CN118135007A - Pose determination method and device, electronic equipment and storage medium - Google Patents

Abstract

The application discloses a pose determining method, a pose determining device, electronic equipment and a readable storage medium, wherein the pose determining method comprises the following steps: acquiring a plurality of components corresponding to a target object, feature points corresponding to each component and feature point position information corresponding to the feature points of each component; determining at least one component set according to the feature point position information of the feature point corresponding to each component; and determining the pose of the target object according to the component pose information of each component in each component set. According to the application, the connection relation among the components in the component set is determined through each component of the target object, the component pose information of each component and the characteristic point position information of the corresponding characteristic points on each component, so that the pose information of the target object can be accurately determined according to the connection relation among the components, and the technical effect of accurately acquiring the pose information of the variable model is achieved.

Description

Pose determination method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of electronic information technology, and more particularly, to a pose determining method, a pose determining device, an electronic device, and a computer readable storage medium.

Background

With the development of technology, pose estimation has many important applications in the fields of automatic driving, robots and the like.

Currently, pose estimation can be performed on a model-fixed object through algorithms such as PVNet, FFB6D and the like. However, when the object is a variable model, the relative pose between the multiple components included in the variable model is not fixed, resulting in difficulty in accurately estimating the pose of the variable model.

Disclosure of Invention

The application provides a pose determining method, a pose determining device, electronic equipment and a computer readable storage medium, so as to improve the defects.

In a first aspect, an embodiment of the present application provides a pose determining method, which is characterized in that the method includes: acquiring a plurality of components corresponding to a target object, feature points corresponding to each component and feature point position information of the feature points corresponding to each component; determining at least one component set according to the feature point position information of the feature points corresponding to each component, wherein feature points matched with each other exist among the components in each component set; and determining the pose of the target object according to the component pose information of each component in each component set.

In a second aspect, an embodiment of the present application further provides a pose determining apparatus, which is used for an electronic device, and the apparatus includes: the device comprises an acquisition module, a storage module and a storage module, wherein the acquisition module is used for acquiring a plurality of components corresponding to a target object, characteristic points corresponding to each component and characteristic point position information corresponding to the characteristic points of each component; the first determining module is used for determining at least one component set according to the feature point position information of the feature point corresponding to each component, wherein feature points matched with each other exist among the components in each component set; and the second determining module is used for determining the pose of the target object according to the component pose information of each component in each component set.

In a third aspect, an embodiment of the present application further provides an electronic device, including: one or more processors; a memory; one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to perform the above-described method.

In a fourth aspect, embodiments of the present application also provide a computer readable storage medium storing program code executable by a processor, the program code when executed by the processor causing the processor to perform the above method.

According to the pose determining method, the pose determining device, the electronic equipment and the computer readable storage medium, the connecting relation among all the parts in the part set is determined through each part of the target object, the part pose information of each part and the characteristic point position information of the corresponding characteristic points on each part, so that the pose information of the target object can be accurately determined according to the connecting relation among all the parts, and the effect of accurately acquiring the pose information of the variable model is achieved.

Additional features and advantages of embodiments of the application will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of embodiments of the application. The objectives and other advantages of embodiments of the application may be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 shows a flow chart of a pose determination method according to an embodiment of the application.

Fig. 2 shows a flow chart of a pose determination method according to a further embodiment of the present application.

Fig. 3 shows a schematic diagram of feature points of a target object in an embodiment of the present application.

Fig. 4 shows a flowchart of a pose determination method according to still another embodiment of the present application.

Fig. 5 shows a block diagram of a pose determining apparatus according to an embodiment of the present application.

Fig. 6 shows a block diagram of an electronic device according to an embodiment of the present application.

Fig. 7 shows a block diagram of a computer-readable storage medium according to an embodiment of the present application.

Detailed Description

In order to make the present application better understood by those skilled in the art, the following description will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. The components of the embodiments of the present application generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the application, as presented in the figures, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by a person skilled in the art without making any inventive effort, are intended to be within the scope of the present application.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only to distinguish the description, and are not to be construed as indicating or implying relative importance.

