KR20210025324A - Device and method for generating animation using deep neural network - Google Patents

Abstract

According to an embodiment of the present invention, provided is an animation generation device which comprises at least one processor. The at least one processor receives location information of two hands and two feet at a starting point and a target point to generate each path through which the two hands, the two feet, and the center of a body move from the starting point to the target point during first time using a path network, obtains the location information of the two hands and the two feet at a second time interval in each of the generated paths, and generates a pose using a pose network based on the location information of the two hands and the two feet in each of the generated paths.

Description

Animation generation device and animation generation method using a deep neural network {DEVICE AND METHOD FOR GENERATING ANIMATION USING DEEP NEURAL NETWORK}

The following embodiments relate to an animation generating apparatus and an animation generating method using a deep neural network.

Traditionally, animation is a film that makes a picture or object appear to move through frame shooting (the way a movie camera does not automatically rotate 24 frames per second like a normal movie, but takes a picture one by one like shooting with a normal camera). It has been known as'genre'. However, animation covers a huge wide range of miscellaneous contents, from cartoons to abstract art, from paper and cell techniques to sand or clay techniques, from puppet films to special effects films made with computer support. Has a category.

Machine learning is a field of artificial intelligence, a field that has evolved from the study of pattern recognition and computer learning theory, and refers to the field of developing algorithms and technologies that enable computers to learn.

The core of machine learning lies in representation and generalization. Representation is the evaluation of data, and generalization is the processing of data that is still unknown. It is also a field of computational learning theory.

Deep learning is defined as a set of machine learning algorithms that attempt a high level of abstraction through a combination of several nonlinear transducers. It can be said that it is a field.

According to an embodiment of the present invention, it is possible to provide an animation generation apparatus and an animation generation method using a deep neural network capable of generating animation based on a path network for generating a path and a pose network for generating a pose.

In addition, according to another embodiment of the present invention, it is possible to provide an animation generating apparatus and an animation generating method using a deep neural network capable of generating a pose in which the positions of hands and feet are the same, but the positions of the body are different.

In addition, according to another embodiment of the present invention, it is possible to provide an animation generation apparatus and an animation generation method using a deep neural network capable of reducing noise by curve fitting a path generated by a path network.

In addition, according to another embodiment of the present invention, an animation generating apparatus and an animation generating method using a deep neural network capable of generating a time-independent pose may be provided.

In addition, according to another embodiment of the present invention, it is possible to provide an animation generating apparatus and an animation generating method using a deep neural network capable of generating fast and accurate motion in real time.

In addition, according to another embodiment of the present invention, it is possible to provide an animation generation apparatus and an animation generation method using a deep neural network capable of generating an animation that can move while accurately holding a specific point with a hand or stepping on a foot.

In addition, according to another embodiment of the present invention, an animation generating apparatus and an animation generating method using a deep neural network that can be changed to be able to compute at once using deep learning an existing numerical analysis method that repeatedly reduces errors. I can.

According to an embodiment of the present invention, an animation generating apparatus using a deep neural network, comprising at least one processor, wherein the at least one processor provides position information of two hands and two feet at a starting point and a target point. Receive input and generate a path for each of the two hands, the feet, and the center of the body to move from the start point to the target point for a first time using a path network, and a second time interval in each of the generated paths As a result, the position information of the two hands and the two feet is obtained, and a pose is generated using a pose network based on the obtained position information of the two hands and the two feet in each of the generated paths.

In addition, the at least one processor may generate an animation by combining the generated pose.

In addition, the location information of the two hands and the two feet may be relative coordinates of the two hands and the two feet with respect to the center of the body.

Also, the second time may be a time obtained by dividing 1 second by a preset number of frames per second of an animation.

In addition, the at least one processor may generate the pose of a number corresponding to a number obtained by multiplying the reciprocal of the second time and the first time.

In addition, the at least one processor may curve fitting each of the generated paths into a smooth curve.

According to another embodiment of the present invention, the center of the two hands, the two feet, and the body is the starting point for a first time using a route network by receiving position information of both hands and feet at a starting point and a target point. The operation of generating each path to move to the target point at and acquiring the location information of the two hands and the two feet at a second time interval in each of the generated paths, and in the obtained respective paths And generating an animation for generating a pose using a pose network based on the location information of the two hands and the two feet.

In addition, the animation generating method may further include an operation of generating an animation for generating an animation by combining the generated pose.

