KR102599219B1 - System for controlling avatar using artificial intelligence hand position recognition - Google Patents

Abstract

The present invention includes an image processing module that detects the user's hand motion and generates the user's hand motion data; an avatar operation module that operates a pre-generated avatar according to a control signal received from the outside; and a processor that generates motion information of the avatar corresponding to the user's hand motion data and operates the avatar based on the generated motion information of the avatar. , Because avatar control commands can be generated through simple calculations from still images without performing heavy artificial intelligence calculations such as the existing LSTM (Long Short Term Memory) deep learning method, it is possible to provide an avatar controller using hand position recognition. It works.

Description

Avatar control system using artificial intelligence hand position recognition {SYSTEM FOR CONTROLLING AVATAR USING ARTIFICIAL INTELLIGENCE HAND POSITION RECOGNITION}

The present invention relates to an avatar control system using artificial intelligence hand position recognition.

Recently, as augmented reality technology has developed, technologies that fuse the real world and the virtual world are being developed. For example, representative examples include a virtual education system, a screen golf system, and a virtual experience simulation system.

These virtual systems are configured to control the content in response to the motion when the user makes a motion while outputting a content image using a display device such as a touch board.

These virtual systems can recognize hands or faces using a tool called mediapipe. Additionally, these tools can recognize finger joints and the entire hand.

However, although it is easy to determine the position of the hand from a single still image, there are limitations in recognizing motion only based on the position of the hand, so artificial intelligence techniques such as LSTM are used to estimate posture.

However, performing posture estimation calculations in real time while displaying more than 60 frames per second on the screen had the problem of placing a large load on the system.

Registered Patent Publication No. 10-1189633 (Publication date: October 10, 2012)

The purpose of the embodiments of the present invention to improve the above-described problems is to provide an artificial intelligence hand that can generate avatar control commands through simple calculations from still images without performing heavy artificial intelligence calculations such as the existing Long Short Term Memory (LSTM) method. It provides an avatar control system using location recognition.

In order to achieve the above object, an avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention includes an image processing module that detects the user's hand motion and generates the user's hand motion data; an avatar operation module that operates a pre-generated avatar according to a control signal received from the outside; and a processor that generates motion information of an avatar corresponding to the user's hand motion data and operates the avatar based on the generated motion information of the avatar.

The image processing module includes an image generator that captures a user image and generates image frames divided into an X-axis (left and right sides of the screen), a Y-axis (top and bottom of the screen), and a Z-axis (depth of the screen); a hand position generator that generates coordinate values corresponding to the position of the user's hand in the video frame; and a motion data generator that calculates the moving direction of the coordinate values and generates preset user's hand motion data based on the calculated moving direction of the coordinate values.

The motion data generator calculates the movement direction using the coordinate values of the user's left hand, right hand, or left and right hands, but when only the user's left hand is moved left and right, the user's motion on the Data is generated, and when only the user's right hand is moved up and down, the user's motion data on the Z axis corresponding to the up and down movement information is generated, and when the user's left and right hands are moved up and down with the user's left and right hands together, up and down It is characterized by generating user motion data on the Y axis corresponding to movement information.

The motion data generator is characterized in that it determines that the user's left and right hands are together if the distance between the user's left and right hands is smaller than the size of the object representing the user's hands.

The motion data generator is characterized in that when the user's left and right hands are brought together and moved to a preset area, the user's motion data indicating a stopping motion is generated.

The avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention provides avatar control commands through simple calculations from still images without performing heavy artificial intelligence calculations such as the existing LSTM (Long Short Term Memory) deep learning method. Since it can be created, an avatar controller using hand position recognition can be provided.

Figure 1 is a block diagram schematically showing an avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention.
FIG. 2 is a diagram showing the configuration of an area for motion recognition in the image generating unit of the image processing module of FIG. 1.
FIG. 3 is a diagram for explaining an algorithm for determining whether the user's two hands are gathered in the motion data generation unit of the image processing module of FIG. 1.
FIG. 4 is a diagram defining user's hand motion data in the motion data generation unit of the image processing module of FIG. 1.
FIGS. 5A to 5D are diagrams illustrating examples of generating user's hand motion data in the motion data generation unit of the image processing module of FIG. 1 .
Figure 6 is a flowchart showing the operation of an avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention.

