System

Abstract

A system is provided.SOLUTION: A system including means for detecting a face of a person in video data in real time, means for converting the detected face of the person into another face using an AI model, and means for distributing the converted video data in real time.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] When live streaming or posting images from the street, people around you may appear in the image, posing a risk of infringement of portrait rights. While a traditional solution is to blur the image, this can create an unnatural feeling in the video and potentially detract from the viewer's experience. Therefore, there is a need for technology that can prevent portrait rights infringement in real time while maintaining the naturalness of the video. [Means for solving the problem]

[0005] The present invention solves the above-mentioned problems with a system that includes a means for detecting human faces in video data in real time, a means for converting the detected human faces into other faces using an AI model, and a means for distributing the converted video data in real time. Furthermore, the present invention includes a means for excluding pre-registered faces of specific people from the conversion targets, and a means for changing hairstyles and clothing and converting specific attributes. This makes it possible to provide high-quality video without compromising the naturalness of the video while protecting the portrait rights of people around the person when live streaming or posting images.

[0006] "Video Data" means a digital representation of visual information captured by a camera or other imaging device.

[0007] "Detection" is the process of identifying specific objects or people within video data using specific algorithms or techniques.

[0008] "Human face" refers to the part of a person's face in video data, and is the area that includes characteristic features such as the eyes, nose, and mouth.

[0009] An "AI model" is a mathematical model that has been trained to perform a specific task using artificial intelligence techniques.

[0010] "Face conversion" is the process of replacing a detected face with a different face image in real time while preserving the original facial features.

[0011] "Real-time" refers to a method that minimizes delays and outputs processing results almost immediately.

[0012] "Distribution" is the process of providing video data to viewers over a network.

[0013] "The face of a specific person registered in advance" refers to facial data that a specific individual, such as a broadcaster, has registered as recognition information in the system before the broadcast begins.

[0014] "Excluded from conversion target" means that a face is excluded from conversion processing based on a specific condition.

[0015] "Changing hairstyle or clothing" refers to replacing the hairstyle or clothing of a person in the video data with a different style.

[0016] "Converting to specific attributes" refers to the process of changing a person in video data to a specified style or attribute, such as an anime character or a foreign character. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0018] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0019] First, the terms used in the following description will be explained.

[0020] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0021] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0022] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0023] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0024] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0025] [First embodiment]

[0026] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0027] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0028] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0029] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0030] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0031] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0032] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0033] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0034] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0035] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0036] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0037] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0038] The system of the present invention uses an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from the street, providing natural-looking images while protecting portrait rights. The implementation of this system follows the steps below.

[0039] First, the user launches the live streaming app and performs the initial setup. The user then scans their face with the camera and registers their facial recognition data as a streamer in the system. This data is then used to exclude the streamer from face conversion later.

[0040] Next, the device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format at a constant frame rate. This stream data is then sent to the server in real time.

[0041] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate facial recognition technology to identify faces in each frame. In doing so, it references the face data of the streamer registered in advance and compares it with faces in the video, excluding any that match from the conversion target.

[0042] The server then uses an AI model to convert the faces that were not rejected into a different face. The AI ​​model, for example, a generative artificial network (GAN), can be used to convert the facial features into natural-looking faces. This model processes in real time, preserving the original facial angle and facial expression.

[0043] Furthermore, the server changes parts other than the face (hairstyle and clothing) and converts them to specific attributes (for example, anime character style or foreign style) based on the option settings selected by the user. Specifically, if the user selects "anime character style," the face and clothing of the person in the video are converted to the corresponding character design.

[0044] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technology to minimize delays, allowing viewers to enjoy natural, high-quality video.

[0045] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the face of the streamer that has been registered in advance, and converts it into a different face using an AI model. This allows viewers to view the converted face or video with specific attributes in real time, allowing live streaming to be carried out safely without violating portrait rights while maintaining the naturalness of the video.

[0046] The processing flow will be explained below.

[0047] Step 1:

[0048] A user launches a live streaming app and completes the initial setup. At this time, the user points their face at the camera and scans it, registering their facial recognition data as a streamer into the system. This will exclude the streamer's face from subsequent facial conversion processing.

[0049] Step 2:

[0050] The device activates the camera and captures video data in real time. The device generates the captured video data in a stream format and continues to acquire video data at a constant frame rate.

[0051] Step 3:

[0052] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection over the network.

[0053] Step 4:

[0054] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video, identifying faces in each frame of the video data and extracting the necessary information.

[0055] Step 5:

[0056] The server matches the faces in the video with pre-registered facial data of the streamer and filters out any matches. This process involves comparing facial feature vectors to reliably identify the streamer's face.

[0057] Step 6:

[0058] The server applies an AI model (such as a generative artificial network (GAN)) to the remaining faces and converts them into a different face in real time. The AI ​​model generates natural-looking facial features that reflect the angle and expression of the original face.

[0059] Step 7:

[0060] The server checks the user's option settings and, if necessary, changes non-facial features (hairstyle and clothing) or converts them to specified attributes (e.g., anime character style or foreign character style), thereby processing the video in the specific style desired by the user.

[0061] Step 8:

[0062] The server then delivers the converted video data to viewers in real time, with minimal latency and using high-speed data processing technology.

[0063] Step 9:

[0064] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0065] Step 10:

[0066] The server continues to process the video data in real time, converting and distributing it consistently until the end of the broadcast, allowing viewers to receive consistently high-quality video.

[0067] Example 1

[0068] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0069] When live streaming or posting images from the street, it is important to provide natural-looking images while avoiding infringing on the portrait rights of people around you. Another challenge is to convert the video in real time and deliver it to viewers with high quality and low latency, without altering the broadcaster's own face.

[0070] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0071] In this invention, the server includes: a means for a user to launch a live streaming app and scan their face to register facial recognition data; a means for the device to launch a camera and capture video data in real time and generate it in a stream format at a constant frame rate; a means for the server to analyze the received video data and detect human faces in the video using a facial recognition algorithm; a means for converting the detected human faces into other faces using a generative AI model; a means for modifying parts other than the face or converting them into specific attributes based on option settings selected by the user; and a means for distributing the converted video data in real time. This makes it possible to distribute high-quality, natural-looking video in real time while protecting portrait rights.

[0072] A "user" is an entity that operates the system and performs initial setup and live streaming.

[0073] A "live streaming app" is application software that users use to stream video in real time.

[0074] "Facial recognition data" refers to image data and characteristic information of a user's face that is scanned with a camera and registered in the system.

[0075] A "terminal" is a device equipped with a camera that captures and transmits video to a server.

[0076] A "camera" is a device that captures video data in real time.

[0077] "Video data" refers to video information captured by a camera.

[0078] "Stream format" is a method of transmitting data continuously in real time.

[0079] The "server" is a computer system that analyzes the received video data, performs facial recognition and conversion, and distributes it in real time.

[0080] A "face recognition algorithm" is a program for detecting and identifying the faces of people in video footage.

[0081] A "generative AI model" is an artificial intelligence model that generates new data based on input data, and in this invention is used to convert a specific face into another face.

[0082] "Optional settings" are additional settings that a user can make to the video being distributed, including changes to parts other than the face and conversion to specific attributes.

[0083] "Real-time distribution" is a method in which converted video data is immediately provided to viewers.

[0084] "Portrait rights" are the right of individuals to prevent their portraits, including their face and figure, from being photographed or used without their permission.

[0085] The system of this invention protects portrait rights and provides natural-looking images by using an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from around the city. The following hardware and software are used to implement this system.

[0086] First, the user launches the live streaming app and performs the initial setup. They then scan their face using the camera on their smartphone or PC, and register this facial recognition data in the system. This data is used by the facial recognition algorithm to exclude the streamer's face from conversion.

[0087] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format (e.g., RTMP protocol) at a frame rate of 30 frames per second and sent to the server.

[0088] The server analyzes the received video data and uses the Python library OpenCV, Dlib, or other facial recognition algorithms to detect faces in the video. At this time, it references the facial data of pre-registered streamers and excludes any matches from the conversion.

[0089] The server converts the detected faces that are not rejected into a new face using a generative anti-aliasing network (GAN). Using deep learning frameworks like TENSORFLOW® and PyTorch, the server uses pre-trained models to process the converted faces in real time so that they appear natural. This model converts the input face into a new face while preserving its original angle and facial expression.

[0090] Furthermore, the server also processes changes to non-facial features (hairstyle and clothing) and conversion to specific attributes (such as anime character style or foreign style) based on the options selected by the user within the app. For example, if the user selects "anime character style," the server uses a style conversion model such as StyleGAN to convert the face and clothing of the person in the video into an anime character design.

[0091] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technologies such as the RTMP protocol to minimize delays. Viewers can enjoy this natural, high-quality video on their smartphones, PCs, or web browsers using viewing apps.

[0092] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the broadcaster's pre-registered face, and converts it to a different face using an AI model. This allows viewers to view the converted face or video with specific attributes converted in real time, allowing live streaming to be carried out safely while maintaining the naturalness of the video and without violating portrait rights.

[0093] Examples of prompts:

[0094] "Generate the program code for a system that transforms the faces of people around you into other faces during a live stream. In this system, the streamer's face will be pre-registered and will not be transformed. The AI ​​model used is a generative adversarial network (GAN), and the transformed footage will be distributed to viewers in real time."

[0095] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0096] Step 1:

[0097] The user launches the live streaming app and scans their face to register facial recognition data. In this step, the user uses the camera on their smartphone or PC to perform the facial scanning operation. The input is the user's facial image data, and the output is facial recognition data registered in the system. This recognition data is used to identify the broadcaster's face and exclude it in the later face conversion step. Specifically, the user follows the app's guide to point the camera at their own face and perform a scan for a few seconds.

[0098] Step 2:

[0099] The device starts the camera and captures video data in real time. In this step, the device captures video at a frame rate of 30 frames per second. The input is video data from the camera, and the output is stream-format data generated in real time. Specifically, the device starts the camera, and the live streaming app captures video data in real time.

[0100] Step 3:

[0101] The video data captured by the terminal is sent to the server in real time. In this step, the video data is sent using a protocol such as RTMP. The input is the captured video data in a stream format, and the output is the video data sent to the server. Specifically, the terminal sends the video data to the server via a network connection.

[0102] Step 4:

[0103] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. The input is the real-time video data sent to the server, and the output is a list of detected faces. Here, the server applies the face detection algorithm using the Python libraries OpenCV and Dlib. Specifically, the server analyzes each frame and identifies the location and features of faces.

[0104] Step 5:

[0105] The server excludes the faces of pre-registered streamers from the detected faces. The input is a list of detected faces and facial recognition data of pre-registered streamers, and the output is a list of faces excluding the streamers' faces. Specifically, the server compares the facial recognition data and removes matching faces from the list.

[0106] Step 6:

[0107] The server uses a generative AI model to convert the detected face into a different face. The input is a list of faces excluding the streamer's face and video data containing that face, and the output is the converted video data. The server uses TensorFlow or PyTorch to apply a pre-trained generative adversarial network (GAN). Specifically, the server performs face conversion in real time, converting the face into a new face while preserving the original facial angle and expression.

[0108] Step 7:

[0109] The server modifies non-facial features or converts specific attributes based on the user's options. The input is video data containing the converted face and the user's options, and the output is video data that has been further converted according to the options. Specifically, the server applies a style transfer model such as StyleGAN to change the hairstyle or clothing of people in the video.

[0110] Step 8:

[0111] The server finally delivers the converted video data to the viewer in real time. The input is the final converted video data, and the output is the real-time video delivered to the viewer. Specifically, the server uses the RTMP protocol to send video data at high speed to the viewer, who then views the video without delay using a viewing app or web browser on their smartphone or PC.

[0112] (Application example 1)

[0113] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0114] When live streaming or posting images from the street or inside a store, there is a risk of violating the portrait rights of other people. In particular, in physical stores, where many customers come and go, the faces of other customers appearing in posted photos and videos can be a problem. This can lead to the risk of violating the privacy of other customers, so there is a need to solve this problem.

[0115] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0116] In this invention, the server includes means for detecting human faces in video data in real time, means for converting the detected human faces into other faces using an AI model, means for distributing the converted video data in real time, means for excluding the faces of specific people registered in advance from the conversion targets, means for changing hairstyles and clothing and converting into specific attributes, and means for processing photos and videos taken by customers in the store in real time and converting the faces of other customers into other faces. This enables customers to safely post photos and videos taken in the store to social media etc.

[0117] "Real-time" refers to data processing and information delivery occurring immediately and without delay.

[0118] "Video data" refers to data containing visual information obtained from a camera or other imaging device.

[0119] "Means for detecting human faces" refers to technology that uses specific algorithms to identify and extract human faces from video data.

[0120] "AI model" refers to a mathematical or algorithmic model trained to perform a specific task using artificial intelligence techniques.

[0121] "Method of converting into another face" refers to technology that converts the detected face of a person into a different facial image using an AI model.

[0122] "Means for delivering converted video data in real time" refers to technology that instantly provides viewers with video data converted by an AI model.

[0123] "Means to exclude the faces of specific people registered in advance from conversion targets" refers to technology that pre-registers the faces of specific people, such as streamers, in a database, and does not perform conversion if that face is detected in the video.

[0124] "Methods for changing hairstyles and clothing" refers to technology that changes the hairstyles and clothing of people in video data to different styles.

