Pose prediction model training method and device, equipment and medium

Abstract

The invention discloses a pose prediction model training method and device, equipment and a medium, and relates to the technical field of data processing. The method comprises the steps that the number of river horse individuals in a river horse population and the number of river horse population iterations are obtained; for any river horse individual, executing a fitness value determination step to obtain a current individual fitness value of the river horse individual; determining the current relative speed change rate of the individual river horse according to the current individual fitness value of the individual river horse and the number of times of repeatedly executing the fitness value determination step; according to the current relative speed change rate of the individual river horse, the initial position of the individual river horse is updated, and the updated position of the individual river horse is obtained; updating the initial position of the individual river horse into the updated position of the individual river horse, and repeatedly executing the fitness value determination step until the number of iterations of the river horse population is reached; and taking the pose prediction model corresponding to the optimal current individual fitness value in the latest obtained current individual fitness values of the river horse individuals as a target pose prediction model.

Description

Technical Field

[0001] The present application relates to the technical field of data processing, and more specifically, to a method, apparatus, device, and medium for training a pose prediction model. Background Art

[0002] Currently, in order to alleviate the problem of limited computing power of head-mounted display devices, remote rendering technology is usually adopted. Specifically, a cloud or a network edge server predicts the pose of a head-mounted display device through a pose prediction model, completes image rendering according to the predicted pose, and transmits the rendered image data back to the head-mounted display device for display by the head-mounted display device.

[0003] For a pose prediction model, during the training process of the pose prediction model, its hyperparameter values are determined according to the traditional hippopotamus optimization algorithm. However, in the traditional hippopotamus optimization algorithm, the position update method of hippopotamus individuals is specifically as follows: the hippopotamus individuals are evenly divided into two parts in order, the first part is updated according to the position update method of the first exploration stage, and the second part is updated according to the position update method of the second exploration stage; for the positions of the updated hippopotamus individuals, the position update method of the exploitation stage is continued to be used for updating. However, in this position update method, fixed grouping restricts the policy space of the agent, reduces the ability of the algorithm to jump out of the local optimum, easily leads to the imbalance between exploration and exploitation of the algorithm, and it is difficult to find the optimal solution. That is, it is difficult to find the optimal hyperparameter values when determining the hyperparameter values of the pose prediction model through the traditional hippopotamus optimization algorithm.

[0004] Therefore, a new method for training a pose prediction model is urgently needed to be proposed. Summary of the Invention

[0005] An object of the present application is to provide a new technical solution for training a pose prediction model.

[0006] According to a first aspect of the present application, there is provided a method for training a pose prediction model based on an improved hippopotamus optimization algorithm, including:

[0007] Obtain the number of hippopotamus individuals in a hippopotamus population and the number of iterations of the hippopotamus population;

[0008] For any hippopotamus individual, perform a fitness value determination step to obtain the current individual fitness value of the hippopotamus individual. The fitness value determination step includes: determining a model to be trained according to the initial position of the hippopotamus individual, training the model to be trained according to a training sample set to obtain a pose prediction model, determining the current individual fitness value of the pose prediction model according to the training sample set. The initial position of the hippopotamus individual is used to represent model hyperparameter values. The training sample set includes multiple groups of training samples, and a group of training samples includes a pose as a sample and a pose as a label;

[0009] Determine the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeating the fitness value determination step;

[0010] Update the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual;

[0011] Update the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeat the fitness value determination step until the execution of the fitness value determination step reaches the iteration number of the hippopotamus population;

[0012] Among the currently obtained current individual fitness values of the hippopotamus individual, use the pose prediction model corresponding to the optimal current individual fitness value as the target pose prediction model.

[0013] Optionally, the determining the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeating the fitness value determination step includes:

[0014] When the number of times of repeating the fitness value determination step is 0, determine that the current relative speed change rate of the hippopotamus individual is infinite;

[0015] When the number of times of repeating the fitness value determination step is greater than 0, determine the current optimal fitness value according to the current individual fitness value of each hippopotamus individual;

[0016] Determine the historical optimal fitness value according to the historical fitness value obtained by each hippopotamus individual in the previous execution of the fitness value determination step;

[0017] Obtain the historical individual fitness value obtained by the hippopotamus individual in the previous execution of the fitness value determination step;

[0018] Determine the current relative speed change rate of the hippopotamus individual according to the historical individual fitness value, the current individual fitness value, the historical optimal fitness value, and the current optimal fitness value.

[0019] Optionally, the updating the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual includes:

[0020] When the current relative speed change rate of the hippopotamus individual is greater than the preset change rate, update the initial position of the hippopotamus individual according to the position update method in the first exploration stage of the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual;

[0021] When the current relative speed change rate of the hippopotamus individual is less than or equal to the preset change rate, update the initial position of the hippopotamus individual according to the position update method in the second exploration stage of the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual;

[0022] For any one of the hippopotamus individuals, update the updated intermediate position of the hippopotamus individual according to the position update method in the exploitation stage of the hippopotamus optimization algorithm to obtain the updated position of the hippopotamus individual.

[0023] Optionally, the updating the initial position of the hippopotamus individual according to the position update formula in the first exploration stage of the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual includes:

[0024] Determine the adaptive weight of the hippopotamus individual according to the current individual fitness value of each hippopotamus individual;

[0025] Update the position update formula in the first exploration stage of the hippopotamus optimization algorithm according to the adaptive weight to obtain the updated position update method in the first exploration stage;

[0026] Update the initial position of the hippopotamus individual according to the updated position update method in the first exploration stage, the updated intermediate position of the hippopotamus individual.

[0027] Optionally, before determining the model to be trained according to the initial position of the hippopotamus individual, the method further includes:

[0028] Determine the random mapping value of the hippopotamus individual according to the random mapping;

[0029] Determine the chaotic mapping value of the hippopotamus individual according to the random mapping value and the chaotic mapping;

[0030] Determine the initial position of the hippopotamus individual according to the chaotic mapping value of the hippopotamus individual.