In addition, the location information of the two hands and the two feet may be relative coordinates of the two hands and the two feet with respect to the center of the body.

Also, the second time may be a time obtained by dividing 1 second by a preset number of frames per second of an animation.

In addition, the animation generating method may generate the number of poses corresponding to a number obtained by multiplying the reciprocal of the second time and the first time.

In addition, the animation generation method may further include an operation of curve fitting each of the generated paths into a smooth curve.

According to an embodiment of the present invention, there is an effect of generating an animation based on a path network for generating a path and a pose network for generating a pose.

In addition, there is an effect of generating a pose where the positions of the hands and feet are the same, but the positions of the body are different.

In addition, there is an effect of reducing noise by curve fitting the path generated by the path network.

In addition, there is an effect of generating a time-independent pose.

In addition, there is an effect of generating fast and accurate animation in real time.

In addition, there is an effect of generating an animation that can move while holding a specific point with your hand or stepping on it.

In addition, there is an effect of being able to change the existing numerical analysis method that repeatedly reduces errors so that it can be calculated at once using deep learning.

1 is a diagram illustrating a configuration of an animation generating apparatus according to an exemplary embodiment.
2 is a flow chart illustrating a method of generating an animation according to an exemplary embodiment.
3 is a diagram illustrating a state in which an animation is generated according to an exemplary embodiment.

Specific structural or functional descriptions of the embodiments according to the concept of the present invention disclosed in the present specification are exemplified only for the purpose of describing the embodiments according to the concept of the present invention, and the embodiments according to the concept of the present invention are It may be implemented in various forms and is not limited to the embodiments described herein.

Since the embodiments according to the concept of the present invention can apply various changes and have various forms, the embodiments will be illustrated in the drawings and described in detail in the present specification. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes all changes, equivalents, or substitutes included in the spirit and scope of the present invention.

Terms such as first or second may be used to describe various constituent elements, but the constituent elements should not be limited by the terms. The terms are only for the purpose of distinguishing one component from other components, for example, without departing from the scope of the rights according to the concept of the present invention, the first component may be referred to as the second component, and similarly The second component may also be referred to as a first component.

When a component is referred to as being "connected" or "connected" to another component, it is understood that it may be directly connected or connected to the other component, but other components may exist in the middle. It should be. On the other hand, when a component is referred to as being "directly connected" or "directly connected" to another component, it should be understood that there is no other component in the middle. Other expressions describing the relationship between components, such as "between" and "directly between" or "adjacent to" and "directly adjacent to" should be interpreted as well.

The terms used in the present specification are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise.

In the present specification, terms such as "comprise" or "have" are intended to designate that the described feature, number, step, action, component, part, or combination thereof is present, but one or more other features or numbers It is to be understood that the possibility of addition or presence of, steps, actions, components, parts, or combinations thereof is not preliminarily excluded.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

Terms as defined in a commonly used dictionary should be construed as having a meaning consistent with the meaning of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in the present specification. Does not.

In the following description, the same identification symbols mean the same configuration, and unnecessary redundant descriptions and descriptions of known technologies will be omitted.

In an embodiment of the present invention,'communication','communication network', and'network' may be used with the same meaning. The three terms refer to wired/wireless local and wide area data transmission/reception networks capable of transmitting and receiving files between a user terminal, a terminal of other users, and a download server.

Hereinafter, the present invention will be described in detail by describing a preferred embodiment of the present invention with reference to the accompanying drawings.

1 is a diagram illustrating a configuration of an animation generating apparatus according to an exemplary embodiment.

Referring to FIG. 1, the animation generating apparatus 100 includes a processor 110, an input/output interface module 120, and a memory 130.

The processor 110, the input/output interface module 120, and the memory 130 constituting the animation generating apparatus 100 are interconnected and can transmit data to each other.

The processor 110 may execute programs or instructions stored in the memory 130. In this case, an operation program (eg, OS) for operating the animation generating apparatus 100 may be stored in the memory 130.

The processor 110 may execute a program for managing information on the animation generating apparatus 100.

The processor 110 may execute a program for managing the operation of the animation generating apparatus 100.

The processor 110 may execute a program for managing the operation of the input/output interface module 120.

Ⅰ) Trajectory creation

The processor 110 may obtain input data through the input/output interface module 120. In this case, the input data may be position information of two hands and two feet at a starting point and a target point, but the input data is not limited thereto.

The processor 110 may generate a path network (eg, a deep neural network (DNN)) that generates continuous animation information.