The present invention as described above will be described in detail through the attached drawings and examples.

It should be noted that the technical terms used in the present invention are only used to describe specific embodiments and are not intended to limit the present invention. In addition, the technical terms used in the present invention, unless specifically defined in a different sense in the present invention, should be interpreted as meanings generally understood by those skilled in the art in the technical field to which the present invention pertains, and are not overly comprehensive. It should not be interpreted in a literal or excessively reduced sense. Additionally, if the technical term used in the present invention is an incorrect technical term that does not accurately express the idea of the present invention, it should be replaced with a technical term that can be correctly understood by a person skilled in the art. In addition, general terms used in the present invention should be interpreted according to the definition in the dictionary or according to the context, and should not be interpreted in an excessively reduced sense.

Additionally, as used in the present invention, singular expressions include plural expressions unless the context clearly dictates otherwise. In the present invention, terms such as “consists of” or “comprises” should not be construed as necessarily including all of the various components or steps described in the invention, and some of the components or steps are included. It may not be possible, or it should be interpreted as including additional components or steps.

Additionally, terms containing ordinal numbers, such as first, second, etc., used in the present invention may be used to describe constituent elements, but the constituent elements should not be limited by the terms. Terms are used only to distinguish one component from another. For example, a first component may be named a second component without departing from the scope of the present invention, and similarly, the second component may also be named a first component.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the attached drawings. However, identical or similar components will be assigned the same reference numbers regardless of the reference numerals, and duplicate descriptions thereof will be omitted.

Additionally, when describing the present invention, if it is determined that a detailed description of related known technologies may obscure the gist of the present invention, the detailed description will be omitted. In addition, it should be noted that the attached drawings are only intended to facilitate easy understanding of the spirit of the present invention, and should not be construed as limiting the spirit of the present invention by the attached drawings.

Figure 1 is a block diagram schematically showing an avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention, and Figure 2 is a block diagram of an area for motion recognition in the image generation unit of the image processing module of Figure 1. It is a diagram showing the configuration, and FIG. 3 is a diagram for explaining an algorithm for determining whether the user's two hands are gathered in the motion data generation unit of the image processing module of FIG. 1, and FIG. 4 is a diagram of the image processing module of FIG. 1. This is a diagram defining the user's hand motion data in the motion data generator, and FIGS. 5A to 5D are diagrams showing an example of generating the user's hand motion data in the motion data generator of the image processing module of FIG. 1, and FIG. 6 is This is a flowchart showing the operation of an avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention.

As shown in Figure 1, the avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention is a system that can generate avatar control commands through simple calculations from still images, and for this purpose, an image processing module ( 10), and includes an avatar operation module 20 and a processor 30.

Although not shown, the avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention may further include an avatar creation module and an output module.

The avatar creation module creates a user's avatar in a virtual reality environment. In this specification, an avatar refers to a virtual character that can be controlled according to user commands in a virtual reality implemented in a virtual space through computer programming rather than in actual reality. The appearance of the avatar may be a human shape with a similar appearance to the user in order to improve the user's immersion and provide smooth biofeedback, but is not limited to a specific shape.

In one embodiment, an avatar may be automatically created based on the user's hand motion data. For example, a change in the user's hand motion can be detected in a still image and the avatar's posture or movement can be changed accordingly. In this way, appropriately changing the shape or attitude of the avatar in the virtual environment according to the user's hand motion data can improve the sense of presence and immersion in virtual reality.

An avatar created in a virtual reality environment performs actions corresponding to commands determined based on the user's hand motion data, as will be described later. Using this, the user can control the virtual reality avatar to perform a specific action by sending a control signal corresponding to the hand motion data even if the user does not actually perform the action, and can perform the evaluation protocol or training protocol provided by the program. there is.

The output module is an output device such as a TV, monitor, or HMD that outputs images of the avatar performing actions in real time. In one embodiment, the output module is a light emitting diode (LED), an organic light emitting device (OLED), a light emitting polymer (LEP), an electroluminescent device (EL), a field emission device (FED), or a polymer light emitting device (PLED). It can be configured with a display to which this is applied.

Additionally, the output module may be configured as a Head Mounted Display (HMD) device that can be worn on the head. HMD devices are next-generation display devices that can be worn on the user's head to view the screen through displays corresponding to both eyes. Typically, the user's head movement can be synchronized through rotation values, including IMU sensors. Through this, users can feel a greater sense of immersion than watching with a conventional monitor device in an Augmented Reality or Virtual Reality environment.