[0031] Optionally, the method further includes:

[0032] Obtain the pose for pose prediction;

[0033] Determine the predicted pose according to the pose for pose prediction and the target pose prediction model.

[0034] Optionally, the obtaining the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population includes:

[0035] Display the set input interface;

[0036] Receive the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population input by the set input interface.

[0037] According to the second aspect of the present application, there is provided a pose prediction model training device based on an improved hippopotamus optimization algorithm, including:

[0038] An acquisition module, configured to acquire the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population;

[0039] An execution module, configured to, for any hippopotamus individual, execute a fitness value determination step to obtain the current individual fitness value of the hippopotamus individual. The fitness value determination step includes: determining a model to be trained according to the initial position of the hippopotamus individual, training the model to be trained according to a training sample set to obtain a pose prediction model, and determining the current individual fitness value of the pose prediction model according to the training sample set. The initial position of the hippopotamus individual is used to represent model hyperparameter values. The training sample set includes multiple groups of training samples, and one group of the training samples includes a pose as a sample and a pose as a label;

[0040] A first determination module, configured to determine the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeating the execution of the fitness value determination step;

[0041] An update module, configured to update the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual;

[0042] An iteration module, configured to update the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeat the execution of the fitness value determination step until the execution of the fitness value determination step reaches the number of iterations of the hippopotamus population;

[0043] A second determination module, configured to use the pose prediction model corresponding to the optimal current individual fitness value among the currently obtained current individual fitness values of the hippopotamus individual as the target pose prediction model.

[0044] According to the third aspect of the present application, there is provided an electronic device, and the electronic device includes the device described in the second aspect;

[0045] Alternatively, the electronic device includes a memory and a processor. The memory is configured to store computer instructions, and the processor is configured to call the computer instructions from the memory to execute the method described in any one of the first aspect.

[0046] According to the fourth aspect of the present application, there is provided a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in any one of the first aspect is implemented.

[0047] The present application provides a method for training a pose prediction model based on an improved hippopotamus optimization algorithm, including: obtaining the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population; for any hippopotamus individual, performing a fitness value determination step to obtain the current individual fitness value of the hippopotamus individual. The fitness value determination step includes: determining a model to be trained according to the initial position of the hippopotamus individual, training the model to be trained according to the training sample set to obtain a pose prediction model, determining the current individual fitness value of the pose prediction model according to the training sample set. The initial position of the hippopotamus individual is used to represent the model hyperparameter value. The training sample set includes multiple groups of training samples, and a group of training samples includes a pose as a sample and a pose as a label; determining the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeatedly performing the fitness value determination step; updating the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual; updating the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeatedly performing the fitness value determination step until the fitness value determination step is performed to reach the number of iterations of the hippopotamus population; using the pose prediction model corresponding to the optimal current individual fitness value among the currently obtained current individual fitness values of the hippopotamus individuals as the target pose prediction model. This method improves the traditional hippopotamus optimization algorithm and trains the pose prediction model based on the improved hippopotamus optimization algorithm, thereby avoiding the problem that it is difficult to find the optimal hyperparameter value when using the traditional hippopotamus optimization algorithm to determine the model hyperparameter value.

[0048] Other features and advantages of the present application will become clear from the following detailed description of exemplary embodiments of the present application with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The drawings incorporated in and constituting a part of this specification illustrate embodiments of the present application and, together with the description, serve to explain the principles of the present application.

[0050] Figure 1 is a block diagram of the hardware configuration of an electronic device for implementing a method for training a pose prediction model based on an improved hippopotamus optimization algorithm according to an embodiment of the present application Figure 1 ;

[0051] Figure 2 is a schematic flowchart of a method for implementing a method for training a pose prediction model based on an improved hippopotamus optimization algorithm according to an embodiment of the present application;

[0052] Figure 3 is a schematic diagram of the architecture of a model to be trained determined according to the initial position of a hippopotamus individual according to an embodiment of the present application;

[0053] Figure 4It is a schematic diagram of the distribution structure of random mapping and chaotic mapping provided according to an embodiment of the present application;

[0054] Figure 5 It is a schematic diagram of the values of the tanh function when a takes different values provided according to an embodiment of the present application;

[0055] Figure 6 It is a schematic diagram of the structure of a device for training a pose prediction model based on an improved hippopotamus optimization algorithm provided according to an embodiment of the present application;

[0056] Figure 7 It is a block diagram of the hardware configuration of an electronic device for implementing a method for training a pose prediction model based on an improved hippopotamus optimization algorithm provided according to an embodiment of the present application Figure 2 . Detailed implementation manners

[0057] Now, various exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings. It should be noted that: Unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions and values set forth in these embodiments do not limit the scope of the present application.

[0058] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present application, its application, or its use.

[0059] Techniques, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered as part of the specification.

[0060] In all the examples shown and discussed here, any specific values should be construed as merely exemplary and not as limitations. Therefore, other examples of the exemplary embodiments may have different values.

[0061] It should be noted that: Similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0062] Figure 1 It is a block diagram of the hardware configuration of an electronic device for implementing a method for training a pose prediction model based on an improved hippopotamus optimization algorithm provided according to an embodiment of the present application Figure 1 .

[0063] The electronic device 1000 can be a terminal or a server. Further, the terminal can be a head-mounted device (such as an AR device, an MR device, and a VR device), a portable computer, a tablet computer, a handheld computer, etc. The server can be a cloud server, etc.