The processor 110 may generate an optimal path from the start point to the target point using the generated path network (eg, a deep neural network (DNN)), but the processor 110 generates the path. How to do this is not limited thereto. In this case, the optimal path may be a natural path according to the terrain from the start point to the target point, but the optimal path is not limited thereto.

The processor 110 generates two paths that the two hands will move from the start point to the target point for a first time using a route network (eg, a deep neural network (DNN)) that generates continuous animation information. can do. In this case, the first time may vary depending on the location of the starting point and the target point, the topography of the starting point and the target point, and a moving distance of two hands.

The processor 110 uses a path network (e.g., a deep neural network (DNN)) that generates continuous animation information to generate two paths that the feet will move from the start point to the target point for the first time. I can. In this case, the first time may vary depending on the location of the starting point and the target point, the topography of the starting point and the target point, and the moving distance of two feet.

The processor 110 uses a path network (e.g., a deep neural network (DNN)) that generates continuous animation information to determine one path in which the center of the body moves from the start point to the target point for the first time. Can be generated. In this case, the first time may vary depending on the location of the starting point and the target point, the topography of the starting point and the target point, and a moving distance of the center of the body.

The processor 110 may generate a path in consideration of the terrain from the start point to the target point using a path network (eg, a deep neural network (DNN)) that generates continuous animation information.

The processor 110 uses a path network (eg, a deep neural network (DNN)) that generates continuous animation information to select a specific point to be held by hand at least one of the two hands. You can create a path through this.

The processor 110 is at least one of the two feet at a specific point where the animation generated using a path network (eg, a deep neural network (DNN)) that generates continuous animation information You can create a path to pass through.

The processor 110 may receive data and generate training data for training a path network (eg, a deep neural network (DNN)). In this case, the data may include positions of two hands and feet at a starting point and a target point, paths of both hands and feet, and animation information. The processor 110 may train the path network (eg, a deep neural network (DNN)) by inputting the generated training data. In this case, the learning data may be data consisting of a set (eg, a set consisting of the positions of two hands and feet, a path of two hands and two feet, and an animation), but the form of the learning data is limited thereto. It is not.

The processor 110 may receive training data for training a path network (eg, a deep neural network (DNN)). In this case, the learning data may be data consisting of a set (eg, a set consisting of the positions of two hands and feet, a path of two hands and two feet, and an animation), but the form of the learning data is limited thereto. It is not.

The processor 110 may train the path network (eg, a deep neural network (DNN)) by inputting the received training data.

가 상기 학습 데이터의 두 손과 두 발의 경로이고,

가 경로 네트워크가 생성한 결과인 경우, 하기 [수학식 1]의 에러 값을 줄이도록 경로 네트워크를 학습시킬 수 있다.The processor 110 is

Is the path of both hands and feet of the training data,

If is the result of generating the path network, the path network may be trained to reduce the error value of [Equation 1] below.

는 뉴럴 네트워크 파라미터

의 L2 레귤레이션이고,

은 실험으로 결정된 값이다.here,

Is the neural network parameter

Is the L2 regulation of,

Is the value determined by experiment.

The path network trained by the processor 110 is a deep neural that calculates the path desired by the user at once using deep learning rather than the conventional numerical analysis method that repeatedly reduces the error between the path desired by the user and the path generated by the device. It can be a network.

Ii) Curve fitting

The processor 110 may curve-fit each path in the form of a curved line with noise generated into a smooth curve.

The processor 110 may transform each path in order to curve fitting each path in the form of a curved line with generated noise into a smooth curve.

Iii) Create a pose

The processor 110 may acquire location information (eg, X, Y, Z coordinate information) of the two hands and the two feet at a second time interval in each generated path. In this case, the second time may be a time (eg, 1/30 second) obtained by dividing 1 second by a preset number of frames per second (eg, 30 frames per second) of the animation, but the second time is not limited thereto. In addition, the location information of the two hands and the two feet may be relative coordinates of the two hands and the two feet with respect to the center of the body.

The processor 110 may generate a pose network (eg, a deep neural network (DNN)) that generates a pose.

The processor 110 uses the generated pose network (e.g., a deep neural network (DNN)) to satisfy the position information of both hands and feet in each of the generated paths. The pose may be generated, but the method of generating the pose by the processor 110 is not limited thereto.

The processor 110 may generate a pose from the starting point of the generated path.

The processor 110 may generate a pose to the target point of the generated path.

