Methods, devices, storage media and products for analyzing the gait behavior of organisms

Abstract

This application discloses a method, apparatus, storage medium, and product for analyzing the gait behavior of an organism. The method includes: acquiring gait data of the organism; performing posture estimation based on the gait data to obtain the organism's movement posture; obtaining gait indices based on the organism's movement posture; and analyzing the organism's gait behavior based on the gait indices. In this embodiment, by acquiring the organism's gait data to obtain its movement posture, and then analyzing it based on gait indices generated from the movement posture, this method effectively reduces errors caused by subjective factors during human analysis, ensuring the accuracy of the data when analyzing the organism's gait behavior using gait indices.

Description

Technical Field

[0001] This application relates to the field of medical research technology, and in particular to a method, device, storage medium and product for analyzing the gait behavior of an organism. Background Technology

[0002] Animal experiments, as a crucial foundation for various life science disciplines, are receiving increasing attention from researchers. Accurate and efficient animal experimental data are essential for subsequent research. Therefore, analyzing animal postures, acquiring experimental data promptly, and establishing a complete and reliable evaluation system are currently the focus of animal experiment research.

[0003] In existing technologies, biological posture analysis often relies on visual observation. This method cannot accurately obtain gait indicators of a certain organism, let alone accurately analyze the gait behavior of the organism. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method, device, storage medium and product for analyzing the gait behavior of organisms, which can obtain the gait data of organisms, obtain the posture of organisms based on the gait data, and obtain the gait behavior of organisms, thereby achieving the purpose of obtaining the behavioral results of organisms, and can effectively reduce the errors caused by human analysis.

[0005] In a first aspect, embodiments of this application provide a method for analyzing the gait behavior of an organism, including:

[0006] Acquire gait data of organisms;

[0007] The attitude is estimated based on the gait data of the organism to obtain the movement attitude of the organism;

[0008] Based on the organism's movement posture, gait indices of the organism are obtained, and the organism's gait behavior is analyzed based on the gait indices.

[0009] In some embodiments, prior to the pose estimation based on the organism's gait data, the method further includes:

[0010] Valid gait data is extracted from the gait data of the organism, wherein the valid gait data includes data whose clarity, resolution, and target position conform to a preset range.

[0011] In some embodiments, the step of performing posture estimation based on the gait data of the organism to obtain the organism's movement posture includes:

[0012] The gait data of the organism is input into the pose estimation network model to obtain the movement pose of the organism.

[0013] In some embodiments, gait data samples of an organism are acquired;

[0014] The gait data samples of the organism are input into the initial pose estimation network model, and the initial pose estimation network model is trained based on the output of the initial pose estimation network model and the loss between the label samples to obtain the pose estimation network model, wherein the label samples include the action poses of the gait data samples of the organism.

[0015] In some embodiments, obtaining gait indices of the organism based on its movement postures, and analyzing the organism's gait behavior based on the gait indices, includes:

[0016] Based on the organism's movement posture, a gait diagram of the organism is obtained;

[0017] The gait indices of the organism are determined based on the gait diagram, and the gait behavior of the organism is analyzed based on the gait indices.

[0018] In some embodiments, analyzing the gait behavior of the organism based on the gait indices includes:

[0019] Based on the gait indices and gait index samples from multiple periods of the organism, the vital signs of the organism corresponding to the gait behavior are analyzed.

[0020] In some embodiments, following the action posture of the organism, the method further includes:

[0021] The organism's movement posture is corrected based on its gait relationship.

[0022] Secondly, embodiments of this application provide a gait behavior analysis device for an organism, comprising: a gait data acquisition unit for acquiring gait data of an organism; a posture estimation unit for performing posture estimation based on the gait data of the organism to obtain the movement posture of the organism; and a gait index calculation unit for obtaining the gait index of the organism based on the movement posture of the organism, and analyzing the gait behavior of the organism based on the gait index.

[0023] Thirdly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, the computer program being used to implement the gait behavior analysis method for organisms as described in the first aspect above.