[0125] "Means of converting to specific attributes" refers to technology that changes the faces and clothing of people in video data to specific designs, such as those of anime characters or foreigners.

[0126] "Store visitors" refers to customers or clients who visit a physical store.

[0127] "Means for processing photos and videos in real time" refers to technology that instantly processes and converts photos and videos taken by customers inside the store.

[0128] "Means of converting the faces of other customers into different faces" refers to technology that uses an AI model to convert the faces of other customers photographed in the store into images of different faces.

[0129] "SNS" is an abbreviation for "social networking service," which refers to a platform where users can share and interact with each other via the Internet.

[0130] The embodiments for carrying out the present invention are described as follows.

[0131] The system of the present invention uses a smartphone to process photos and videos taken in a physical store in real time, protecting the portrait rights of other customers. This system is realized using the following hardware and software components.

[0132] Hardware

[0133] Smartphone: The device on which this system is implemented, capturing and processing video data.

[0134] Camera module: A smartphone's built-in camera that takes photos and videos in real time.

[0135] software

[0136] An open-source face recognition library (e.g., OpenCV 4.5.3) for detecting human faces in video data.

[0137] Generative Adversarial Network (GAN) models (e.g., StyleGAN2): Used to transform a detected face into another face in real time.

[0138] Real-time video streaming library (e.g., FFmpeg 4.4): Used to instantly deliver converted video data to viewers.

[0139] System Operation

[0140] 1. The user launches the smartphone app and performs the initial setup, which includes registering the user's own facial data. This facial data is later used as an exclusion target during the facial conversion process.

[0141] 2. The camera module is activated and the device (smartphone) captures video data in real time. The video data is acquired at a constant frame rate and generated in a stream format.

[0142] 3. The acquired video data is analyzed in real time using a facial recognition library such as OpenCV. The server detects faces in the video and compares them with pre-registered user facial data, excluding any matches from the conversion target.

[0143] 4. The detected face is transformed into another face in real time using a generative adversarial network (GAN) model such as StyleGAN2. This transformation process is fast and preserves the original face's natural angle and facial expression.

[0144] 5. Based on the options selected by the user, the server changes non-facial features (such as hairstyle and clothing) and converts the user to specific attributes. For example, it is possible to convert the user to an anime character or a foreign character.

[0145] 6. The converted video data is then distributed to viewers in real time via a video streaming library such as FFmpeg, allowing viewers to enjoy high-quality, seamless video.

[0146] Specific examples

[0147] For example, if a user takes a photo inside a physical store, the system can detect the faces of other customers and convert them into different faces before distributing the video, thereby avoiding portrait rights issues. This allows users to post photos and videos to social media with peace of mind.

[0148] Prompt Sentence Examples

[0149] "Please create a camera app that protects portrait rights by converting the faces of other customers in real time into videos taken in a store."

[0150] In this way, the present invention simultaneously realizes protection of portrait rights and provision of natural-looking images in brick-and-mortar stores and public places.

[0151] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0152] Step 1:

[0153] The user launches the smartphone app and performs the initial setup. Here, the user scans their face data with the camera and registers it in the system. The user's facial image data is used as input, and this is the output that is registered in the facial recognition database. This data is later used as an exclusion target during the face conversion process.

[0154] Step 2:

[0155] The device activates the camera module and captures video data in real time. The input is video data acquired from the camera, and the output is video data in a stream format generated at a constant frame rate. This video data is immediately sent to the server.

[0156] Step 3:

[0157] The server analyzes the received video data using an open-source face recognition library (e.g., OpenCV 4.5.3). The input is stream-format video data, and the output is data indicating the position of people's faces in the video. The server detects faces in the video based on this data.

[0158] Step 4:

[0159] The server compares the detected person's face with the user's face data registered in advance. The input is face position data and the user's face recognition data, and the output is a list of faces to be excluded from conversion. This prevents conversion even if the user's face data is included in the video.

[0160] Step 5:

[0161] The server uses a generative adversarial network (GAN) model (e.g., StyleGAN2) to convert the detected face into a different face in real time. The input is the recognized face position data and facial image data in the video, and the output is the converted facial image data. This data processing allows the conversion to be performed while maintaining the natural angle and expression of the original face.

[0162] Step 6:

[0163] The server changes parts of the character other than the face (such as hairstyle or clothing) based on the options selected by the user. The input is the user's options and video data, and the output is the video data after the changes. This data calculation results in a transformation to make the character look like an anime character or a foreign character, depending on the settings.

[0164] Step 7:

[0165] The server then uses a real-time video streaming library (e.g., FFmpeg 4.4) to deliver the converted video data to viewers. The input is the converted video data, and the output is the converted video delivered in real time. This allows viewers to enjoy high-quality video without delay.

[0166] These processing steps enable the system to safely broadcast live footage from within a physical store, providing natural-looking footage without violating the portrait rights of other customers.

[0167] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0168] The system of the present invention has the function of detecting human faces in video data in real time, converting the detected human faces into other faces using an AI model, and distributing the converted video data in real time. It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for changing hairstyles, clothing, and converting specific attributes. It also combines an emotion engine that recognizes user emotions, and can perform conversion processing and attribute changes within the video data based on the recognized emotions.

[0169] The implementation of this system is described in detail below.

[0170] The user launches the live streaming app and performs the initial setup. The streamer then points their face at the camera and scans it, which is then registered as facial recognition data in the system. This data is then used to exclude the streamer from face conversion later.

[0171] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format and sent to the system at a constant frame rate.

[0172] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate technology to identify faces in each frame. It then references the facial data of streamers registered in advance and excludes faces that match from the conversion target.

[0173] The server then applies an AI model (e.g., a generative artificial network, or GAN) to the detected face to transform it into a different face in real time. The AI ​​model generates natural-looking facial features while preserving the original facial angle and expression.

[0174] Furthermore, an emotion engine is introduced to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The analysis results are classified into emotion categories such as joy, sadness, and anger.

[0175] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, additional processing can be performed to make the face in the video brighter or to change it into a smiling face. The attributes of the face and clothing generated by the AI ​​model can also be changed according to the emotion. For example, if the user selects an anime-style setting and is sad, detailed expressions such as the character shedding tears can be added.

[0176] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0177] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0178] As described above, this system performs video processing that reflects the user's emotions, making it possible to provide viewers with consistently high-quality, natural-looking video.

[0179] The processing flow will be explained below.

[0180] Step 1:

[0181] The user launches the live streaming app and performs the initial setup. The user then points their face at the camera, scans it, and registers facial recognition data into the system. This data is used to identify the face as the streamer.

[0182] Step 2:

[0183] The device activates the camera for live streaming and captures video data in real time, which is generated in a stream format at a constant frame rate.

[0184] Step 3:

[0185] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection with the server via the network.

[0186] Step 4:

[0187] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. High-precision facial recognition technology is used to identify faces in each frame of the video data.

[0188] Step 5:

[0189] The server compares the detected face data with the face data of the streamer registered in advance and excludes any matches from the conversion process. This means that the streamer's face is excluded from the conversion process.

[0190] Step 6:

[0191] The server then applies an AI model to the remaining faces, transforming them into a different face in real time. The AI ​​model (e.g., a generative adversarial network: GAN) generates natural-looking facial features while preserving the original facial angle and expression.

[0192] Step 7:

[0193] The device uses an emotion engine to recognize the user's emotions and analyzes them in real time through cameras and sensors, categorizing the results into emotion categories such as joy, sadness, and anger.

[0194] Step 8:

[0195] The server adjusts the transformation process in the video data based on the user's emotions recognized by the emotion engine. For example, if the server recognizes that the user is happy, it may transform the face of a person in the video into a brighter expression.

[0196] Step 9:

[0197] The server changes the facial and clothing attributes generated by the AI ​​model based on the user's emotions recognized by the emotion engine. For example, if the user selects the "anime character" setting, character expressions corresponding to the emotion will be added.

[0198] Step 10:

[0199] The server then delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0200] Step 11:

[0201] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0202] Step 12:

[0203] The server processes the video data in real time, converting and delivering it consistently until the end of the broadcast, allowing viewers to receive high-quality, natural video continuously.

[0204] Example 2

[0205] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0206] In real-time video streaming, there is a demand for providing viewers with higher quality and more natural-looking images. However, in technology for converting faces in streamed video, there is a lack of methods for excluding the streamer's own face from the conversion target, or for processing video that reflects the user's emotions. Therefore, there is a need for the development of a system that can appropriately convert the faces of people other than the streamer during live streaming, and further recognize the user's emotions and reflect them in the video.

[0207] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0208] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for distributing the converted video data in real time, means for identifying a user's emotion using an emotion engine, and means for adjusting the conversion process in the video data based on the identified emotion. This makes it possible to naturally convert the faces of people other than the broadcaster in real time during broadcasting, and further to reflect the user's emotion in the video.

[0209] "Real-time" refers to the ability of a system to operate such that data is processed and responded to almost immediately, implying very little processing delay.

[0210] "Video data" means data that describes visual information captured by a camera or other imaging device.

[0211] "Human face" refers to the facial features of an individual human being within video data, and specifically refers to the area that can be identified by facial recognition technology.

[0212] "Means for detecting" refers to a combination of hardware and software capable of identifying and extracting a particular target.

[0213] "Generative AI models" refer to algorithms that use artificial intelligence to generate new data and features, and specifically include generative artificial network (GAN).

[0214] "Means for converting" refers to software or hardware that has the functionality to change data of one format into another format.

[0215] "Distribution means" refers to a system capable of transmitting data to a specific receiver and making it available in a usable form.

[0216] "Emotion engine" refers to a combination of software and hardware for recognizing and analyzing user emotions in real time.

[0217] "User emotion" refers to the user's psychological state as recognized from the video and audio data being distributed, and includes emotion categories such as joy, sadness, and anger.

[0218] "Adjusting means" refers to hardware and software that has the functionality to change the system's behavior or output based on specific conditions.

[0219] The system of the present invention has the function of detecting human faces in video data in real time during live streaming, converting the detected faces into other faces using a generative AI model, and distributing the converted video data in real time.It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for recognizing user emotions and changing the attributes of the video based on the emotions.

[0220] The user launches the live streaming app and performs the initial setup. For example, the streamer points the camera at their face and scans it, registering it as facial recognition data in the system. This data is later used to exclude the streamer from face conversion. The user scans their face using a smartphone or computer, following the app's instructions.

[0221] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the system at a certain frame rate (e.g., 30 fps). At this time, the device uses a high-resolution camera to capture clear video.

[0222] The server analyzes the received video data and uses a facial recognition algorithm to detect faces of people appearing in the video. The facial detection algorithm uses open-source facial recognition libraries (e.g., dlib or OpenCV) to perform highly accurate face detection. The server performs face detection for each series of frames and excludes matching faces from conversion by referencing the facial data of broadcasters registered in advance.

[0223] The server then applies a generative AI model to the detected face, transforming it into a different face in real time. This model uses generative artificial intelligence (GAN) or deep learning. For example, GANs preserve the original facial features while generating new facial features. During this process, the server carefully transforms the face to maintain the original facial angle and facial expression.

[0224] Furthermore, an emotion engine is implemented to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The emotion engine uses facial expression recognition technology and voice analysis technology to classify the user's emotions into joy, sadness, anger, etc. The analysis results are sent to the server in real time.

[0225] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, the server adds a process to brighten the face in the video and transform it into a smiling face. The attributes of the face and clothing generated by the generative AI model are also changed according to the user's emotions. For example, if the user selects an anime-style setting and is sad, the character will shed tears.

[0226] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay via live streaming platforms (e.g., YouTube® Live or Twitch).

[0227] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0228] Example prompt sentence:

[0229] Start a live stream at a tourist spot, and have it detect people's faces in real time and convert their expressions into happy expressions. The streamer's own face will not be converted, ensuring a fun-filled stream for your viewers.

[0230] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0231] Step 1:

[0232] The user launches the live streaming app and performs the initial setup. The user then points the camera at their face to scan it and registers it as facial recognition data in the system. The facial image captured by the camera is used as input, and a facial recognition algorithm is applied to obtain the broadcaster's facial data. As output, the broadcaster's facial data is stored in the system's database.

[0233] Step 2:

[0234] The device activates the camera for live streaming and captures video data in real time. As input, it uses the video stream from the camera and captures data at a constant frame rate (e.g., 30 fps). As output, the captured video data is generated in a stream format and sent to the server.

[0235] Step 3:

[0236] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. As input, it uses the received video frames and applies a facial recognition algorithm (e.g., dlib or OpenCV). As output, it generates location information and feature point data of the detected faces.

[0237] Step 4:

[0238] The server references the face data of the streamer registered in advance and compares it with the detected face. As input, it uses the face data of the streamer registered in step 1 and the detected face data obtained in step 3. As output, it generates a list in which matching faces are excluded from conversion.

[0239] Step 5:

[0240] The server applies a generative AI model to the detected face and transforms it into a different face in real time. It uses facial position information and feature point data as input and uses a generative artificial network (GAN). The output is image data of the transformed face. During this process, the original facial angle and facial expression are maintained.

[0241] Step 6:

[0242] The emotion engine analyzes the user's emotions in real time using a camera or sensor on the device that can identify the user's emotions. As input, it uses data obtained from the camera or sensor (facial expressions and voice data) and applies emotion analysis algorithms. As output, it generates an analysis result of emotion categories such as joy, sadness, anger, etc. and sends it to the server.