The image processing module 10 is a device that detects the user's hand motion and generates the user's hand motion data, and includes an image generator 11, a hand position generator 12, and a motion data generator 13. do.

The image generator 11 can use a normal camera, and as shown in FIG. 2, it is a device that captures a user and generates an image frame based on a still image, with X-axis (left and right sides of the screen), Y It is a device that generates video frames divided into the axis (top and bottom of the screen) and Z-axis (depth of the screen).

The hand position generator 12 is a device that generates coordinate values corresponding to the position of the user's hand among the image frames generated by the image generator 11. Specifically, the hand position generator 12 captures an image of the user's hand among an image frame composed of a still image and generates coordinate values (i.e., hand coordinates) corresponding to the motion of the hand. At this time, the coordinates of the hand include finger joints and wrist joints, allowing the overall size and finger length of the hand to be confirmed. In addition, the hand position generator 12 tracks the movement of the fingertips and the points of each joint to track the movement of the hand and generates coordinate values.

The hand position generator 12 includes a preprocessor (not shown) that defines the connection relationship between image frames transmitted from the image generator 11, and learning information learned in advance from preprocessed image frames. A hand recognition unit (not shown) that recognizes the hand, and a movement coordinate calculation unit (not shown) that performs coordinate adjustment for the recognized hand and calculates the movement coordinates of the hand from the displacement between video frames based on pre-learned learning information. (not shown).

Here, in recognizing the hand, it is desirable to first learn the hand, store the learned results, and then recognize the hand from this stored learning information. Meanwhile, along with the recognition of the hand, learning and recognition of the head, torso, arms, and legs are also carried out, and their recognition can be used to accurately determine the position of the hand.

In the hand position generator 12 configured as described above, when recognizing a hand and calculating coordinates, the presence or absence of the hand is confirmed at a previously predicted position using an estimation technique for estimating movement. In each video frame, the hand is estimated and searched through a hand movement estimation technique, and the movement coordinates of the searched hand are calculated. In other words, motion estimation is performed by extracting the hand motion vector component from the previous video frame and extracting the hand motion vector component from the current video frame, and thus the hand position in the next video frame can be estimated. Additionally, this estimation is made based on pre-learned learning information. If the hand does not exist at the estimated position, a radius is set around the estimated position and a search for the hand is performed within that radius. At this time, if the arm is searched, the position of the hand can be more easily estimated based on the structure of the human body.

The motion data generator 13 is a device that calculates the movement direction of the coordinate value generated by the hand position generator 12 and generates preset user's hand motion data based on the movement direction of the calculated coordinate value. am.

The motion data generator 13 receives the coordinate values generated by the hand position generator 12, processes changes in movement position, etc., and also generates an event (control) from the change in movement position, and generates an event (control) of the transmitted hand. It recognizes the hand shape from coordinates and generates the user's hand motion data.

That is, the motion data generator 13 receives the coordinate values generated by the hand position generator 12, processes them through the processor 30, and generates the user's corresponding command (i.e., avatar control signal). Determine the hand motion data. Related information such as analysis algorithms or look-up tables necessary for command decision may be stored in memory along with the hand position-based control program.

Additionally, the motion data generator 13 performs frequency filtering on the avatar control signal through the processor 30. If you want to remove high-frequency noise, you can use a low-pass filter, if you want to select a specific frequency area, you can use a band-pass filter, or if you want to remove a specific frequency area, you can use a low-pass filter. A bandstop filter can be used.

The motion data generator 13 extracts a feature vector based on the data power calculated including the logarithmic value and variance value for the corresponding frequency of the filtered signal. For example, the processor 30 can quantitatively analyze the proportion of signal components of a specific frequency in signal data based on fast Fourier transform (FFT).

In addition, by applying a signal processing function to the avatar control signal, feature vectors can be extracted by calculating mean, deviation, RMS (root mean square), skewness, kurtosis, DF (dominant frequency), etc., and the extracted feature vector data is It is used to determine the corresponding user command.