[0064] The electronic device 1000 may include a processor 1100, a memory 1200, an interface device 1300, a communication device 1400, a display device 1500, an input device 1600, a speaker 1700, a microphone 1800, and so on. Among them, the processor 1100 may be a central processing unit CPU, a microcontroller unit MCU, etc. The memory 1200 includes, for example, a ROM (read-only memory), a RAM (random access memory), a non-volatile memory such as a hard disk, etc. The interface device 1300 includes, for example, a USB interface, a headphone interface, etc. The communication device 1400 can perform wired or wireless communication, for example. The display device 1500 is, for example, a liquid crystal display screen, a touch display screen, etc. The input device 1600 may include, for example, a touch screen, a keyboard, etc. The user can input / output voice information through the speaker 1700 and the microphone 1800.

[0065] Although Figure 1 multiple devices are shown for the electronic device 1000 in

[0066] In the embodiments applied to the present application, the memory 1200 of the electronic device 1000 is used to store instructions for controlling the processor 1100 to execute the pose prediction model training method based on the improved hippopotamus optimization algorithm provided by the embodiments of the present application.

[0067] In the above description, those skilled in the art can design instructions according to the solutions disclosed in the present application. How the instructions control the processor to operate is well known in the art, so it will not be described in detail here.

[0068] The present application provides a pose prediction model training method based on an improved hippopotamus optimization algorithm, and this method is applied to an electronic device as shown in Figure 1 As shown in Figure 2 As shown, the pose prediction model training method based on the improved hippopotamus optimization algorithm provided by the present application includes the following steps S2100 to step S2600.

[0069] Step S2100, obtain the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population.

[0070] Among them, the number of hippopotamus individuals is the number of candidate solutions in the search space. In this embodiment, the number of hippopotamus individuals is the number of groups of hyperparameters.

[0071] The number of iterations of the hippopotamus population is the number of times the positions of the hippopotamus individuals are updated, and is used as the convergence condition for updating the positions of the hippopotamus individuals.

[0072] In one embodiment of the present application, the hippopotamus individual data and the hippopotamus population iteration number in the hippopotamus population are respectively a default empirical value.

[0073] In another embodiment of the present application, in order to allow the user to input the number of hippopotamus individuals and the hippopotamus population iteration number that meet their own needs, the electronic device provides a setting input interface. The user can input the number of hippopotamus individuals and the hippopotamus population iteration number that meet their own needs through this setting input interface. In this regard, the above step S2100 is specifically implemented through the following steps S2110 and S2111.

[0074] Step S2110, display the setting input interface.

[0075] Step S2111, receive the number of hippopotamus individuals in the hippopotamus population and the hippopotamus population iteration number input through the setting input interface.

[0076] In one example, the setting input interface may specifically include a first interface and a second interface. Among them, the first interface is used for the user to input the number of hippopotamus individuals that meet their own needs, and the second interface is used for the user to input the hippopotamus population iteration number that meet their own needs.

[0077] Step S2200, for any hippopotamus individual, execute the fitness value determination step to obtain the current individual fitness value of the hippopotamus individual.

[0078] Among them, the fitness value determination step includes the following steps S2201 to S2203.

[0079] Step S2201, determine the model to be trained according to the initial position of the hippopotamus individual.

[0080] Among them, the initial position of the hippopotamus individual is used to represent the model hyperparameter value. Based on this, the model can be constructed according to the model hyperparameter value represented by the initial position of the hippopotamus individual to obtain the model to be trained. The initial position of the hippopotamus individual is usually represented by a vector.

[0081] It should be noted that when the above step S2201 is executed for the first time, the model hyperparameter value represented by the initial position of the hippopotamus individual is specifically the hyperparameter value in the initial state of the model, that is, an initial hyperparameter value.

[0082] In one example, the initial position of the hippopotamus individual can be: a vector with an input layer including 5 neurons, hidden layers being hidden layer 1 and hidden layer 2, hidden layer 1 including 5 neurons, hidden layer 2 including 5 neurons, and an output layer including 5 neurons. On this basis, such as Figure 3As shown in the figure, the architecture of the model to be trained determined based on the initial position of the hippopotamus individual is as follows: an input layer including 5 neurons, a hidden layer 1 including 5 neurons, a hidden layer 2 including 5 neurons, and an output layer including 5 neurons are connected in sequence. Among them, the input of one neuron is the output of the neurons in the previous layer.

[0083] In an embodiment of the present application, the initial position of the hippopotamus individual can be determined by means of random mapping as in the traditional hippopotamus optimization algorithm.

[0084] Due to the large randomness of random mapping, the random values determined by random mapping are unevenly distributed. This further leads to the problem of unstable optimization convergence speed in the subsequent traditional hippopotamus optimization algorithm. Therefore, in an embodiment of the present application, before the above step S2201, the pose prediction model training method based on the improved hippopotamus optimization algorithm further includes the following steps S2201-1 to step S2201-3 to implement.

[0085] Step S2201-1: Determine the random mapping value of the hippopotamus individual according to random mapping.

[0086] Step S2201-2: Determine the chaotic mapping value of the hippopotamus individual according to the random mapping value and chaotic mapping.

[0087] In this embodiment, for any hippopotamus individual, first determine the random mapping value corresponding to the hippopotamus individual through random mapping. Further, through chaotic mapping, map the chaotic mapping value corresponding to the random mapping value.

[0088] In an embodiment of the present application, the chaotic mapping can specifically be Cubic chaotic mapping or Tent chaotic mapping.

[0089] Taking the chaotic mapping in step S2201-2 as Tent mapping as an example, the chaotic mapping can be represented by the following formula one.

[0090]

[0091] Among them, represents the random mapping value of the i-th hippopotamus individual, represents the chaotic mapping value of the i-th hippopotamus individual, μ represents the control parameter in the Tent mapping, and 0 < μ < 1. When μ = 1 / 2, the result of the Tent mapping is uniformly distributed.

[0092] Step S2201-3: Determine the initial position of the hippopotamus individual according to the chaotic mapping value of the hippopotamus individual.

[0093] In the traditional hippopotamus optimization algorithm, the initial position of the hippopotamus individual is determined according to the following formula two.