[0243] Step 7:

[0244] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process in the video data. The inputs are the emotion analysis results and the transformed facial image data. The output is video data that has been adjusted according to the emotion. For example, if the user is happy, the face in the video is made brighter and transformed into a smiling face.

[0245] Step 8:

[0246] The server delivers the converted video data to viewers in real time. The server uses the adjusted video data as input and sends the data via a live streaming platform. The output is high-quality, latency-free video delivered to viewers, allowing them to enjoy a natural, real-time experience.

[0247] (Application example 2)

[0248] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0249] Providing a consistently visually pleasing experience to viewers is challenging in current live streaming and video distribution services. In particular, there is a need for technology that can flexibly address the needs of privacy protection and emotion-based video conversion. Furthermore, there is no system that can exclude specific people's faces from conversion and dynamically process video based on emotions in real time.

[0250] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0251] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for delivering the converted video data in real time, and means for adjusting the conversion processing in the video data based on the recognized emotion using an emotion engine that recognizes emotion. This makes it possible to provide viewers with consistently high-quality, natural video and realize video processing that corresponds to the user's emotion in real time.

[0252] "Real-time" refers to a state in which results are available immediately at the moment a process or operation is performed.

[0253] "Video Data" refers to video and image information captured by a camera or other device.

[0254] "Means for detecting faces" refers to technologies and algorithms that identify and locate human faces within video data.

[0255] "Generative AI models" refer to artificial intelligence models that learn from specific input data and generate new data. This includes technologies such as generative artificial network (GAN).

[0256] "Face conversion method" refers to the technology or algorithm used to replace a detected person's face with a different face.

[0257] "Means of distribution" refers to the technology and infrastructure used to transmit and display the converted video data to viewers in real time via the Internet.

[0258] An "emotion engine" refers to the technology or algorithms used to identify and categorize people's emotions.

[0259] "Means for adjusting the transformation process" refers to technologies and algorithms that dynamically change and optimize the way video data is transformed based on emotion or other criteria.

[0260] "Means for excluding the faces of specific people from conversion targets" refers to technologies or algorithms that detect the faces of specific people registered in advance and exclude them from real-time conversion targets.

[0261] "Means for changing hairstyles and clothing, and converting to specific attributes" refers to technologies and algorithms that change the hairstyles and clothing of people in video data to different ones and convert them to specific attributes (for example, to look like an anime character).

[0262] The present invention provides a system that detects human faces in video data in real time, converts the detected faces into different faces using a generative AI model, and distributes the video in real time. Specific embodiments are described below.

[0263] First, the user launches the live streaming application on their device, which can be a smartphone, smart glasses, or a head-mounted display. The application has a function that allows the streamer to point the camera at their face, scan it, and register it as facial recognition data in the system. This data is used to exclude the streamer's face from conversion targets.

[0264] Next, the device activates the camera and captures video data in real time. The captured video data is sent to the server as a stream at a fixed frame rate. The server analyzes the received video data and uses dlib's facial recognition algorithm to detect human faces in the video.

[0265] The server applies a generative AI model (e.g., a generative adversarial network: GAN) to the detected face and converts it into a different face in real time. This generative AI model generates natural-looking facial features and can preserve the original facial angle and expression. It also references the facial data of pre-registered streamers and excludes matching faces from the conversion target.

[0266] Furthermore, the server introduces an emotion engine to recognize emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time and adjust the conversion process within the video data based on the analysis results. For example, if the user is happy, the server can convert the face in the video into a bright, smiling face.

[0267] The server then delivers the converted video data to viewers in real time, allowing viewers to instantly enjoy high-quality, natural-looking video and a visual experience that reflects the broadcaster's intentions and emotions.

[0268] As a concrete example, consider a case where a user is live streaming at a tourist spot and the emotion engine detects "joy" during the stream. In this case, the faces of all other people in the stream will be converted to bright, smiling faces. An example of a prompt sentence is as follows:

[0269] Example prompt sentence:

[0270] "Identify the faces of people in a photo, recognize the emotions on those faces, and convert them into appropriate face images based on that. 1. Face recognition: using dlib 2. Emotion recognition: using EmotionRecognizer 3. Emotion-based face transformation: using GAN to generate natural facial expressions"

[0271] In this way, the system is able to respond to the user's emotions and provide high quality, visually pleasing video in real time.

[0272] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0273] Step 1:

[0274] The user launches the live streaming application on their device. They point the camera at their own face, scan it, and register it as facial recognition data in the system. This data is sent from the device to the server and used to exclude the broadcaster's face from conversion. The input is facial image data from the camera, and the output is facial recognition data.

[0275] Step 2:

[0276] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the server at a constant frame rate. The input is video data from the camera, and the output is video data in stream format.

[0277] Step 3:

[0278] The server prepares to analyze the received video data. It uses dlib's facial recognition algorithm to detect faces in the video. It references the face data of pre-registered streamers and excludes matching faces from the conversion target. The input is stream-format video data, and the output is the face coordinates and the streamer's face data detection results.

[0279] Step 4:

[0280] The server applies a generative AI model (e.g., a generative artificial network (GAN)) to the detected face and converts it into a different face in real time, preserving the original facial angle and expression. The input is the face coordinates and facial image data, and the output is the converted facial image data.

[0281] Step 5:

[0282] The emotion engine analyzes the user's emotions in real time using cameras and sensors that can identify the user's emotions. The analysis results are classified into emotion categories such as joy, sadness, and anger. The input is data from the cameras and sensors, and the output is emotion category data.

[0283] Step 6:

[0284] The server adjusts the transformation process in the video data based on the emotion category. For example, if the user is happy, additional processing is performed to transform the face in the video into a brighter, smiling face. The input is emotion category data and transformed face image data, and the output is adjusted face image data.

[0285] Step 7:

[0286] The server delivers the converted video data to viewers in real time, allowing viewers to enjoy natural, high-quality video without delay. The input is adjusted video data, and the output is video data delivered in real time.

[0287] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0288] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0289] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0290] [Second embodiment]

[0291] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0292] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0293] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0294] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0295] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0296] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0297] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0298] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0299] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0300] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0301] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0302] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0303] The system of the present invention uses an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from the street, providing natural-looking images while protecting portrait rights. The implementation of this system follows the steps below.

[0304] First, the user launches the live streaming app and performs the initial setup. The user then scans their face with the camera and registers their facial recognition data as a streamer in the system. This data is then used to exclude the streamer from face conversion later.

[0305] Next, the device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format at a constant frame rate. This stream data is then sent to the server in real time.

[0306] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate facial recognition technology to identify faces in each frame. In doing so, it references the face data of the streamer registered in advance and compares it with faces in the video, excluding any that match from the conversion target.

[0307] The server then uses an AI model to convert the faces that were not rejected into a different face. The AI ​​model, for example, a generative artificial network (GAN), can be used to convert the facial features into natural-looking faces. This model processes in real time, preserving the original facial angle and facial expression.

[0308] Furthermore, the server changes parts other than the face (hairstyle and clothing) and converts them to specific attributes (for example, anime character style or foreign style) based on the option settings selected by the user. Specifically, if the user selects "anime character style," the face and clothing of the person in the video are converted to the corresponding character design.

[0309] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technology to minimize delays, allowing viewers to enjoy natural, high-quality video.

[0310] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the face of the streamer that has been registered in advance, and converts it into a different face using an AI model. This allows viewers to view the converted face or video with specific attributes in real time, allowing live streaming to be carried out safely without violating portrait rights while maintaining the naturalness of the video.

[0311] The processing flow will be explained below.

[0312] Step 1:

[0313] A user launches a live streaming app and completes the initial setup. At this time, the user points their face at the camera and scans it, registering their facial recognition data as a streamer into the system. This will exclude the streamer's face from subsequent facial conversion processing.

[0314] Step 2:

[0315] The device activates the camera and captures video data in real time. The device generates the captured video data in a stream format and continues to acquire video data at a constant frame rate.

[0316] Step 3:

[0317] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection over the network.

[0318] Step 4:

[0319] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video, identifying faces in each frame of the video data and extracting the necessary information.

[0320] Step 5:

[0321] The server matches the faces in the video with pre-registered facial data of the streamer and filters out any matches. This process involves comparing facial feature vectors to reliably identify the streamer's face.

[0322] Step 6:

[0323] The server applies an AI model (such as a generative artificial network (GAN)) to the remaining faces and converts them into a different face in real time. The AI ​​model generates natural-looking facial features that reflect the angle and expression of the original face.

[0324] Step 7:

[0325] The server checks the user's option settings and, if necessary, changes non-facial features (hairstyle and clothing) or converts them to specified attributes (e.g., anime character style or foreign character style), thereby processing the video in the specific style desired by the user.

[0326] Step 8:

[0327] The server then delivers the converted video data to viewers in real time, with minimal latency and using high-speed data processing technology.

[0328] Step 9:

[0329] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0330] Step 10:

[0331] The server continues to process the video data in real time, converting and distributing it consistently until the end of the broadcast, allowing viewers to receive consistently high-quality video.

[0332] Example 1

[0333] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0334] When live streaming or posting images from the street, it is important to provide natural-looking images while avoiding infringing on the portrait rights of people around you. Another challenge is to convert the video in real time and deliver it to viewers with high quality and low latency, without altering the broadcaster's own face.

[0335] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0336] In this invention, the server includes: a means for a user to launch a live streaming app and scan their face to register facial recognition data; a means for the device to launch a camera and capture video data in real time and generate it in a stream format at a constant frame rate; a means for the server to analyze the received video data and detect human faces in the video using a facial recognition algorithm; a means for converting the detected human faces into other faces using a generative AI model; a means for modifying parts other than the face or converting them into specific attributes based on option settings selected by the user; and a means for distributing the converted video data in real time. This makes it possible to distribute high-quality, natural-looking video in real time while protecting portrait rights.

[0337] A "user" is an entity that operates the system and performs initial setup and live streaming.

[0338] A "live streaming app" is application software that users use to stream video in real time.

[0339] "Facial recognition data" refers to image data and characteristic information of a user's face that is scanned with a camera and registered in the system.

[0340] A "terminal" is a device equipped with a camera that captures and transmits video to a server.

[0341] A "camera" is a device that captures video data in real time.

[0342] "Video data" refers to video information captured by a camera.

[0343] "Stream format" is a method of transmitting data continuously in real time.

[0344] The "server" is a computer system that analyzes the received video data, performs facial recognition and conversion, and distributes it in real time.

[0345] A "face recognition algorithm" is a program for detecting and identifying the faces of people in video footage.

[0346] A "generative AI model" is an artificial intelligence model that generates new data based on input data, and in this invention is used to convert a specific face into another face.

[0347] "Optional settings" are additional settings that a user can make to the video being distributed, including changes to parts other than the face and conversion to specific attributes.

[0348] "Real-time distribution" is a method in which converted video data is immediately provided to viewers.

[0349] "Portrait rights" are the right of individuals to prevent their portraits, including their face and figure, from being photographed or used without their permission.

[0350] The system of this invention protects portrait rights and provides natural-looking images by using an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from around the city. The following hardware and software are used to implement this system.

[0351] First, the user launches the live streaming app and performs the initial setup. They then scan their face using the camera on their smartphone or PC, and register this facial recognition data in the system. This data is used by the facial recognition algorithm to exclude the streamer's face from conversion.

[0352] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format (e.g., RTMP protocol) at a frame rate of 30 frames per second and sent to the server.

[0353] The server analyzes the received video data and uses the Python library OpenCV, Dlib, or other facial recognition algorithms to detect faces in the video. At this time, it references the facial data of pre-registered streamers and excludes any matches from the conversion.

[0354] The server then uses a generative anti-aliasing network (GAN) to transform any detected faces that are not rejected into a new face. Using deep learning frameworks like TensorFlow and PyTorch, the server uses a pre-trained model to process the transformed faces in real time so that they appear natural. This model converts the input face into a new face while preserving its original angle and facial expression.

[0355] Furthermore, the server also processes changes to non-facial features (hairstyle and clothing) and conversion to specific attributes (such as anime character style or foreign style) based on the options selected by the user within the app. For example, if the user selects "anime character style," the server uses a style conversion model such as StyleGAN to convert the face and clothing of the person in the video into an anime character design.

[0356] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technologies such as the RTMP protocol to minimize delays. Viewers can enjoy this natural, high-quality video on their smartphones, PCs, or web browsers using viewing apps.

[0357] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the broadcaster's pre-registered face, and converts it to a different face using an AI model. This allows viewers to view the converted face or video with specific attributes converted in real time, allowing live streaming to be carried out safely while maintaining the naturalness of the video and without violating portrait rights.

[0358] Examples of prompts:

[0359] "Generate the program code for a system that transforms the faces of people around you into other faces during a live stream. In this system, the streamer's face will be pre-registered and will not be transformed. The AI ​​model used is a generative adversarial network (GAN), and the transformed footage will be distributed to viewers in real time."

[0360] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0361] Step 1:

[0362] The user launches the live streaming app and scans their face to register facial recognition data. In this step, the user uses the camera on their smartphone or PC to perform the facial scanning operation. The input is the user's facial image data, and the output is facial recognition data registered in the system. This recognition data is used to identify the broadcaster's face and exclude it in the later face conversion step. Specifically, the user follows the app's guide to point the camera at their own face and perform a scan for a few seconds.

