CN115188063A - Running posture analysis method, device, treadmill and storage medium based on treadmill - Google Patents

Abstract

The embodiment of the application discloses a running posture analysis method and device based on a running machine, the running machine and a storage medium. According to the technical scheme provided by the embodiment of the application, the foot key points of a certain number of frames of video images in the running video are determined by acquiring the running video in real time and detecting based on the skeleton key points; confirming the pixel coordinate height of the foot key point in the video images of a certain number of frames, and generating a mapping relation between the pixel coordinate height and the corresponding time point according to the pixel coordinate height and the time point corresponding to the video images of the certain number of frames; and calculating corresponding running posture parameters based on the mapping relation, and outputting the running posture parameters. By adopting the technical means, running posture analysis is carried out through the contact process curve of the left foot and the right foot, running posture analysis based on running video images can be realized, extra burden on a user caused by detection equipment is avoided, and running exercise experience of the user is optimized.

Description

Running posture analysis method, device, treadmill and storage medium based on treadmill

technical field

The embodiments of the present application relate to the technical field of computer vision, and in particular, to a treadmill-based running posture analysis method, device, treadmill, and storage medium.

Background technique

As running becomes more and more popular, treadmills have become a convenient choice for more and more people. At the same time, the diversified functions that treadmills can provide has also become an important trade-off point for people to choose treadmill products. At present, in order to provide users with a richer functional experience when running, there are some smart wearable sports devices on the market that provide analysis information for the user's running posture, such as smart running shoes and running wizards. While using the treadmill, the user performs running by wearing these smart wearable sports devices, and uses the smart wearable sports devices to collect relevant sensor data, thereby obtaining the user's running posture analysis information.

However, wearing a smart wearable sports device will inevitably increase the extra burden on the user during the running exercise, resulting in inconvenience throughout the running exercise process and affecting the user's running experience.

SUMMARY OF THE INVENTION

The embodiments of the present application provide a running posture analysis method, device, treadmill, and storage medium based on a treadmill, which can perform real-time detection and analysis of the user's running posture without increasing the burden on the user, and solve the problem of running posture detection and analysis during the user's running process. .

In a first aspect, an embodiment of the present application provides a treadmill-based running posture analysis method, including:

Acquire the running video in real time, and determine the foot key points of the video images of a certain number of frames in the running video based on the skeleton key point detection;

Confirm the pixel coordinate height of the key point of the foot in the video image of a certain number of frames, and generate the pixel coordinate height and the corresponding time point according to the pixel coordinate height and the video image of a certain number of frames. The mapping relationship between time points;

The corresponding running posture parameters are calculated based on the mapping relationship, and the running posture parameters are output.

In a second aspect, an embodiment of the present application provides a treadmill-based running posture analysis device, including:

an acquisition module, configured to acquire the running video in real time, and determine the foot key points of the video images of a certain number of frames in the running video based on the skeleton key point detection;

The generation module is used to confirm the pixel coordinate height of the key point of the foot in the video image of a certain number of frames, and generate the pixel coordinate height according to the pixel coordinate height and the time point corresponding to the video image of a certain number of frames. The mapping relationship between the pixel coordinate height and the corresponding time point;

An output module, configured to calculate the corresponding running posture parameters based on the mapping relationship, and output the running posture parameters.

In a third aspect, an embodiment of the present application provides a treadmill, including:

memory and one or more processors;

the memory for storing one or more programs;

When the one or more programs are executed by the one or more processors, the one or more processors implement the treadmill-based running posture analysis method as described in the first aspect.

In a fourth aspect, embodiments of the present application provide a storage medium containing computer-executable instructions, the computer-executable instructions, when executed by a computer processor, are used to execute the treadmill-based running according to the first aspect Attitude analysis method.

In the embodiment of the present application, the running video is acquired in real time, and the foot key points of the video image of a certain number of frames in the running video are determined based on the detection of skeleton key points; the pixel coordinate height of the foot key point in the video image of a certain number of frames is confirmed, according to The pixel coordinate height and the time point corresponding to a certain number of frames of video images are used to generate a mapping relationship between the pixel coordinate height and the corresponding time point; the corresponding running posture parameters are calculated based on the mapping relationship, and the running posture parameters are output. Using the above technical means, by detecting the key points of the footsteps in the video image, generating the curve of the left and right feet touching the ground based on the key points of the footsteps, and analyzing the running posture according to the curve of the left and right feet touching the ground, the running posture analysis based on the running video image can be realized. Avoid the extra burden of detection equipment on users, and optimize the user's running experience.

In addition, the embodiment of the present application calculates the running posture parameters by determining the key points of the feet in the process of the left and right feet touching the ground, which can ensure the accurate detection of the key points of the skeleton, avoid the situation that the key points of the feet are affected by occlusion and cause detection errors, and make the running posture parameters. The calculation is more accurate; by selecting the key points of the left and right feet touching the ground from the running video to calculate the running posture parameters, the number of selected key points can be reduced, thereby reducing the calculation amount of running posture parameters and improving the efficiency of running posture analysis. On the other hand, by selecting the key points of the left and right feet touching the ground from the running video to calculate the running posture parameters, it can also ensure the rapid extraction of the key points of the foot, reduce the error effect caused by the running video recording delay, and further improve The accuracy of running posture parameter calculation.

In addition, in the embodiment of the present application, the running posture analysis is performed by combining the detection of the key points of the footsteps and the contacting process curves of the left and right feet, which can realize the accurate calculation of the running posture parameters and provide a more accurate running posture analysis result.

Description of drawings

1 is a flowchart of a treadmill-based running posture analysis method provided in Embodiment 1 of the present application;

Fig. 2 is the running video shooting schematic diagram in the first embodiment of the present application;

3 is a schematic diagram of a left foot contacting process curve diagram in Embodiment 1 of the present application;

4 is a schematic diagram of a process curve of the right foot touching the ground in Embodiment 1 of the present application;

Fig. 5 is the running posture parameter calculation flow chart in the first embodiment of the present application;

6 is a schematic diagram of the horizontal distance between the key point of the foot and the camera in Embodiment 1 of the present application;

7 is a schematic structural diagram of a treadmill-based running posture analysis device provided in Embodiment 2 of the present application;

FIG. 8 is a schematic structural diagram of a treadmill provided in Embodiment 3 of the present application.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present application clearer, the specific embodiments of the present application will be further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application. In addition, it should be noted that, for the convenience of description, the drawings only show some but not all of the contents related to the present application. Before discussing the exemplary embodiments in greater detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart depicts various operations (or steps) as a sequential process, many of the operations may be performed in parallel, concurrently, or concurrently. Additionally, the order of operations can be rearranged. The process may be terminated when its operation is complete, but may also have additional steps not included in the figures. The processes may correspond to methods, functions, procedures, subroutines, subroutines, and the like.