The processor 110 may continuously operate a pause network (eg, a deep neural network (DNN)) by a number corresponding to a number obtained by multiplying the reciprocal of the second time and the first time. In this case, the second time may be a time (eg, 1/30 second) obtained by dividing 1 second by a preset number of frames per second (eg, 30 frames per second) of the animation, but the second time is not limited thereto.

The processor 110 may generate a number of poses corresponding to a number obtained by multiplying the reciprocal of the second time and the first time using a pose network (eg, a deep neural network (DNN)). In this case, the second time may be a time (eg, 1/30 second) obtained by dividing 1 second by a preset number of frames per second (eg, 30 frames per second) of the animation, but the second time is not limited thereto.

The processor 110 may receive data and generate training data for training a pose network (eg, a deep neural network (DNN)). In this case, the data may include positions of two hands and feet at a starting point and a target point, paths of both hands and feet, and animation information.

In order to consider a case where the positions of the two hands and the two feet are the same, but the poses are different, the processor 110 is It is possible to generate training data including location information and location information of both hands and feet in each path acquired at the set point in time.

For example, in order to consider a case in which the positions of the two hands and the two feet are the same, but the poses are different, the processor 110 has two hands in each path acquired at a preset time point and two previous time points having a second time interval. And position information of two feet, position information of two hands and feet in each path obtained at two viewpoints after having the second time interval at the preset time point, and each route obtained at the preset time point It is possible to generate training data that includes the location information of both hands and feet in.

However, generating the learning data is only an example, and the present invention is not limited or limited by the example.

The processor 110 may train the pose network (eg, a deep neural network (DNN)) by inputting the generated training data. In this case, the learning data may be data consisting of a set (eg, a set consisting of the positions of two hands and feet, a path of two hands and two feet, and an animation), but the form of the learning data is limited thereto. It is not.

The processor 110 may receive training data for training a pose network (eg, a deep neural network (DNN)). In this case, the learning data may be data consisting of a set (eg, a set consisting of the positions of two hands and feet, a path of two hands and two feet, and an animation), but the form of the learning data is limited thereto. It is not.

The processor 110 may train the pose network (eg, a deep neural network (DNN)) by inputting the received training data.

는 생성된 포즈이고,

가 그라운드 트루스(ground truth)인 경우, 하기 [수학식 2]의 에러 값을 줄이도록 포즈 네트워크를 학습시킬 수 있다.The processor 110 is

Is the created pose,

When is a ground truth, the pose network may be trained to reduce the error value of [Equation 2] below.

The pose network trained by the processor 110 is a deep neural that calculates the pose desired by the user at once using deep learning rather than the conventional numerical analysis method that repeatedly reduces the error between the pose desired by the user and the pose generated by the device. It can be a network.

Iv) Animation creation

The processor 110 may generate an animation by sequentially combining the generated poses.

The processor 110 may match the generated pose and the frame constituting the animation on a one-to-one basis, and may generate the animation by combining the frame and the matched pose.

The input/output interface module 120 may be connected to an external device (eg, a server) through a network.

The input/output interface module 120 may acquire input data.

The input/output interface module 120 may acquire data for generating training data.

The input/output interface module 120 may acquire learning data.

The input/output interface module 120 may obtain a user's input.

The input/output interface module 120 may display the generated animation.

The input/output interface module 120 may be provided integrally with the animation generating device 100.

The input/output interface module 120 may be provided separately from the animation generating apparatus 100.

The input/output interface module 120 may be a separate device to be communicatively connected with the animation generating device 100.

The input/output interface module 120 may include a port (eg, a USB port) for connecting to an external device.

The input/output interface module 120 may include a monitor, a touch screen, a mouse, an electronic pen, a microphone, a keyboard, a speaker, an earphone, a headphone, or a touch pad.

The memory 130 may store input data acquired through the input/output interface module 120.

The memory 130 may store a path generated by the processor 110 using a path network.

The memory 130 may store data acquired through the input/output interface module 120 to generate learning data.

The memory 130 may store training data generated by the processor 110.

The memory 130 may store the received training data.

The memory 130 may store a pose generated by the processor 110 using a pose network.

The memory 130 may store an animation generated by the processor 110.

The memory 130 may store a route network generated by the processor 110.

The memory 130 may store the pause network generated by the processor 110.

The term'module' used herein denotes a logical structural unit, and it is obvious to those skilled in the art that the present invention is not necessarily a physically classified component.