[0094] X(:, i) = X min (i) + r1 × (X max (i) - X min (i)) (Equation 2)

[0095] Among them, X(:, i) represents the initial position of the i-th hippopotamus individual, r1 is the random mapping value of the i-th hippopotamus individual, X max (i) represents the upper limit value of the i-th hippopotamus individual, X min (i) represents the lower limit value of the i-th hippopotamus individual, X max (i) and X min (i) can be set according to experience.

[0096] In this embodiment, the specific implementation of the above step S2201-3 is to replace r1 in the above Equation 2 with the chaotic mapping value of the i-th hippopotamus individual.

[0097] Taking the chaotic mapping as the Tent mapping as an example, as Figure 4 shown, the chaotic mapping values obtained based on the chaotic mapping are more evenly distributed compared to the random mapping values obtained by random mapping. On this basis, the initial position distribution of the hippopotamus individuals obtained through the above steps S2201-1 to S2201-3 is more uniform, which can increase the exploration difficulty and reduce the convergence speed.

[0098] Step S2202, training the model to be trained according to the training sample set to obtain a pose prediction model.

[0099] Among them, the training sample set includes multiple groups of training samples, and a group of training samples includes a pose as a sample and a pose as a label.

[0100] In this embodiment, according to the traditional model training method, the model to be trained can be trained according to the training sample set to obtain a pose prediction model.

[0101] Step S2203, determining the current individual fitness value of the pose prediction model according to the training sample set.

[0102] In this embodiment, for a hippopotamus individual, for each training sample, the pose as a sample in the training sample is input into the trained pose prediction model to obtain the predicted pose corresponding to the training sample; according to the preset fitness function, the predicted pose, and the pose as a label in the training sample, the fitness value corresponding to the training sample is determined; the average or median of the fitness values of all training samples is used as the current individual fitness value of the hippopotamus individual. Among them, the preset fitness function can be a function for calculating the Euler distance.

[0103] Step S2300: Determine the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeatedly executing the fitness value determination step.

[0104] In an embodiment of the present application, the above step S2300 is specifically implemented through the following steps S2310 to S2350.

[0105] Step S2310: When the number of times of repeatedly executing the fitness value determination step is 0, determine that the current relative speed change rate of the hippopotamus individual is infinite.

[0106] In this embodiment, when the above step S2200 is executed for the first time, the number of times of repeatedly executing the fitness value determination step is 0. When the above step S2200 is executed for the second time, the number of times of repeatedly executing the fitness value determination step is 1. And so on, the number of times of repeatedly executing the fitness value determination step can be determined.

[0107] When the number of times of repeatedly executing the fitness value determination step is 0, the current relative speed change rate of the hippopotamus individual is determined based on the above step S2310. When the number of times of repeatedly executing the fitness value determination step is greater than 0, the current relative change rate of the hippopotamus individual is determined based on the following steps S2320 to S2350.

[0108] Step S2320: When the number of times of repeatedly executing the fitness value determination step is greater than 0, determine the current optimal fitness value according to the current individual fitness value of each hippopotamus individual.

[0109] In this embodiment, the current optimal fitness value is selected from the current individual fitness values of all hippopotamus individuals obtained after the above step S2200 is currently executed for any hippopotamus individual.

[0110] Step S2330: Determine the historical optimal fitness value according to the historical fitness value obtained by each hippopotamus individual in the previous execution of the fitness value determination step.

[0111] The historical optimal fitness value is selected from the current individual fitness values of all hippopotamus individuals obtained after the previous execution of the above step S2200 for any hippopotamus individual.

[0112] Step S2340: Obtain the historical individual fitness value obtained by the hippopotamus individual in the previous execution of the fitness value determination step.

[0113] For any hippopotamus individual, the current individual fitness value obtained after the previous execution of the above step S2200 for this hippopotamus individual is determined as the historical individual fitness value of this hippopotamus individual.

[0114] Step S2350: Determine the current relative speed change rate of the hippopotamus individual according to the historical individual fitness value, the current individual fitness value, the historical optimal fitness value, and the current optimal fitness value.

[0115] Based on the above steps S2310 to S2350, in this embodiment, the above step S2300 can be specifically implemented by the following formula three.

[0116]

[0117] Where t represents the current number of times of repeating the above step S2200, and the maximum value of t is the iteration times of the hippopotamus population; represents the current relative speed change rate of the i-th hippopotamus individual; f(·) represents the preset fitness function; B t represents the current optimal fitness value; B t-1 represents the historical optimal fitness value; represents the current individual fitness value of the i-th hippopotamus individual, represents the historical individual fitness value of the i-th hippopotamus individual.

[0118] Based on the above steps S2310 to S2350, the present application provides a method for determining the current relative speed change rate of the hippopotamus individual.

[0119] Step S2400: Update the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual.

[0120] In an embodiment of the present application, the initial position of the hippopotamus individual can be updated according to the specific value of the current relative speed change rate. For example, when the current relative speed change rate represents a large change amplitude, the initial position of the hippopotamus individual is updated by means of a large position adjustment. When the current relative speed change rate represents a small change amplitude, the initial position of the hippopotamus individual is updated by means of a small position adjustment.

[0121] Compared with the traditional hippopotamus optimization algorithm, in the pose prediction model training method based on the improved hippopotamus optimization algorithm provided by the present application, the position is no longer updated in groups according to the order of the hippopotamus individuals as in the traditional hippopotamus algorithm, but depends on the current relative speed change rate of each hippopotamus individual to update the position. On this basis, the problem of difficult to find the optimal hyperparameter value when using the traditional hippopotamus optimization algorithm to determine the model hyperparameter value can be avoided.

[0122] Therefore, the present application improves the traditional hippopotamus optimization algorithm and trains the pose prediction model based on the improved hippopotamus optimization algorithm, so as to avoid the problem that it is difficult to find the optimal hyperparameter value when using the traditional hippopotamus optimization algorithm to determine the model hyperparameter value.