Pose is a relative concept for representing a relatively rigid transformation relationship between two coordinate systems. For the pose of a rigid object, 6 parameters are typically used for description, including the relative position coordinates (X, Y, Z) describing the translational motion and three euler angles (roll, pitch, yaw) describing the relative rotational motion, while the goal of pose estimation is to find the 6 parameters describing the translation and rotation. Pose estimation has many important applications in the fields of autopilot, robotics, etc. In pose estimation, it is necessary for an estimated object to have enough relevant knowledge about the object, for example, for an actual object it will often be necessary to have a three-dimensional model matching it, and its pose is determined in combination with extracting information from the sensor. In practical applications it may take a lot of time and effort to manually register. Specifically, taking as an example the deep learning technology which is increasingly widely used in recent years, for a fixed model, the deep learning model can be trained in a large amount, and then the pose of the model in the scene can be identified according to a color image or in combination with a depth map.

The existing pose estimation algorithm generally estimates the pose of a fixed object, and there are algorithms for estimating the pose by using deep learning, such as a pixel voting network (PVNet), a full-flow bidirectional fusion network (FFB 6D), and the like, and algorithms for estimating the pose by using template matching.

However, with the development of science and technology, more and more pose estimation objects are variable objects, such as dolls with variable joints, mobile phone camera holders, and the like. The inventors found in the study that when an object is a variable model, the relative pose between the multiple components included in the variable model is not fixed, resulting in difficulty in accurately estimating the pose of the variable model. Taking a doll with movable joints as an example, the relative pose among a plurality of parts is not fixed, when a plurality of identical parts exist in a scene, the relation among the parts cannot be determined, and the pose of the whole object cannot be determined.

Therefore, in order to overcome the above-mentioned drawbacks, embodiments of the present application provide a pose determining method, apparatus, electronic device, and computer readable storage medium, which are capable of determining a connection relationship between each component in a component set based on pose information of each component of a target object, component pose information of each component, and feature point position information of each corresponding feature point on each component, so as to accurately determine pose information of the target object.

Referring to fig. 1, fig. 1 shows a flow chart of a pose determining method according to an embodiment of the present application, for an electronic device, the method includes:

s101, acquiring a plurality of components corresponding to the target object, feature points corresponding to each component and feature point position information of the feature points corresponding to each component.

In this embodiment, the target object may be a variable model, which refers to an object in which the relative pose between the respective parts of the object is not fixed. Each component on the variable model is acquired, and feature points are set on each component, wherein the feature points can represent the relation between the component and the component related to the feature points, and the feature points can be points of connection parts between the components or points of contact parts between the components. At least one characteristic point is arranged on each component, the characteristic point is a preset point, the position information of the characteristic point does not change along with the change of the relative pose among the components of the target object, and the position information of the characteristic point can be three-dimensional coordinate information of the characteristic point on a world coordinate system.

S102, determining at least one component set according to the feature point position information of the feature point corresponding to each component, wherein feature points matched with each other exist among the components in each component set.

The feature points being matched with each other may be that feature points are close to each other or that feature points are coincident. The component set refers to a set of components where feature points with matching relations are located, and the components with matching relations are taken as a component set and are not divided in a physical sense. The feature points matched with each other exist among the components in each component set, so that the association relationship among the components in the component set can be determined, and the association relationship among the components can be a connection relationship or a contact relationship.

Taking a joint movable doll as an example, two arms of the doll are two identical parts, the relative pose between the arms and a trunk is variable, each of the two arms is provided with a characteristic point, the trunk is provided with two characteristic points matched with the characteristic points on the two arms, so that the trunk and the two arms are respectively a part set, two part sets exist between the trunk and the arms, no mutually matched characteristic points exist between the two arms, the two arms are not a part set, and the two arms and the trunk are respectively associated, wherein the association is a connection relationship.

S103, determining the pose of the target object according to the component pose information of each component in each component set.

In this embodiment, the component pose information of each component may refer to pose information of the component in a world coordinate system, and according to the component pose information of each component in each component set of the target object, a relationship between all components of the target object may be obtained, and based on the relationship between components of the target object and the pose information of each component, the pose of the target object is determined.

For example, taking a doll with movable joints as an example, it is known from the above that the trunk and the two arms are respectively the same component set, and if the two arms and the trunk are one target object, the pose of the target object can be obtained based on the connection relationship between the two arms and the trunk and the pose information of the two arms and the trunk.