The treadmill-based running posture analysis method provided by the present application aims to capture the running video of the user through a camera when the user uses the treadmill to run, and then analyzes the running posture based on the detection of key points of the foot combined with the curve of the left and right feet touching the ground. , calculate the corresponding running posture parameters, so as to avoid the extra burden of the detection equipment on the user, optimize the user's running experience, and provide more accurate running posture analysis results. As for the traditional treadmill, in order to analyze the running posture of the user, the user needs to wear a corresponding intelligent wearable sports device to detect and analyze the running posture parameters. Since the smart wearable sports device needs to detect running posture parameters, it is bound to integrate a large number of sensing devices, which makes the smart wearable sports device too heavy. When a user wears such a smart wearable sports device for running, it will undoubtedly hinder the running, increase the burden of the user's running, and then affect the user's running experience. Based on this, a treadmill-based running posture analysis method according to an embodiment of the present application is provided to solve the problem of running posture detection and analysis based on a treadmill, and to avoid additional burdens of the user during running.

Example 1:

FIG. 1 shows a flowchart of a treadmill-based running posture analysis method provided in Embodiment 1 of the present application. The treadmill-based running posture analysis method provided in this embodiment may be performed by a treadmill-based running posture analysis device. Execution, the treadmill-based running posture analysis device may be implemented by means of software and/or hardware, and the treadmill-based running posture analysis device may be composed of two or more physical entities, or may be constituted by one physical entity. Generally speaking, the running posture analysis device based on the treadmill may be a computing device with a processor such as a treadmill.

The following description takes the treadmill as the subject for executing the treadmill-based running posture analysis method as an example.

1 , the treadmill-based running posture analysis method specifically includes:

S110. Acquire the running video in real time, and determine the foot key points of the video images of a certain number of frames in the running video based on skeleton key point detection.

Specifically, the embodiment of the present application is based on computer vision technology, by collecting user running video images, and performing detection and analysis based on each frame of video images, thereby obtaining a running posture analysis result of the corresponding user. Among them, the running video of the user is captured in real time by the camera, and the treadmill obtains the running video of the current set time period, and then detects the key points of the skeleton based on each frame of video image in the running video, and determines that the key points of the user's feet are in each frame of the video. position on the image. It can be understood that the change of the position of the key points of the user's feet on each frame of video image can be used to determine the change of the running posture of the user's feet from touching the ground to the air during the running process. Running posture analysis of running videos.

Exemplarily, referring to FIG. 2 , a schematic diagram of shooting a running video according to an embodiment of the present application is provided. As shown in FIG. 2 , the camera 11 is set on the treadmill, and shoots the running video from a perspective facing the front of the user 12 . When the user starts the treadmill to run, the camera simultaneously shoots the user's running video. The treadmill pre-sets the running video extraction cycle. According to the extraction cycle, the camera uploads the running video of the corresponding time period every set time period (such as 10S) for analysis of the user's running posture at the current moment.

Further, when the camera is shooting running videos, it generally only captures the video images of the lower body of the user during running (that is, the video images of the lower body from the hip bones, knees, ankles, and finally to the toes) of the human body. The running video of the corresponding period of time, based on this running video, call the underlying skeleton key point extraction algorithm to detect the human skeleton key points frame by frame video images, and determine the foot key points. Specifically, the skeleton key point extraction algorithm obtains the key point heat map of the lower body by detecting the skeleton key points, and further decodes the key point heat map to obtain the pixel coordinate information of the key points.

Specifically, the key point heat map decoding method of the embodiment of the present application adopts the wave peak point method. For each key point heat map, the peak value of its Gaussian point is first calculated and obtained, and the coordinates of the peak value are the corresponding heat map. The pixel coordinates of the key point in the original image. In the case where there are multiple Gaussian point peaks in the heat map, the largest peak point is obtained by traversal first, and the Gaussian point of the second highest peak that overlaps with the Gaussian point of the highest peak is saved. Further, by calculating the distance between the coordinates of the highest peak and the coordinates of the overlapping peaks, the coordinates of the highest peak are fine-tuned by the coordinates of the overlapping peaks, so as to obtain the final key point coordinates. It should be noted that there are many technical means for detecting key points of human skeleton based on images in the prior art, and the embodiments of the present application do not impose fixed restrictions on specific detection algorithms, which will not be repeated here.

More specifically, in this embodiment of the present application, when the foot key points of a certain number of frames of video images in the running video are determined based on the skeleton key point detection, the left foot and/or the left foot and/or the left foot are selected from the running video based on the skeleton key point detection. The video images of a certain number of frames corresponding to the process of the right foot touching the ground, and the pixel coordinates of the key points of the foot are determined from the video images. The touchdown process is the process from when the user's foot just starts touching the ground to preparing to leave the ground. Since the user is running on the track, on the running video image, the lowest point that the user's feet can reach is the position of the track. Based on this, through video image detection, when it is determined that the pixel coordinate height of the lowest point of the user's foot on the video image is equal to the pixel coordinate height of the runway, it is considered that the current user's foot touches the ground; When the pixel coordinate height is lower than the pixel coordinate height of the runway, it is considered that the current user's feet are off the ground. Based on the above detection principle, it is possible to determine each time period in which the user's foot touches the ground in the current running video, and then extract a certain number of frames of video images corresponding to each time period from the running video, and then perform human skeleton key points on the video image. Detect, and select the pixel coordinates of the key points of the foot from the detected key points to calculate the running posture parameters.