2 is a flow chart illustrating a method of generating an animation according to an exemplary embodiment.

2, the animation generating apparatus receives position information of two hands and two feet at a starting point and a target point, and the center of the two hands, the two feet, and the body is the target at the starting point for a first time. Each path to be moved to the point is generated (200).

At this time, the animation generating device generates a natural path according to the terrain from the start point to the target point at once using a route network (e.g., a deep neural network (DNN)) that generates continuous animation information. I can.

The animation generating device curve fitting each path into a smooth curve (210).

In this case, the animation generating apparatus may transform each path in order to curve fitting each path in a curved shape with noise into a smooth curve.

Acquires the position information of the two hands and the two feet at a second time interval in each path generated by the animation generating device, and based on the obtained position information of the two hands and the two feet in each of the generated paths Create a pose with a (220).

In this case, the animation generating apparatus may generate a pose desired by the user at once using a pose network (eg, a deep neural network (DNN)) that generates a pose.

Also, the location information of the two hands and the two feet may be relative coordinates of the two hands and the two feet with respect to the center of the body.

The animation generating device combines the generated pose to generate an animation (230).

In this case, the animation generating apparatus may match the generated pose with a frame constituting the animation on a one-to-one basis, and may generate an animation by combining the frame and the matched pose.

3 is a diagram illustrating a state in which an animation is generated according to an exemplary embodiment.

Referring to FIG. 3, the apparatus for generating animation may receive position information 300 of two hands and feet at a starting point and a target point. In this case, the location information may be X, Y, and Z coordinates of eight points at which two hands and two feet are located at a starting point and a target point, but the location information is not limited thereto.

The animation generating apparatus may generate the path 320 using the path network 310 that generates continuous animation information.

The animation generating apparatus includes two paths (321, 322) that the two hands will move from the start point to the target point for a first time (for example, 2 seconds), and the two feet from the start point to the target point are the first time. (E.g., 2 seconds) two paths to move (321, 322) and the center of the body from the start point to the target point during the first time (e.g., 2 seconds) to create one path (352) I can. In this case, when the number of frames per second of the preset animation is 30 frames, each of the five generated paths may be expressed as X, Y, and Z coordinates of 60 frames.

The animation generating apparatus may curve-fit 330 each path in which noise exists into a smooth curve in order to generate an animation having a natural motion.

The animation generating apparatus includes position information (351, 352) of the two hands and position information (353) of the two feet in the same time period in the five paths (341, 342, 343, 344, 345) where the curve fitting is completed (340). 354). In this case, the position information 351 and 352 of the two hands and the position information 353 and 354 of the two feet may be relative coordinates of the two hands and the two feet with respect to the center 355 of the body in the same time period.

The animation generating apparatus may acquire position information 351 and 352 of the two hands and position information 353 and 354 of the two feet in each path at intervals of a second time (eg, 1/30 second).

The animation generating device generates a pose 370 desired by the user based on the obtained position information (351, 352) of the two hands and the position information (353, 354) of the two hands in each of the generated paths. can do.

The animation generating apparatus may generate the pose 370 using the pose network 360 that generates the pose.

The animation generating apparatus includes position information (351, 352) of the two hands and position information (353, 354) of the two hands in each of the acquired paths at intervals of a second time (for example, 1/30 second). One pose 370 corresponding to may be generated. In this case, the generated one pose may be expressed as 23 joint information and rotation information of 9 bones.

The animation generating apparatus may operate the pose network 360 by a number (e.g., 60) corresponding to the number (e.g., 60) multiplied by the reciprocal of the second time (e.g., 1/30 second) and the first time (e.g., 2 seconds). .

The animation generating apparatus may generate a pose of a number (eg, 60) corresponding to a number obtained by multiplying the reciprocal of the second time (eg, 1/30 second) and the first time (eg, 2 seconds).

The animation generating apparatus may match the generated pose and the frame constituting the animation on a one-to-one basis, and may generate the animation 380 by combining the frame and the matched pose.

In the above, even if all the constituent elements constituting the embodiments of the present invention have been described as being combined into one or operating in combination, the present invention is not necessarily limited to these embodiments. That is, within the scope of the object of the present invention, all the constituent elements may be selectively combined into at least one and operated.

In addition, although all of the components may be implemented as one independent hardware, a program module that performs some or all functions combined in one or more hardware by selectively combining some or all of the components. It may be implemented as a computer program having Codes and code segments constituting the computer program may be easily inferred by those skilled in the art of the present invention.