In this embodiment, the connection relationship between the components in the component set is determined by each component of the target object, the component pose information of each component, and the feature point position information of each corresponding feature point on each component, so that the pose information of the target object can be accurately determined according to the connection relationship between the components, and the technical effect of accurately acquiring the pose information of the variable model is achieved.

Referring to fig. 2, fig. 2 shows a flowchart of a pose determining method according to another embodiment of the present application, for an electronic device, the method includes:

S201, acquiring a plurality of components corresponding to the target object.

The description of S201 refers to the description of S101 above, and is not repeated here.

S202, if any two components have a connection relationship, acquiring points connecting any two components as characteristic points of any two components, and acquiring characteristic point position information corresponding to each characteristic point.

In the present embodiment, a point at which two members having a connection relationship are connected is taken as a feature point, for example, a member a is connected to a member B, and a feature point at a connection portion of the member a and the member B is both a feature point on the member a and a feature point on the member B.

S203, according to the feature point position information of the feature point corresponding to each component, if feature points with the feature point position information matched with each other exist in any two components, acquiring any two components as a component set.

The feature point position information being matched with each other may mean that there are equal values in the feature point position information of the feature points, or that the feature point position information of the feature points is close, or that the feature point position information of the feature points is equal within a certain error range. If there are feature points in which feature point position information is matched with each other in any two parts, the two feature points are matched with each other, so that the two parts are matched with each other, and the two parts can be used as a part set.

S204, determining the connection relation of the components in each component set according to the characteristic point position information of the characteristic points matched with each other in each component set.

The feature point may refer to a point of a connection portion between two members having a connection relationship, and the feature point is a feature point on either a first member or a second member of the two members having a connection relationship, the feature point on the first member and the feature point on the second member being matched with each other, and feature point position information of the feature point on the first member and feature point position information of the feature point on the second member being matched with each other.

That is, feature point position information of feature points matching each other exists in each component set, and connection relations exist between the components in this component set. For example, when there are a plurality of identical parts in a scene, the connection relationship between the parts is determined, and for two connected parts, part 1 and part 2, if their pose is known, the feature point position information of the feature point P between them can be obtained by either part 1 or part 2. So for each component 1, the feature point position information of the P point is found first, then for each component 2, the feature point position information of the P point corresponding thereto is found, if the two feature point position information match each other, the two components are connected, and if the distance is far, there is no connection relationship between them.

S205, determining the pose of the target object according to the connection relation of all the components in each component set.

In this embodiment, the target object is a whole, and there is a phenomenon that some parts exist in multiple parts among the parts of the target object, and by means of the connection relationships between the parts repeatedly appearing in the multiple parts, the connection relationships between all the parts of the whole target object can be determined, and based on the part pose information of the parts inside the target object, the pose of the target object can be determined.

For example, referring to fig. 3, fig. 3 shows a schematic diagram of feature points of a target object in the present embodiment. After the two heads, the four upper limbs, the four lower limbs and the component pose information of the two torsos in the scene are obtained, feature points calculated according to the component pose information are marked on the graph in fig. 3, black points in the graph are positions of the feature points determined according to the pose information of the torsos, white points in the graph are positions of the feature points determined according to the pose information of the heads, the upper limbs and the lower limbs, and the pose of the bear model is determined according to the positions of the feature points of the components, for example, the bear model of the skewed head on the left side of the graph in fig. 3 and the bear model of the normal standing on the right side of the graph in fig. 3.

It can be seen that the same feature point is matched with the feature point position information of the feature point obtained according to two different parts where the same feature point is located, the connection relation between the parts where the feature point is located can be determined according to the relation between the feature point position information of the feature point, and corresponding other parts can be found through the feature points respectively, so that two complete models can be obtained.

In this embodiment, by acquiring the feature points and the feature point position information of the feature points corresponding to each component of the target object, the connection relationship between the components in the component set is determined, so that the connection relationship between the components of the target object can be accurately determined, and the accuracy of determining the pose information of the target object is higher.

Referring to fig. 4, fig. 4 shows a flowchart of a pose determining method according to still another embodiment of the present application, for an electronic device, the method includes:

S401, acquiring a plurality of components corresponding to the target object and feature points corresponding to each component.

