CN111047690A

CN111047690A - Model construction method and apparatus, storage medium, and electronic apparatus

Info

Publication number: CN111047690A
Application number: CN201911344760.3A
Authority: CN
Inventors: 刘思阳
Original assignee: Beijing QIYI Century Science and Technology Co Ltd
Current assignee: Beijing QIYI Century Science and Technology Co Ltd
Priority date: 2019-12-23
Filing date: 2019-12-23
Publication date: 2020-04-21
Anticipated expiration: 2039-12-23
Also published as: CN111047690B

Abstract

The application provides a model construction method and device, a storage medium and an electronic device, wherein the method comprises the following steps: acquiring target model parameters of a three-dimensional object model of a target object, wherein the target model parameters comprise first axis angle information of joint points of the three-dimensional object model and first shape information for controlling the body type of the three-dimensional object model; inputting the first axis angle information and the first shape information into a target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axis angle information and the first shape information are matched with a target object; and under the condition that the target identification result indicates that the first axis angle information and the first shape information are matched with the target object, constructing a three-dimensional object model according to the target model parameters. The method and the device solve the problem that the constructed 3D object model is unreasonable due to the limitation of the object in the related art.

Description

Model construction method and apparatus, storage medium, and electronic apparatus

Technical Field

The present application relates to the field of computers, and in particular, to a model building method and apparatus, a storage medium, and an electronic apparatus.

Background

3D (3 Dimensions) object reconstruction (e.g., human body reconstruction), which is a task in computer vision, is to reconstruct or restore a 3D model of an object's pose from a single picture or video, and can be applied to a variety of application fields, such as avatars, interactive games, and the like. If the accuracy of the reconstruction algorithm is continuously improved, the method can replace the traditional motion capture system.

The 3D object model may be constructed using model parameters, which may be set as desired. However, due to the limitations of the object itself, some model parameters are not realistic, resulting in the problem that the constructed 3D object model is not reasonable.

Disclosure of Invention

The embodiment of the application provides a model construction method and device, a storage medium and an electronic device, so as to at least solve the problem that a constructed 3D object model is unreasonable due to the limitation of an object in the related art.

According to an aspect of an embodiment of the present application, there is provided a model building method including: acquiring target model parameters of a three-dimensional object model of a target object, wherein the target model parameters comprise first axis angle information of joint points of the three-dimensional object model and first shape information for controlling the body type of the three-dimensional object model; inputting the first axis angle information and the first shape information into a target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axis angle information and the first shape information are matched with a target object; and under the condition that the target identification result indicates that the first axis angle information and the first shape information are matched with the target object, constructing a three-dimensional object model according to the target model parameters.

According to another aspect of embodiments of the present application, there is provided a model building apparatus including: a first acquisition unit configured to acquire target model parameters of a three-dimensional object model of a target object, wherein the target model parameters include first axis angle information of joint points of the three-dimensional object model and first shape information for controlling a body type of the three-dimensional object model; a first input unit, configured to input the first axis angle information and the first shape information to a target discriminator to obtain a target discrimination result output by the target discriminator, where the target discrimination result is used to indicate whether the first axis angle information and the first shape information match the target object; and the construction unit is used for constructing the three-dimensional object model according to the target model parameters under the condition that the target identification result indicates that the first axis angle information and the first shape information are matched with the target object.

Optionally, the first input unit includes: the first input module is used for inputting the first axis angle information into the first discriminator to obtain a first discrimination result output by the first discriminator, wherein the first discrimination result is used for indicating whether the first axis angle information is matched with the target object; the second input module is used for inputting the first shape information into the second discriminator to obtain a second discrimination result output by the second discriminator, wherein the second discrimination result is used for indicating whether the first shape information is matched with the target object or not; a third input module, configured to input the first authentication result and the second authentication result to a third authenticator, so as to obtain a target authentication result output by the third authenticator, where the target authenticator includes: a first discriminator, a second discriminator, and a third discriminator.

Optionally, the first input module comprises: the input sub-module is used for respectively inputting a plurality of first axis angle information of a plurality of joint points to a plurality of first sub-discriminators to obtain a plurality of first sub-discrimination results under the condition that the joint points comprise a plurality of joint points, wherein the joint points correspond to the first sub-discriminators one by one, and each first sub-discrimination result is used for indicating whether the first axis angle information of one joint point is matched with a target object or not; a second input sub-module, configured to input the first axis angle information of the joint points to a second sub-discriminator to obtain a second sub-discrimination result output by the second sub-discriminator, where the second sub-discrimination result is used to indicate whether the first axis angle information of the joint points matches the target object, and the first discriminator includes: a plurality of first sub-discriminators and second sub-discriminators, the first discrimination result including: a plurality of first sub-authentication results and second sub-authentication results.

Optionally, the apparatus further comprises: a second obtaining unit, configured to obtain m target training sample sets before inputting the first axis angle information and the first shape information into the target discriminator to obtain a target discrimination result output by the target discriminator, where each target training sample set includes a target positive sample set and a target negative sample set, the target positive sample set includes n target positive samples, the target negative sample set includes n target negative samples, the target positive samples are model parameters including second axis angle information and second shape information that match the target object, the target negative samples are model parameters including third axis angle information and third shape information that do not match the target object, m is a positive integer greater than or equal to 1, and n is a positive integer greater than 1; the training unit is used for respectively training the m first initial discriminators by using m target training sample sets to obtain m candidate discriminators, wherein the m target training sample sets correspond to the m first initial discriminators one by one; and the selecting unit is used for selecting the target discriminator from the m candidate discriminators.

Optionally, the second obtaining unit includes: the first acquisition module is used for acquiring m target positive sample groups; a second obtaining module, configured to repeatedly perform the following steps until m × n target negative samples are obtained: acquiring m initial training sample groups, wherein each initial training sample group comprises a target positive sample group and a random negative sample group, the random negative sample group comprises n random negative samples, and the random negative samples are model parameters containing randomly generated fourth axis angle information and fourth shape information; respectively training m second initial discriminators by using m initial training sample sets to obtain m reference discriminators, wherein the m initial training sample sets correspond to the m second initial discriminators one by one; sequentially using other reference discriminators except the target reference discriminator to discriminate each random negative sample in the target random negative sample group to obtain a sample discrimination result of each random negative sample, wherein the target reference discriminator is a reference discriminator selected from m reference discriminators, the target random negative sample group is a random negative sample group in an initial training sample group corresponding to the target reference discriminator, and the sample discrimination result is used for indicating whether the random negative sample is a negative sample; indicating the sample identification result as a random negative sample of the negative sample, and determining the sample identification result as a target negative sample; and the third acquisition module is used for acquiring m target negative sample groups according to the m multiplied by n target negative samples.