[0024] Fourthly, embodiments of this application provide a computer program product having a computer program stored thereon, the computer program being used to implement the gait behavior analysis method for organisms as described in the first aspect above.

[0025] The gait behavior analysis method, apparatus, storage medium, and product for organisms provided in this application first acquire gait data of the organism to obtain its movement posture, and then analyze it based on gait indices generated from the movement posture. This method effectively reduces errors caused by subjective factors during human analysis, ensuring the accuracy of the data when analyzing the organism's gait behavior using gait indices. Attached Figure Description

[0026] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0027] Figure 1 This is a flowchart of a method for analyzing the gait behavior of an organism according to an embodiment of this application;

[0028] Figure 2 This is a schematic diagram illustrating the acquisition of gait data of an organism according to an embodiment of this application;

[0029] Figure 3 This is a schematic diagram of the gait of an organism according to an embodiment of this application;

[0030] Figure 4 This is a gait analysis diagram of an organism according to an embodiment of this application;

[0031] Figure 5 This is a gait analysis diagram of an organism according to another embodiment of this application;

[0032] Figure 6 This is a structural block diagram of the gait behavior analysis device for an organism according to an embodiment of this application;

[0033] Figure 7 This is a schematic diagram of the effective gait range of an organism according to an embodiment of this application;

[0034] Figure 8 This is a schematic diagram showing the height of key points of an organism above the ground in an embodiment of this application;

[0035] Figure 9 This is a diagram showing the vertical distance between the hip and ankle of an organism according to an embodiment of this application. Detailed Implementation

[0036] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the relevant disclosure and not intended to limit the scope of the disclosure. Furthermore, it should be noted that, for ease of description, only the parts relevant to the disclosure are shown in the accompanying drawings.

[0037] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0038] The following describes, with reference to the accompanying drawings, a method, apparatus, storage medium, and product for analyzing the gait behavior of organisms according to embodiments of the present invention.

[0039] Figure 1 This is a flowchart of a biological gait behavior analysis method according to an embodiment of this application, as shown below. Figure 1 As shown, a method for analyzing the gait behavior of an organism according to an embodiment of this application includes the following steps:

[0040] S101, acquire gait data of the organism.

[0041] Specifically, it involves acquiring gait data of organisms.

[0042] For example, gait data can be obtained by capturing videos of an organism. Here, we will use the gait of a monkey's lower limbs as an example. Figure 2 As shown, video data is obtained by recording the monkey's lower limbs at a fixed angle, and the movement trajectory information of key parts of the monkey's lower limbs is obtained to obtain gait data.

[0043] In some embodiments, a high-speed camera and a treadmill can also be used to collect gait data of the organism on the treadmill. This allows the camera to remain stable, ensuring higher accuracy of the gait data obtained through video recording.

[0044] S102, based on the gait data of the organism, perform posture estimation to obtain the movement posture of the organism.

[0045] Specifically, based on pre-obtained gait data, deep learning methods are used to estimate the acquired gait data of the organism, thereby obtaining the organism's movement posture. This is achieved through learning from a large amount of data to ensure the accuracy of the state estimation.

[0046] In some embodiments, before performing posture estimation based on the gait data of the organism, the method further includes: extracting valid gait data from the gait data of the organism, wherein the valid gait data includes data whose sharpness, resolution, and target position conform to a preset range.

[0047] Specifically, not all gait data obtained through this method meets the requirements for attitude estimation. Some gait data is invalid due to low video clarity, incomplete video recognition, or key position shifts caused by motion capture equipment malfunctions. For example, foot positions may not be accurately located, or the gait data may not conform to conventional patterns. Therefore, it is necessary to acquire valid gait video data and remove abnormal data to reduce computational load and data errors. The preset data range can be dynamically adjusted according to the implementation analysis requirements.

[0048] In some embodiments, the step of performing posture estimation based on the gait data of the organism to obtain the organism's movement posture includes:

[0049] The gait data of the organism is input into the pose estimation network model to obtain the movement pose of the organism.