The description of S401 refers to the description of S101 above, and is not repeated here.

S402, acquiring a depth image and a color image of a target object; processing the color image corresponding to the target model through the component mask corresponding to each component to obtain the color image of each component; and processing the depth image corresponding to the target model through the component mask corresponding to each component to obtain the depth image of each component.

A color image refers to an image in which each pixel is composed of R, G, B components, wherein R, G, B is described by different gray levels, corresponding to the three primary colors of human vision, namely red, green, and blue. The depth image (DEPTH IMAGES), also called a range image (RANGE IMAGES), refers to an image that uses the distance (depth) values of points in the scene acquired by the image acquisition unit as pixel values, and directly reflects the geometry of the visible surface of the scene.

A mask refers to a region or process that uses a selected image, graphic, or object to block (in whole or in part) a processed image to control image processing. And (3) obtaining a color image and a depth image of the whole target object, carrying out Mask processing on the target object to obtain the color image and the depth image of each component of the target object, and generating a Mask image with the same size as the depth image and the color image for each component, wherein only the pixel of the component to be obtained is 1, and the rest is 0, so that the depth image and the color image corresponding to the corresponding component can be obtained.

For example, taking a doll with movable joints as an example, a color image of the doll head is acquired, a region except the head of the color image of the doll needs to be shielded, the color image of the doll head is obtained, and the depth image of each part of the target object is acquired similarly.

S403, inputting the color image and the depth image of each component into a pose estimation model to obtain the component pose information of each component.

As one implementation mode, the pose estimation model can be obtained by training based on a deep learning algorithm, the pose estimation model can also be obtained by training based on a template matching algorithm, and the specific model can be selected according to different sensors used.

In some embodiments, component pose is determined based on a deep learning network FFB6D, FFB6D is a full-flow bi-directional fusion network, FFB6D network requires color images and depth images as input, a mask map and a real pose R _i,T_i of the component during training, where R represents a rotation matrix, T represents a translation vector, and i represents an ith component of the sample object. Since the real data acquisition process is relatively time consuming, a portion of the simulated data may be added to the training. For a real dataset, the color image may be acquired by a common camera; the depth map can be obtained by a depth camera, and three main types of depth camera schemes exist in the current mainstream: structured light, binocular vision, time of flight (TOF) light; the point cloud information of the model can be obtained through the depth map, and then the model can be manually matched with the three-dimensional model to obtain the pose R _i,T_i of the component. The simulation dataset may directly use software to generate the above information.

Firstly, key points are selected on a three-dimensional model of a sample object, wherein the key points can be selected by adopting a FPS (Farthest Point Sampling) furthest point sampling method, the principle of the FPS is that the furthest points from the existing sampling point set are selected continuously and iteratively, all points in a space can be covered as far as possible, the number of the selected key points is not less than 3, and at least 3 points are not collinear. Secondly, training a model, respectively sending a color image and a depth image/point cloud into a CNN and a point cloud network by FFB6D to extract features, fusing the two features to obtain features of each point, sending the features to a segmentation module and a key point estimation module, wherein the segmentation module can obtain the position of each part of the model in the image, and the key point estimation module can give an estimation of the three-dimensional position of each part relative to the part. From this we can get an estimate of the location of the keypoints on the component. And finally, obtaining pose information of the component through a one-to-one correspondence relationship between the key points estimated by the deep learning model and the key points selected on the three-dimensional model.

The point cloud is a data set, data acquisition is carried out on a target object according to a measuring instrument in 3D engineering, a massive point set of target surface characteristics can be obtained, each point contains X, Y, Z geometric coordinates, intensity values, classification values and other information, the points are combined together to form a point cloud, and the point cloud can restore the three-dimensional effect of the target object more truly, so that visualization is realized. The CNN is a convolutional neural network, and the basic structure of the CNN comprises two layers, namely a feature extraction layer and a feature mapping layer.

S404, obtaining feature point relative position information of feature points of each component, wherein the feature point relative position information of the feature points of each component refers to the relative position information between the feature points and the component, and determining the feature point position information of the feature points of each component according to the component pose information of each component and the feature point relative position information of the feature points of each component.