[0123] Step S2500: Update the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeat the fitness value determination step until the fitness determination step is executed for the number of iterations of the hippopotamus population.

[0124] In this embodiment, after updating the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, repeat the above step S2200 until the number of times of repeating the above step S2200 reaches the number of iterations of the hippopotamus population.

[0125] After updating the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, at this time, when repeating the above step S2200, after step S2201, the hyperparameter value of the model to be trained is specifically the hyperparameter value represented by the updated position.

[0126] In one example, if the updated position is a vector indicating that the input layer includes 5 neurons, the hidden layers are hidden layer 1 and hidden layer 2 respectively, hidden layer 1 includes 6 neurons, hidden layer 2 includes 6 neurons, and the output layer includes 5 neurons. On this basis, the architecture of the model to be trained determined based on the initial position updated according to the updated position of the hippopotamus individual is: an input layer including 5 neurons, a hidden layer 1 including 6 neurons, a hidden layer 2 including 6 neurons, and an output layer including 5 neurons are connected in sequence, where the input of one neuron is the output of the neurons in the previous layer.

[0127] It should be noted that the above steps S2200 to S2500 are executed for any hippopotamus individual.

[0128] Step S2600: Use the pose prediction model corresponding to the optimal current individual fitness value among the currently obtained current individual fitness values of the hippopotamus individuals as the target pose prediction model.

[0129] In this embodiment, for each hippopotamus individual, after repeating the above step S2200 for the number of iterations of the hippopotamus population, the latest current individual fitness value is obtained. Further, use the pose prediction model corresponding to the optimal current individual fitness value among the latest current individual fitness values of all hippopotamus individuals as the target pose prediction model. It can be understood that the target pose prediction model is the pose prediction model when the improved hippopotamus optimization algorithm converges. Therefore, accurate pose prediction can be achieved based on the target pose prediction model.

[0130] In one embodiment of the present application, when the preset fitness function used to determine the current individual fitness value is a function for calculating the Euler distance, the minimum value in the current individual fitness value is used as the optimal current individual fitness value.

[0131] The present application provides a method for training a pose prediction model based on an improved hippopotamus optimization algorithm, including: obtaining the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population; for any hippopotamus individual, performing a fitness value determination step to obtain the current individual fitness value of the hippopotamus individual. The fitness value determination step includes: determining a model to be trained according to the initial position of the hippopotamus individual, training the model to be trained according to the training sample set to obtain a pose prediction model, determining the current individual fitness value of the pose prediction model according to the training sample set. The initial position of the hippopotamus individual is used to represent the model hyperparameter value. The training sample set includes multiple groups of training samples. One group of training samples includes a pose as a sample and a pose as a label; determining the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeatedly performing the fitness value determination step; updating the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual; updating the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeatedly performing the fitness value determination step until the number of iterations of the hippopotamus population is reached when performing the fitness value determination step; using the pose prediction model corresponding to the optimal current individual fitness value among the currently obtained current individual fitness values of the hippopotamus individual as the target pose prediction model. This method improves the traditional hippopotamus optimization algorithm and trains the pose prediction model based on the improved hippopotamus optimization algorithm, thereby avoiding the problem that it is difficult to find the optimal hyperparameter value when using the traditional hippopotamus optimization algorithm to determine the model hyperparameter value.

[0132] In one embodiment of the present application, the method for training a pose prediction model based on the improved hippopotamus optimization algorithm provided by the present application further includes the following steps S2700 and S2800.

[0133] Step S2700: Obtain the pose for pose prediction.

[0134] Step S2800: Determine the predicted pose according to the pose for pose prediction and the target pose prediction model.

[0135] In this embodiment, based on the above steps S2100 to S2600, a target pose prediction model for realizing accurate pose prediction can be obtained. On this basis, when the pose for pose prediction is input into the target pose prediction model, the target pose prediction model can output the predicted pose.

[0136] In one embodiment of the present application, the above step S2400 is specifically implemented through the following steps S2410 to S2430.

[0137] Step S2410, when the current relative speed change rate of the hippopotamus individual is greater than the preset change rate, update the initial position of the hippopotamus individual according to the position update method in the first exploration stage of the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual.

[0138] In this embodiment, the preset change rate Rate1 represents the critical value of the current relative speed change rate in the first exploration stage and the second exploration stage, and the preset change rate can be set according to experience. In one example, the value of the preset change rate can be 1.

[0139] Moreover, the updated intermediate position of the hippopotamus individual is an intermediate quantity in a population iteration process and needs to be continuously updated through the following step S2430.

[0140] The traditional hippopotamus optimization algorithm includes two exploration stages, namely exploration stage 1 and exploration stage 2. In this embodiment, exploration stage 1 is denoted as the first exploration stage, and exploration stage 2 is denoted as the second exploration stage.

[0141] In one embodiment of the present application, the position update method in the first exploration stage of the traditional hippopotamus optimization algorithm is specifically as shown in the following formula four. That is, when the current relative speed change rate Rate is greater than the preset change rate Rate1, the initial position of the hippopotamus individual is updated through the following formula four.

[0142] X_P1(i,:)=X(i,:)+r2×(D hippo -I1X(i,:)) (Formula Four)

[0143] Wherein, X_P1(i,:) represents the updated intermediate position of the i-th hippopotamus individual when the current relative speed change rate of the hippopotamus individual is greater than the preset change rate; X(i,:) represents the current initial position to be updated of the i-th hippopotamus individual, r2 represents a random number in the interval [0,1]; D hippo represents the initial position of the current optimal hippopotamus individual, that is, the initial position of the hippopotamus individual corresponding to the currently obtained optimal current individual fitness value; I1 represents a random number in the interval [1,2].

[0144] In one embodiment of the present application, the above step S2410 can also be specifically implemented through the following steps S2411 to S2413.