[0050] Specifically, a large amount of gait data is used to train the pose estimation network model. The gait data of the organism is input into the trained pose estimation network model to obtain the organism's movement posture. This pose estimation network model can be trained using limited training data.

[0051] For example, a trained pose estimation network model can obtain the action pose of an organism from its motion video, i.e., gait data. This action pose includes the location information and confidence level of the organism's gait key points. In other implementations, the gait key point locations can also be plotted in the organism's motion video.

[0052] In other implementations, the pose estimation network model also includes a gait index calculation module.

[0053] In some embodiments, gait data samples of an organism are acquired;

[0054] The gait data samples of the organism are input into the initial pose estimation network model, and the initial pose estimation network model is trained based on the output of the initial pose estimation network model and the loss between the label samples to obtain the pose estimation network model, wherein the label samples include the action poses of the gait data samples of the organism.

[0055] Specifically, the first step is to acquire gait data samples of the organism, which consist of pre-prepared gait videos and corresponding labels. These gait data samples are then input into the pose estimation network model for training. If the difference between the output of the pose estimation network model and the sample labels is within a preset range, the model is considered successfully trained.

[0056] Here, when acquiring video annotation data of biological gait, the amount of annotation can be reduced by preprocessing, that is, by extracting frames and cropping the acquired video data to ensure that the organism is located in the center of the video, and using the K-means algorithm to extract key frames from the video data for annotation. The generated training data contains biological poses in various situations.

[0057] In another embodiment, the DeepLabCut toolkit is used, with ResNet50 as the baseline. Both training and prediction data are cropped to the range of biological activity. The ratio of validation set to training set is set to 0.15; the batch size is set to 1; the learning rate is set to a constant 0.001; the total training steps for the pose estimation network model are set to 60,000 steps, with a learning rate of 0.005 for steps 0-14,000, 0.02 for steps 14,000-30,000, 0.002 for steps 30,000-45,000, and 0.001 for steps 45,000-60,000. During training, the pose estimation network model is saved every 2,500 steps, for a total of 12 models. New pose estimation network models overwrite previous ones. Finally, all models are evaluated, and the optimal pose estimation network model is selected.

[0058] In some embodiments, the pose estimation network model includes DeepLabCut, HRNet, OpenPose, or Hourglass.

[0059] Specifically, taking DeepLabCut as an example, DeepLabCut constructs a pose estimation network model. DeepLabCut is typically used for animal pose estimation. It builds upon existing animal pose estimation algorithms and can train the pose estimation network model using limited training data. DeepLabCut supports pose estimation for various organisms, such as horses, monkeys, and rats. HRNet, OpenPose, or Hourglas can also be used to construct pose estimation network models.

[0060] S103, Based on the organism's movement posture, obtain the organism's gait index, and analyze the organism's gait behavior based on the gait index.

[0061] Specifically, based on the acquired biological movement postures, gait indices of the organism are obtained, and the current gait behavior of the organism is analyzed based on the obtained gait indices.

[0062] In some embodiments, before obtaining the gait index of the organism, the method further includes: extracting the gait start frame from the organism's movement posture and removing abnormal gait frames.

[0063] Specifically, this study uses monkey toes for calculation within the effective gait interval. The calculation of the effective gait interval includes three aspects: finding the effective gait interval, screening for gait anomalies, and calculating indicators. The effective gait interval is determined based on the confidence level of the predicted location using the formula:

[0064] Calculate the weighted toe distance from the origin and plot the distance curve. Remove small peaks through multiple smoothing operations. Then, recursively find the frames containing peak values ​​and remove abnormal frames with excessively small peak values. Finally, select the effective gait interval based on the gait interval time, where (x i ,y i () represents the toe coordinates in the i-th frame, w is the weight parameter, and coef is the confidence score of the toe keypoint. The starting frame of this effective gait interval is obtained by... Figure 7 As shown.