In the above, the running posture parameters are calculated by determining the key points of the left and right feet touching the ground, which can ensure the accurate detection of the key points of the skeleton, avoid the detection error caused by the occlusion of the key points of the feet, and make the calculation of the running posture parameters more accurate. ; By selecting the key points of the left and right feet touching the ground from the running video to calculate the running posture parameters, the number of selected key points can be reduced, thereby reducing the calculation amount of running posture parameters and improving the efficiency of running posture analysis. On the other hand, by selecting the key points of the left and right feet touching the ground from the running video to calculate the running posture parameters, it can also ensure the rapid extraction of the key points of the foot, reduce the error effect caused by the running video recording delay, and further improve The accuracy of running posture parameter calculation.

It should be noted that, based on the detection of the above-mentioned foot key points, the pixel coordinates of the corresponding foot key points on each frame of video image can be determined, and the foot key points include two or two groups of foot key points corresponding to the left and right feet. It can be understood that each frame of video image includes corresponding time point information, and then according to the coordinate height of the pixel coordinates, the height position of the user's feet at the corresponding time point can be determined. That is, the "y" value in each pixel coordinate represents the pixel coordinate height of the key point of the foot. In the pixel coordinate system, the larger the "y" value corresponding to the vertical axis, the lower the position of the key point of the user's foot, and the closer it is Running track for treadmill.

S120. Confirm the pixel coordinate height of the key point of the foot in the video image of a certain number of frames, and generate the pixel coordinate height according to the pixel coordinate height and the time point corresponding to the video image of a certain number of frames The mapping relationship with the corresponding time point.

After the key points of the foot are determined according to the detection of the key points of the human skeleton, the embodiment of the present application determines the "y" value of each pixel coordinate, that is, the height of the pixel coordinate, according to the pixel coordinates of the key point of the foot in each frame of video image. It can be understood that the pixel coordinate height has a corresponding time point information in each frame of video image, then by determining the time point and pixel coordinate corresponding to a certain number of frames of video images for the key points of the feet during the process of the left and right feet touching the ground. height, the corresponding mapping relationship can be generated.

Specifically, in this embodiment of the present application, the video images corresponding to the process of the left foot and/or the right foot touching the ground are selected, and the key points of the foot are determined therefrom. Correspondingly, when constructing a mapping relationship in this embodiment of the present application, the following steps are included:

Corresponding to the height of the pixel coordinates of the key point of the foot of the left foot, a first mapping relationship is generated with the time point corresponding to the video image;

A second mapping relationship is generated corresponding to the height of the pixel coordinates of the key point of the foot of the right foot and the time point corresponding to the video image.

By constructing the corresponding mapping relationship corresponding to the touchdown process of the left foot and the right foot respectively, the mapping between the pixel coordinate height of the key point of the left foot and the key point of the right foot during the touchdown process of the left and right feet and the corresponding time point can be determined respectively. Therefore, it is convenient for the subsequent calculation of the relevant running posture parameters corresponding to the left and right feet respectively.

It should be noted that, the mapping relationship in this embodiment of the present application may be in the form of an information sequence, a function curve, or the like. The specific form of the mapping relationship is not fixedly limited in this application, and details are not described here.

Further, the embodiment of the present application uses a function curve as an example to describe the mapping relationship. Since the mapping relationship includes a first mapping relationship corresponding to the left foot key point and a second mapping relationship corresponding to the right foot key point, when the mapping relationship is represented by a function curve, the first mapping relationship is represented as the left foot touch The ground process curve, and the second mapping relationship is expressed as the right foot touching the ground process curve.

It can be understood that, when constructing the left foot contacting process curve and the right foot contacting process curve in this embodiment of the present application, only the pixel coordinate heights of the corresponding key points of the foot during the contacting process of the left and right feet are used to construct the function curve. Among them, by determining each time period in which the user's foot touches the ground in the current running video, and then determining the pixel coordinate height of the key point of the foot and the corresponding time point in the video image corresponding to each time period, the left and right The construction of the foot contact process curve.

In one embodiment, the touchdown process of the left and right feet can also be detected through the cooperation of the sensing devices arranged on the track. Among them, a pressure sensing device is installed on the runway corresponding to the left and right feet, and the pressure sensing device detects the pressure information on the runway in real time. When the pressure sensing device detects that the instantaneous value of the current pressure is greater than the initial value (that is, the detection value in the unpressurized state), it indicates that the foot of the current user is in contact with the runway, and when the pressure sensing device detects that the instantaneous value of the current pressure returns to the initial value value, it indicates that the current user's feet leave the track. Based on the above detection principle, the time period during which the user's left and right feet touch the ground (ie, touch the runway) can be determined. Combining this time period with the running video, the video image of the process of the user's left and right feet touching the ground can be determined. Then, the pixel coordinate height of the key point of the foot is determined from this part of the video image, and combined with the corresponding time point information, the curve of the left and right feet touching the ground can be constructed.

Further, with reference to Figure 3, the pixel coordinate height of the key point of the left foot of each frame of the image is used as the value of the y-axis on the function curve, and the corresponding time point is used as the value of the x-axis on the function curve, so that the following can be obtained. Figure 3 shows the curve of the left foot touching the ground. Similarly, as shown in Figure 4, the pixel coordinate height of the key point of the right foot of each frame of the image is taken as the value of the y-axis on the function curve, and the corresponding time point is taken as the value of the x-axis on the function curve, so that Obtain the curve of the right foot touching the ground as shown in Figure 4. It can be understood that in the pixel coordinate system, the larger the value of the y-axis, the lower the current key point of the foot, that is, the closer to the runway. According to the whole touchdown process of the foot from just touching the ground to preparing to leave the track during running, it can be determined that when the foot just touches the ground, the position of the key point of the foot should be at the lowest point in the image pixel (that is, the y value is the largest ), and its pixel coordinate height should correspond to the peaks on the curve of the left and right feet touching the ground. Similarly, when the foot is ready to leave the track, the position of the key point of the foot should be at the highest point in the image pixel (that is, the y value is the smallest), and its pixel coordinate height should correspond to the trough on the curve of the left and right feet touching the ground.