[0145] Step S2411, determine the adaptive weight of the hippopotamus individual according to the current individual fitness value of each hippopotamus individual.

[0146] In one embodiment of the present application, the adaptive weight of any hippopotamus individual can be specifically determined by the following formula five.

[0147]

[0148] Where f represents the current individual fitness value of the hippopotamus individual; f avg represents the average value of the current individual fitness values of all hippopotamus individuals; f min represents the minimum value of the current individual fitness values of all hippopotamus individuals; f max represents the maximum value of the current individual fitness values of all hippopotamus individuals; ω min represents the minimum weight value; ω max represents the maximum weight value; a represents an adjustable parameter; ω represents the adaptive weight of the hippopotamus individual.

[0149] It should be noted that when the value of a is different, the value range of tanh is different. For example, in this embodiment, when a takes different values, the values of the tanh function are as Figure 5 shown. Based on Figure 5 it can be known that the larger the value of a, the slower the change speed of tanh. Therefore, in this embodiment, the value rule of a is when taking the maximum value, tanh converges, that is, tanh approaches 1.

[0150] Based on the above formula five, it can be realized that when the adaptive weight is relatively large, the global search ability of the algorithm is relatively strong, and when the adaptive weight is relatively small, the algorithm can perform fine search around the optimal solution to accelerate the convergence speed. In this embodiment, when the current relative speed change rate of the hippopotamus individual is greater than the preset change rate, the adaptive weight is determined according to the current individual fitness value of the hippopotamus individual to adjust the global search and local exploration capabilities. If the current individual fitness value of the hippopotamus individual is lower than the average value of the current individual fitness values of all hippopotamus individuals, the adaptive fitness weight value increases to enhance the global exploration ability. If the current individual fitness value of the hippopotamus individual is higher than the average value of the current individual fitness values of all hippopotamus individuals, the adaptive fitness weight decreases to enhance the local exploration ability.

[0151] Step S2412, according to the adaptive weight, update the position update formula in the first exploration stage of the hippopotamus optimization algorithm to obtain the updated position update formula in the first exploration stage.

[0152] In this embodiment, the above step S2412 can be specifically implemented by the following formula six.

[0153] X_P1(i,:)=ω(t)×X(i,:)+r2×(D hippo-I1X(i,:)) (Formula VI)

[0154] Where, w(t) represents the adaptive weight when repeating the above step S2200 for the current number of times.

[0155] Step S2413: Update the initial position of the hippopotamus individual and the updated intermediate position of the hippopotamus individual according to the updated first exploration stage position update formula.

[0156] Step S2420: When the current relative speed change rate of the hippopotamus individual is less than or equal to the preset change rate, update the initial position of the hippopotamus individual according to the second exploration stage position update formula in the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual.

[0157] In this embodiment, the second exploration stage position update formula in the hippopotamus optimization algorithm is shown as the following Formula VII. That is, when the current relative speed change rate Rate is less than or equal to the preset change rate Rate1, the initial position of the hippopotamus individual is updated by the following Formula VII.

[0158]

[0159] Where, X_P2(i,:) represents the updated intermediate position of the i-th hippopotamus individual when the current relative speed change rate of the hippopotamus individual is less than or equal to the preset change rate; r3 represents a random number in the interval [-1, 1]; b, c, d, and g are all uniformly distributed random numbers, the value range of b is [2, 4], the value range of c is [1, 1.5], the value range of d is [2, 3], and the value range of g is [2, 4]; Predator j represents the position of the predator in the search space, which is a random position, and the specific value-taking method can refer to the traditional hippopotamus optimization algorithm; represents the factor for the hippopotamus to take defensive behavior to protect itself from predator attacks, and the specific value-taking method can refer to the traditional hippopotamus optimization algorithm; F i represents the objective function value, and the specific value-taking method can refer to the traditional hippopotamus optimization algorithm; represents a random vector with Levy distribution, which is used to describe the position mutation of the predator when attacking the hippopotamus, and the specific value-taking method can refer to the traditional hippopotamus optimization algorithm.

[0160] Step S2430: For any hippopotamus individual, update the updated intermediate position of the hippopotamus individual according to the exploitation stage position update formula in the hippopotamus optimization algorithm to obtain the updated position of the hippopotamus individual.

[0161] In the development stage of the Hippopotamus Optimization Algorithm, the position update formula is as shown in Formula VIII below. That is, based on the above steps S2310 and S2320, for the updated intermediate position of each hippopotamus individual, it is further updated through Formula VIII below.

[0162]

[0163] Among them, X_P3(i,:) represents the updated position of the i-th hippopotamus individual; r4 represents a random number within the interval [0,1]; r5 represents a random number conforming to the normal distribution.

[0164] It should be noted that the above steps S2410 to S2430 belong to the complete update of the position of the hippopotamus individual, that is, it belongs to a complete iteration of the hippopotamus population.

[0165] The present application also provides a pose prediction model training device 600 based on an improved Hippopotamus Optimization Algorithm, as Figure 6 shown, including:

[0166] An acquisition module 610, configured to acquire the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population;

[0167] An execution module 620, configured to, for any hippopotamus individual, execute a fitness value determination step to obtain the current individual fitness value of the hippopotamus individual. The fitness value determination step includes: determining a model to be trained according to the initial position of the hippopotamus individual, training the model to be trained according to a training sample set to obtain a pose prediction model, determining the current individual fitness value of the pose prediction model according to the training sample set. The initial position of the hippopotamus individual is used to represent the model hyperparameter value, and the training sample set includes multiple groups of training samples. One group of the training samples includes a pose as a sample and a pose as a label;

[0168] A first determination module 630, configured to determine the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times of repeating the execution of the fitness value determination step;

[0169] An update module 640, configured to update the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual;

[0170] An iteration module 650, configured to update the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeat the execution of the fitness value determination step until the execution of the fitness value determination step reaches the number of iterations of the hippopotamus population;

[0171] The second determination module 660 is configured to use the pose prediction model corresponding to the optimal current individual fitness value among the currently obtained current individual fitness values of the hippopotamus individuals as the target pose prediction model.