As an implementation manner, component pose information of the components in the world coordinate system is obtained, the component pose information can be R _i,T_i, for a variable model, although relative pose between the components is not fixed, position information of the feature points relative to the components where the components are located is not changed, and accurate feature point position information of the feature points, namely the positions of the feature points in the world coordinate system, can be obtained based on the component pose information of the components and the feature point relative position information of the feature points.

S405, determining at least one component set according to the feature point position information of the feature point corresponding to each component, wherein feature points matched with each other exist among the components in each component set;

s406, determining the pose of the target object according to the component pose information of each component in each component set.

The descriptions of S405 to S406 refer to the descriptions of S102 to S103 above, and are not repeated here.

In this embodiment, component pose information of the components is determined based on the pose estimation model, feature point position information of the feature points is determined based on the acquired component pose information of each component of the target object and feature point relative position information of the feature points, and whether the component where the feature points are located is a component set is determined by comparing the feature point position information of the feature points, so that whether the feature points are matched with each other is determined more clearly, the relationship between the components can be clearly determined, and the pose of the target object can be accurately acquired.

Referring to fig. 5, fig. 5 is a block diagram illustrating a pose determining apparatus according to an embodiment of the present application. For an electronic device, the apparatus 500 comprises:

An obtaining module 501, configured to obtain a plurality of components corresponding to a target object, feature points corresponding to each component, and feature point position information corresponding to the feature points of each component;

A first determining module 502, configured to determine at least one component set according to feature point location information of feature points corresponding to each component, where feature points matched with each other exist between components in each component set;

A second determining module 503, configured to determine the pose of the target object according to the component pose information of each component in each component set.

Optionally, the first determining module 502 is further configured to obtain any two components as a component set if feature points with feature point position information matched with each other exist in any two components.

Optionally, the obtaining module 501 is further configured to obtain component pose information of each component and feature point relative position information of a feature point of each component, where the feature point relative position information of the feature point of each component refers to relative position information between the feature point and the component; and determining the feature point position information of the feature point of each component according to the component pose information of each component and the feature point relative position information of the feature point of each component.

Optionally, the acquiring module 501 is further configured to acquire a color image and a depth image of each component; and inputting the color image and the depth image of each component into a pose estimation model to obtain the component pose information of each component.

Optionally, the acquiring module 501 is further configured to acquire a depth image and a color image of the target object; processing the color image corresponding to the target model through the component mask corresponding to each component to obtain the color image of each component; and processing the depth image corresponding to the target model through the component mask corresponding to each component to obtain the depth image of each component.

Optionally, the second determining module 503 is further configured to determine a connection relationship of the components in each component set according to the feature point position information of the feature points matched with each other in each component set; and determining the pose of the target object according to the connection relation of the components in each component set.

Optionally, the obtaining module 501 is further configured to obtain, if any two components have a connection relationship, a point connecting any two components as a feature point of any two components.

It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process of the apparatus and modules described above may refer to the corresponding process in the foregoing method embodiment, which is not repeated herein.

In addition, each function in each embodiment of the present application may be integrated into one processing module, each module may exist alone physically, or two or more modules may be integrated into one module. The integrated modules may be implemented in hardware or in software functional modules.

Referring to fig. 6, a block diagram of an electronic device according to an embodiment of the application is shown. The electronic device 600 may be a smart phone, tablet, electronic book, or other electronic device capable of running applications. The electronic device 600 of the present application may include one or more of the following components: a processor 610, a memory 620, and one or more applications. Wherein one or more application programs may be stored in the memory 620 and configured to be executed by the one or more processors 610, the one or more program(s) configured to perform the methods as described in the foregoing method embodiments.

Processor 610 may include one or more processing cores. The processor 610 utilizes various interfaces and lines to connect various portions of the overall electronic device 600, perform various functions of the electronic device 600, and process data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 620, and invoking data stored in the memory 620. Alternatively, the processor 610 may be implemented in at least one hardware form of digital signal Processing (DIGITAL SIGNAL Processing, DSP), field-Programmable gate array (Field-Programmable GATE ARRAY, FPGA), programmable logic array (Programmable Logic Array, PLA). The processor 610 may integrate one or a combination of several of a central processing unit (Central Processing Unit, CPU), an image processor (GraphicsProcessing Unit, GPU), a modem, etc. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for being responsible for rendering and drawing of display content; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the processor 610 and may be implemented solely by a single communication chip.