Optionally, the second obtaining module includes: the training submodule is used for training a current initial discriminator in the m second initial discriminators by using a current initial training sample group in the m initial training sample groups to obtain a candidate reference discriminator, wherein the discrimination accuracy of the candidate reference discriminator on the samples in the current initial training sample group is greater than or equal to a first accuracy threshold; and the first determining sub-module is used for determining the candidate reference discriminator as the reference discriminator when the discrimination accuracy of the candidate reference discriminator on other target positive samples is greater than or equal to a second accuracy threshold, wherein the other target positive samples are target positive samples contained in other initial training sample groups except the current initial training sample group in the m initial training sample groups.

Optionally, the second obtaining module includes: the identification submodule is used for identifying the target random negative sample in the target random negative sample group by using a plurality of other reference identifiers to obtain a plurality of sub-sample identification results, wherein each sub-sample identification result is used for indicating whether the target random negative sample identified by one other reference identifier is a negative sample or not; and the second determining submodule is used for determining the sample identification result of the target random negative sample according to the plurality of sub-sample identification results.

According to a further embodiment of the present invention, a computer-readable storage medium is also provided, in which a computer program is stored, wherein the computer program is configured to carry out the steps of any of the above-described method embodiments when executed.

According to yet another embodiment of the present invention, there is also provided an electronic device, including a memory in which a computer program is stored and a processor configured to execute the computer program to perform the steps in any of the above method embodiments.

According to the method and the device, the target model parameters of the three-dimensional object model of the target object are obtained in a mode of judging whether the model parameters are matched with the target object or not according to the shaft angle information and the shape information of the joint points in the model parameters, wherein the target model parameters comprise first shaft angle information of the joint points of the three-dimensional object model and first shape information used for controlling the body type of the three-dimensional object model; inputting the first axis angle information and the first shape information into a target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axis angle information and the first shape information are matched with a target object; under the condition that the target identification result indicates that the first axis angle information and the first shape information are matched with the target object, the three-dimensional object model is constructed according to the target model parameters, and because the discriminator is used for distinguishing the axis angle information (rotation information) and the shape information of the joint points in the model parameters before the model is constructed, whether the model parameters of the three-dimensional object model are matched with the target object (such as a human body) (whether the model parameters are actions capable of being made by the target object) or not can be judged, the technical effect of improving the rationality of the constructed three-dimensional object model is achieved, and the problem that the constructed 3D object model is unreasonable due to the limitation of the object in the related technology is solved.

Drawings

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

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.

FIG. 1 is a block diagram of an alternative server hardware configuration according to an embodiment of the present application;

FIG. 2 is a flow chart of an alternative model building method according to an embodiment of the present application;

FIG. 3 is a schematic diagram of an alternative model construction method according to an embodiment of the present application;

FIG. 4 is a schematic diagram of another alternative model construction method according to an embodiment of the present application;

FIG. 5 is a schematic diagram of yet another alternative model construction method according to an embodiment of the present application;

FIG. 6 is a flow chart of another alternative model building method according to an embodiment of the present application; and the number of the first and second groups,

fig. 7 is a block diagram of an alternative model building apparatus according to an embodiment of the present application.

Detailed Description

The invention will be described in detail hereinafter with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.

It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.

According to an aspect of an embodiment of the present application, there is provided a model construction method. Alternatively, the method may be performed in a server, a user terminal or a similar computing device. Taking an example of an application running on a server, fig. 1 is a block diagram of a hardware structure of an optional server according to an embodiment of the present application. As shown in fig. 1, the server 10 may include one or more processors 102 (only one is shown in fig. 1) (the processor 102 may include, but is not limited to, a processing device such as an MCU (micro controller Unit) or an FPGA (Field Programmable gate array)) and a memory 104 for storing data, and optionally, a transmission device 106 for communication function and an input/output device 108. It will be understood by those skilled in the art that the structure shown in fig. 1 is only an illustration, and is not intended to limit the structure of the server. For example, the server 10 may also include more or fewer components than shown in FIG. 1, or have a different configuration than shown in FIG. 1.

The memory 104 can be used for storing computer programs, for example, software programs and modules of application software, such as computer programs corresponding to the model building method in the embodiment of the present application, and the processor 102 executes various functional applications and data processing by running the computer programs stored in the memory 104, so as to implement the method described above. The memory 104 may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 104 may further include memory located remotely from processor 102, which may be connected to server 10 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission device 106 is used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the server 10. In one example, the transmission device 106 includes a NIC (Network Interface Controller) that can be connected to other Network devices through a base station to communicate with the internet. In one example, the transmission device 106 may be an RF (Radio Frequency) module, which is used for communicating with the internet in a wireless manner.

In this embodiment, a model building method operating on the server is provided, and fig. 2 is a flowchart of an alternative model building method according to an embodiment of the present application, and as shown in fig. 2, the flowchart includes the following steps:

step S202, obtaining target model parameters of a three-dimensional object model of a target object, wherein the target model parameters comprise first axial angle information of joint points of the three-dimensional object model and first shape information for controlling the body type of the three-dimensional object model;

step S204, inputting the first axial angle information and the first shape information into a target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axial angle information and the first shape information are matched with a target object;

in step S206, in a case where the target discrimination result indicates that the first axis angle information and the first shape information match the target object, a three-dimensional object model is constructed according to the target model parameters.

Optionally, the executing subject of the above steps may be a server, a user terminal, and the like, but is not limited thereto, and other apparatuses capable of performing model building may be all used to execute the method in the embodiment of the present application.

Alternatively, the model building method in the embodiment of the present application may be applied to, but not limited to, an integration algorithm in an AR (Augmented Reality) solution at a mobile end, a basic algorithm of an avatar, or a motion capture scheme in animation production and movie production, and the like.

According to the embodiment, whether the model parameters are matched with the target object or not is determined according to the shaft angle information and the shape information of the joint points in the model parameters, and the discriminator is used for distinguishing the shaft angle information and the shape information of the joint points in the model parameters before the model is built, so that whether the model parameters of the three-dimensional object model are matched with the target object or not can be determined, the problem that the built 3D object model is unreasonable due to the limitation of the object in the related technology is solved, and the rationality of the built three-dimensional object model is improved.

The model construction method in the embodiment of the present application is described below with reference to fig. 2.

In step S202, target model parameters of a three-dimensional object model of a target object are acquired, wherein the target model parameters include first axis angle information of joint points of the three-dimensional object model and first shape information for controlling a body type of the three-dimensional object model.

