key points of a human body in the current motion video frame comprises: Determine a key point confidence map and a key point offset error map corresponding to the current motion video frame, wherein the key point confidence map is used to represent the position of each human key point in the image, and the key point offset error map is used to represent the error compensation of each human key point position; The human body key points in the current motion video frame are determined according to the key point confidence map and the key point offset error map.

8. The motion counting method according to claim 7, characterized in that: The determining of the key point confidence map and the key point offset error map corresponding to the current motion video frame includes: Performing feature extraction on the current motion video frame based on a pre-trained image feature extraction model to obtain image features corresponding to the current motion video frame; Based on a pre-trained key point information output model, a key point confidence map and a key point offset error map corresponding to the current motion video frame are determined from the image features.

9. The motion counting method according to any one of claims 1 to 5, characterized in that: The first to Nth standard process postures corresponding to the target motion completing one action cycle are determined by the following method: Obtain the 1st to Nth standard process postures corresponding to a target motion preset by the user to complete an action cycle; Alternatively, the first to Nth standard process postures corresponding to a target motion pre-configured by the motion application program to complete an action cycle are obtained.

10. The motion counting method according to any one of claims 1 to 5, characterized in that: The N standard process postures are: key process postures among the standard process postures when the target moves to complete one action cycle, or the standard process postures when the target moves to complete one action cycle.

11. A motion counting device, characterized in that: The device comprises: A video acquisition unit, used to acquire a current motion video frame of a user; A recognition unit, used to recognize body posture information in the current motion video frame; A counting unit is used to detect whether the body posture information corresponding to each uncounted motion video frame before the current motion video frame and the body posture information corresponding to the current motion video frame have completed the target motion and completed one action cycle corresponding to the standard process postures from the 1st to the Nth, and determine whether to add 1 to the count of the user completing the target motion according to the detection result, wherein the uncounted video frame is a video frame for which the target motion count is not added 1.

12. An electronic device, characterized in that: include: A processor, a memory, and computer program instructions stored on the memory and executable on the processor; When the processor executes the computer program instructions, the method according to any one of claims 1 to 10 is implemented.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method according to any one of claims 1 to 10.

Images