Exemplarily, when constructing the curve of the left and right feet touching the ground, the pixel coordinate sequence of the key points of the foot is constructed by determining and extracting the coordinates of the key points of the foot during the grounding process of the left and right feet and the corresponding time point. Among them, according to the duration of the running video, the pixel coordinate sequence corresponding to the time length is divided according to the left and right feet, and the pixel coordinate sequence corresponding to the key points of the left and right feet is obtained. Then, from the pixel coordinate sequence, record the pixel coordinate height of the key point of the foot at each time point, as the coordinate value of the y-axis of the function curve (y-axis is the vertical axis of the image pixel coordinate system), so that the corresponding The left foot touchdown process curve of the left foot key point and the right foot touchdown process curve corresponding to the right foot key point.

In one embodiment, when confirming that the foot key point is at the pixel coordinate height of each frame of the video image, the treadmill determines one of the foot key points of the video image, from the foot key point The corresponding pixel coordinate height is determined from the pixel coordinates of the point; or, a plurality of the foot key points of the video image are determined, and the average pixel coordinate height of the plurality of the foot key points is calculated as the corresponding pixel coordinate height. It can be understood that, when detecting and determining the key points of the user's foot, the number of the key points of the user's foot may be one (such as ankle) or multiple (such as ankle, instep, toe, etc.), according to the actual foot. If only one foot key point is detected, the pixel coordinate height is directly determined according to the pixel coordinates of the detected foot key point. If multiple key points of the foot are detected, the average value is obtained according to the pixel coordinate height value of each key point of the foot, and the average value is used as the pixel coordinate height value for constructing the left and right foot touchdown process curve. By adaptively determining the height value of the pixel coordinates to construct the curve of the left and right feet touching the ground, the information recorded by the curve can be more accurate, and the subsequently obtained running posture parameters can be more accurate.

S130. Calculate the corresponding running posture parameter based on the mapping relationship, and output the running posture parameter.

Further, based on the above mapping relationship, the running parameters can be calculated correspondingly. The mapping relationship takes the left and right foot contact process curves as an example, calculates the corresponding running posture parameters based on the left foot contact process curve and the right foot contact process curve, and outputs the running posture parameters.

Specifically, based on the ground contact process curve of the left and right feet determined in the above step S120, the embodiment of the present application uses the ground contact process curve to calculate the running posture parameters, and further outputs and displays the calculated running posture parameters on the display screen of the treadmill to provide visual running posture analysis results, thereby optimizing the user's running experience.

Specifically, referring to FIG. 5, the corresponding running posture parameters are calculated based on the left foot contact process curve and the right foot contact process curve, and output the running posture parameters, including:

S1301. Determine the peak information and the trough information of the left foot contacting process curve and the right foot contacting process curve;

S1302. Calculate and output and display corresponding running posture parameters based on the wave crest information and the wave trough information, where the running posture parameters include left foot contact time, right foot contact time, ground contact balance parameters, air time, ground contact Airborne ratio and stride parameters.

Based on the left and right foot touchdown process curves, the peak information and wave trough information of the left foot touchdown process curve and the right foot touchdown process curve are determined. Since the peak information identifies the pixel coordinate height of the key point of the foot when the foot just touches the runway, the trough information identifies the pixel coordinate height of the key point of the foot when the foot is about to leave the runway. According to the characteristics of this curve, it can be determined that each section of the curve from the peak to the trough represents the corresponding foot contact process, and the user's left foot contact time and right foot contact time can be calculated accordingly.

Specifically, the calculation formula of the left foot contact time is:

表示所述左脚触地过程曲线在第i次左脚触地过程中波谷所对应的时间点，

表示所述左脚触地过程曲线在第i次左脚触地过程中波峰所对应的时间点，

表示所述左脚触地过程曲线在第i次左脚触地过程的触地时长，

表示所述左脚触地时间，n_l为左脚触地次数；in,

represents the time point corresponding to the trough of the left foot contacting process curve in the i-th left foot contacting process,

represents the time point corresponding to the peak of the left foot contacting process curve in the i-th left foot contacting process,

represents the touchdown duration of the left foot touchdown process curve in the i-th left foot touchdown process,

represents the time of the left foot touching the ground, n _l is the number of times the left foot touches the ground;

The calculation formula of the right foot contact time is:

表示所述右脚触地过程曲线在第i次右脚触地过程中波谷所对应的时间点，

表示所述右脚触地过程曲线在第i次右脚触地过程中波峰所对应的时间点，

表示所述右脚触地过程曲线在第i次右脚触地过程的触地时长，

表示所述右脚触地时间，n_r为右脚触地次数。in,

represents the time point corresponding to the trough of the right foot contacting process curve in the i-th right foot contacting process,

represents the time point corresponding to the peak of the right foot contacting process curve in the i-th right foot contacting process,

represents the touchdown duration of the right foot touchdown process curve in the i-th right foot touchdown process,

represents the time of the right foot touching the ground, and n _r is the number of times the right foot touches the ground.

Based on the above calculation formula, for the unilateral foot, during the running process, the key point of the foot often falls at the lowest point in the image pixel (the y value is the largest) when the foot just touches the ground, and then the foot is just ready to leave the ground. The key point often falls on the highest point in the image pixel (with the smallest y value), that is, in the curve of the left and right feet touching the ground, the time difference between each pair of wave peaks and wave troughs can be considered as the ground contact time during running. Therefore, the duration of each touchdown is determined by the time difference between each group of peaks and troughs in the touchdown process curve of the left and right feet, and then the average value is obtained according to the number of touchdowns, and the touchdown time of the left foot and the right foot corresponding to the left and right feet can be obtained. touchdown time.

Further, based on the ground contact time of the left foot and the ground contact time of the right foot obtained by the above calculation, the embodiment of the present application corresponds to the calculation of the ground contact balance parameter, and the calculation formula of the ground contact balance parameter is:

where α represents the touchdown balance parameter.

The ground contact balance parameter represents the ground contact time ratio of the left and right feet, and based on the above ground contact balance parameter, the difference between the left and right feet contact time can be determined. It can be understood that, if the value of the ground contact balance parameter α is closer to 1, it means that the left and right foot contact time is relatively balanced, otherwise, it means that the difference between the left and right foot contact time is large.