The three-dimensional object model may correspond to a target object. The target object may be a human-shaped object, an animal object, or other object having a joint. The target object may contain one or more joints, and the joint points in the three-dimensional object model may correspond to one or more of the one or more joints.

For example, the human body may have L joints, the number of movable joints is M, and the number of joint points of the 3D humanoid model may be N, where N ≦ M < L.

Model parameters for constructing the three-dimensional object model may include, but are not limited to, at least one of: pose information (e.g., axis angle information of joint points), shape information (for controlling the body type of the three-dimensional object model), and lens information (for controlling the size of the three-dimensional object model).

For example, the Pose information may be represented as a Pose matrix P (24 × 3), where 24 is 24 joint points and each row (1 × 3) represents the axis angle information of one joint point. The Shape information may be represented as a Shape matrix S (1 × 10), where each dimension in the Shape matrix is used to control one body type parameter, such as thickness of arms, waist, legs, height, etc., and the lens information may be represented as Cams (1 × 3), each dimension representing: human body scaling factor s_b(ii) a X-axis displacement o of human body_xDisplacement on the x-axis relative to the origin; human y-axis displacement o_yDisplacement in the y-axis relative to the origin.

To construct a three-dimensional object model of a target object, target model parameters of the three-dimensional object model may first be obtained, which may include: the first axis angle information of the joint points of the three-dimensional object model, the first shape information, may further include: and (4) lens information.

The method for obtaining the target model parameters may be random obtaining (for example, shaft angle information, shape information, and lens information of joint points of a three-dimensional object model are randomly generated), may be received from other devices, and may also be input by a user through an interactive interface, and the method for obtaining the target model parameters may be set as needed, which is not described herein again.

In step S204, the first axis angle information and the first shape information are input to the target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axis angle information and the first shape information match with the target object.

After obtaining the target model parameters, the first axis angle information and the first shape information may be discriminated using a target discriminator to obtain a target discrimination result indicating whether the first axis angle information and the first shape information match the target object. The target authentication result may be an identifier indicating whether the two match, for example, the identifier value is a first value (e.g., 1) indicating that the two match, the identifier value is a second value (e.g., 0) indicating that the two do not match, the target authentication result may be a probability indicating a degree of match, if the probability of match is greater than or equal to a match threshold, the two match, and if the probability of match is less than the match threshold, the two do not match.

The object discriminator may be a discriminator (classifier) whose inputs are the first axis angle information and the first shape information and whose output is the object discrimination result. The target discriminator may be a combination of plural discriminators, and the rotation information and the shape information of the joint point may be discriminated using plural simple discriminators. And fusing through a plurality of simple discriminators, and finally giving a score to the current three-dimensional model parameter (target model parameter).

As an alternative embodiment, inputting the first axis angle information and the first shape information to the object discriminator to obtain the object discrimination result output by the object discriminator includes: inputting the first axis angle information into a first discriminator to obtain a first discrimination result output by the first discriminator, wherein the first discrimination result is used for indicating whether the first axis angle information is matched with the target object; inputting the first shape information into a second discriminator to obtain a second discrimination result output by the second discriminator, wherein the second discrimination result is used for indicating whether the first shape information is matched with the target object; inputting the first authentication result and the second authentication result into a third authenticator to obtain a target authentication result output by the third authenticator, wherein the target authenticator comprises: a first discriminator, a second discriminator, and a third discriminator.

When the object model parameter discrimination is performed using a plurality of discriminators (classifiers), the discriminators can be divided into two layers: the first layer comprising a first discriminator and a second discriminator, the second layer comprising a third discriminator, wherein,

(1) the first discriminator may discriminate the first axis angle information to obtain a first discrimination result indicating whether the first axis angle information matches the target object;

(2) the second discriminator may discriminate the first shape information to obtain a second discrimination result indicating whether the first shape information matches the target object;

(3) the third discriminator may discriminate the first discrimination result and the second discrimination result to obtain a target discrimination result, which is used to synthesize the first discrimination result and the second discrimination result and indicate whether the first axis angle information and the first shape information match the target object.

Through the embodiment, the first axis angle information and the first shape information are identified through the two layers of identifiers, so that the first axis angle information and the first shape information can be comprehensively identified, and the accuracy of the identification result is improved.

The number of the joint points of the three-dimensional object model may be one or more, and in the case where the number of the joint points of the three-dimensional object model is plural, the first axis angle information may be discriminated using a plurality of discriminators.

As an alternative embodiment, inputting the first axis angle information to the first discriminator, and obtaining the first discrimination result output by the first discriminator comprises: under the condition that the joint points comprise a plurality of joint points, respectively inputting a plurality of first axis angle information of the joint points into a plurality of first sub-discriminators to obtain a plurality of first sub-discrimination results, wherein the joint points correspond to the first sub-discriminators one by one, and each first sub-discrimination result is used for indicating whether the first axis angle information of one joint point is matched with a target object or not; inputting a plurality of first axis angle information of a plurality of joint points into a second sub-discriminator to obtain a second sub-discrimination result output by the second sub-discriminator, wherein the second sub-discrimination result is used for indicating whether the plurality of first axis angle information of the plurality of joint points are matched with the target object or not, and the first discriminator comprises: a plurality of first sub-discriminators and second sub-discriminators, the first discrimination result including: a plurality of first sub-authentication results and second sub-authentication results.

In the case that the joint point includes a plurality of joint points, the first axis angle information of each joint point may be respectively identified by using a plurality of first sub-identifiers to obtain a plurality of first sub-identification results, so as to determine whether the first axis angle information of each joint point matches the target object; the first axis angle information of the plurality of joint points as a whole may be discriminated using the second sub-discriminator to obtain a second sub-discrimination result to determine whether the first axis angle information of the plurality of joint points as a whole matches the target object.

The number of joint points included in the three-dimensional object model may be the same as or different from the number of joint points used for identification. For example, the information on the axial angle of the joint point (pelvic bone, Pelvis) as the origin may not be identified.

For example, the inputs to the 3D human model are: a rotation matrix R (24 × 3) representing joint point rotation information, and a shape matrix S (10 × 1) representing a shape of the 3D human body model.

For the joint point rotation matrix, 24 shallow classifiers can be used for identification, wherein the 24 shallow classifiers are composed of two fully connected layers, 23 (without considering the Pelvis node) inputs the axial angle information (1 × 3) of each joint point, 23 1-dimensional outputs are output, the input of one classifier is 72-dimensional (rotation matrix R flattening) or 69-dimensional (rotation matrix R flattening after removing the axial angle information of the Pelvis node), and the output is one-dimensional.

For the shape matrix S, a classifier can be used for discrimination, with the input being a 10-dimensional shape matrix and the output being 1-dimensional.