[0172] In an embodiment of the present application, the first determination module 630 is specifically configured to:

[0173] When the number of times of repeating the execution of the fitness value determination step is 0, determine that the current relative speed change rate of the hippopotamus individual is infinite;

[0174] When the number of times of repeating the execution of the fitness value determination step is greater than 0, determine the current optimal fitness value according to the current individual fitness value of each hippopotamus individual;

[0175] Determine the historical optimal fitness value according to the historical fitness value obtained by each hippopotamus individual in the previous execution of the fitness value determination step;

[0176] Obtain the historical individual fitness value obtained by the hippopotamus individual in the previous execution of the fitness value determination step;

[0177] Determine the current relative speed change rate of the hippopotamus individual according to the historical individual fitness value, the current individual fitness value, the historical optimal fitness value, and the current optimal fitness value.

[0178] In an embodiment of the present application, the update module 640 is specifically configured to:

[0179] When the current relative speed change rate of the hippopotamus individual is greater than the preset change rate, update the initial position of the hippopotamus individual according to the position update method in the first exploration stage of the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual;

[0180] When the current relative speed change rate of the hippopotamus individual is less than or equal to the preset change rate, update the initial position of the hippopotamus individual according to the position update method in the second exploration stage of the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual;

[0181] For any hippopotamus individual, update the updated intermediate position of the hippopotamus individual according to the position update method in the exploitation stage of the hippopotamus optimization algorithm to obtain the updated position of the hippopotamus individual.

[0182] In an embodiment of the present application, the update module 640 is specifically configured to: determine the adaptive weight of each hippopotamus individual according to the current individual fitness value of each hippopotamus individual;

[0183] Update the position update formula in the first exploration stage of the hippopotamus optimization algorithm according to the adaptive weight to obtain the updated position update method in the first exploration stage;

[0184] Update the initial position of the hippopotamus individual and the updated intermediate position of the hippopotamus individual according to the updated position update method in the first exploration stage.

[0185] In an embodiment of the present application, the pose prediction model training device 600 based on the improved hippopotamus optimization algorithm provided by the present application further includes:

[0186] A third determination module, configured to determine the random mapping value of the hippopotamus individual according to the random mapping;

[0187] Determine the chaotic mapping value of the hippopotamus individual according to the random mapping value and the chaotic mapping;

[0188] Determine the initial position of the hippopotamus individual according to the chaotic mapping value of the hippopotamus individual.

[0189] In an embodiment of the present application, the pose prediction model training device 600 based on the improved hippopotamus optimization algorithm provided by the present application further includes:

[0190] A fourth determination module, configured to obtain the pose for pose prediction;

[0191] Determine the predicted pose according to the pose for pose prediction and the target pose prediction model.

[0192] In an embodiment of the present application, the acquisition module 610 is specifically configured to display a set input interface;

[0193] Receive the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population input by the set input interface.

[0194] The present application also provides an electronic device 700, and the electronic device 700 includes any one of the pose prediction model training devices 600 based on the improved hippopotamus optimization algorithm provided in the above device embodiment;

[0195] Or, as Figure 7 shown, the electronic device 700 includes a memory 710 and a processor 720, the memory 710 is used to store computer instructions, and the processor 720 is used to call the computer instructions from the memory 710 to execute any one of the pose prediction model training methods based on the improved hippopotamus optimization algorithm provided in the above method embodiment.

[0196] The present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements any one of the pose prediction model training methods based on the improved hippopotamus optimization algorithm provided by the above method embodiments.

[0197] The present application may be a system, a method, and / or a computer program product. The computer program product may include a computer-readable storage medium having thereon computer-readable program instructions for causing a processor to implement various aspects of the present application.

[0198] A computer-readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disc (DVD), a memory stick, a floppy disk, a mechanically encoded device such as a punch card or raised structures in a groove having instructions stored thereon, and any suitable combination of the foregoing. The computer-readable storage medium as used herein is not construed as an instantaneous signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated through a waveguide or other transmission medium (e.g., an optical pulse through an optical fiber cable), or an electrical signal transmitted through a wire.

[0199] The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to various computing / processing devices, or downloaded to an external computer or external storage device through a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network may include a copper transmission cable, an optical fiber transmission, a wireless transmission, a router, a firewall, a switch, a gateway computer, and / or an edge server. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing / processing device.

[0200] The computer program instructions for performing the operations of this application may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine - related instructions, microcode, firmware instructions, state - setting data, or source code or object code written in any combination of one or more programming languages, including object - oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer - readable program instructions may be executed entirely on the user's computer, partially on the user's computer, executed as a stand - alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider). In some embodiments, by using the state information of the computer - readable program instructions to customize an electronic circuit, such as a programmable logic circuit, a field - programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer - readable program instructions to implement various aspects of this application.

[0201] Aspects of the present application are described herein with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer - readable program instructions.

[0202] These computer - readable program instructions can be provided to a processor of a general - purpose computer, a special - purpose computer, or other programmable data - processing apparatus to produce a machine such that the instructions, when executed by the processor of the computer or other programmable data - processing apparatus, create a means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer - readable program instructions can also be stored in a computer - readable storage medium, and these instructions cause a computer, a programmable data - processing apparatus, and / or other devices to work in a particular manner. Thus, the computer - readable medium storing the instructions includes a manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0203] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device, causing a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process such that the instructions executed on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in one or more boxes of the flowchart and / or block diagram.

[0204] The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of an instruction, which contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two consecutive blocks may in fact be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block of the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or acts, or by a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that implementation by hardware, implementation by software, and implementation by a combination of software and hardware are equivalent.