On the other hand, based on the touchdown time of the above-mentioned left and right feet, the embodiment of the present application further performs the calculation of the air time during the current user's running exercise, and the calculation formula of the air time is as follows:

Wherein, N represents the left and right foot rotation times after the left foot touchdown process curve and the right foot touchdown process curve are aligned, T represents the total duration of the running video, t ^tot represents the average time of left and right foot rotation, t ^air represents the air time.

Specifically, the flight time refers to the average time that the user's feet leave the ground during running. Based on the above-mentioned left foot contacting process curve and right foot contacting process curve, the curves are first aligned according to the time point information, and then the left foot touches the ground. In the ground process curve and the right foot contact process curve, the process from a peak to a trough of the curve is used as a ground contact process. Then, after aligning the left foot contacting process curve and the right foot contacting process curve, the process between the previous peak-to-valley and the next peak-to-valley in the curve is used as a rotation of the left and right feet. Use this to count the number of left and right foot rotations in the running video. Based on the number of left and right foot rotations and the total duration of the running video, the average time for one left and right foot rotation can be determined. Then, the difference between the average time t ^tot of the left and right foot rotation and the sum of the left and right feet touching the ground obtained by the above calculation can be calculated, and the average time t ^air for the user's feet to leave the ground during the running process can be calculated.

Further, based on the above-mentioned flight time and the contact time of the left and right feet, the embodiment of the present application correspondingly calculates the touch-to-air ratio during the running process of the user, and the calculation formula of the touch-to-air ratio is:

where β represents the touchdown airborne ratio.

The touch-to-air ratio identifies the ratio of the user’s left and right feet touching the ground to the air-to-air time. It is understandable that when the touch-to-air ratio reaches the standard value of touch and air, it means that the current user’s touch-to-air time and air-to-air ratio are in the process of running. The time distribution of time is accurate and meets the needs of the correct running posture. On the contrary, it indicates that the current running posture is wrong, and it is necessary to adjust the time distribution of the touchdown time and the air time during the running movement.

In addition, the embodiment of the present application also calculates the corresponding stride parameter based on the above-mentioned left and right foot contacting process curve, and the calculation formula of the stride parameter is:

为所述步幅参数。Among them, h is the normalized parameter of the vertical distance between the camera and the runway, cy is the normalized parameter of the pixel center point, f _y is the normalized parameter of the camera focal length, and _y ^peak-l is the process of the left foot touching the ground After the curve and the right foot touchdown process curve are aligned, the normalization parameter of the pixel coordinate height of the corresponding wave crest for one left foot touchdown, z ^peak-l is the key point of the foot corresponding to the wave crest for one left foot touchdown and The horizontal distance of the camera, y ^trough-r is the normalization parameter of the height of the pixel coordinates corresponding to the wave valley after the left foot touchdown process curve and the right foot touchdown process curve are aligned, z ^trough-r is the horizontal distance between the key point of the foot corresponding to the trough of a right foot touchdown and the camera, and y ^peak-r is the alignment of the left foot touchdown process curve and the right foot touchdown process curve, a right foot touchdown process curve is aligned. The normalization parameter of the height of the pixel coordinates of the corresponding wave crest, z ^peak-r is the horizontal distance between the key point of the foot corresponding to the wave crest and the camera when a right foot touches the ground, y ^trough-l is the left foot touching the ground After the process curve is aligned with the process curve of the right foot touching the ground, the normalization parameter of the pixel coordinate height of a trough corresponding to a left foot touching the ground, z ^trough-l is the key point of the foot corresponding to a trough when a left foot touches the ground The horizontal distance from the camera, z _i ^trough -zi ^peak represents the horizontal distance between the key point of the foot corresponding to the trough of the right foot and the key point of the foot corresponding to the peak of the left foot when the left and right feet touch the ground at the same time for the _i -th time, or the left foot The horizontal distance between the key point of the foot corresponding to the trough and the key point of the foot corresponding to the crest of the right foot, N represents the number of left and right foot rotations after the left foot touchdown process curve and the right foot touchdown process curve are aligned,

is the stride parameter.

It can be understood that the stride parameter is the average stride in the current running process of the user. As shown in Figure 6, o is the camera position, z is the horizontal distance between the key point of the foot and the camera, h is the vertical distance between the camera and the runway, c _y is the pixel center point, f _y is the camera focal length, and y is the left foot When touching the ground or the right foot touches the ground, the key point of the foot corresponds to the pixel coordinate height of the peak or trough of the wave, then based on the coordinate system transformation, the geometric relationship formula can be obtained:

Correspondingly, based on the above geometric relationship formula, the horizontal distance between the key point of the foot and the camera can be obtained. Then, according to the pixel coordinate height of the key point of the foot corresponding to the peak or trough when the left foot touches the ground or the right foot touches the ground, it can be obtained that the right foot is just about to be lifted and the left foot is just touched the ground, or the left foot is about to be lifted at the same time. The horizontal distance between the key point of the foot and the camera when the right foot just touches the ground, the difference between the two is the stride of a left and right foot rotation. Finally, the above stride parameter can be obtained by taking the average value based on the stride of each left and right caster rotation.

After that, based on the left foot contact time, right foot contact time, ground contact balance parameters, flight time, ground contact-to-air ratio and stride parameters calculated above, they are output to the display screen of the treadmill for display. At this time, the user can intuitively know his own running posture based on the running posture parameters displayed above, and adjust the running posture in time when the running posture is wrong, so as to achieve a better running experience.

In the above, by acquiring the running video in real time, the key points of the feet of the video images of a certain number of frames in the running video are determined based on the detection of skeleton key points; The height and the time points corresponding to the video images of a certain number of frames are used to generate the mapping relationship between the pixel coordinate height and the corresponding time point; the corresponding running posture parameters are calculated based on the mapping relationship, and the running posture parameters are output. Using the above technical means, by detecting the key points of the footsteps in the video image, generating the curve of the left and right feet touching the ground based on the key points of the footsteps, and analyzing the running posture according to the curve of the left and right feet touching the ground, the running posture analysis based on the running video image can be realized. Avoid the extra burden of detection equipment on users, and optimize the user's running experience. In addition, in the embodiment of the present application, the running posture analysis is performed by combining the detection of the key points of the footsteps and the contacting process curves of the left and right feet, which can realize the accurate calculation of the running posture parameters and provide a more accurate running posture analysis result.