And then using a total classifier, inputting 25-dimensional vectors formed by splicing the results of the 25 classifiers, and outputting one-dimensional vectors, namely the probability that the model parameters are legal parameters.

By the embodiment, the accuracy of the identification result can be improved by identifying the shaft angle information of the plurality of joint points by using a plurality of classifiers.

It should be noted that whether the axis angle information and/or the shape information matches the target object may be considered as a prediction, and whether the three-dimensional object model corresponding to the axis angle information and/or the shape information is an image that the target object can actually achieve.

The human body model is generally driven by a human body model driving algorithm by adopting a joint point driving mode, namely, the human body model is driven by utilizing the rotation information of the joint point. But the range of oscillation of many joints is limited due to the structure of the human body. And the parameter of the Shape for controlling the height and the weight of the three-dimensional object model is related, such as the big belly and the slender waist are impossible to appear. Although the constraint of the axial angle information and the constraint of the shape information of the joint point can be set manually by experience, the constraints set manually are difficult and are easily influenced by personal experience, so that the constraint setting is not accurate.

The identification model provided in the embodiment of the application can identify whether the parameters of the three-dimensional object model conform to the object structure. Since the constraint of the shaft angle information and the constraint of the shape information are difficult to be artificially defined, the initial discriminator may be trained using a training sample group to obtain the target discriminator.

The initial discriminator may be trained using a training sample set comprising a positive sample set and a negative sample set to obtain the target discriminator. Alternatively, a plurality of training sample groups including a positive sample group and a negative sample group may be used to train a plurality of initial discriminators to obtain a plurality of candidate discriminators, and one candidate discriminator with better performance may be selected from the plurality of candidate discriminators as the target discriminator.

As an alternative embodiment, before the first axis angle information and the first shape information are input to the target discriminator to obtain the target discrimination result output by the target discriminator, m target training sample sets may be obtained; training m first initial discriminators by using m target training sample groups respectively to obtain m candidate discriminators; and selecting a target discriminator from the m candidate discriminators.

M (e.g., 4) target training sample sets may be obtained, where m is a positive integer greater than or equal to 1, and each target training sample set may include: one target positive sample group and one target negative sample group.

The target positive sample group comprises n target positive samples, and the target positive samples are model parameters containing second axis angle information and second shape information matched with the target object; the target negative sample group includes n target negative samples, and the target negative samples are model parameters including third axis angle information and third shape information that do not match the target object. Wherein n is a positive integer greater than 1.

After m target training sample sets are obtained, m first initial discriminators may be trained using the m target training sample sets, respectively, to obtain m candidate discriminators. The process of model training may refer to related technologies, which are not described herein. For a trained candidate discriminator, an accuracy exceeding a predetermined threshold (e.g., 90%) may be obtained on the target positive samples, or the target positive samples and the target negative samples in other training sample sets.

After m candidate discriminators are obtained, one candidate discriminator may be selected from the m candidate discriminators as a target discriminator. The selection mode can be as follows: and randomly selecting, or selecting according to the accuracy rate obtained by identifying the target positive sample or the target negative sample in the training sample group or other training sample groups.

According to the embodiment, the initial discriminators are trained through a plurality of groups of training samples to obtain a plurality of candidate discriminators, so that the target discriminator can be selected from the candidate discriminators, the performance of the obtained target discriminator can be improved, and the discrimination accuracy of the obtained target discriminator can be improved.

The positive samples that conform to the model parameters of the anatomy are easy to obtain, while the negative samples that do not conform to the model parameters of the anatomy are not easy to obtain. The classification problem of lack of negative samples can be solved by training a plurality of models in groups and voting and screening samples.

As an alternative embodiment, in order to obtain m target training sample sets, m target positive sample sets may be obtained first. The target positive sample group may be obtained by acquiring data of the target object by using an acquisition device, or may be obtained by other methods, which is not specifically limited in this embodiment.

For example, 4n positive samples may be taken and randomly divided into four groups of positive samples, each group of n positive samples.

Then, the following steps are repeatedly executed until m × n target negative samples are acquired:

step 1, m initial training sample groups are obtained, wherein each initial training sample group comprises a target positive sample group and a random negative sample group, each random negative sample group comprises n random negative samples, and each random negative sample is a model parameter containing randomly generated fourth axis angle information and fourth shape information.

In addition to the m target positive sample groups, m random negative sample groups may be obtained, each random negative sample group including n random negative samples, and each random negative sample is a model parameter including randomly generated fourth axis angle information and fourth shape information.

The way to obtain the random negative sample set may be: and randomly generating m multiplied by n samples, taking the m multiplied by n samples as random negative samples, and randomly dividing the random negative samples into m random negative sample groups, wherein each random negative sample group comprises n random negative samples.

For example, 4n samples are randomly generated and considered to be all negative samples, and the 4n samples are randomly divided into 4 random negative sample groups, each group also being n negative samples.

After m target positive sample groups and m random negative sample groups are obtained, one target positive sample group and one random negative sample group are combined to obtain m initial training sample groups.

And 2, using m initial training sample groups to respectively train the m second initial discriminators to obtain m reference discriminators, wherein the m initial training sample groups correspond to the m second initial discriminators one by one.

The m second initial discriminators may be trained separately using the acquired m initial training sample sets, resulting in each reference discriminator. In a round of training, an initial training sample set is used to train a second initial discriminator.

For example, as shown in fig. 3, four discrimination models can be trained simultaneously to obtain 4 reference discrimination models (discriminators) when four sets of positive and negative samples are obtained.

In order to ensure the performance of the trained reference discriminator, for the trained discrimination model, the performance of the current discrimination model may be evaluated using a target positive sample group in the other initial training sample groups except the initial training sample group corresponding to the current discrimination model.

As an alternative embodiment, the training of the m second initial discriminators with the m initial training sample sets respectively to obtain the m reference discriminators includes: training a current initial discriminator in m second initial discriminators by using a current initial training sample group in m initial training sample groups to obtain a candidate reference discriminator, wherein the discrimination accuracy of the candidate reference discriminator on samples in the current initial training sample group is greater than or equal to a first accuracy threshold; and determining the candidate reference discriminator as the reference discriminator when the discrimination accuracy of the candidate reference discriminator on other target positive samples is greater than or equal to a second accuracy threshold, wherein the other target positive samples are target positive samples contained in other initial training sample groups except the current initial training sample group in the m initial training sample groups.

For the current initial training sample set of the m initial training sample sets and the current initial discriminator of the m second initial discriminators, multi-round training may be performed on the current initial discriminator using the current initial training sample set, and model parameters may be adjusted until the discrimination accuracy of the adjusted discrimination model pair on samples (including target positive samples and random negative samples) in the current initial training sample set is greater than or equal to a first accuracy threshold. The adjusted authentication model is a candidate reference authenticator.