[0205] The embodiments of the present application have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, the practical application, or the technical improvement of the technology in the market, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein. The scope of the present application is defined by the appended claims.

Claims

1. A posture prediction model training method based on an improved Hippo optimization algorithm, characterized in that: include: Get the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population; For any hippopotamus individual, a fitness value determination step is performed to obtain the current individual fitness value of the hippopotamus individual, wherein the fitness value determination step comprises: determining a model to be trained according to the initial position of the hippopotamus individual, training the model to be trained according to a training sample set to obtain a posture prediction model, and determining the current individual fitness value of the posture prediction model according to the training sample set, wherein the initial position of the hippopotamus individual is used to characterize a model hyperparameter value, wherein the training sample set comprises a plurality of groups of training samples, wherein a group of the training samples comprises a posture as a sample and a posture as a label; Determine the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times the fitness value determination step is repeated; According to the current relative speed change rate of the hippopotamus individual, the initial position of the hippopotamus individual is updated to obtain the updated position of the hippopotamus individual; Updating the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeating the fitness value determination step until the fitness value determination step is executed for the number of iterations of the hippopotamus population; The posture prediction model corresponding to the best current individual fitness value among the latest obtained current individual fitness values ​​of the hippopotamus individuals is used as the target posture prediction model.

2. The method according to claim 1, characterized in that: Determining the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times the fitness value determination step is repeated includes: When the number of times the fitness value determination step is repeated is 0, determining that the current relative speed change rate of the hippopotamus individual is infinite; When the number of times the fitness value determination step is repeated is greater than 0, determining the current optimal fitness value according to the current individual fitness value of each of the hippopotamus individuals; Determine the historical optimal fitness value according to the historical fitness value obtained by performing the fitness value determination step for each hippopotamus individual last time; Obtaining a historical individual fitness value obtained by performing the fitness value determination step for the hippopotamus individual last time; The current relative speed change rate of the hippopotamus individual is determined according to the historical individual fitness value, the current individual fitness value, the historical optimal fitness value and the current optimal fitness value.

3. The method according to claim 1, characterized in that The updating of the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual to obtain the updated position of the hippopotamus individual includes: When the current relative speed change rate of the hippopotamus individual is greater than the preset change rate, the initial position of the hippopotamus individual is updated according to the position update method of the first exploration phase in the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual; When the current relative speed change rate of the hippopotamus individual is less than or equal to the preset change rate, the initial position of the hippopotamus individual is updated according to the position update method of the second exploration phase in the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual; For any of the hippopotamus individuals, the updated intermediate position of the hippopotamus individual is updated according to the position update method in the development phase of the hippopotamus optimization algorithm to obtain the updated position of the hippopotamus individual.

4. The method according to claim 3, characterized in that The method of updating the initial position of the hippopotamus individual according to the position update formula of the first exploration phase in the hippopotamus optimization algorithm to obtain the updated intermediate position of the hippopotamus individual includes: Determine the adaptive weight of each hippopotamus individual according to the current individual fitness value of the hippopotamus individual; According to the adaptive weight, updating the position update formula of the first exploration phase in the Hippo optimization algorithm to obtain an updated position update method of the first exploration phase; According to the updated position updating method of the first exploration phase, the initial position of the hippopotamus individual and the updated intermediate position of the hippopotamus individual are updated.

5. The method according to claim 1, characterized in that Before determining the model to be trained according to the initial position of the hippopotamus individual, the method further comprises: According to the random mapping, determining the random mapping value of the hippopotamus individual; Determining the chaotic mapping value of the hippopotamus individual according to the random mapping value and the chaotic mapping; The initial position of the hippopotamus individual is determined according to the chaotic mapping value of the hippopotamus individual.

6. The method according to claim 1, characterized in that The method further comprises: Get the pose for pose prediction; A predicted pose is determined according to the pose used for pose prediction and the target pose prediction model.

7. The method according to claim 1, characterized in that The method of obtaining the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population includes: Display setting input interface; The number of hippopotamus individuals in the hippopotamus population and the number of hippopotamus population iterations inputted from the setting input interface are received.

8. A posture prediction model training device based on an improved Hippo optimization algorithm, characterized in that: include: An acquisition module is used to obtain the number of hippopotamus individuals in the hippopotamus population and the number of iterations of the hippopotamus population; An execution module is used to execute a fitness value determination step for any hippopotamus individual to obtain a current individual fitness value of the hippopotamus individual, wherein the fitness value determination step comprises: determining a model to be trained according to an initial position of the hippopotamus individual, training the model to be trained according to a training sample set to obtain a posture prediction model, and determining a current individual fitness value of the posture prediction model according to the training sample set, wherein the initial position of the hippopotamus individual is used to characterize a model hyperparameter value, wherein the training sample set comprises a plurality of groups of training samples, wherein a group of the training samples comprises a posture as a sample and a posture as a label; A first determination module is used to determine the current relative speed change rate of the hippopotamus individual according to the current individual fitness value of the hippopotamus individual and the number of times the fitness value determination step is repeated; An updating module, used for updating the initial position of the hippopotamus individual according to the current relative speed change rate of the hippopotamus individual, so as to obtain the updated position of the hippopotamus individual; An iteration module, used for updating the initial position of the hippopotamus individual to the updated position of the hippopotamus individual, and repeating the fitness value determination step until the fitness value determination step is executed to reach the number of iterations of the hippopotamus population; The second determination module is used to use the posture prediction model corresponding to the best current individual fitness value among the latest current individual fitness values ​​of the hippopotamus individuals as the target posture prediction model.

9. An electronic device, characterized in that: The electronic device comprises the apparatus as claimed in claim 8; Alternatively, the electronic device includes a memory and a processor, the memory is used to store computer instructions, and the processor is used to call the computer instructions from the memory to execute the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, which implements the method according to any one of claims 1 to 7 when executed by a processor.

Images