[0065] In practical applications, some data may be considered anomalous. For example, anomalous gait data includes keypoint positional relationship filtering and distance filtering: Based on keypoint positional relationships, such as the angles formed by the hip-knee and knee-condyle in the "monkey" shape, and the angles formed by the knee-condyle and condyle-toe in the "monkey" shape, all should be less than 180° and their orientations should be opposite. Taking the hip as the origin, the knee should be in the first or fourth quadrant; taking the knee as the origin, the condyle should be in the third or fourth quadrant, and so on. Based on keypoint distance relationships, the average distance between connected keypoints in the gait data is calculated. If the keypoint distance in a frame of gait data is greater than 1.5 times the average distance or less than 0.3 times the average distance, then that frame is considered anomalous.

[0066] The specific calculation of the indicators includes: OSCT (one step cycle time), OSCD (one step cycle distance), maximum and minimum knee and condylar joint angles, step height, and other measurement indicators. OSCT is the time required for one step, and OSCD is the distance of one step. The formulas are as follows:

[0067]

[0068] Step height is the maximum height the condyle is raised during a single step.

[0069] Where OSCT is the ratio of the number of frames representing a complete gait to the frame rate in the gait data of an organism, V is the speed of the treadmill, and x start and x end These are the x-axis coordinates of the toes at the start and end of the gait of the organism, respectively. Ratio is the ratio of video pixels to the actual distance, and θ is the angle between the treadmill plane and the x-axis.

[0070] For example, using a monkey as an example, this method employs a deep neural network, pre-trained with a large amount of biological posture data to obtain gait indices. These gait indices can then be used to analyze the gait recovery of the monkey's right hind limb. This method can accurately estimate the posture of collected monkey gait videos and generate gait maps based on the estimation results. Furthermore, by combining multi-period, multi-sample analysis, it can effectively summarize the organism's gait behavior. Compared to existing technologies, using posture estimation for gait recovery analysis offers higher accuracy and provides quantification capabilities.

[0071] In some embodiments, obtaining gait indices of the organism based on its movement postures, and analyzing the organism's gait behavior based on the gait indices, includes:

[0072] Based on the organism's movement posture, a gait diagram of the organism is obtained;

[0073] The gait indices of the organism are determined based on the gait diagram, and the gait behavior of the organism is analyzed based on the gait indices.

[0074] Specifically, by obtaining the location of key parts of an organism, determining its movement posture, and then generating a gait diagram of the organism, such as... Figure 3 As shown; hip-ankle vertical distance diagram, as... Figure 9 As shown, gait parameters of an organism, such as the position of the knee joint, are determined based on the gait diagram, and corresponding charts are generated to analyze the gait.

[0075] For example, taking a monkey's lower limbs as an example, by obtaining the positions of the hip, knee, ankle, and toes, a gait diagram of the monkey is obtained, as shown in the figure below. Figure 3 As shown, key points of an organism are acquired. Taking a monkey's lower limb as an example, the key points are: hip, knee, ankle, and toes. The positions of these key points at different time periods are obtained to generate corresponding analysis results, such as obtaining parameters like the monkey's lower limb movement speed. These parameters are then analyzed to determine the recovery status of the monkey's lower limb nerves, muscles, and bones. Figure 4 This is a graph showing the relationship between OSCD in an organism over a period of time at a treadmill speed of 1 km / h. Figure 5 The graph shows the relationship between OSCT and treadmill speed at 2 km / h over a certain period of time. The analysis results are as follows: Figure 4 and Figure 5 As shown; Figure 8 This is a schematic diagram showing the height of key points above the ground.

[0076] In some embodiments, analyzing the gait behavior of the organism based on the gait indices includes:

[0077] Based on the gait indices and gait index samples from multiple periods of the organism, the vital signs of the organism corresponding to the gait behavior are analyzed.

[0078] Specifically, multiple samples of the organism are analyzed to obtain gait indices at different times. This method allows for gait analysis of the organism's historical states, providing more accurate data.