After obtaining the candidate reference discriminator, the accuracy of the candidate reference discriminator in discriminating other target positive samples may be further determined, where the other target positive samples are target positive samples in other initial training sample sets except for the current initial training sample set.

The candidate reference discriminator may be determined to be the reference discriminator if the accuracy of the candidate reference discriminator in discriminating the other target positive samples is greater than or equal to the second accuracy threshold. If the accuracy of the candidate reference discriminator for discriminating the other target positive samples is less than the second accuracy threshold, step 1 may be performed again to obtain m initial training sample sets again.

It should be noted that the first accuracy threshold and the second accuracy threshold may be the same value (for example, both are 90%), or may be different values (for example, the first accuracy threshold is 90%, the second accuracy threshold is 80%), and the first accuracy threshold and the second accuracy threshold may be fixed or may be modified according to the configuration instruction, which is not specifically limited in this embodiment.

For example, for 4 identification models that are trained respectively, each identification model needs to obtain 90% accuracy on the other three sets of positive samples, otherwise, the data is regenerated and retrained.

By the embodiment, the candidate reference discriminators are screened by using the target positive samples in other initial training sample sets, so that the performance of the obtained reference discriminators can be ensured, and the discrimination accuracy of the target discriminators can be improved.

And 3, sequentially using other reference discriminators except the target reference discriminator to discriminate each random negative sample in the target random negative sample group to obtain a sample discrimination result of each random negative sample, wherein the target reference discriminator is a reference discriminator selected from m reference discriminators, the target random negative sample group is a random negative sample group in the initial training sample group corresponding to the target reference discriminator, and the sample discrimination result is used for indicating whether the random negative sample is a negative sample.

After m reference discriminators are obtained, the random negative samples in each random negative sample group can be classified, and the negative samples with higher confidence coefficient are screened out.

The random negative sample groups in each of the m initial training sample groups may be sequentially processed, for example, one initial training sample group may be sequentially selected from the m initial training sample groups as a target initial training sample group, and the reference discriminator corresponding to the target initial training sample group is a selected target reference discriminator.

For the target random negative sample group in the target initial training sample group, other reference discriminators except the target reference discriminator may be used to discriminate each random negative sample in the target random negative sample group to obtain a sample discrimination result of each random negative sample, where the sample discrimination result is used to indicate whether the random negative sample is a negative sample.

As an alternative embodiment, the sequentially using other reference discriminators except the target reference discriminator to discriminate each random negative sample in the target random negative sample group, and obtaining the sample discrimination result of each random negative sample includes: respectively identifying the target random negative samples in the target random negative sample group by using a plurality of other reference identifiers to obtain a plurality of sub-sample identification results, wherein each sub-sample identification result is used for indicating whether the target random negative sample identified by one other reference identifier is a negative sample; and determining the sample identification result of the target random negative sample according to the plurality of sub-sample identification results.

In the case that there are a plurality of other reference discriminators, the plurality of other reference discriminators may be used to discriminate the target random negative sample in the target random negative sample group, respectively, to obtain a plurality of sub-sample discrimination results, where each random negative sample has at most (m-1) sub-sample discrimination results, and each sub-sample discrimination result is used to indicate whether the target random negative sample discriminated by one of the other reference discriminators is a negative sample.

After obtaining the plurality of sub-sample authentication results, the sample authentication result of the target random negative sample may be determined according to the plurality of sub-sample authentication results. There are various ways to determine the sample discrimination result of the target random negative sample based on the plurality of sub-sample discrimination results. For example, a target random negative example is determined to be a negative example only if the plurality of subsample authentication results all indicate that the target random negative example is a negative example, otherwise, the random negative example is determined to be a positive example. For another example, when the plurality of sub-sample authentication results exceeds a predetermined number (e.g., more than half) of the sub-authentication results indicates that the target random negative sample is a negative sample, the target random negative sample is determined as a negative sample, otherwise, the random negative sample is determined as a positive sample.

For example, as shown in fig. 4, when all four models satisfy the model condition (90% accuracy is obtained on the other three sets of positive samples), the other three trained models are used to classify the current set of random negative samples, each random negative sample obtains three results through the other three models, and the three results are voted to obtain the final type (positive or negative). By labeling the randomly generated samples in this way, a batch of negative samples with slightly higher confidence relative to the randomly generated negative samples can be obtained. By repeating the previous steps, a negative sample with a higher confidence of a total of 4n can be obtained.

By the embodiment, the random negative sample is labeled by integrating the identification results of a plurality of other reference identifiers, so that the confidence of the obtained target negative sample can be improved.

And 4, indicating the sample identification result as a random negative sample of the negative sample, and determining the sample identification result as a target negative sample.

After determining the sample authentication result for each random negative sample, the sample authentication result may be indicated as a random negative sample of the negative samples, determined as a target negative sample, and the random negative samples with the sample authentication result indicated as a positive sample are filtered out.

And (4) repeatedly executing the steps 1 to 4 until the number of the obtained target negative samples reaches m multiplied by n.

For the obtained m × n target negative samples, the target negative samples can be randomly divided into m groups, each group includes n target negative samples, and thus m target negative sample groups are obtained.

For example, after 4n negative samples with a higher confidence are obtained, the 4n negative samples may be equally divided, the same process may be continuously performed, and such a process may be repeated multiple times, so that an identification model may be obtained, where the identification model may be one of the models 1 to 4 that is selected to have the best effect.

By the embodiment, the difficulty in obtaining the negative sample can be solved by randomly generating the random negative sample and screening the random negative sample to obtain the target negative sample, so that the identification accuracy of the target identifier is improved.

In step S206, in a case where the target discrimination result indicates that the first axis angle information and the first shape information match the target object, a three-dimensional object model is constructed in accordance with the target model parameters.

After obtaining the target identification result, if the target identification result indicates that the first axis angle information and the first shape information match the target object, a three-dimensional object model may be constructed according to the target model parameters, and the three-dimensional object model may be obtained as shown in fig. 5 or obtained after rendering on the basis of fig. 5. If the target discrimination result indicates that the first axis angle information and the first shape information do not match the target object, the construction of the three-dimensional object model using the target model parameters may be prohibited, and prompt information may be displayed on the display device to prompt that the target model parameters are not suitable for constructing the three-dimensional object model, and modification of the model parameters may also be prompted.

The model construction method described above is explained below with reference to an alternative example. The model construction method in this example is a human body reconstruction parameter rationalization identification method, and rotation information and shape information of a joint point are distinguished by designing a plurality of simple discriminators. And finally giving a score to the current 3D model parameter group through fusion of a plurality of simple discriminators, thereby identifying whether the human body reconstruction parameters are reasonable or not.