Meanwhile, when preprocessing of the avatar control signal is completed, the motion data generator 13 can generate a plurality of classifiers through a classifier generation module. The processor 30 may generate a classifier based on a plurality of classification algorithms. A plurality of classifiers classify the user's hand motion data and determine whether the data corresponds to a specific class.

Based on this, the motion data generator 13 selects an action corresponding to the real-time user's hand motion data from among a plurality of actions and finally determines the user's command. Specifically, the processor 30 may calculate each movement of the input real-time user's hand motion data as an output value such as probability or score based on the classifier, and the processor 30 may calculate the output value such as probability or score. The command can be determined by selecting an action with .

Depending on the embodiment, various types of commands may be stored in advance to correspond to the user's hand motion data. For example, in addition to commands that simply move the avatar forward, backward, left, and right, commands such as sitting/standing, or joint-related movements such as lifting a foot up or down, or bending or straightening a knee, can also be specified as commands. The present invention can simultaneously acquire various types of user's hand position signals and match them with detailed operation commands to implement various natural types of movements in virtual reality. The signal processing algorithm for identifying the user's intention for each command and reducing the malfunction rate is as described above.

To explain again, the motion data generator 13 defines the user's operation command according to the user's hand position, as shown in FIG. 4, and uses the user's left hand (L), right hand (R), or left hand ( The direction of movement is calculated using the coordinate values of the right hand (L) and right hand (R).

In detail, the motion data generator 13 defines the coordinate axis movement direction as follows.

X axis : Direction of moving left and right on the screen, moving to the right +

Y axis : Direction of moving up and down on the screen, going up is +

Z axis : Direction of movement towards the depth of the screen, when the screen moves towards the viewer, +

X-axis direction control

Adjust by moving your left hand left and right (if your left hand is at -X, it moves left, and when it is at +X, it moves right).

Y-axis direction control

Adjust by keeping both hands together and moving them up and down (if both hands are positioned at +Y, move upward, and if both hands are positioned at -Y, move downward).

Z-axis direction control

Adjust by moving your right hand up and down (if your right hand is at +Z, it moves to the front of the screen, and when it is at -Z, it moves to the back of the screen).

stop motion

Move to the stop zone with both hands together and it will stop.

As shown in FIG. 5A, when only the user's left hand is moved left or right, the motion data generator 13 generates the user's motion data on the X-axis corresponding to the left and right movement information.

Additionally, as shown in FIG. 5B, the motion data generator 13 generates the user's motion data on the Z-axis corresponding to the vertical movement information when only the user's right hand is moved up and down.

In addition, as shown in FIG. 5C, when the user's left and right hands are moved up and down, the motion data generator 13 generates the user's motion data on the Y axis corresponding to the up and down movement information. Create.

Meanwhile, the motion data generator 13 determines that the user's left and right hands are together if the distance between the user's left and right hands is smaller than the size of the object representing the user's hands. That is, the motion data generator 13 determines that the two hands are brought together when the distance ( d) between the two hands is smaller than the size (2 * r) of the object representing the hand, as shown in FIG. 3 and Equation 1. do.

[Equation 1]

d < 2* r

As shown in FIG. 5D, the motion data generator 13 generates user motion data indicating a stop motion when the user's left and right hands are brought together and moved to a preset area (stop). .

The avatar operation module 20 is a device that operates a pre-created avatar according to control signals received from the outside. That is, the avatar operation module 20 operates the avatar created by the avatar creation module based on the control signal received from the processor 30 and outputs it through the output module.

The avatar operation module 20 receives a determined command from the processor 30 and controls the avatar to perform an operation corresponding to the command through the processor 30. The user's operational intention is converted into a state and determined as a command. (For example, moving left and right, moving forward and backward, moving up and down, etc.)

Meanwhile, even in the same operation, the magnitude of rotation and speed of operation may be different, and may be adjusted according to the amplitude of the control signal in addition to the state of the command. For example, if a control signal with a specific frequency is detected, and the signal of the corresponding frequency and region matches the command to “move forward” the avatar, the avatar motion module 20 controls the avatar to move forward. At this time, the larger the amplitude of the control signal, the faster the avatar can move forward, and the smaller the amplitude, the slower the avatar can move forward.

The processor 30 generates avatar motion information corresponding to the user's hand motion data generated by the image processing module 10, and operates the avatar through the avatar motion module based on the generated avatar motion information. I order it.

The avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention configured as described above generates a hand position from the user's image frame captured by the image generator 11, as shown in FIG. 6. It is captured through the unit 12 and the coordinate value corresponding to the position of the hand is recognized (S10).

Then, after calculating the distance between both hands through the motion data generator 12 (S20), if it is determined to be one hand, calculate the front, rear, left, and right hand positions (S30), and if it is determined to be moving left and right, the processor 30 ) moves the avatar left and right (S31), and if it is determined to be moving forward and backward, the avatar is moved back and forth through the processor 30 (S32).

In addition, after calculating the distance between both hands through the motion data generator 12 (S20), if it is determined that there are two hands, the up and down hand positions are calculated (S40), and if it is determined that the hand is moving up and down, the avatar is generated through the processor 30. Move up and down (S41).

According to the avatar control system using artificial intelligence hand position recognition according to an embodiment of the present invention configured as described above, simple Since avatar control commands can be generated through calculation, an avatar controller using hand position recognition can be provided.

Meanwhile, an anti-pollution layer made of an anti-pollution coating composition may be applied to the surface of the image processing module 10 to effectively prevent adhesion and remove contaminants.

The anti-pollution coating composition contains sodium sesquicarbonate and butyl carbitol in a molar ratio of 1:0.01 to 1:2, and the total content of sodium sesquicarbonate and butyl carbitol is 1 to 10% by weight based on the total aqueous solution. .

The molar ratio of sodium sesquicarbonate and butyl carbitol is preferably 1:0.01 to 1:2. If the molar ratio is outside the above range, the applicability of the anti-pollution layer may decrease or moisture adsorption on the surface may increase after application. There is a problem with the coating film being removed.

The sodium sesquicarbonate and butyl carbitol are preferably used in an amount of 1 to 10% by weight in the total composition aqueous solution. If the amount is less than 1% by weight, there is a problem that the applicability of the anti-fouling coating layer is reduced, and if it exceeds 10% by weight, the coating film Crystal precipitation is likely to occur due to an increase in thickness.

Meanwhile, it is preferable to apply the anti-pollution composition on the surface of the image processing module 10 by spraying. In addition, the final coating film thickness on the surface of the image processing module 10 is preferably 700 to 2500 Å, and more preferably 900 to 2000 Å. If the thickness of the coating film is less than 700 Å, there is a problem of deterioration in the case of high temperature heat treatment, and if it exceeds 2500 Å, there is a disadvantage in that crystal precipitation on the coating surface is likely to occur.

Additionally, this anti-pollution coating composition can be prepared by adding 0.1 mol of sodium sesquicarbonate and 0.05 mol of butyl carbitol to 1000 ml of distilled water and then stirring.

The reason for limiting the ratio of the components and the thickness of the coating film to the above values is that the present inventor analyzed the test results after repeated failures and found that the ratio showed the optimal anti-contamination application effect.

A heat dissipating coating agent is applied to the surface of the processor 30 to prevent the surface of the processor 30 from being excessively heated as heat emitted from the processor 30 cannot be sufficiently dissipated and heat can be effectively dissipated.

This heat dissipation coating composition contains 58% by weight of potassium silicate, 11% by weight of chromium oxide, 13% by weight of graphite, 8% by weight of silicon nitride, 3% by weight of sodium hydroxide (NaOH), 3% by weight of titanium oxide, 2% by weight of polyamide wax, It consists of 2% by weight of 3-aminopropyl trimethoxysilane.

Potassium silicate acts as a binder resin, chromium oxide acts as wear resistance, graphite has excellent thermal conductivity and electrical properties, silicon nitride improves strength and prevents cracking, sodium hydroxide acts as a dispersant, and titanium oxide provides weather resistance. For this purpose, polyamide wax acts as an anti-settling agent, and 3-aminopropyl trimethoxysilane serves as an adhesion enhancer.

The reason for limiting the constituent materials and components and limiting the mixing ratio as described above is that the present inventor analyzed the test results after repeated failures and found that the optimal effect was achieved with the above-mentioned constituents and numerical ratio. indicated.

The surface of the avatar operation module 20 may be coated with an environmental fragrance material mixed with functional oil that helps sterilize and relieve user stress.

The mixing ratio of the fragrance material and the functional oil is 95 to 97% by weight of the fragrance material and 3 to 5% by weight of the functional oil, and the functional oil is 50% by weight of angelica oil and citronella oil. ) consists of 50% by weight.