On the basis of the above-mentioned embodiment, the embodiment of the present application further includes:

The touchdown balance parameter, the touchdown-to-air ratio and the stride parameter are compared with corresponding set thresholds, and a corresponding running posture symmetry analysis result is output.

The treadmill of the embodiment of the present application is preset with threshold information corresponding to the touchdown balance parameter, touchdown air ratio and stride parameter, and outputs the corresponding running posture symmetry analysis result to prompt the current user's running posture to correspond to this situation. It can be understood that the closer the touchdown balance parameter is to 1, the closer the touchdown time of the left and right feet, and the better the symmetry of the running posture. Similarly, when the touchdown-to-air ratio or stride parameter meets the corresponding threshold, it indicates that the running posture is better. On the contrary, when the above-mentioned touchdown balance parameter, touchdown-to-air ratio or stride parameter and the corresponding threshold information are detected When the difference is large (reaches the set deviation value), it indicates that the user is currently running with a wrong running posture. At this time, in order to guide the user to perform a healthy running exercise, the treadmill prompts the user to adjust the running posture, so as to provide a better running experience.

Embodiment 2:

On the basis of the foregoing embodiment, FIG. 7 is a schematic structural diagram of a running posture analysis device based on a treadmill provided in Embodiment 2 of the present application. Referring to FIG. 7 , the apparatus for analyzing running posture based on a treadmill provided in this embodiment specifically includes: an acquisition module 21 , a generation module 22 and an output module 23 .

Wherein, the acquisition module 21 is used to acquire the running video in real time, and determine the foot key points of the video images of a certain number of frames in the running video based on the skeleton key point detection;

The generation module 22 is used to confirm the pixel coordinate height of the key point of the foot in the video image of a certain number of frames, and generate the pixel coordinate height according to the pixel coordinate height and the time point corresponding to the video image of a certain number of frames. The mapping relationship between the pixel coordinate height and the corresponding time point;

The output module 23 is configured to calculate the corresponding running posture parameters based on the mapping relationship, and output the running posture parameters.

On the basis of the above embodiment, the generating module 22 includes:

a determining unit, configured to determine one key point of the foot of the video image, and determine the corresponding pixel coordinate height from the pixel coordinates of the key point of the foot; or, determine a plurality of the feet of the video image The key points of the foot are calculated, and the average value of the height of the pixel coordinates of the plurality of key points of the foot is calculated as the corresponding height of the pixel coordinates.

On the basis of the above embodiment, the acquisition module 21 includes:

Based on skeleton key point detection, the video images of a certain number of frames corresponding to the process of the left foot and/or the right foot touching the ground are selected from the running video, and the pixel coordinates of the key points of the foot are determined from the video images.

On the basis of the above embodiment, the generating module 22 includes:

Corresponding to the height of the pixel coordinates of the key point of the foot of the left foot, a first mapping relationship is generated with the time point corresponding to the video image;

A second mapping relationship is generated corresponding to the height of the pixel coordinates of the key point of the foot of the right foot and the time point corresponding to the video image.

On the basis of the above embodiment, the first mapping relationship is represented as a left foot contacting process curve, and the second mapping relationship is represented as a right foot contacting process curve;

Correspondingly, the output module 23 includes:

a curve generating unit, configured to determine peak information and trough information of the left foot contacting process curve and the right foot contacting process curve;

A display unit, configured to calculate and output and display corresponding running posture parameters based on the wave crest information and the wave trough information, where the running posture parameters include left foot contact time, right foot contact time, ground contact balance parameters, and air time , touchdown air ratio and stride parameters.

On the basis of the above embodiment, the calculation formula of the left foot contact time is:

表示所述左脚触地过程曲线在第i次左脚触地过程中波谷所对应的时间点，

表示所述左脚触地过程曲线在第i次左脚触地过程中波峰所对应的时间点，

表示所述左脚触地过程曲线在第i次左脚触地过程的触地时长，

表示所述左脚触地时间，n_l为左脚触地次数；in,

represents the time point corresponding to the trough of the left foot contacting process curve in the i-th left foot contacting process,

represents the time point corresponding to the peak of the left foot contacting process curve in the i-th left foot contacting process,

represents the touchdown duration of the left foot touchdown process curve in the i-th left foot touchdown process,

represents the time of the left foot touching the ground, n _l is the number of times the left foot touches the ground;

The calculation formula of the right foot contact time is:

表示所述右脚触地过程曲线在第i次右脚触地过程中波谷所对应的时间点，

表示所述右脚触地过程曲线在第i次右脚触地过程中波峰所对应的时间点，

表示所述右脚触地过程曲线在第i次右脚触地过程的触地时长，

表示所述右脚触地时间，n_r为右脚触地次数。in,

represents the time point corresponding to the trough of the right foot contacting process curve in the i-th right foot contacting process,

represents the time point corresponding to the peak of the right foot contacting process curve in the i-th right foot contacting process,

represents the touchdown duration of the right foot touchdown process curve in the i-th right foot touchdown process,

represents the time of the right foot touching the ground, and n _r is the number of times the right foot touches the ground.

On the basis of the above embodiment, the calculation formula of the ground contact balance parameter is:

where α represents the touchdown balance parameter.

On the basis of the above embodiment, the calculation formula of the flight time is:

Wherein, N represents the left and right foot rotation times after the left foot touchdown process curve and the right foot touchdown process curve are aligned, T represents the total duration of the running video, t ^tot represents the average time of left and right foot rotation, t ^air represents the air time.

On the basis of the above embodiment, the calculation formula of the air-to-ground ratio is:

where β represents the touchdown airborne ratio.