[0363] Step 2:

[0364] The device starts the camera and captures video data in real time. In this step, the device captures video at a frame rate of 30 frames per second. The input is video data from the camera, and the output is stream-format data generated in real time. Specifically, the device starts the camera, and the live streaming app captures video data in real time.

[0365] Step 3:

[0366] The video data captured by the terminal is sent to the server in real time. In this step, the video data is sent using a protocol such as RTMP. The input is the captured video data in a stream format, and the output is the video data sent to the server. Specifically, the terminal sends the video data to the server via a network connection.

[0367] Step 4:

[0368] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. The input is the real-time video data sent to the server, and the output is a list of detected faces. Here, the server applies the face detection algorithm using the Python libraries OpenCV and Dlib. Specifically, the server analyzes each frame and identifies the location and features of faces.

[0369] Step 5:

[0370] The server excludes the faces of pre-registered streamers from the detected faces. The input is a list of detected faces and facial recognition data of pre-registered streamers, and the output is a list of faces excluding the streamers' faces. Specifically, the server compares the facial recognition data and removes matching faces from the list.

[0371] Step 6:

[0372] The server uses a generative AI model to convert the detected face into a different face. The input is a list of faces excluding the streamer's face and video data containing that face, and the output is the converted video data. The server uses TensorFlow or PyTorch to apply a pre-trained generative adversarial network (GAN). Specifically, the server performs face conversion in real time, converting the face into a new face while preserving the original facial angle and expression.

[0373] Step 7:

[0374] The server modifies non-facial features or converts specific attributes based on the user's options. The input is video data containing the converted face and the user's options, and the output is video data that has been further converted according to the options. Specifically, the server applies a style transfer model such as StyleGAN to change the hairstyle or clothing of people in the video.

[0375] Step 8:

[0376] The server finally delivers the converted video data to the viewer in real time. The input is the final converted video data, and the output is the real-time video delivered to the viewer. Specifically, the server uses the RTMP protocol to send video data at high speed to the viewer, who then views the video without delay using a viewing app or web browser on their smartphone or PC.

[0377] (Application example 1)

[0378] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0379] When live streaming or posting images from the street or inside a store, there is a risk of violating the portrait rights of other people. In particular, in physical stores, where many customers come and go, the faces of other customers appearing in posted photos and videos can be a problem. This can lead to the risk of violating the privacy of other customers, so there is a need to solve this problem.

[0380] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0381] In this invention, the server includes means for detecting human faces in video data in real time, means for converting the detected human faces into other faces using an AI model, means for distributing the converted video data in real time, means for excluding the faces of specific people registered in advance from the conversion targets, means for changing hairstyles and clothing and converting into specific attributes, and means for processing photos and videos taken by customers in the store in real time and converting the faces of other customers into other faces. This enables customers to safely post photos and videos taken in the store to social media etc.

[0382] "Real-time" refers to data processing and information delivery occurring immediately and without delay.

[0383] "Video data" refers to data containing visual information obtained from a camera or other imaging device.

[0384] "Means for detecting human faces" refers to technology that uses specific algorithms to identify and extract human faces from video data.

[0385] "AI model" refers to a mathematical or algorithmic model trained to perform a specific task using artificial intelligence techniques.

[0386] "Method of converting into another face" refers to technology that converts the detected face of a person into a different facial image using an AI model.

[0387] "Means for delivering converted video data in real time" refers to technology that instantly provides viewers with video data converted by an AI model.

[0388] "Means to exclude the faces of specific people registered in advance from conversion targets" refers to technology that pre-registers the faces of specific people, such as streamers, in a database, and does not perform conversion if that face is detected in the video.

[0389] "Methods for changing hairstyles and clothing" refers to technology that changes the hairstyles and clothing of people in video data to different styles.

[0390] "Means of converting to specific attributes" refers to technology that changes the faces and clothing of people in video data to specific designs, such as those of anime characters or foreigners.

[0391] "Store visitors" refers to customers or clients who visit a physical store.

[0392] "Means for processing photos and videos in real time" refers to technology that instantly processes and converts photos and videos taken by customers inside the store.

[0393] "Means of converting the faces of other customers into different faces" refers to technology that uses an AI model to convert the faces of other customers photographed in the store into images of different faces.

[0394] "SNS" is an abbreviation for "social networking service," which refers to a platform where users can share and interact with each other via the Internet.

[0395] The embodiments for carrying out the present invention are described as follows.

[0396] The system of the present invention uses a smartphone to process photos and videos taken in a physical store in real time, protecting the portrait rights of other customers. This system is realized using the following hardware and software components.

[0397] Hardware

[0398] Smartphone: The device on which this system is implemented, capturing and processing video data.

[0399] Camera module: A smartphone's built-in camera that takes photos and videos in real time.

[0400] software

[0401] An open-source face recognition library (e.g., OpenCV 4.5.3) for detecting human faces in video data.

[0402] Generative Adversarial Network (GAN) models (e.g., StyleGAN2): Used to transform a detected face into another face in real time.

[0403] Real-time video streaming library (e.g., FFmpeg 4.4): Used to instantly deliver converted video data to viewers.

[0404] System Operation

[0405] 1. The user launches the smartphone app and performs the initial setup, which includes registering the user's own facial data. This facial data is later used as an exclusion target during the facial conversion process.

[0406] 2. The camera module is activated and the device (smartphone) captures video data in real time. The video data is acquired at a constant frame rate and generated in a stream format.

[0407] 3. The acquired video data is analyzed in real time using a facial recognition library such as OpenCV. The server detects faces in the video and compares them with pre-registered user facial data, excluding any matches from the conversion target.

[0408] 4. The detected face is transformed into another face in real time using a generative adversarial network (GAN) model such as StyleGAN2. This transformation process is fast and preserves the original face's natural angle and facial expression.

[0409] 5. Based on the options selected by the user, the server changes non-facial features (such as hairstyle and clothing) and converts the user to specific attributes. For example, it is possible to convert the user to an anime character or a foreign character.

[0410] 6. The converted video data is then distributed to viewers in real time via a video streaming library such as FFmpeg, allowing viewers to enjoy high-quality, seamless video.

[0411] Specific examples

[0412] For example, if a user takes a photo inside a physical store, the system can detect the faces of other customers and convert them into different faces before distributing the video, thereby avoiding portrait rights issues. This allows users to post photos and videos to social media with peace of mind.

[0413] Prompt Sentence Examples

[0414] "Please create a camera app that protects portrait rights by converting the faces of other customers in real time into videos taken in a store."

[0415] In this way, the present invention simultaneously realizes protection of portrait rights and provision of natural-looking images in brick-and-mortar stores and public places.

[0416] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0417] Step 1:

[0418] The user launches the smartphone app and performs the initial setup. Here, the user scans their face data with the camera and registers it in the system. The user's facial image data is used as input, and this is the output that is registered in the facial recognition database. This data is later used as an exclusion target during the face conversion process.

[0419] Step 2:

[0420] The device activates the camera module and captures video data in real time. The input is video data acquired from the camera, and the output is video data in a stream format generated at a constant frame rate. This video data is immediately sent to the server.

[0421] Step 3:

[0422] The server analyzes the received video data using an open-source face recognition library (e.g., OpenCV 4.5.3). The input is stream-format video data, and the output is data indicating the position of people's faces in the video. The server detects faces in the video based on this data.

[0423] Step 4:

[0424] The server compares the detected person's face with the user's face data registered in advance. The input is face position data and the user's face recognition data, and the output is a list of faces to be excluded from conversion. This prevents conversion even if the user's face data is included in the video.

[0425] Step 5:

[0426] The server uses a generative adversarial network (GAN) model (e.g., StyleGAN2) to convert the detected face into a different face in real time. The input is the recognized face position data and facial image data in the video, and the output is the converted facial image data. This data processing allows the conversion to be performed while maintaining the natural angle and expression of the original face.

[0427] Step 6:

[0428] The server changes parts of the character other than the face (such as hairstyle or clothing) based on the options selected by the user. The input is the user's options and video data, and the output is the video data after the changes. This data calculation results in a transformation to make the character look like an anime character or a foreign character, depending on the settings.

[0429] Step 7:

[0430] The server then uses a real-time video streaming library (e.g., FFmpeg 4.4) to deliver the converted video data to viewers. The input is the converted video data, and the output is the converted video delivered in real time. This allows viewers to enjoy high-quality video without delay.

[0431] These processing steps enable the system to safely broadcast live footage from within a physical store, providing natural-looking footage without violating the portrait rights of other customers.

[0432] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0433] The system of the present invention has the function of detecting human faces in video data in real time, converting the detected human faces into other faces using an AI model, and distributing the converted video data in real time. It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for changing hairstyles, clothing, and converting specific attributes. It also combines an emotion engine that recognizes user emotions, and can perform conversion processing and attribute changes within the video data based on the recognized emotions.

[0434] The implementation of this system is described in detail below.

[0435] The user launches the live streaming app and performs the initial setup. The streamer then points their face at the camera and scans it, which is then registered as facial recognition data in the system. This data is then used to exclude the streamer from face conversion later.

[0436] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format and sent to the system at a constant frame rate.

[0437] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate technology to identify faces in each frame. It then references the facial data of streamers registered in advance and excludes faces that match from the conversion target.

[0438] The server then applies an AI model (e.g., a generative artificial network, or GAN) to the detected face to transform it into a different face in real time. The AI ​​model generates natural-looking facial features while preserving the original facial angle and expression.

[0439] Furthermore, an emotion engine is introduced to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The analysis results are classified into emotion categories such as joy, sadness, and anger.

[0440] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, additional processing can be performed to make the face in the video brighter or to change it into a smiling face. The attributes of the face and clothing generated by the AI ​​model can also be changed according to the emotion. For example, if the user selects an anime-style setting and is sad, detailed expressions such as the character shedding tears can be added.

[0441] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0442] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0443] As described above, this system performs video processing that reflects the user's emotions, making it possible to provide viewers with consistently high-quality, natural-looking video.

[0444] The processing flow will be explained below.

[0445] Step 1:

[0446] The user launches the live streaming app and performs the initial setup. The user then points their face at the camera, scans it, and registers facial recognition data into the system. This data is used to identify the face as the streamer.

[0447] Step 2:

[0448] The device activates the camera for live streaming and captures video data in real time, which is generated in a stream format at a constant frame rate.

[0449] Step 3:

[0450] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection with the server via the network.

[0451] Step 4:

[0452] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. High-precision facial recognition technology is used to identify faces in each frame of the video data.

[0453] Step 5:

[0454] The server compares the detected face data with the face data of the streamer registered in advance and excludes any matches from the conversion process. This means that the streamer's face is excluded from the conversion process.

[0455] Step 6:

[0456] The server then applies an AI model to the remaining faces, transforming them into a different face in real time. The AI ​​model (e.g., a generative adversarial network: GAN) generates natural-looking facial features while preserving the original facial angle and expression.

[0457] Step 7:

[0458] The device uses an emotion engine to recognize the user's emotions and analyzes them in real time through cameras and sensors, categorizing the results into emotion categories such as joy, sadness, and anger.

[0459] Step 8:

[0460] The server adjusts the transformation process in the video data based on the user's emotions recognized by the emotion engine. For example, if the server recognizes that the user is happy, it may transform the face of a person in the video into a brighter expression.

[0461] Step 9:

[0462] The server changes the facial and clothing attributes generated by the AI ​​model based on the user's emotions recognized by the emotion engine. For example, if the user selects the "anime character" setting, character expressions corresponding to the emotion will be added.

[0463] Step 10:

[0464] The server then delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0465] Step 11:

[0466] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0467] Step 12:

[0468] The server processes the video data in real time, converting and delivering it consistently until the end of the broadcast, allowing viewers to receive high-quality, natural video continuously.

[0469] Example 2

[0470] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0471] In real-time video streaming, there is a demand for providing viewers with higher quality and more natural-looking images. However, in technology for converting faces in streamed video, there is a lack of methods for excluding the streamer's own face from the conversion target, or for processing video that reflects the user's emotions. Therefore, there is a need for the development of a system that can appropriately convert the faces of people other than the streamer during live streaming, and further recognize the user's emotions and reflect them in the video.

[0472] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0473] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for distributing the converted video data in real time, means for identifying a user's emotion using an emotion engine, and means for adjusting the conversion process in the video data based on the identified emotion. This makes it possible to naturally convert the faces of people other than the broadcaster in real time during broadcasting, and further to reflect the user's emotion in the video.

[0474] "Real-time" refers to the ability of a system to operate such that data is processed and responded to almost immediately, implying very little processing delay.

[0475] "Video data" means data that describes visual information captured by a camera or other imaging device.

[0476] "Human face" refers to the facial features of an individual human being within video data, and specifically refers to the area that can be identified by facial recognition technology.

[0477] "Means for detecting" refers to a combination of hardware and software capable of identifying and extracting a particular target.

[0478] "Generative AI models" refer to algorithms that use artificial intelligence to generate new data and features, and specifically include generative artificial network (GAN).

[0479] "Means for converting" refers to software or hardware that has the functionality to change data of one format into another format.

[0480] "Distribution means" refers to a system capable of transmitting data to a specific receiver and making it available in a usable form.