Memory 620 may include random access Memory (Random Access Memory, RAM) or Read-Only Memory (ROM). Memory 620 may be used to store instructions, programs, code sets, or instruction sets. The memory 620 may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the various method embodiments described below, etc. The storage data area may also store data created by the electronic device 600 in use (e.g., phonebook, audiovisual data, chat log data), and the like.

Referring to fig. 7, fig. 7 shows a block diagram of a computer-readable storage medium according to an embodiment of the present application. The computer readable storage medium 700 has stored therein program code that can be invoked by a processor to perform the methods described in the method embodiments described above.

The computer readable storage medium 700 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read only memory), an EPROM, a hard disk, or a ROM. Optionally, computer readable storage medium 700 comprises a non-volatile computer readable storage medium (non-transitorycomputer-readable storage medium). The computer readable storage medium 700 has memory space for program code 710 that performs any of the method steps described above. The program code can be read from or written to one or more computer program products. Program code 710 may be compressed, for example, in a suitable form.

In summary, in the pose determining method, the pose determining device, the electronic device and the computer readable storage medium provided by the present application, in this embodiment, by acquiring pose information of each component of a target object, component pose information of each component, and feature point position information of each corresponding feature point on each component, a relationship between each component of the target object can be determined, and when a plurality of identical components exist in a scene, the relationship between each component and the target object can be accurately determined.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting; although the application has been described in detail with reference to the foregoing embodiments, it will be appreciated by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not drive the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims (10)

1. A pose determination method, the method comprising:

Acquiring a plurality of components corresponding to a target object, feature points corresponding to each component and feature point position information corresponding to the feature points of each component;

Determining at least one component set according to the feature point position information of the feature point corresponding to each component, wherein feature points matched with each other exist among the components in each component set;

and determining the pose of the target object according to the component pose information of each component in each component set.

2. The pose determination method according to claim 1, wherein the component set acquisition method comprises:

if feature points with the position information of the feature points matched with each other exist between any two parts, the any two parts are obtained to be used as a part set.

3. The pose determination method according to claim 1, wherein the feature point position information acquisition method of the feature point of each of the parts includes:

Acquiring component pose information of each component and feature point relative position information of feature points of each component, wherein the feature point relative position information of the feature points of each component refers to the relative position information between the feature points and the component;

And determining the feature point position information of the feature points of each component according to the component pose information of each component and the feature point relative position information of the feature points of each component.

4. A method according to claim 3, wherein said obtaining component pose information for each of said components comprises:

acquiring a color image and a depth image of each component;

And inputting the color image and the depth image of each component into a pose estimation model to obtain the component pose information of each component.

5. The method of claim 4, wherein said acquiring a color image and a depth image of each of said components comprises:

Acquiring a depth image and a color image of the target object;

Processing the color image corresponding to the target model through a component mask corresponding to each component to obtain a color image of each component;

And processing the depth image corresponding to the target model through the component mask corresponding to each component to obtain the depth image of each component.

6. The method of claim 1, wherein determining the pose of the target object based on the component pose information for each component in each set of components comprises:

Determining the connection relation of the components in each component set according to the feature point position information of the feature points matched with each other in each component set;

and determining the pose of the target object according to the connection relation of the components in each component set.

7. The method according to claim 1, wherein the feature point acquisition method includes:

If any two parts have a connection relationship, a point connecting the any two parts is obtained as a characteristic point of the any two parts.

8. A pose determination apparatus for an electronic device, the apparatus comprising:

the device comprises an acquisition module, a storage module and a storage module, wherein the acquisition module is used for acquiring a plurality of components corresponding to a target object, characteristic points corresponding to each component and characteristic point position information corresponding to the characteristic points of each component;

The first determining module is used for determining at least one component set according to the feature point position information of the feature point corresponding to each component, and feature points matched with each other exist among the components in each component set;

And the second determining module is used for determining the pose of the target object according to the component pose information of each component in each component set.

9. An electronic device, comprising:

One or more processors;

A memory;

One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to perform the method of any of claims 1-7.

10. A computer readable storage medium, characterized in that the computer readable storage medium stores a program code executable by a processor, which program code, when executed by the processor, causes the processor to perform the method of any of claims 1-7.