Here, it is preferable that the functional oil is mixed in an amount of 3 to 5% by weight based on the fragrance material. If the mixing ratio of the functional oil is less than 3% by weight, the effect is minimal, and if the mixing ratio of the functional oil exceeds 3 to 5% by weight, the effect is not significantly improved and the economic feasibility is low. Angelica oil is effective in relieving stress, relieving tension, and sterilizing, while citronella oil purifies and uplifts the mind psychologically and is effective in treating headaches, depression, and neuralgia. Therefore, as the fragrance material mixed with these functional oils is coated on the surface of the avatar operation module 20, effects such as sterilizing the surface of the avatar operation module 20 and reducing the user's stress, etc. can be obtained. .

The reason for limiting the components and limiting the mixing ratio for environmental fragrance materials and functional oils is that the present inventor analyzed the test results after repeated failures and found that the optimal composition and numerical ratio was determined. showed the effect.

Meanwhile, a vibration absorbing part made of rubber may be further installed at the lower part of the image processing module 10.

This vibration absorbing part may be made of a rubber material, and the raw material content ratio of this vibration absorbing part is 60% by weight of rubber, 8% by weight of dibutylthiourea, 6% by weight of calcium stearate, 19% by weight of carbon black, and 3C (N- Mix 3% by weight of PHENYL-N'-ISOPROPYL-P-PHENYLENEDIAMINE) and 4% by weight of organic peroxide.

Dibutylthiourea is added to accelerate vulcanization, calcium stearate is added to act as a softener, and carbon black is added to increase or improve wear resistance and thermal conductivity.

3C (N-PHENYL-N'-ISOPROPYL- P-PHENYLENEDIAMINE) is added as an antioxidant, and organic peroxide is added to act as an accelerator.

Therefore, the present invention improves durability by increasing the elasticity, toughness, and rigidity of the vibration absorbing part, and thus increases the lifespan of the vibration absorbing part.

The tensile strength of the rubber material is 155Kg/㎠.

The reason for limiting the rubber material and components and limiting the mixing ratio values, etc. is that the present inventor analyzed the test results after repeated failures and found that the optimal effect was achieved with the above-mentioned components and numerical ratios. indicated.

In the above, preferred embodiments according to the present invention are shown and described. However, the present invention is not limited to the above-described embodiments, and various modifications may be made by anyone skilled in the art without departing from the gist of the present invention appended to the claims. .

10: Image processing module 11: Image generation unit
12: Hand position generation unit 13: Motion data generation unit
20: Avatar operation module 30: Processor

Claims (5)

An image processing module that detects the user's hand motion and generates the user's hand motion data;
an avatar operation module that operates a pre-generated avatar according to a control signal received from the outside; and
a processor that generates motion information of an avatar corresponding to the user's hand motion data and operates the avatar based on the generated motion information of the avatar;
The image processing module is,
An image generator that captures the user and generates image frames divided into the X-axis (left and right of the screen), Y-axis (top and bottom of the screen), and Z-axis (depth of the screen); a hand position generator that generates coordinate values corresponding to the position of the user's hand in the video frame; and a motion data generator that calculates a moving direction of the coordinate value and generates preset user's hand motion data based on the calculated moving direction of the coordinate value;
The motion data generator
Calculate the direction of movement using the coordinate values of the user's left hand, right hand, or left and right hands,
When only the user's left hand is moved left or right, the user's motion data on the X-axis corresponding to the left and right movement information is generated,
When only the user's right hand is moved up and down, the user's motion data on the Z axis corresponding to the up and down movement information is generated,
An avatar control system using artificial intelligence hand position recognition, characterized in that when the user's left and right hands are moved up and down with the user's left and right hands together, the user's motion data on the Y axis corresponding to the up and down movement information is generated.
delete delete According to paragraph 1,
The motion data generator
An avatar control system using artificial intelligence hand position recognition, characterized in that it is determined that the user's left and right hands are together if the distance between the user's left and right hands is smaller than the size of the object representing the user's hands.
According to paragraph 1,
The motion data generator
An avatar control system using artificial intelligence hand position recognition, characterized in that it generates user motion data indicating a stopping motion when the user's left and right hands are brought together and moved to a preset area.

Images