[0079] In some embodiments, following the action posture of the organism, the method further includes:

[0080] The organism's movement posture is corrected based on its gait relationship.

[0081] Specifically, the organism can include several animals, such as dogs, cats, monkeys, etc. After gait indexes are calculated, the movement posture is corrected to ensure the accuracy of the gait calculation. Because some gait data is not clear enough due to low gait video quality, or because the video cannot be fully recognized (e.g., the foot position cannot be located), the recognition results may be biased. Therefore, it is necessary to correct the organism's posture to obtain accurate results.

[0082] In one embodiment, such as Figure 6 As shown, Figure 6 This is a structural block diagram of a gait behavior analysis device for an organism according to this application. The device includes a gait data acquisition unit 210, a posture estimation unit 220, and a gait index calculation unit 230.

[0083] Gait data acquisition unit 210 is used to acquire gait data of an organism;

[0084] The attitude estimation unit 220 is used to perform attitude estimation based on the gait data of the organism to obtain the action attitude of the organism.

[0085] The gait index calculation unit 230 is used to obtain the gait index of the organism based on the organism's movement posture, and to analyze the gait behavior of the organism based on the gait index.

[0086] In one embodiment, the posture estimation unit 220 further includes: performing posture estimation based on the gait data of the organism to obtain the movement posture of the organism, including:

[0087] The gait data of the organism is input into the pose estimation network model to obtain the movement pose of the organism.

[0088] In one embodiment, the gait index calculation unit 230 includes:

[0089] Based on the organism's movement posture, a gait diagram of the organism is obtained;

[0090] The gait indices of the organism are determined based on the gait diagram, and the gait behavior of the organism is analyzed based on the gait indices.

[0091] In one embodiment, the gait index calculation unit 230 includes:

[0092] Based on the gait indices and gait index samples from multiple periods of the organism, the vital signs of the organism corresponding to the gait behavior are analyzed.

[0093] In summary, the gait behavior analysis device for organisms provided in this application first acquires the gait data of the organism to obtain its movement posture, and then analyzes it based on gait indices generated from the movement posture. This method effectively reduces errors caused by subjective factors during human analysis, ensuring the accuracy of the data when analyzing the organism's gait behavior using gait indices.

[0094] In one embodiment, a non-transitory computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it performs the following steps: acquiring gait data of an organism; performing posture estimation based on the gait data of the organism to obtain the movement posture of the organism; obtaining gait indices of the organism based on the movement posture of the organism; and analyzing the gait behavior of the organism based on the gait indices.

[0095] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: inputting the gait data of the organism into a pose estimation network model to obtain the action pose of the organism.

[0096] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: obtaining a gait diagram of the organism based on the organism's movement posture;

[0097] The gait indices of the organism are determined based on the gait diagram, and the gait behavior of the organism is analyzed based on the gait indices.

[0098] In one embodiment, the vital signs of the organism corresponding to the gait behavior are analyzed based on the gait index and gait index samples from multiple periods of the organism.

[0099] In summary, this application provides a non-transitory computer-readable storage medium that first acquires the gait data of an organism to obtain its movement posture, and then analyzes the data based on gait indices generated from the movement posture. This method effectively reduces errors caused by subjective factors during human analysis, ensuring the accuracy of the data when analyzing the organism's gait behavior using gait indices.

[0100] In one embodiment, a computer program product is provided, which, when executed by a processor of a mobile terminal, enables a communication device to perform the following steps: acquiring gait data of an organism; performing posture estimation based on the gait data of the organism to obtain the organism's action posture; obtaining gait indices of the organism based on the organism's action posture; and analyzing the gait behavior of the organism based on the gait indices.

[0101] In one embodiment, the step of estimating the posture based on the gait data of the organism to obtain the movement posture of the organism includes: inputting the gait data of the organism into a posture estimation network model to obtain the movement posture of the organism.