After the human body reconstruction parameters (target model parameters) are identified, a plurality of identification models can be trained in groups, randomly generated negative samples are voted and screened, and classification models can be trained without real negative samples.

When model training is carried out, firstly, 4n real positive samples are obtained and are randomly divided into four positive sample groups, and each group of n positive samples; then, 4n samples are randomly generated and considered as negative samples, and are randomly divided into four random negative sample groups, and each random negative sample group also has n negative samples. In this way, four sets of positive and negative samples can be acquired, and four discriminative models (models 1 to 4) can be trained simultaneously by the four sets of positive and negative samples. The trained identification model needs to obtain 90% accuracy on other three groups of positive samples, otherwise, the data needs to be regenerated and retrained.

When the four trained discrimination models all meet the above conditions, the current random negative sample group can be classified by using other three trained discrimination models, each random negative sample obtains three results through the other three models, and the three results are voted to obtain the final type (positive or negative). By labeling the randomly generated samples in this way, a batch of negative samples with slightly higher confidence may be obtained.

In the same way, a little more negative example with a total of 4n confidence may be obtained. For a point-negative sample with high 4n confidence, the same flow can be divided equally and repeated for multiple times, and an identification model (the best one among the models 1 to 4) can be obtained.

After the identification model is obtained, the identification model may be used to identify the model parameters. As shown, the model construction method in this example may include the following steps:

step S602, obtaining model parameters of the 3D human body model.

Step S604, inputting the model parameters into the two-layer classifier to obtain the probability that the model parameters output by the two-layer classifier are legal parameters.

And step S606, under the condition that the probability that the model parameter is a legal parameter is greater than or equal to the target probability, the model parameter is used for constructing the 3D human body model.

By the method, the efficiency of identifying the human body reconstruction parameters can be improved, and the rationality of the constructed 3D human body model is further improved.

Through the above description of the embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but the former is a better implementation mode in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.

According to another aspect of the embodiments of the present application, there is provided a model construction apparatus for implementing the above-described model construction method. Optionally, the apparatus is used to implement the above embodiments and preferred embodiments, and details are not repeated for what has been described. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.

Fig. 7 is a block diagram of an alternative model building apparatus according to an embodiment of the present application, as shown in fig. 7, the apparatus including:

(1) a first obtaining unit 702 configured to obtain target model parameters of a three-dimensional object model of a target object, wherein the target model parameters include first axis angle information of joint points of the three-dimensional object model and first shape information for controlling a body type of the three-dimensional object model;

(2) a first input unit 704, connected to the first obtaining unit 702, configured to input the first axis angle information and the first shape information into the target discriminator to obtain a target discrimination result output by the target discriminator, where the target discrimination result is used to indicate whether the first axis angle information and the first shape information match the target object;

(3) and a constructing unit 706 connected to the first input unit 704, for constructing a three-dimensional object model according to the target model parameters in case that the target discrimination result indicates that the first axis angle information and the first shape information match the target object.

Alternatively, the first obtaining unit 702 may be used in step S202 in the foregoing embodiment, the first input unit 704 may be used in step S204 in the foregoing embodiment, and the constructing unit 706 may be used to execute step S206 in the foregoing embodiment.

As an alternative embodiment, the first input unit 704 includes:

(1) the first input module is used for inputting the first axis angle information into the first discriminator to obtain a first discrimination result output by the first discriminator, wherein the first discrimination result is used for indicating whether the first axis angle information is matched with the target object;

(2) the second input module is used for inputting the first shape information into the second discriminator to obtain a second discrimination result output by the second discriminator, wherein the second discrimination result is used for indicating whether the first shape information is matched with the target object or not;

(3) a third input module, configured to input the first authentication result and the second authentication result to a third authenticator, so as to obtain a target authentication result output by the third authenticator, where the target authenticator includes: a first discriminator, a second discriminator, and a third discriminator.

As an alternative embodiment, the first input module comprises:

(1) the first input submodule is used for respectively inputting a plurality of first axis angle information of a plurality of joint points into a plurality of first sub-discriminators to obtain a plurality of first sub-discrimination results under the condition that the joint points comprise a plurality of joint points, wherein the joint points correspond to the first sub-discriminators one by one, and each first sub-discrimination result is used for indicating whether the first axis angle information of one joint point is matched with a target object or not;

(2) a second input sub-module, configured to input the first axis angle information of the joint points to a second sub-discriminator to obtain a second sub-discrimination result output by the second sub-discriminator, where the second sub-discrimination result is used to indicate whether the first axis angle information of the joint points matches the target object, and the first discriminator includes: a plurality of first sub-discriminators and second sub-discriminators, the first discrimination result including: a plurality of first sub-authentication results and second sub-authentication results.

As an alternative embodiment, the apparatus further comprises:

(1) a second obtaining unit, configured to obtain m target training sample sets before inputting the first axis angle information and the first shape information into the target discriminator to obtain a target discrimination result output by the target discriminator, where each target training sample set includes a target positive sample set and a target negative sample set, the target positive sample set includes n target positive samples, the target negative sample set includes n target negative samples, the target positive samples are model parameters including second axis angle information and second shape information that match the target object, the target negative samples are model parameters including third axis angle information and third shape information that do not match the target object, m is a positive integer greater than or equal to 1, and n is a positive integer greater than 1;

(2) the training unit is used for respectively training the m first initial discriminators by using m target training sample sets to obtain m candidate discriminators, wherein the m target training sample sets correspond to the m first initial discriminators one by one;

and the selecting unit is used for selecting the target discriminator from the m candidate discriminators.

As an alternative embodiment, the second obtaining unit includes:

(1) the first acquisition module is used for acquiring m target positive sample groups;

(2) a second obtaining module, configured to repeatedly perform the following steps until m × n target negative samples are obtained: acquiring m initial training sample groups, wherein each initial training sample group comprises a target positive sample group and a random negative sample group, the random negative sample group comprises n random negative samples, and the random negative samples are model parameters containing randomly generated fourth axis angle information and fourth shape information; respectively training m second initial discriminators by using m initial training sample sets to obtain m reference discriminators, wherein the m initial training sample sets correspond to the m second initial discriminators one by one; sequentially using other reference discriminators except the target reference discriminator to discriminate each random negative sample in the target random negative sample group to obtain a sample discrimination result of each random negative sample, wherein the target reference discriminator is a reference discriminator selected from m reference discriminators, the target random negative sample group is a random negative sample group in an initial training sample group corresponding to the target reference discriminator, and the sample discrimination result is used for indicating whether the random negative sample is a negative sample; indicating the sample identification result as a random negative sample of the negative sample, and determining the sample identification result as a target negative sample;

(3) and the third acquisition module is used for acquiring m target negative sample groups according to the m multiplied by n target negative samples.