[0481] "Emotion engine" refers to a combination of software and hardware for recognizing and analyzing user emotions in real time.

[0482] "User emotion" refers to the user's psychological state as recognized from the video and audio data being distributed, and includes emotion categories such as joy, sadness, and anger.

[0483] "Adjusting means" refers to hardware and software that has the functionality to change the system's behavior or output based on specific conditions.

[0484] The system of the present invention has the function of detecting human faces in video data in real time during live streaming, converting the detected faces into other faces using a generative AI model, and distributing the converted video data in real time.It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for recognizing user emotions and changing the attributes of the video based on the emotions.

[0485] The user launches the live streaming app and performs the initial setup. For example, the streamer points the camera at their face and scans it, registering it as facial recognition data in the system. This data is later used to exclude the streamer from face conversion. The user scans their face using a smartphone or computer, following the app's instructions.

[0486] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the system at a certain frame rate (e.g., 30 fps). At this time, the device uses a high-resolution camera to capture clear video.

[0487] The server analyzes the received video data and uses a facial recognition algorithm to detect faces of people appearing in the video. The facial detection algorithm uses open-source facial recognition libraries (e.g., dlib or OpenCV) to perform highly accurate face detection. The server performs face detection for each series of frames and excludes matching faces from conversion by referencing the facial data of broadcasters registered in advance.

[0488] The server then applies a generative AI model to the detected face, transforming it into a different face in real time. This model uses generative artificial intelligence (GAN) or deep learning. For example, GANs preserve the original facial features while generating new facial features. During this process, the server carefully transforms the face to maintain the original facial angle and facial expression.

[0489] Furthermore, an emotion engine is implemented to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The emotion engine uses facial expression recognition technology and voice analysis technology to classify the user's emotions into joy, sadness, anger, etc. The analysis results are sent to the server in real time.

[0490] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, the server adds a process to brighten the face in the video and transform it into a smiling face. The attributes of the face and clothing generated by the generative AI model are also changed according to the user's emotions. For example, if the user selects an anime-style setting and is sad, the character will shed tears.

[0491] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay via live streaming platforms (such as YouTube Live or Twitch).

[0492] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0493] Example prompt sentence:

[0494] Start a live stream at a tourist spot, and have it detect people's faces in real time and convert their expressions into happy expressions. The streamer's own face will not be converted, ensuring a fun-filled stream for your viewers.

[0495] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0496] Step 1:

[0497] The user launches the live streaming app and performs the initial setup. The user then points the camera at their face to scan it and registers it as facial recognition data in the system. The facial image captured by the camera is used as input, and a facial recognition algorithm is applied to obtain the broadcaster's facial data. As output, the broadcaster's facial data is stored in the system's database.

[0498] Step 2:

[0499] The device activates the camera for live streaming and captures video data in real time. As input, it uses the video stream from the camera and captures data at a constant frame rate (e.g., 30 fps). As output, the captured video data is generated in a stream format and sent to the server.

[0500] Step 3:

[0501] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. As input, it uses the received video frames and applies a facial recognition algorithm (e.g., dlib or OpenCV). As output, it generates location information and feature point data of the detected faces.

[0502] Step 4:

[0503] The server references the face data of the streamer registered in advance and compares it with the detected face. As input, it uses the face data of the streamer registered in step 1 and the detected face data obtained in step 3. As output, it generates a list in which matching faces are excluded from conversion.

[0504] Step 5:

[0505] The server applies a generative AI model to the detected face and transforms it into a different face in real time. It uses facial position information and feature point data as input and uses a generative artificial network (GAN). The output is image data of the transformed face. During this process, the original facial angle and facial expression are maintained.

[0506] Step 6:

[0507] The emotion engine analyzes the user's emotions in real time using a camera or sensor on the device that can identify the user's emotions. As input, it uses data obtained from the camera or sensor (facial expressions and voice data) and applies emotion analysis algorithms. As output, it generates an analysis result of emotion categories such as joy, sadness, anger, etc. and sends it to the server.

[0508] Step 7:

[0509] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process in the video data. The inputs are the emotion analysis results and the transformed facial image data. The output is video data that has been adjusted according to the emotion. For example, if the user is happy, the face in the video is made brighter and transformed into a smiling face.

[0510] Step 8:

[0511] The server delivers the converted video data to viewers in real time. The server uses the adjusted video data as input and sends the data via a live streaming platform. The output is high-quality, latency-free video delivered to viewers, allowing them to enjoy a natural, real-time experience.

[0512] (Application example 2)

[0513] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0514] Providing a consistently visually pleasing experience to viewers is challenging in current live streaming and video distribution services. In particular, there is a need for technology that can flexibly address the needs of privacy protection and emotion-based video conversion. Furthermore, there is no system that can exclude specific people's faces from conversion and dynamically process video based on emotions in real time.

[0515] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0516] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for delivering the converted video data in real time, and means for adjusting the conversion processing in the video data based on the recognized emotion using an emotion engine that recognizes emotion. This makes it possible to provide viewers with consistently high-quality, natural video and realize video processing that corresponds to the user's emotion in real time.

[0517] "Real-time" refers to a state in which results are available immediately at the moment a process or operation is performed.

[0518] "Video Data" refers to video and image information captured by a camera or other device.

[0519] "Means for detecting faces" refers to technologies and algorithms that identify and locate human faces within video data.

[0520] "Generative AI models" refer to artificial intelligence models that learn from specific input data and generate new data. This includes technologies such as generative artificial network (GAN).

[0521] "Face conversion method" refers to the technology or algorithm used to replace a detected person's face with a different face.

[0522] "Means of distribution" refers to the technology and infrastructure used to transmit and display the converted video data to viewers in real time via the Internet.

[0523] An "emotion engine" refers to the technology or algorithms used to identify and categorize people's emotions.

[0524] "Means for adjusting the transformation process" refers to technologies and algorithms that dynamically change and optimize the way video data is transformed based on emotion or other criteria.

[0525] "Means for excluding the faces of specific people from conversion targets" refers to technologies or algorithms that detect the faces of specific people registered in advance and exclude them from real-time conversion targets.

[0526] "Means for changing hairstyles and clothing, and converting to specific attributes" refers to technologies and algorithms that change the hairstyles and clothing of people in video data to different ones and convert them to specific attributes (for example, to look like an anime character).

[0527] The present invention provides a system that detects human faces in video data in real time, converts the detected faces into different faces using a generative AI model, and distributes the video in real time. Specific embodiments are described below.

[0528] First, the user launches the live streaming application on their device, which can be a smartphone, smart glasses, or a head-mounted display. The application has a function that allows the streamer to point the camera at their face, scan it, and register it as facial recognition data in the system. This data is used to exclude the streamer's face from conversion targets.

[0529] Next, the device activates the camera and captures video data in real time. The captured video data is sent to the server as a stream at a fixed frame rate. The server analyzes the received video data and uses dlib's facial recognition algorithm to detect human faces in the video.

[0530] The server applies a generative AI model (e.g., a generative adversarial network: GAN) to the detected face and converts it into a different face in real time. This generative AI model generates natural-looking facial features and can preserve the original facial angle and expression. It also references the facial data of pre-registered streamers and excludes matching faces from the conversion target.

[0531] Furthermore, the server introduces an emotion engine to recognize emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time and adjust the conversion process within the video data based on the analysis results. For example, if the user is happy, the server can convert the face in the video into a bright, smiling face.

[0532] The server then delivers the converted video data to viewers in real time, allowing viewers to instantly enjoy high-quality, natural-looking video and a visual experience that reflects the broadcaster's intentions and emotions.

[0533] As a concrete example, consider a case where a user is live streaming at a tourist spot and the emotion engine detects "joy" during the stream. In this case, the faces of all other people in the stream will be converted to bright, smiling faces. An example of a prompt sentence is as follows:

[0534] Example prompt sentence:

[0535] "Identify the faces of people in a photo, recognize the emotions on those faces, and convert them into appropriate face images based on that. 1. Face recognition: using dlib 2. Emotion recognition: using EmotionRecognizer 3. Emotion-based face transformation: using GAN to generate natural facial expressions"

[0536] In this way, the system is able to respond to the user's emotions and provide high quality, visually pleasing video in real time.

[0537] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0538] Step 1:

[0539] The user launches the live streaming application on their device. They point the camera at their own face, scan it, and register it as facial recognition data in the system. This data is sent from the device to the server and used to exclude the broadcaster's face from conversion. The input is facial image data from the camera, and the output is facial recognition data.

[0540] Step 2:

[0541] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the server at a constant frame rate. The input is video data from the camera, and the output is video data in stream format.

[0542] Step 3:

[0543] The server prepares to analyze the received video data. It uses dlib's facial recognition algorithm to detect faces in the video. It references the face data of pre-registered streamers and excludes matching faces from the conversion target. The input is stream-format video data, and the output is the face coordinates and the streamer's face data detection results.

[0544] Step 4:

[0545] The server applies a generative AI model (e.g., a generative artificial network (GAN)) to the detected face and converts it into a different face in real time, preserving the original facial angle and expression. The input is the face coordinates and facial image data, and the output is the converted facial image data.

[0546] Step 5:

[0547] The emotion engine analyzes the user's emotions in real time using cameras and sensors that can identify the user's emotions. The analysis results are classified into emotion categories such as joy, sadness, and anger. The input is data from the cameras and sensors, and the output is emotion category data.

[0548] Step 6:

[0549] The server adjusts the transformation process in the video data based on the emotion category. For example, if the user is happy, additional processing is performed to transform the face in the video into a brighter, smiling face. The input is emotion category data and transformed face image data, and the output is adjusted face image data.

[0550] Step 7:

[0551] The server delivers the converted video data to viewers in real time, allowing viewers to enjoy natural, high-quality video without delay. The input is adjusted video data, and the output is video data delivered in real time.

[0552] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0553] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0554] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0555] [Third embodiment]

[0556] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0557] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0558] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0559] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0560] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0561] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0562] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0563] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0564] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0565] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0566] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0567] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0568] The system of the present invention uses an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from the street, providing natural-looking images while protecting portrait rights. The implementation of this system follows the steps below.

[0569] First, the user launches the live streaming app and performs the initial setup. The user then scans their face with the camera and registers their facial recognition data as a streamer in the system. This data is then used to exclude the streamer from face conversion later.

[0570] Next, the device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format at a constant frame rate. This stream data is then sent to the server in real time.

[0571] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate facial recognition technology to identify faces in each frame. In doing so, it references the face data of the streamer registered in advance and compares it with faces in the video, excluding any that match from the conversion target.

[0572] The server then uses an AI model to convert the faces that were not rejected into a different face. The AI ​​model, for example, a generative artificial network (GAN), can be used to convert the facial features into natural-looking faces. This model processes in real time, preserving the original facial angle and facial expression.

[0573] Furthermore, the server changes parts other than the face (hairstyle and clothing) and converts them to specific attributes (for example, anime character style or foreign style) based on the option settings selected by the user. Specifically, if the user selects "anime character style," the face and clothing of the person in the video are converted to the corresponding character design.

[0574] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technology to minimize delays, allowing viewers to enjoy natural, high-quality video.

[0575] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the face of the streamer that has been registered in advance, and converts it into a different face using an AI model. This allows viewers to view the converted face or video with specific attributes in real time, allowing live streaming to be carried out safely without violating portrait rights while maintaining the naturalness of the video.

[0576] The processing flow will be explained below.

[0577] Step 1:

[0578] A user launches a live streaming app and completes the initial setup. At this time, the user points their face at the camera and scans it, registering their facial recognition data as a streamer into the system. This will exclude the streamer's face from subsequent facial conversion processing.

[0579] Step 2:

[0580] The device activates the camera and captures video data in real time. The device generates the captured video data in a stream format and continues to acquire video data at a constant frame rate.

[0581] Step 3:

[0582] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection over the network.

[0583] Step 4:

[0584] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video, identifying faces in each frame of the video data and extracting the necessary information.

[0585] Step 5:

[0586] The server matches the faces in the video with pre-registered facial data of the streamer and filters out any matches. This process involves comparing facial feature vectors to reliably identify the streamer's face.

[0587] Step 6:

[0588] The server applies an AI model (such as a generative artificial network (GAN)) to the remaining faces and converts them into a different face in real time. The AI ​​model generates natural-looking facial features that reflect the angle and expression of the original face.

[0589] Step 7:

[0590] The server checks the user's option settings and, if necessary, changes non-facial features (hairstyle and clothing) or converts them to specified attributes (e.g., anime character style or foreign character style), thereby processing the video in the specific style desired by the user.

[0591] Step 8:

[0592] The server then delivers the converted video data to viewers in real time, with minimal latency and using high-speed data processing technology.

[0593] Step 9:

[0594] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0595] Step 10:

[0596] The server continues to process the video data in real time, converting and distributing it consistently until the end of the broadcast, allowing viewers to receive consistently high-quality video.

[0597] Example 1

[0598] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0599] When live streaming or posting images from the street, it is important to provide natural-looking images while avoiding infringing on the portrait rights of people around you. Another challenge is to convert the video in real time and deliver it to viewers with high quality and low latency, without altering the broadcaster's own face.