Such a computer program is stored in a computer-readable storage medium, and is read and executed by a computer, thereby implementing an embodiment of the present invention. The storage medium of the computer program may include a magnetic recording medium, an optical recording medium, and the like.

In addition, the terms such as "include", "consist of" or "have" described above mean that the corresponding component may be included unless otherwise stated, excluding other components. It should not be construed as being able to further include other components.

All terms, including technical or scientific terms, unless otherwise defined, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms generally used, such as terms defined in the dictionary, should be interpreted as being consistent with the meaning in the context of the related technology, and are not interpreted as ideal or excessively formal meanings unless explicitly defined in the present invention.

The methods disclosed in the present invention comprise one or more actions or steps for achieving the above-described method. Method actions and/or steps may be interchanged with each other without departing from the scope of the claims. In other words, unless a specific order for the actions or steps is specified, the order and/or use of specific actions and/or steps may be modified without departing from the scope of the claims.

As used herein, a phrase referring to “at least one of” in a list of items refers to any combination of these items, including single members. As an example, “at least one of a, b, or c:” means a, b, c, ab, ac, bc, and abc, as well as any combination with multiples of the same element (e.g., aa , aaa, aab, aac, abb, acc, bb, bbb, bbc, cc, and ccc or any other ordering of a, b, and c).

As used herein, the term "determining" encompasses a wide variety of actions. For example, “determining” may include computing, computing, processing, deriving, examining, looking up (eg, looking up in a table, database, or other data structure), identifying, and the like. . Further, “determining” may include receiving (eg, receiving information), accessing (accessing data in a memory), and the like. Also, “determining” may include resolving, choosing, choosing, establishing, and the like.

The above description is merely illustrative of the technical idea of the present invention, and those of ordinary skill in the art to which the present invention pertains will be able to make various modifications and variations without departing from the essential characteristics of the present invention.

Accordingly, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain the technical idea, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be construed by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

100... animation generating device

Claims (12)

In the animation generating device,
Including at least one processor,
The at least one processor,
Each path that the center of the two hands, the two feet, and the body will move from the starting point to the target point during the first time using a route network by receiving position information of both hands and feet at a starting point and a target point And create
A pose network is obtained based on the location information of the two hands and the two feet at a second time interval in each of the generated paths, and the obtained position information of the two hands and the two feet in each of the generated paths An animation generation device using a deep neural network that generates a pose using. The method of claim 1,
The at least one processor,
An animation generation apparatus using a deep neural network that combines the generated poses to generate animation. The method of claim 1,
The location information of the two hands and the two feet,
An animation generating apparatus using a deep neural network, which is the relative coordinates of the two hands and the two feet with respect to the center of the body. The method of claim 1,
The second time,
An animation generation device using a deep neural network, which is the time obtained by dividing 1 second by the number of frames per second of the preset animation. The method of claim 1,
The at least one processor,
An animation generating apparatus using a deep neural network that generates the poses of a number corresponding to a number obtained by multiplying the reciprocal of the second time and the first time. The method of claim 1,
The at least one processor,
An animation generation apparatus using a deep neural network for curve fitting each of the generated paths into a smooth curve. Each path that the center of the two hands, the two feet, and the body will move from the starting point to the target point during the first time using a route network by receiving position information of both hands and feet at a starting point and a target point Generating an operation; And
A pose network is obtained based on the location information of the two hands and the two feet at a second time interval in each of the generated paths, and the obtained position information of the two hands and the two feet in each of the generated paths Animation creation using a deep neural network that creates a pose using
Animation generation method comprising a. The method of claim 7,
The animation creation method,
An operation of generating an animation by combining the generated poses to generate an animation
Animation generation method using a deep neural network further comprising. The method of claim 7,
The location information of the two hands and the two feet,
An animation generation method using a deep neural network, which is the relative coordinates of the two hands and the two feet with respect to the center of the body. The method of claim 7,
The second time,
Animation creation method using a deep neural network, which is the time obtained by dividing 1 second by the number of frames per second of the preset animation. The method of claim 7,
The animation creation method,
An animation generation method using a deep neural network for generating the number of poses corresponding to a number obtained by multiplying the reciprocal of the second time and the first time. The method of claim 7,
The animation creation method,
Curve fitting operation of each of the generated paths with a smooth curve
Animation generation method using a deep neural network further comprising.

Images