As an alternative embodiment, the second obtaining module includes:

(1) the training submodule is used for training a current initial discriminator in the m second initial discriminators by using a current initial training sample group in the m initial training sample groups to obtain a candidate reference discriminator, wherein the discrimination accuracy of the candidate reference discriminator on the samples in the current initial training sample group is greater than or equal to a first accuracy threshold;

(2) and the first determining sub-module is used for determining the candidate reference discriminator as the reference discriminator when the discrimination accuracy of the candidate reference discriminator on other target positive samples is greater than or equal to a second accuracy threshold, wherein the other target positive samples are target positive samples contained in other initial training sample groups except the current initial training sample group in the m initial training sample groups.

As an alternative embodiment, the second obtaining module includes:

(1) the identification submodule is used for identifying the target random negative sample in the target random negative sample group by using a plurality of other reference identifiers to obtain a plurality of sub-sample identification results, wherein each sub-sample identification result is used for indicating whether the target random negative sample identified by one other reference identifier is a negative sample or not;

(2) and the second determining submodule is used for determining the sample identification result of the target random negative sample according to the plurality of sub-sample identification results.

It should be noted that, the above modules may be implemented by software or hardware, and for the latter, the following may be implemented, but not limited to: the modules are all positioned in the same processor; alternatively, the modules are respectively located in different processors in any combination.

According to yet another aspect of embodiments herein, there is provided a computer-readable storage medium. Optionally, the storage medium has a computer program stored therein, where the computer program is configured to execute the steps in any one of the methods provided in the embodiments of the present application when the computer program is executed.

Alternatively, in the present embodiment, the storage medium may be configured to store a computer program for executing the steps of:

s1, obtaining target model parameters of a three-dimensional object model of the target object, wherein the target model parameters comprise first axis angle information of joint points of the three-dimensional object model and first shape information for controlling the body type of the three-dimensional object model;

s2, inputting the first axis angle information and the first shape information into the target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axis angle information and the first shape information are matched with the target object;

s3, in a case where the target discrimination result indicates that the first axis angle information and the first shape information match the target object, constructing a three-dimensional object model according to the target model parameters.

Optionally, in this embodiment, the storage medium may include, but is not limited to: a variety of media that can store computer programs, such as a usb disk, a ROM (Read-only Memory), a RAM (Random Access Memory), a removable hard disk, a magnetic disk, or an optical disk.

According to still another aspect of an embodiment of the present application, there is provided an electronic apparatus including: a processor (which may be the processor 102 in fig. 1) and a memory (which may be the memory 104 in fig. 1) having a computer program stored therein, the processor being configured to execute the computer program to perform the steps of any of the above methods provided in embodiments of the present application.

Optionally, the electronic apparatus may further include a transmission device (the transmission device may be the transmission device 106 in fig. 1) and an input/output device (the input/output device may be the input/output device 108 in fig. 1), wherein the transmission device is connected to the processor, and the input/output device is connected to the processor.

Optionally, in this embodiment, the processor may be configured to execute the following steps by a computer program:

Optionally, for an optional example in this embodiment, reference may be made to the examples described in the above embodiment and optional implementation, and this embodiment is not described herein again.

It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and alternatively, they may be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, and in some cases, the steps shown or described may be performed in an order different than that described herein, or they may be separately fabricated into individual integrated circuit modules, or multiple ones of them may be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method of model construction, comprising:

acquiring target model parameters of a three-dimensional object model of a target object, wherein the target model parameters comprise first axis angle information of joint points of the three-dimensional object model and first shape information for controlling the body type of the three-dimensional object model;

inputting the first axis angle information and the first shape information into a target discriminator to obtain a target discrimination result output by the target discriminator, wherein the target discrimination result is used for indicating whether the first axis angle information and the first shape information are matched with the target object;

and under the condition that the target identification result indicates that the first axis angle information and the first shape information are matched with the target object, constructing the three-dimensional object model according to the target model parameters.

2. The method of claim 1, wherein inputting the first axis angle information and the first shape information to the target discriminator, and obtaining the target discrimination result output by the target discriminator comprises:

inputting the first axis angle information into a first discriminator to obtain a first discrimination result output by the first discriminator, wherein the first discrimination result is used for indicating whether the first axis angle information is matched with the target object;

inputting the first shape information into a second discriminator to obtain a second discrimination result output by the second discriminator, wherein the second discrimination result is used for indicating whether the first shape information is matched with the target object;

inputting the first authentication result and the second authentication result into a third authenticator, and obtaining the target authentication result output by the third authenticator, wherein the target authenticator comprises: the first discriminator, the second discriminator, and the third discriminator.

3. The method of claim 2, wherein inputting the first axis angle information to the first discriminator, obtaining the first discrimination output by the first discriminator comprises:

when the joint point comprises a plurality of joint points, respectively inputting a plurality of first axis angle information of the joint points to a plurality of first sub-discriminators to obtain a plurality of first sub-discrimination results, wherein the joint points correspond to the first sub-discriminators one by one, and each first sub-discrimination result is used for indicating whether the first axis angle information of one joint point is matched with the target object;

inputting a plurality of first axis angle information of a plurality of joint points into a second sub-discriminator to obtain a second sub-discrimination result output by the second sub-discriminator, wherein the second sub-discrimination result is used for indicating whether the plurality of first axis angle information of the plurality of joint points is matched with the target object, and the first discriminator comprises: a plurality of the first sub-discriminators and the second sub-discriminators, the first discrimination result including: a plurality of the first sub-authentication results and the second sub-authentication results.

4. The method according to any one of claims 1 to 3, wherein before the first axis angle information and the first shape information are input to a target discriminator to obtain a target discrimination result output by the target discriminator, the method further comprises:

acquiring m target training sample sets, wherein each target training sample set comprises a target positive sample set and a target negative sample set, the target positive sample set comprises n target positive samples, the target negative sample set comprises n target negative samples, the target positive samples are model parameters containing second axis angle information and second shape information matched with the target object, the target negative samples are model parameters containing third axis angle information and third shape information unmatched with the target object, m is a positive integer greater than or equal to 1, and n is a positive integer greater than 1;

respectively training m first initial discriminators by using m target training sample groups to obtain m candidate discriminators, wherein the m target training sample groups correspond to the m first initial discriminators one by one;

and selecting the target discriminator from the m candidate discriminators.