[0600] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0601] In this invention, the server includes: a means for a user to launch a live streaming app and scan their face to register facial recognition data; a means for the device to launch a camera and capture video data in real time and generate it in a stream format at a constant frame rate; a means for the server to analyze the received video data and detect human faces in the video using a facial recognition algorithm; a means for converting the detected human faces into other faces using a generative AI model; a means for modifying parts other than the face or converting them into specific attributes based on option settings selected by the user; and a means for distributing the converted video data in real time. This makes it possible to distribute high-quality, natural-looking video in real time while protecting portrait rights.

[0602] A "user" is an entity that operates the system and performs initial setup and live streaming.

[0603] A "live streaming app" is application software that users use to stream video in real time.

[0604] "Facial recognition data" refers to image data and characteristic information of a user's face that is scanned with a camera and registered in the system.

[0605] A "terminal" is a device equipped with a camera that captures and transmits video to a server.

[0606] A "camera" is a device that captures video data in real time.

[0607] "Video data" refers to video information captured by a camera.

[0608] "Stream format" is a method of transmitting data continuously in real time.

[0609] The "server" is a computer system that analyzes the received video data, performs facial recognition and conversion, and distributes it in real time.

[0610] A "face recognition algorithm" is a program for detecting and identifying the faces of people in video footage.

[0611] A "generative AI model" is an artificial intelligence model that generates new data based on input data, and in this invention is used to convert a specific face into another face.

[0612] "Optional settings" are additional settings that a user can make to the video being distributed, including changes to parts other than the face and conversion to specific attributes.

[0613] "Real-time distribution" is a method in which converted video data is immediately provided to viewers.

[0614] "Portrait rights" are the right of individuals to prevent their portraits, including their face and figure, from being photographed or used without their permission.

[0615] The system of this invention protects portrait rights and provides natural-looking images by using an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from around the city. The following hardware and software are used to implement this system.

[0616] First, the user launches the live streaming app and performs the initial setup. They then scan their face using the camera on their smartphone or PC, and register this facial recognition data in the system. This data is used by the facial recognition algorithm to exclude the streamer's face from conversion.

[0617] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format (e.g., RTMP protocol) at a frame rate of 30 frames per second and sent to the server.

[0618] The server analyzes the received video data and uses the Python library OpenCV, Dlib, or other facial recognition algorithms to detect faces in the video. At this time, it references the facial data of pre-registered streamers and excludes any matches from the conversion.

[0619] The server then uses a generative anti-aliasing network (GAN) to transform any detected faces that are not rejected into a new face. Using deep learning frameworks like TensorFlow and PyTorch, the server uses a pre-trained model to process the transformed faces in real time so that they appear natural. This model converts the input face into a new face while preserving its original angle and facial expression.

[0620] Furthermore, the server also processes changes to non-facial features (hairstyle and clothing) and conversion to specific attributes (such as anime character style or foreign style) based on the options selected by the user within the app. For example, if the user selects "anime character style," the server uses a style conversion model such as StyleGAN to convert the face and clothing of the person in the video into an anime character design.

[0621] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technologies such as the RTMP protocol to minimize delays. Viewers can enjoy this natural, high-quality video on their smartphones, PCs, or web browsers using viewing apps.

[0622] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the broadcaster's pre-registered face, and converts it to a different face using an AI model. This allows viewers to view the converted face or video with specific attributes converted in real time, allowing live streaming to be carried out safely while maintaining the naturalness of the video and without violating portrait rights.

[0623] Examples of prompts:

[0624] "Generate the program code for a system that transforms the faces of people around you into other faces during a live stream. In this system, the streamer's face will be pre-registered and will not be transformed. The AI ​​model used is a generative adversarial network (GAN), and the transformed footage will be distributed to viewers in real time."

[0625] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0626] Step 1:

[0627] The user launches the live streaming app and scans their face to register facial recognition data. In this step, the user uses the camera on their smartphone or PC to perform the facial scanning operation. The input is the user's facial image data, and the output is facial recognition data registered in the system. This recognition data is used to identify the broadcaster's face and exclude it in the later face conversion step. Specifically, the user follows the app's guide to point the camera at their own face and perform a scan for a few seconds.

[0628] Step 2:

[0629] The device starts the camera and captures video data in real time. In this step, the device captures video at a frame rate of 30 frames per second. The input is video data from the camera, and the output is stream-format data generated in real time. Specifically, the device starts the camera, and the live streaming app captures video data in real time.

[0630] Step 3:

[0631] The video data captured by the terminal is sent to the server in real time. In this step, the video data is sent using a protocol such as RTMP. The input is the captured video data in a stream format, and the output is the video data sent to the server. Specifically, the terminal sends the video data to the server via a network connection.

[0632] Step 4:

[0633] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. The input is the real-time video data sent to the server, and the output is a list of detected faces. Here, the server applies the face detection algorithm using the Python libraries OpenCV and Dlib. Specifically, the server analyzes each frame and identifies the location and features of faces.

[0634] Step 5:

[0635] The server excludes the faces of pre-registered streamers from the detected faces. The input is a list of detected faces and facial recognition data of pre-registered streamers, and the output is a list of faces excluding the streamers' faces. Specifically, the server compares the facial recognition data and removes matching faces from the list.

[0636] Step 6:

[0637] The server uses a generative AI model to convert the detected face into a different face. The input is a list of faces excluding the streamer's face and video data containing that face, and the output is the converted video data. The server uses TensorFlow or PyTorch to apply a pre-trained generative adversarial network (GAN). Specifically, the server performs face conversion in real time, converting the face into a new face while preserving the original facial angle and expression.

[0638] Step 7:

[0639] The server modifies non-facial features or converts specific attributes based on the user's options. The input is video data containing the converted face and the user's options, and the output is video data that has been further converted according to the options. Specifically, the server applies a style transfer model such as StyleGAN to change the hairstyle or clothing of people in the video.

[0640] Step 8:

[0641] The server finally delivers the converted video data to the viewer in real time. The input is the final converted video data, and the output is the real-time video delivered to the viewer. Specifically, the server uses the RTMP protocol to send video data at high speed to the viewer, who then views the video without delay using a viewing app or web browser on their smartphone or PC.

[0642] (Application example 1)

[0643] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0644] When live streaming or posting images from the street or inside a store, there is a risk of violating the portrait rights of other people. In particular, in physical stores, where many customers come and go, the faces of other customers appearing in posted photos and videos can be a problem. This can lead to the risk of violating the privacy of other customers, so there is a need to solve this problem.

[0645] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0646] In this invention, the server includes means for detecting human faces in video data in real time, means for converting the detected human faces into other faces using an AI model, means for distributing the converted video data in real time, means for excluding the faces of specific people registered in advance from the conversion targets, means for changing hairstyles and clothing and converting into specific attributes, and means for processing photos and videos taken by customers in the store in real time and converting the faces of other customers into other faces. This enables customers to safely post photos and videos taken in the store to social media etc.

[0647] "Real-time" refers to data processing and information delivery occurring immediately and without delay.

[0648] "Video data" refers to data containing visual information obtained from a camera or other imaging device.

[0649] "Means for detecting human faces" refers to technology that uses specific algorithms to identify and extract human faces from video data.

[0650] "AI model" refers to a mathematical or algorithmic model trained to perform a specific task using artificial intelligence techniques.

[0651] "Method of converting into another face" refers to technology that converts the detected face of a person into a different facial image using an AI model.

[0652] "Means for delivering converted video data in real time" refers to technology that instantly provides viewers with video data converted by an AI model.

[0653] "Means to exclude the faces of specific people registered in advance from conversion targets" refers to technology that pre-registers the faces of specific people, such as streamers, in a database, and does not perform conversion if that face is detected in the video.

[0654] "Methods for changing hairstyles and clothing" refers to technology that changes the hairstyles and clothing of people in video data to different styles.

[0655] "Means of converting to specific attributes" refers to technology that changes the faces and clothing of people in video data to specific designs, such as those of anime characters or foreigners.

[0656] "Store visitors" refers to customers or clients who visit a physical store.

[0657] "Means for processing photos and videos in real time" refers to technology that instantly processes and converts photos and videos taken by customers inside the store.

[0658] "Means of converting the faces of other customers into different faces" refers to technology that uses an AI model to convert the faces of other customers photographed in the store into images of different faces.

[0659] "SNS" is an abbreviation for "social networking service," which refers to a platform where users can share and interact with each other via the Internet.

[0660] The embodiments for carrying out the present invention are described as follows.

[0661] The system of the present invention uses a smartphone to process photos and videos taken in a physical store in real time, protecting the portrait rights of other customers. This system is realized using the following hardware and software components.

[0662] Hardware

[0663] Smartphone: The device on which this system is implemented, capturing and processing video data.

[0664] Camera module: A smartphone's built-in camera that takes photos and videos in real time.

[0665] software

[0666] An open-source face recognition library (e.g., OpenCV 4.5.3) for detecting human faces in video data.

[0667] Generative Adversarial Network (GAN) models (e.g., StyleGAN2): Used to transform a detected face into another face in real time.

[0668] Real-time video streaming library (e.g., FFmpeg 4.4): Used to instantly deliver converted video data to viewers.

[0669] System Operation

[0670] 1. The user launches the smartphone app and performs the initial setup, which includes registering the user's own facial data. This facial data is later used as an exclusion target during the facial conversion process.

[0671] 2. The camera module is activated and the device (smartphone) captures video data in real time. The video data is acquired at a constant frame rate and generated in a stream format.

[0672] 3. The acquired video data is analyzed in real time using a facial recognition library such as OpenCV. The server detects faces in the video and compares them with pre-registered user facial data, excluding any matches from the conversion target.

[0673] 4. The detected face is transformed into another face in real time using a generative adversarial network (GAN) model such as StyleGAN2. This transformation process is fast and preserves the original face's natural angle and facial expression.

[0674] 5. Based on the options selected by the user, the server changes non-facial features (such as hairstyle and clothing) and converts the user to specific attributes. For example, it is possible to convert the user to an anime character or a foreign character.

[0675] 6. The converted video data is then distributed to viewers in real time via a video streaming library such as FFmpeg, allowing viewers to enjoy high-quality, seamless video.

[0676] Specific examples

[0677] For example, if a user takes a photo inside a physical store, the system can detect the faces of other customers and convert them into different faces before distributing the video, thereby avoiding portrait rights issues. This allows users to post photos and videos to social media with peace of mind.

[0678] Prompt Sentence Examples

[0679] "Please create a camera app that protects portrait rights by converting the faces of other customers in real time into videos taken in a store."

[0680] In this way, the present invention simultaneously realizes protection of portrait rights and provision of natural-looking images in brick-and-mortar stores and public places.

[0681] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0682] Step 1:

[0683] The user launches the smartphone app and performs the initial setup. Here, the user scans their face data with the camera and registers it in the system. The user's facial image data is used as input, and this is the output that is registered in the facial recognition database. This data is later used as an exclusion target during the face conversion process.

[0684] Step 2:

[0685] The device activates the camera module and captures video data in real time. The input is video data acquired from the camera, and the output is video data in a stream format generated at a constant frame rate. This video data is immediately sent to the server.

[0686] Step 3:

[0687] The server analyzes the received video data using an open-source face recognition library (e.g., OpenCV 4.5.3). The input is stream-format video data, and the output is data indicating the position of people's faces in the video. The server detects faces in the video based on this data.

[0688] Step 4:

[0689] The server compares the detected person's face with the user's face data registered in advance. The input is face position data and the user's face recognition data, and the output is a list of faces to be excluded from conversion. This prevents conversion even if the user's face data is included in the video.

[0690] Step 5:

[0691] The server uses a generative adversarial network (GAN) model (e.g., StyleGAN2) to convert the detected face into a different face in real time. The input is the recognized face position data and facial image data in the video, and the output is the converted facial image data. This data processing allows the conversion to be performed while maintaining the natural angle and expression of the original face.

[0692] Step 6:

[0693] The server changes parts of the character other than the face (such as hairstyle or clothing) based on the options selected by the user. The input is the user's options and video data, and the output is the video data after the changes. This data calculation results in a transformation to make the character look like an anime character or a foreign character, depending on the settings.

[0694] Step 7:

[0695] The server then uses a real-time video streaming library (e.g., FFmpeg 4.4) to deliver the converted video data to viewers. The input is the converted video data, and the output is the converted video delivered in real time. This allows viewers to enjoy high-quality video without delay.

[0696] These processing steps enable the system to safely broadcast live footage from within a physical store, providing natural-looking footage without violating the portrait rights of other customers.

[0697] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0698] The system of the present invention has the function of detecting human faces in video data in real time, converting the detected human faces into other faces using an AI model, and distributing the converted video data in real time. It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for changing hairstyles, clothing, and converting specific attributes. It also combines an emotion engine that recognizes user emotions, and can perform conversion processing and attribute changes within the video data based on the recognized emotions.

[0699] The implementation of this system is described in detail below.

[0700] The user launches the live streaming app and performs the initial setup. The streamer then points their face at the camera and scans it, which is then registered as facial recognition data in the system. This data is then used to exclude the streamer from face conversion later.

[0701] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format and sent to the system at a constant frame rate.

[0702] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate technology to identify faces in each frame. It then references the facial data of streamers registered in advance and excludes faces that match from the conversion target.

[0703] The server then applies an AI model (e.g., a generative artificial network, or GAN) to the detected face to transform it into a different face in real time. The AI ​​model generates natural-looking facial features while preserving the original facial angle and expression.