[0102] In one embodiment, obtaining gait indices of the organism based on its movement postures, and analyzing the organism's gait behavior based on the gait indices, includes:

[0103] Based on the organism's movement posture, a gait diagram of the organism is obtained;

[0104] The gait indices of the organism are determined based on the gait diagram, and the gait behavior of the organism is analyzed based on the gait indices.

[0105] In one embodiment, analyzing the gait behavior of the organism based on the gait index includes: analyzing the vital signs of the organism corresponding to the gait behavior based on the gait index and gait index samples from multiple periods of the organism.

[0106] In summary, this product first acquires the organism's gait data to obtain its movement posture, and then analyzes it based on gait indices generated from the movement posture. This method effectively reduces errors caused by subjective factors in human analysis, ensuring the accuracy of the data when analyzing the organism's gait behavior using gait indices.

[0107] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static random access memory (SRAM) and dynamic random access memory (DRAM), etc.

[0108] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method of analyzing gait behavior of a living body, characterized by, The method comprises the following steps: acquiring gait data of a living body; the acquiring gait data of a living body comprises: acquiring video data of the living body by using a high-speed camera arranged on a treadmill, and acquiring the gait data based on key position motion trajectory information of the living body in the video data; performing posture estimation based on the gait data of the living body to obtain a motion posture of the living body; before the posture estimation based on the gait data of the living body, the method further comprises: extracting valid gait data from the gait data of the living body, wherein the valid gait data comprises data with a clear definition, a resolution target position meeting a preset range, etc.; based on the motion posture of the living body, obtaining gait indicators of the living body, and analyzing gait behavior of the living body according to the gait indicators; the posture estimation based on the gait data of the living body to obtain the motion posture of the living body comprises: inputting the gait data of the living body into a posture estimation network model to obtain the motion posture of the living body, wherein the motion posture comprises key point position information and confidence of the living body gait; before obtaining the gait indicators of the living body, extracting a gait starting frame from the motion posture of the living body and eliminating abnormal gait frames, wherein the elimination of the abnormal gait frames is achieved by calculation in a valid gait interval, and the calculation in the valid gait interval comprises: selecting a valid gait interval, screening gait abnormal data, and calculating indicators; the selection of the valid gait interval comprises: calculating a distance weighted by a distance of a toe of the living body from an origin according to a confidence of a predicted position and drawing a distance curve, removing small peaks by multiple smoothing operations, and selecting the valid gait interval according to a gait interval time by recursively finding frames where the peaks are located and eliminating abnormal frames with small peaks; the calculation of the distance of the toe of the living body from the origin is represented by the following formula: wherein, , ) are used to represent the i-th frame toe coordinates, w is used to represent the weight parameter, and coef is used to represent the toe key point confidence. the screening of the gait abnormal data comprises key point position relationship screening and distance screening, wherein, in the distance screening, the average distance of connected key points in the gait data of the living body is calculated according to the distance relationship of the key points, and when the distance of one frame of key points in the gait data is greater than 1.5 times of the average distance or less than 0.3 times of the average distance, it is determined that the distance of the one frame of key points is abnormal data; the calculation of the indicators comprises at least a time required for one step, a distance of one step, maximum and minimum values of knee and condyle joint angles, and step height; the time required for one step is a ratio of a frame number of a complete gait to a frame rate in the gait data of the living body, the step height is a maximum height of a condyle lifted in one step, and the distance of one step is calculated by the following formula: wherein OSCD is used to represent the distance of the step, OSCT is used to represent the time required for the step, V is used to represent the speed of the treadmill, and respectively represent the x-axis coordinates of the toes of the organism at the beginning and end of the gait, and Ratio is used to represent the ratio of video pixels to real distance, is used to represent the angle between the plane of the treadmill and the x-axis.

2. The biological gait analysis method according to claim 1, wherein the method further comprises the following steps of training the posture estimation network model: acquiring gait data samples of a living body; inputting the gait data samples of the living body into an initial posture estimation network model, and training the initial posture estimation network model according to a loss between an output of the initial posture estimation network model and a label sample to obtain the posture estimation network model, wherein the label sample comprises a motion posture of the gait data samples of the living body.