5. The method of claim 4, wherein obtaining the set of m target training samples comprises:

acquiring m target positive sample groups;

repeatedly executing the following steps until m × n target negative samples are acquired: acquiring m initial training sample groups, wherein each initial training sample group comprises one target positive sample group and one random negative sample group, each random negative sample group comprises n random negative samples, and each random negative sample is a model parameter containing randomly generated fourth axis angle information and fourth shape information; using m initial training sample groups to respectively train m second initial discriminators to obtain m reference discriminators, wherein the m initial training sample groups correspond to the m second initial discriminators one by one; sequentially using other reference discriminators except for a target reference discriminator to discriminate each random negative sample in a target random negative sample group to obtain a sample discrimination result of each random negative sample, wherein the target reference discriminator is a reference discriminator selected from m reference discriminators, the target random negative sample group is a random negative sample group in an initial training sample group corresponding to the target reference discriminator, and the sample discrimination result is used for indicating whether the random negative sample is a negative sample; determining the sample discrimination result as a random negative sample of negative samples, determining as the target negative sample;

and acquiring m target negative sample groups according to the m multiplied by n target negative samples.

6. The method of claim 5, wherein training m of the second initial discriminators using m of the initial training sample sets respectively, resulting in m of the reference discriminators comprises:

training a current initial discriminator in m second initial discriminators by using a current initial training sample group in m initial training sample groups to obtain a candidate reference discriminator, wherein the discrimination accuracy of the candidate reference discriminator on samples in the current initial training sample group is greater than or equal to a first accuracy threshold;

and determining the candidate reference discriminator as the reference discriminator when the discrimination accuracy of the candidate reference discriminator on other target positive samples is greater than or equal to a second accuracy threshold, wherein the other target positive samples are target positive samples contained in other initial training sample groups except the current initial training sample group in the m initial training sample groups.

7. The method of claim 5, wherein the step of sequentially using the other reference discriminators except the target reference discriminator to discriminate each random negative example in the target random negative example group to obtain the example discrimination result of each random negative example comprises:

respectively identifying the target random negative samples in the target random negative sample group by using a plurality of other reference identifiers to obtain a plurality of sub-sample identification results, wherein each sub-sample identification result is used for indicating whether the target random negative sample identified by one of the other reference identifiers is a negative sample;

and determining the sample identification result of the target random negative sample according to a plurality of the sub-sample identification results.

8. A model building apparatus, comprising:

a first acquisition unit configured to acquire target model parameters of a three-dimensional object model of a target object, wherein the target model parameters include first axis angle information of joint points of the three-dimensional object model and first shape information for controlling a body type of the three-dimensional object model;

a first input unit, configured to input the first axis angle information and the first shape information to a target discriminator to obtain a target discrimination result output by the target discriminator, where the target discrimination result is used to indicate whether the first axis angle information and the first shape information match the target object;

a construction unit configured to construct the three-dimensional object model according to the target model parameters in a case where the target discrimination result indicates that the first axis angle information and the first shape information match the target object.

9. The apparatus of claim 8, wherein the first input unit comprises:

a first input module, configured to input the first axis angle information to a first discriminator to obtain a first discrimination result output by the first discriminator, where the first discrimination result is used to indicate whether the first axis angle information matches the target object;

the second input module is used for inputting the first shape information into a second discriminator to obtain a second discrimination result output by the second discriminator, wherein the second discrimination result is used for indicating whether the first shape information is matched with the target object;

a third input module, configured to input the first authentication result and the second authentication result to a third authenticator, so as to obtain the target authentication result output by the third authenticator, where the target authenticator includes: the first discriminator, the second discriminator, and the third discriminator.

10. The apparatus of claim 9, wherein the first input module comprises:

a first input sub-module, configured to, when the joint point includes a plurality of joint points, respectively input the first axis angle information of the joint points to a plurality of first sub-discriminators to obtain a plurality of first sub-discrimination results, where the joint points correspond to the first sub-discriminators one by one, and each of the first sub-discrimination results is used to indicate whether the first axis angle information of one joint point matches the target object;

a second input sub-module, configured to input a plurality of first axis angle information of a plurality of joint points to a second sub-discriminator to obtain a second sub-discrimination result output by the second sub-discriminator, where the second sub-discrimination result is used to indicate whether the plurality of first axis angle information of the plurality of joint points match the target object, and the first discriminator includes: a plurality of the first sub-discriminators and the second sub-discriminators, the first discrimination result including: a plurality of the first sub-authentication results and the second sub-authentication results.

11. The apparatus of any one of claims 8 to 10, further comprising:

a second obtaining unit, configured to obtain m target training sample sets before inputting the first axis angle information and the first shape information into a target discriminator to obtain a target discrimination result output by the target discriminator, where each target training sample set includes a target positive sample set and a target negative sample set, the target positive sample set includes n target positive samples, the target negative sample set includes n target negative samples, the target positive sample is a model parameter including second axis angle information and second shape information that match the target object, the target negative sample is a model parameter including third axis angle information and third shape information that do not match the target object, m is a positive integer greater than or equal to 1, and n is a positive integer greater than 1;

a training unit, configured to use m target training sample sets to respectively train m first initial discriminators to obtain m candidate discriminators, where the m target training sample sets correspond to the m first initial discriminators one to one;

12. The apparatus of claim 11, wherein the second obtaining unit comprises:

a first obtaining module, configured to obtain m target positive sample groups;

a second obtaining module, configured to repeatedly perform the following steps until m × n target negative samples are obtained: acquiring m initial training sample groups, wherein each initial training sample group comprises one target positive sample group and one random negative sample group, each random negative sample group comprises n random negative samples, and each random negative sample is a model parameter containing randomly generated fourth axis angle information and fourth shape information; using m initial training sample groups to respectively train m second initial discriminators to obtain m reference discriminators, wherein the m initial training sample groups correspond to the m second initial discriminators one by one; sequentially using other reference discriminators except for a target reference discriminator to discriminate each random negative sample in a target random negative sample group to obtain a sample discrimination result of each random negative sample, wherein the target reference discriminator is a reference discriminator selected from m reference discriminators, the target random negative sample group is a random negative sample group in an initial training sample group corresponding to the target reference discriminator, and the sample discrimination result is used for indicating whether the random negative sample is a negative sample; determining the sample discrimination result as a random negative sample of negative samples, determining as the target negative sample;

and the third acquisition module is used for acquiring m target negative sample groups according to the m multiplied by n target negative samples.

13. A computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to carry out the method of any one of claims 1 to 7 when executed.

14. An electronic device comprising a memory and a processor, characterized in that the memory has stored therein a computer program, the processor being arranged to execute the method of any of claims 1 to 7 by means of the computer program.