On the basis of the above embodiment, the calculation formula of the stride parameter is:

为所述步幅参数。Among them, h is the normalized parameter of the vertical distance between the camera and the runway, cy is the normalized parameter of the pixel center point, f _y is the normalized parameter of the camera focal length, and _y ^peak-l is the process of the left foot touching the ground After the curve and the right foot touchdown process curve are aligned, the normalization parameter of the pixel coordinate height of the corresponding wave crest for one left foot touchdown, z ^peak-l is the key point of the foot corresponding to the wave crest for one left foot touchdown and The horizontal distance of the camera, y ^trough-r is the normalization parameter of the height of the pixel coordinates corresponding to the wave valley after the left foot touchdown process curve and the right foot touchdown process curve are aligned, z ^trough-r is the horizontal distance between the key point of the foot corresponding to the trough of a right foot touchdown and the camera, and y ^peak-r is the alignment of the left foot touchdown process curve and the right foot touchdown process curve, a right foot touchdown process curve is aligned. The normalization parameter of the height of the pixel coordinates of the corresponding wave crest, z ^peak-r is the horizontal distance between the key point of the foot corresponding to the wave crest and the camera when a right foot touches the ground, y ^trough-l is the left foot touching the ground After the process curve is aligned with the process curve of the right foot touching the ground, the normalization parameter of the pixel coordinate height of a trough corresponding to a left foot touching the ground, z ^trough-l is the key point of the foot corresponding to a trough when a left foot touches the ground The horizontal distance from the camera, z _i ^trough -zi ^peak represents the horizontal distance between the key point of the foot corresponding to the trough of the right foot and the key point of the foot corresponding to the peak of the left foot when the left and right feet touch the ground at the same time for the _i -th time, or the left foot The horizontal distance between the key point of the foot corresponding to the trough and the key point of the foot corresponding to the crest of the right foot, N represents the number of left and right foot rotations after the left foot touchdown process curve and the right foot touchdown process curve are aligned,

is the stride parameter.

On the basis of the above-mentioned embodiment, the running posture analysis device based on the treadmill further includes:

An analysis module, configured to compare the touchdown balance parameter, the touchdown-to-air ratio and the stride parameter with a corresponding set threshold, and output a corresponding running posture symmetry analysis result.

In the above, by acquiring the running video in real time, the key points of the feet of the video images of a certain number of frames in the running video are determined based on the detection of skeleton key points; The height and the time points corresponding to the video images of a certain number of frames are used to generate the mapping relationship between the pixel coordinate height and the corresponding time point; the corresponding running posture parameters are calculated based on the mapping relationship, and the running posture parameters are output. Using the above technical means, by detecting the key points of the footsteps in the video image, generating the curve of the left and right feet touching the ground based on the key points of the footsteps, and analyzing the running posture according to the curve of the left and right feet touching the ground, the running posture analysis based on the running video image can be realized. Avoid the extra burden of detection equipment on users, and optimize the user's running experience.

In addition, the embodiment of the present application calculates the running posture parameters by determining the key points of the feet when the left and right feet touch the ground, which can ensure the accurate detection of the key points of the skeleton, avoid the situation that the key points of the feet are affected by occlusion and cause detection errors, and make the running posture parameters. The calculation is more accurate; by selecting the key points of the feet when the left and right feet touch the ground from the running video to calculate the running posture parameters, the number of selected key points can be reduced, thereby reducing the calculation amount of running posture parameters and improving the efficiency of running posture analysis. On the other hand, by selecting the key points of the feet when the left and right feet touch the ground from the running video to calculate the running posture parameters, it can also ensure the rapid extraction of the key points of the feet, reduce the error effect caused by the running video recording delay, and further improve The accuracy of running posture parameter calculation.

In addition, in the embodiment of the present application, the running posture analysis is performed by combining the detection of the key points of the footsteps and the contacting process curves of the left and right feet, which can realize the accurate calculation of the running posture parameters and provide a more accurate running posture analysis result.

The treadmill-based running posture analysis device provided in the second embodiment of the present application can be used to execute the treadmill-based running posture analysis method provided in the above-mentioned first embodiment, and has corresponding functions and beneficial effects.

Embodiment three:

The third embodiment of the present application provides a treadmill. Referring to FIG. 8 , the treadmill includes: a processor 31 , a memory 32 , a communication module 33 , an input device 34 and an output device 35 . The number of processors 31 in the treadmill may be one or more, and the number of memories 32 in the treadmill may be one or more. The processor 31 , the memory 32 , the communication module 33 , the input device 34 and the output device 35 of the treadmill can be connected through a bus or in other ways.

As a computer-readable storage medium, the memory 32 can be used to store software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the treadmill-based running posture analysis method described in any embodiment of the present application (for example, The acquisition module, the generation module and the output module in the treadmill-based running posture analysis device). The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the device, and the like. Additionally, memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, the memory may further include memory located remotely from the processor, the remote memory being connectable to the device through a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

The communication module 33 is used for data transmission.

The processor 31 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory, ie, implements the above-mentioned treadmill-based running posture analysis method.

The input device 34 may be used to receive input numerical or character information and to generate key signal input related to user settings and function control of the device. The output device 35 may include a display device such as a display screen.

The treadmill provided above can be used to execute the treadmill-based running posture analysis method provided in the first embodiment, and has corresponding functions and beneficial effects.

Embodiment 4:

Embodiments of the present application further provide a storage medium containing computer-executable instructions, where the computer-executable instructions are used to execute a treadmill-based running posture analysis method when executed by a computer processor. The posture analysis method includes: acquiring a running video in real time, and determining the key points of the feet of the video images of a certain number of frames in the running video based on the detection of skeleton key points; confirming that the key points of the feet are in the video images of the certain number of frames. The height of the pixel coordinates, according to the height of the pixel coordinates and the time points corresponding to the video images of a certain number of frames, generate the mapping relationship between the height of the pixel coordinates and the corresponding time points; calculate the corresponding running based on the mapping relationship attitude parameters, and output the running attitude parameters.

storage medium - any of various types of memory devices or storage devices. The term "storage medium" is intended to include: installation media, such as CD-ROMs, floppy disks, or tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc. ; non-volatile memory, such as flash memory, magnetic media (eg hard disk or optical storage); registers or other similar types of memory elements, etc. The storage medium may also include other types of memory or combinations thereof. In addition, the storage medium may be located in the first computer system in which the program is executed, or may be located in a second, different computer system connected to the first computer system through a network such as the Internet. The second computer system may provide program instructions to the first computer for execution. The term "storage medium" may include two or more storage media residing in different locations (eg, in different computer systems connected by a network). The storage medium may store program instructions (eg, embodied as a computer program) executable by one or more processors.