3. The biological gait analysis method according to claim 1, wherein The gait index of the organism is obtained based on the action posture of the organism, and the gait behavior of the organism is analyzed according to the gait index, including: A gait graph of the organism is obtained based on the action posture of the organism; The gait index of the organism is determined according to the gait graph, and the gait behavior of the organism is analyzed according to the gait index.

4. The biological gait analysis method according to claim 3, wherein The gait behavior of the organism is analyzed according to the gait index, including: The signs of the organism corresponding to the gait behavior are analyzed according to the gait index and gait index samples of multiple periods of the organism.

5. The biological gait analysis method according to claim 1, wherein After the action posture of the organism is obtained based on the action posture of the organism, the action posture of the organism is also corrected according to the gait relationship of the organism. It includes:

6. A gait behavior analysis device of a living body, characterized by comprising: A gait data acquisition unit is configured to obtain gait data of an organism; The gait data of the organism is obtained by using a high-speed camera arranged on a treadmill to obtain video data of the organism, and the gait data is obtained based on key part motion trajectory information of the organism in the video data; A posture estimation unit is configured to perform posture estimation based on the gait data of the organism to obtain the action posture of the organism; before the posture estimation based on the gait data of the organism, the effective gait data is extracted from the gait data of the organism, wherein the effective gait data includes data with a clarity, resolution, and target position that meet a preset range; A gait index calculation unit is configured to obtain the gait index of the organism based on the action posture of the organism, and analyze the gait behavior of the organism according to the gait index; The action posture of the organism is obtained by inputting the gait data of the organism into a posture estimation network model, including the position information and confidence of the gait key points of the organism; Before the gait index of the organism is obtained, the gait start frame is extracted from the action posture of the organism and the abnormal gait frame is removed, and the removal of the abnormal gait frame is realized by calculation in the effective gait interval, and the calculation steps in the effective gait interval include selecting the effective gait interval, screening the gait abnormal data, and calculating the index; The selection process of the effective gait interval includes: calculating the distance of the organism's toes from the origin point and drawing a distance curve based on the confidence of the predicted position, removing small peaks through multiple smoothing operations, and removing abnormal frames with small peak values by recursively finding the frames where the peak values are located, and selecting the effective gait interval according to the gait interval time, and the calculation of the distance of the organism's toes from the origin point is represented by the following formula: ​ wherein, , ) are used to represent the i-th frame toe coordinates, w is used to represent the weight parameter, and coef is used to represent the toe key point confidence. The screening of the gait abnormality data comprises key point position relationship screening and distance screening, wherein in the distance screening, the average distance of connected key points in the organism gait data is calculated according to the distance relationship of the key points, and when the distance of one frame of key points in the gait data is greater than 1.5 times of the average distance or less than 0.3 times of the average distance, it is determined that the distance of the one frame of key points is abnormal data; The measurement index of the index calculation comprises at least time required for one step, distance of one step, maximum and minimum values of knee and condyle joint angle, and step height; the time required for one step is the ratio of the frame number of complete gait in the gait data of the organism to the frame rate, the step height is the maximum height of the condyle lifted in the process of one step, and the distance of one step is calculated by the following formula: wherein OSCD is used to represent the distance of the step, OSCT is used to represent the time required for the step, V is used to represent the speed of the treadmill, and respectively represent the x-axis coordinates of the toes of the organism at the beginning and end of the gait, and Ratio is used to represent the ratio of video pixels to real distance, is used to represent the angle between the plane of the treadmill and the x-axis. 7.A computer readable storage medium having stored thereon a computer program for implementing the gait behavior analysis method of an organism according to any one of claims 1-5.

8. A computer program product, characterised in that, A computer readable storage medium having stored thereon a computer program for implementing the gait behavior analysis method of an organism according to any one of claims 1-5.

Images