[0704] Furthermore, an emotion engine is introduced to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The analysis results are classified into emotion categories such as joy, sadness, and anger.

[0705] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, additional processing can be performed to make the face in the video brighter or to change it into a smiling face. The attributes of the face and clothing generated by the AI ​​model can also be changed according to the emotion. For example, if the user selects an anime-style setting and is sad, detailed expressions such as the character shedding tears can be added.

[0706] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0707] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0708] As described above, this system performs video processing that reflects the user's emotions, making it possible to provide viewers with consistently high-quality, natural-looking video.

[0709] The processing flow will be explained below.

[0710] Step 1:

[0711] The user launches the live streaming app and performs the initial setup. The user then points their face at the camera, scans it, and registers facial recognition data into the system. This data is used to identify the face as the streamer.

[0712] Step 2:

[0713] The device activates the camera for live streaming and captures video data in real time, which is generated in a stream format at a constant frame rate.

[0714] Step 3:

[0715] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection with the server via the network.

[0716] Step 4:

[0717] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. High-precision facial recognition technology is used to identify faces in each frame of the video data.

[0718] Step 5:

[0719] The server compares the detected face data with the face data of the streamer registered in advance and excludes any matches from the conversion process. This means that the streamer's face is excluded from the conversion process.

[0720] Step 6:

[0721] The server then applies an AI model to the remaining faces, transforming them into a different face in real time. The AI ​​model (e.g., a generative adversarial network: GAN) generates natural-looking facial features while preserving the original facial angle and expression.

[0722] Step 7:

[0723] The device uses an emotion engine to recognize the user's emotions and analyzes them in real time through cameras and sensors, categorizing the results into emotion categories such as joy, sadness, and anger.

[0724] Step 8:

[0725] The server adjusts the transformation process in the video data based on the user's emotions recognized by the emotion engine. For example, if the server recognizes that the user is happy, it may transform the face of a person in the video into a brighter expression.

[0726] Step 9:

[0727] The server changes the facial and clothing attributes generated by the AI ​​model based on the user's emotions recognized by the emotion engine. For example, if the user selects the "anime character" setting, character expressions corresponding to the emotion will be added.

[0728] Step 10:

[0729] The server then delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0730] Step 11:

[0731] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0732] Step 12:

[0733] The server processes the video data in real time, converting and delivering it consistently until the end of the broadcast, allowing viewers to receive high-quality, natural video continuously.

[0734] Example 2

[0735] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0736] In real-time video streaming, there is a demand for providing viewers with higher quality and more natural-looking images. However, in technology for converting faces in streamed video, there is a lack of methods for excluding the streamer's own face from the conversion target, or for processing video that reflects the user's emotions. Therefore, there is a need for the development of a system that can appropriately convert the faces of people other than the streamer during live streaming, and further recognize the user's emotions and reflect them in the video.

[0737] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0738] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for distributing the converted video data in real time, means for identifying a user's emotion using an emotion engine, and means for adjusting the conversion process in the video data based on the identified emotion. This makes it possible to naturally convert the faces of people other than the broadcaster in real time during broadcasting, and further to reflect the user's emotion in the video.

[0739] "Real-time" refers to the ability of a system to operate such that data is processed and responded to almost immediately, implying very little processing delay.

[0740] "Video data" means data that describes visual information captured by a camera or other imaging device.

[0741] "Human face" refers to the facial features of an individual human being within video data, and specifically refers to the area that can be identified by facial recognition technology.

[0742] "Means for detecting" refers to a combination of hardware and software capable of identifying and extracting a particular target.

[0743] "Generative AI models" refer to algorithms that use artificial intelligence to generate new data and features, and specifically include generative artificial network (GAN).

[0744] "Means for converting" refers to software or hardware that has the functionality to change data of one format into another format.

[0745] "Distribution means" refers to a system capable of transmitting data to a specific receiver and making it available in a usable form.

[0746] "Emotion engine" refers to a combination of software and hardware for recognizing and analyzing user emotions in real time.

[0747] "User emotion" refers to the user's psychological state as recognized from the video and audio data being distributed, and includes emotion categories such as joy, sadness, and anger.

[0748] "Adjusting means" refers to hardware and software that has the functionality to change the system's behavior or output based on specific conditions.

[0749] The system of the present invention has the function of detecting human faces in video data in real time during live streaming, converting the detected faces into other faces using a generative AI model, and distributing the converted video data in real time.It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for recognizing user emotions and changing the attributes of the video based on the emotions.

[0750] The user launches the live streaming app and performs the initial setup. For example, the streamer points the camera at their face and scans it, registering it as facial recognition data in the system. This data is later used to exclude the streamer from face conversion. The user scans their face using a smartphone or computer, following the app's instructions.

[0751] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the system at a certain frame rate (e.g., 30 fps). At this time, the device uses a high-resolution camera to capture clear video.

[0752] The server analyzes the received video data and uses a facial recognition algorithm to detect faces of people appearing in the video. The facial detection algorithm uses open-source facial recognition libraries (e.g., dlib or OpenCV) to perform highly accurate face detection. The server performs face detection for each series of frames and excludes matching faces from conversion by referencing the facial data of broadcasters registered in advance.

[0753] The server then applies a generative AI model to the detected face, transforming it into a different face in real time. This model uses generative artificial intelligence (GAN) or deep learning. For example, GANs preserve the original facial features while generating new facial features. During this process, the server carefully transforms the face to maintain the original facial angle and facial expression.

[0754] Furthermore, an emotion engine is implemented to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The emotion engine uses facial expression recognition technology and voice analysis technology to classify the user's emotions into joy, sadness, anger, etc. The analysis results are sent to the server in real time.

[0755] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, the server adds a process to brighten the face in the video and transform it into a smiling face. The attributes of the face and clothing generated by the generative AI model are also changed according to the user's emotions. For example, if the user selects an anime-style setting and is sad, the character will shed tears.

[0756] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay via live streaming platforms (such as YouTube Live or Twitch).

[0757] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0758] Example prompt sentence:

[0759] Start a live stream at a tourist spot, and have it detect people's faces in real time and convert their expressions into happy expressions. The streamer's own face will not be converted, ensuring a fun-filled stream for your viewers.

[0760] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0761] Step 1:

[0762] The user launches the live streaming app and performs the initial setup. The user then points the camera at their face to scan it and registers it as facial recognition data in the system. The facial image captured by the camera is used as input, and a facial recognition algorithm is applied to obtain the broadcaster's facial data. As output, the broadcaster's facial data is stored in the system's database.

[0763] Step 2:

[0764] The device activates the camera for live streaming and captures video data in real time. As input, it uses the video stream from the camera and captures data at a constant frame rate (e.g., 30 fps). As output, the captured video data is generated in a stream format and sent to the server.

[0765] Step 3:

[0766] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. As input, it uses the received video frames and applies a facial recognition algorithm (e.g., dlib or OpenCV). As output, it generates location information and feature point data of the detected faces.

[0767] Step 4:

[0768] The server references the face data of the streamer registered in advance and compares it with the detected face. As input, it uses the face data of the streamer registered in step 1 and the detected face data obtained in step 3. As output, it generates a list in which matching faces are excluded from conversion.

[0769] Step 5:

[0770] The server applies a generative AI model to the detected face and transforms it into a different face in real time. It uses facial position information and feature point data as input and uses a generative artificial network (GAN). The output is image data of the transformed face. During this process, the original facial angle and facial expression are maintained.

[0771] Step 6:

[0772] The emotion engine analyzes the user's emotions in real time using a camera or sensor on the device that can identify the user's emotions. As input, it uses data obtained from the camera or sensor (facial expressions and voice data) and applies emotion analysis algorithms. As output, it generates an analysis result of emotion categories such as joy, sadness, anger, etc. and sends it to the server.

[0773] Step 7:

[0774] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process in the video data. The inputs are the emotion analysis results and the transformed facial image data. The output is video data that has been adjusted according to the emotion. For example, if the user is happy, the face in the video is made brighter and transformed into a smiling face.

[0775] Step 8:

[0776] The server delivers the converted video data to viewers in real time. The server uses the adjusted video data as input and sends the data via a live streaming platform. The output is high-quality, latency-free video delivered to viewers, allowing them to enjoy a natural, real-time experience.

[0777] (Application example 2)

[0778] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0779] Providing a consistently visually pleasing experience to viewers is challenging in current live streaming and video distribution services. In particular, there is a need for technology that can flexibly address the needs of privacy protection and emotion-based video conversion. Furthermore, there is no system that can exclude specific people's faces from conversion and dynamically process video based on emotions in real time.

[0780] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0781] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for delivering the converted video data in real time, and means for adjusting the conversion processing in the video data based on the recognized emotion using an emotion engine that recognizes emotion. This makes it possible to provide viewers with consistently high-quality, natural video and realize video processing that corresponds to the user's emotion in real time.

[0782] "Real-time" refers to a state in which results are available immediately at the moment a process or operation is performed.

[0783] "Video Data" refers to video and image information captured by a camera or other device.

[0784] "Means for detecting faces" refers to technologies and algorithms that identify and locate human faces within video data.

[0785] "Generative AI models" refer to artificial intelligence models that learn from specific input data and generate new data. This includes technologies such as generative artificial network (GAN).

[0786] "Face conversion method" refers to the technology or algorithm used to replace a detected person's face with a different face.

[0787] "Means of distribution" refers to the technology and infrastructure used to transmit and display the converted video data to viewers in real time via the Internet.

[0788] An "emotion engine" refers to the technology or algorithms used to identify and categorize people's emotions.

[0789] "Means for adjusting the transformation process" refers to technologies and algorithms that dynamically change and optimize the way video data is transformed based on emotion or other criteria.

[0790] "Means for excluding the faces of specific people from conversion targets" refers to technologies or algorithms that detect the faces of specific people registered in advance and exclude them from real-time conversion targets.

[0791] "Means for changing hairstyles and clothing, and converting to specific attributes" refers to technologies and algorithms that change the hairstyles and clothing of people in video data to different ones and convert them to specific attributes (for example, to look like an anime character).

[0792] The present invention provides a system that detects human faces in video data in real time, converts the detected faces into different faces using a generative AI model, and distributes the video in real time. Specific embodiments are described below.

[0793] First, the user launches the live streaming application on their device, which can be a smartphone, smart glasses, or a head-mounted display. The application has a function that allows the streamer to point the camera at their face, scan it, and register it as facial recognition data in the system. This data is used to exclude the streamer's face from conversion targets.

[0794] Next, the device activates the camera and captures video data in real time. The captured video data is sent to the server as a stream at a fixed frame rate. The server analyzes the received video data and uses dlib's facial recognition algorithm to detect human faces in the video.

[0795] The server applies a generative AI model (e.g., a generative adversarial network: GAN) to the detected face and converts it into a different face in real time. This generative AI model generates natural-looking facial features and can preserve the original facial angle and expression. It also references the facial data of pre-registered streamers and excludes matching faces from the conversion target.

[0796] Furthermore, the server introduces an emotion engine to recognize emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time and adjust the conversion process within the video data based on the analysis results. For example, if the user is happy, the server can convert the face in the video into a bright, smiling face.

[0797] The server then delivers the converted video data to viewers in real time, allowing viewers to instantly enjoy high-quality, natural-looking video and a visual experience that reflects the broadcaster's intentions and emotions.

[0798] As a concrete example, consider a case where a user is live streaming at a tourist spot and the emotion engine detects "joy" during the stream. In this case, the faces of all other people in the stream will be converted to bright, smiling faces. An example of a prompt sentence is as follows:

[0799] Example prompt sentence:

[0800] "Identify the faces of people in a photo, recognize the emotions on those faces, and convert them into appropriate face images based on that. 1. Face recognition: using dlib 2. Emotion recognition: using EmotionRecognizer 3. Emotion-based face transformation: using GAN to generate natural facial expressions"

[0801] In this way, the system is able to respond to the user's emotions and provide high quality, visually pleasing video in real time.

[0802] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0803] Step 1:

[0804] The user launches the live streaming application on their device. They point the camera at their own face, scan it, and register it as facial recognition data in the system. This data is sent from the device to the server and used to exclude the broadcaster's face from conversion. The input is facial image data from the camera, and the output is facial recognition data.

[0805] Step 2:

[0806] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the server at a constant frame rate. The input is video data from the camera, and the output is video data in stream format.

[0807] Step 3:

[0808] The server prepares to analyze the received video data. It uses dlib's facial recognition algorithm to detect faces in the video. It references the face data of pre-registered streamers and excludes matching faces from the conversion target. The input is stream-format video data, and the output is the face coordinates and the streamer's face data detection results.

[0809] Step 4:

[0810] The server applies a generative AI model (e.g., a generative artificial network (GAN)) to the detected face and converts it into a different face in real time, preserving the original facial angle and expression. The input is the face coordinates and facial image data, and the output is the converted facial image data.

[0811] Step 5:

[0812] The emotion engine analyzes the user's emotions in real time using cameras and sensors that can identify the user's emotions. The analysis results are classified into emotion categories such as joy, sadness, and anger. The input is data from the cameras and sensors, and the output is emotion category data.

[0813] Step 6:

[0814] The server adjusts the transformation process in the video data based on the emotion category. For example, if the user is happy, additional processing is performed to transform the face in the video into a brighter, smiling face. The input is emotion category data and transformed face image data, and the output is adjusted face image data.