Of course, a storage medium containing computer-executable instructions provided by the embodiments of the present application is not limited to the above-mentioned treadmill-based running posture analysis method, and the computer-executable instructions of the storage medium provided by any embodiment of the present application can also be executed. Related operations in the treadmill-based running posture analysis method.

The treadmill-based running posture analysis device, storage medium, and treadmill provided in the foregoing embodiments can execute the treadmill-based running posture analysis method provided by any embodiment of the present application, and the technical details not described in detail in the foregoing embodiments , please refer to the treadmill-based running posture analysis method provided in any embodiment of the present application.

The above are only the preferred embodiments of the present application and the applied technical principles. The present application is not limited to the specific embodiments described herein, and various obvious changes, readjustments and substitutions that can be made by those skilled in the art will not depart from the protection scope of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments, and can also include more other equivalent embodiments without departing from the concept of the present application. The scope is determined by the scope of the claims.

Claims (13)

1. A running posture analysis method based on a treadmill is characterized by comprising the following steps:

acquiring a running video in real time, and determining foot key points of video images of a certain number of frames in the running video based on skeleton key point detection;

confirming the pixel coordinate height of the foot key point in the video images of a certain number of frames, and generating a mapping relation between the pixel coordinate height and the corresponding time point according to the pixel coordinate height and the time point corresponding to the video images of a certain number of frames;

and calculating corresponding running posture parameters based on the mapping relation, and outputting the running posture parameters.

2. The treadmill-based gesture analysis method of claim 1, wherein identifying pixel coordinate heights of the foot keypoints in the video images of a number of frames comprises:

determining one foot key point of the video image, and determining the corresponding pixel coordinate height from the pixel coordinates of the foot key point; or,

determining a plurality of foot key points of the video image, and calculating the pixel coordinate height mean value of the foot key points as the corresponding pixel coordinate height.

3. A treadmill-based running gesture analysis method as recited in claim 1, wherein determining foot keypoints for video images of a number of frames in the running video based on skeletal keypoint detection comprises:

selecting a certain number of frames of the video images corresponding to the left foot and/or right foot touchdown process from the running video based on skeleton key point detection, and determining pixel coordinates of foot key points from the video images.

4. The running gesture analysis method based on the treadmill according to claim 3, wherein generating the mapping relationship between the pixel coordinate height and the corresponding time point according to the pixel coordinate height and the time point corresponding to the video image of a certain number of frames comprises:

generating a first mapping relation between the pixel coordinate height of the foot key point corresponding to the left foot and the time point corresponding to the video image;

and generating a second mapping relation between the pixel coordinate height of the foot key point corresponding to the right foot and the time point corresponding to the video image.

5. The treadmill based running gesture analysis method of claim 4, wherein the first mapping is represented as a left foot touchdown process curve and the second mapping is represented as a right foot touchdown process curve;

correspondingly, calculating a corresponding running posture parameter based on the mapping relation, and outputting the running posture parameter, including:

determining peak information and trough information of the left foot touchdown process curve and the right foot touchdown process curve;

and calculating and outputting and displaying corresponding running posture parameters based on the peak information and the valley information, wherein the running posture parameters comprise left foot touchdown time, right foot touchdown time, touchdown balance parameters, flight time, touchdown flight ratio and stride parameters.

6. A running posture analysis method based on a treadmill as recited in claim 5, wherein the left foot contact time calculation formula is:

wherein,

represents the time point corresponding to the wave trough of the curve of the left foot touchdown process in the ith left foot touchdown process,

represents the time point corresponding to the peak of the curve of the left foot touchdown process in the ith left foot touchdown process,

represents the touchdown time length of the left foot touchdown process curve in the ith left foot touchdown process,

representing said left foot contact time, n _l The number of left foot touchdowns;

the right foot touchdown time calculation formula is as follows:

wherein,

represents the time point corresponding to the wave trough of the curve of the right foot touchdown process in the ith right foot touchdown process,

represents the time point corresponding to the peak of the curve of the right foot touchdown process in the ith right foot touchdown process,

represents the contact duration of the curve of the right foot contact process in the ith right foot contact process,

representing said right foot contact time, n _r The number of right foot touchdowns.

7. The running posture analysis method based on a treadmill of claim 6, wherein the touchdown balance parameter calculation formula is:

wherein α represents the touchdown balance parameter.

8. A treadmill based running gesture analysis method according to claim 6, wherein the flight time is calculated by the formula:

wherein N represents the left and right caster changing times after the left foot touchdown process curve and the right foot touchdown process curve are aligned, T represents the total duration of the running video, and T represents the total duration of the running video ^tot Mean time to left and right foot rotation, t ^air Representing the flight time.

9. A running posture analyzing method based on a treadmill as recited in claim 8, wherein the formula for calculating the touchdown-to-empty ratio is:

wherein β represents the touchdown-to-empty ratio.

10. The running posture analysis method based on a treadmill of claim 5, after calculating corresponding running posture parameters based on the mapping relationship and outputting the running posture parameters, further comprising:

and comparing the touchdown balance parameter, the touchdown flight-space ratio and the stride parameter with corresponding set thresholds, and outputting corresponding running posture symmetry analysis results.

11. A running posture analysis device based on a running machine is characterized by comprising:

the system comprises an acquisition module, a judgment module and a display module, wherein the acquisition module is used for acquiring a running video in real time and determining foot key points of video images of a certain number of frames in the running video based on skeleton key point detection;

the generating module is used for confirming the pixel coordinate heights of the foot key points in the video images of a certain number of frames and generating the mapping relation between the pixel coordinate heights and the corresponding time points according to the pixel coordinate heights and the time points corresponding to the video images of the certain number of frames;

and the output module is used for calculating the corresponding running posture parameters based on the mapping relation and outputting the running posture parameters.

12. A treadmill, comprising:

a memory and one or more processors;

the memory for storing one or more programs;

when executed by the one or more processors, cause the one or more processors to implement a treadmill based gesture analysis method as recited in any of claims 1-10.

13. A storage medium containing computer executable instructions for performing the treadmill based running gesture analysis method of any one of claims 1-10 when executed by a computer processor.

Images