[0815] Step 7:

[0816] The server delivers the converted video data to viewers in real time, allowing viewers to enjoy natural, high-quality video without delay. The input is adjusted video data, and the output is video data delivered in real time.

[0817] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0818] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0819] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[0820] [Fourth embodiment]

[0821] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0822] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0823] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0824] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[0825] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0826] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0827] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0828] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[0829] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0830] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0831] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0832] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0833] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0834] The system of the present invention uses an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from the street, providing natural-looking images while protecting portrait rights. The implementation of this system follows the steps below.

[0835] First, the user launches the live streaming app and performs the initial setup. The user then scans their face with the camera and registers their facial recognition data as a streamer in the system. This data is then used to exclude the streamer from face conversion later.

[0836] Next, the device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format at a constant frame rate. This stream data is then sent to the server in real time.

[0837] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate facial recognition technology to identify faces in each frame. In doing so, it references the face data of the streamer registered in advance and compares it with faces in the video, excluding any that match from the conversion target.

[0838] The server then uses an AI model to convert the faces that were not rejected into a different face. The AI ​​model, for example, a generative artificial network (GAN), can be used to convert the facial features into natural-looking faces. This model processes in real time, preserving the original facial angle and facial expression.

[0839] Furthermore, the server changes parts other than the face (hairstyle and clothing) and converts them to specific attributes (for example, anime character style or foreign style) based on the option settings selected by the user. Specifically, if the user selects "anime character style," the face and clothing of the person in the video are converted to the corresponding character design.

[0840] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technology to minimize delays, allowing viewers to enjoy natural, high-quality video.

[0841] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the face of the streamer that has been registered in advance, and converts it into a different face using an AI model. This allows viewers to view the converted face or video with specific attributes in real time, allowing live streaming to be carried out safely without violating portrait rights while maintaining the naturalness of the video.

[0842] The processing flow will be explained below.

[0843] Step 1:

[0844] A user launches a live streaming app and completes the initial setup. At this time, the user points their face at the camera and scans it, registering their facial recognition data as a streamer into the system. This will exclude the streamer's face from subsequent facial conversion processing.

[0845] Step 2:

[0846] The device activates the camera and captures video data in real time. The device generates the captured video data in a stream format and continues to acquire video data at a constant frame rate.

[0847] Step 3:

[0848] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection over the network.

[0849] Step 4:

[0850] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video, identifying faces in each frame of the video data and extracting the necessary information.

[0851] Step 5:

[0852] The server matches the faces in the video with pre-registered facial data of the streamer and filters out any matches. This process involves comparing facial feature vectors to reliably identify the streamer's face.

[0853] Step 6:

[0854] The server applies an AI model (such as a generative artificial network (GAN)) to the remaining faces and converts them into a different face in real time. The AI ​​model generates natural-looking facial features that reflect the angle and expression of the original face.

[0855] Step 7:

[0856] The server checks the user's option settings and, if necessary, changes non-facial features (hairstyle and clothing) or converts them to specified attributes (e.g., anime character style or foreign character style), thereby processing the video in the specific style desired by the user.

[0857] Step 8:

[0858] The server then delivers the converted video data to viewers in real time, with minimal latency and using high-speed data processing technology.

[0859] Step 9:

[0860] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0861] Step 10:

[0862] The server continues to process the video data in real time, converting and distributing it consistently until the end of the broadcast, allowing viewers to receive consistently high-quality video.

[0863] Example 1

[0864] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0865] When live streaming or posting images from the street, it is important to provide natural-looking images while avoiding infringing on the portrait rights of people around you. Another challenge is to convert the video in real time and deliver it to viewers with high quality and low latency, without altering the broadcaster's own face.

[0866] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0867] In this invention, the server includes: a means for a user to launch a live streaming app and scan their face to register facial recognition data; a means for the device to launch a camera and capture video data in real time and generate it in a stream format at a constant frame rate; a means for the server to analyze the received video data and detect human faces in the video using a facial recognition algorithm; a means for converting the detected human faces into other faces using a generative AI model; a means for modifying parts other than the face or converting them into specific attributes based on option settings selected by the user; and a means for distributing the converted video data in real time. This makes it possible to distribute high-quality, natural-looking video in real time while protecting portrait rights.

[0868] A "user" is an entity that operates the system and performs initial setup and live streaming.

[0869] A "live streaming app" is application software that users use to stream video in real time.

[0870] "Facial recognition data" refers to image data and characteristic information of a user's face that is scanned with a camera and registered in the system.

[0871] A "terminal" is a device equipped with a camera that captures and transmits video to a server.

[0872] A "camera" is a device that captures video data in real time.

[0873] "Video data" refers to video information captured by a camera.

[0874] "Stream format" is a method of transmitting data continuously in real time.

[0875] The "server" is a computer system that analyzes the received video data, performs facial recognition and conversion, and distributes it in real time.

[0876] A "face recognition algorithm" is a program for detecting and identifying the faces of people in video footage.

[0877] A "generative AI model" is an artificial intelligence model that generates new data based on input data, and in this invention is used to convert a specific face into another face.

[0878] "Optional settings" are additional settings that a user can make to the video being distributed, including changes to parts other than the face and conversion to specific attributes.

[0879] "Real-time distribution" is a method in which converted video data is immediately provided to viewers.

[0880] "Portrait rights" are the right of individuals to prevent their portraits, including their face and figure, from being photographed or used without their permission.

[0881] The system of this invention protects portrait rights and provides natural-looking images by using an AI model to convert the faces of people around you into different faces in real time when live streaming or posting images from around the city. The following hardware and software are used to implement this system.

[0882] First, the user launches the live streaming app and performs the initial setup. They then scan their face using the camera on their smartphone or PC, and register this facial recognition data in the system. This data is used by the facial recognition algorithm to exclude the streamer's face from conversion.

[0883] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format (e.g., RTMP protocol) at a frame rate of 30 frames per second and sent to the server.

[0884] The server analyzes the received video data and uses the Python library OpenCV, Dlib, or other facial recognition algorithms to detect faces in the video. At this time, it references the facial data of pre-registered streamers and excludes any matches from the conversion.

[0885] The server then uses a generative anti-aliasing network (GAN) to transform any detected faces that are not rejected into a new face. Using deep learning frameworks like TensorFlow and PyTorch, the server uses a pre-trained model to process the transformed faces in real time so that they appear natural. This model converts the input face into a new face while preserving its original angle and facial expression.

[0886] Furthermore, the server also processes changes to non-facial features (hairstyle and clothing) and conversion to specific attributes (such as anime character style or foreign style) based on the options selected by the user within the app. For example, if the user selects "anime character style," the server uses a style conversion model such as StyleGAN to convert the face and clothing of the person in the video into an anime character design.

[0887] Finally, the server delivers the converted video data to viewers in real time, using high-speed data processing technologies such as the RTMP protocol to minimize delays. Viewers can enjoy this natural, high-quality video on their smartphones, PCs, or web browsers using viewing apps.

[0888] As a concrete example, consider the case where a user is live streaming at a tourist spot. At this time, tourist spots are frequented by many people, and each person has their own portrait rights. When the live stream begins, the device captures video in real time using its camera, and the video is sent to a server. The server detects faces in the video, removes the broadcaster's pre-registered face, and converts it to a different face using an AI model. This allows viewers to view the converted face or video with specific attributes converted in real time, allowing live streaming to be carried out safely while maintaining the naturalness of the video and without violating portrait rights.

[0889] Examples of prompts:

[0890] "Generate the program code for a system that transforms the faces of people around you into other faces during a live stream. In this system, the streamer's face will be pre-registered and will not be transformed. The AI ​​model used is a generative adversarial network (GAN), and the transformed footage will be distributed to viewers in real time."

[0891] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0892] Step 1:

[0893] The user launches the live streaming app and scans their face to register facial recognition data. In this step, the user uses the camera on their smartphone or PC to perform the facial scanning operation. The input is the user's facial image data, and the output is facial recognition data registered in the system. This recognition data is used to identify the broadcaster's face and exclude it in the later face conversion step. Specifically, the user follows the app's guide to point the camera at their own face and perform a scan for a few seconds.

[0894] Step 2:

[0895] The device starts the camera and captures video data in real time. In this step, the device captures video at a frame rate of 30 frames per second. The input is video data from the camera, and the output is stream-format data generated in real time. Specifically, the device starts the camera, and the live streaming app captures video data in real time.

[0896] Step 3:

[0897] The video data captured by the terminal is sent to the server in real time. In this step, the video data is sent using a protocol such as RTMP. The input is the captured video data in a stream format, and the output is the video data sent to the server. Specifically, the terminal sends the video data to the server via a network connection.

[0898] Step 4:

[0899] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. The input is the real-time video data sent to the server, and the output is a list of detected faces. Here, the server applies the face detection algorithm using the Python libraries OpenCV and Dlib. Specifically, the server analyzes each frame and identifies the location and features of faces.

[0900] Step 5:

[0901] The server excludes the faces of pre-registered streamers from the detected faces. The input is a list of detected faces and facial recognition data of pre-registered streamers, and the output is a list of faces excluding the streamers' faces. Specifically, the server compares the facial recognition data and removes matching faces from the list.

[0902] Step 6:

[0903] The server uses a generative AI model to convert the detected face into a different face. The input is a list of faces excluding the streamer's face and video data containing that face, and the output is the converted video data. The server uses TensorFlow or PyTorch to apply a pre-trained generative adversarial network (GAN). Specifically, the server performs face conversion in real time, converting the face into a new face while preserving the original facial angle and expression.

[0904] Step 7:

[0905] The server modifies non-facial features or converts specific attributes based on the user's options. The input is video data containing the converted face and the user's options, and the output is video data that has been further converted according to the options. Specifically, the server applies a style transfer model such as StyleGAN to change the hairstyle or clothing of people in the video.

[0906] Step 8:

[0907] The server finally delivers the converted video data to the viewer in real time. The input is the final converted video data, and the output is the real-time video delivered to the viewer. Specifically, the server uses the RTMP protocol to send video data at high speed to the viewer, who then views the video without delay using a viewing app or web browser on their smartphone or PC.

[0908] (Application example 1)

[0909] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0910] When live streaming or posting images from the street or inside a store, there is a risk of violating the portrait rights of other people. In particular, in physical stores, where many customers come and go, the faces of other customers appearing in posted photos and videos can be a problem. This can lead to the risk of violating the privacy of other customers, so there is a need to solve this problem.

[0911] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0912] In this invention, the server includes means for detecting human faces in video data in real time, means for converting the detected human faces into other faces using an AI model, means for distributing the converted video data in real time, means for excluding the faces of specific people registered in advance from the conversion targets, means for changing hairstyles and clothing and converting into specific attributes, and means for processing photos and videos taken by customers in the store in real time and converting the faces of other customers into other faces. This enables customers to safely post photos and videos taken in the store to social media etc.

[0913] "Real-time" refers to data processing and information delivery occurring immediately and without delay.

[0914] "Video data" refers to data containing visual information obtained from a camera or other imaging device.

[0915] "Means for detecting human faces" refers to technology that uses specific algorithms to identify and extract human faces from video data.

[0916] "AI model" refers to a mathematical or algorithmic model trained to perform a specific task using artificial intelligence techniques.

[0917] "Method of converting into another face" refers to technology that converts the detected face of a person into a different facial image using an AI model.

[0918] "Means for delivering converted video data in real time" refers to technology that instantly provides viewers with video data converted by an AI model.

[0919] "Means to exclude the faces of specific people registered in advance from conversion targets" refers to technology that pre-registers the faces of specific people, such as streamers, in a database, and does not perform conversion if that face is detected in the video.

[0920] "Methods for changing hairstyles and clothing" refers to technology that changes the hairstyles and clothing of people in video data to different styles.

[0921] "Means of converting to specific attributes" refers to technology that changes the faces and clothing of people in video data to specific designs, such as those of anime characters or foreigners.

[0922] "Store visitors" refers to customers or clients who visit a physical store.

[0923] "Means for processing photos and videos in real time" refers to technology that instantly processes and converts photos and videos taken by customers inside the store.

[0924] "Means of converting the faces of other customers into different faces" refers to technology that uses an AI model to convert the faces of other customers photographed in the store into images of different faces.

[0925] "SNS" is an abbreviation for "social networking service," which refers to a platform where users can share and interact with each other via the Internet.

[0926] The embodiments for carrying out the present invention are described as follows.

[0927] The system of the present invention uses a smartphone to process photos and videos taken in a physical store in real time, protecting the portrait rights of other customers. This system is realized using the following hardware and software components.

[0928] Hardware

[0929] Smartphone: The device on which this system is implemented, capturing and processing video data.

[0930] Camera module: A smartphone's built-in camera that takes photos and videos in real time.

[0931] software

[0932] An open-source face recognition library (e.g., OpenCV 4.5.3) for detecting human faces in video data.

[0933] Generative Adversarial Network (GAN) models (e.g., StyleGAN2): Used to transform a detected face into another face in real time.

[0934] Real-time video streaming library (e.g., FFmpeg 4.4): Used to instantly deliver converted video data to viewers.

[0935] System Operation

[0936] 1. The user launches the smartphone app and performs the initial setup, which includes registering the user's own facial data. This facial data is later used as an exclusion target during the facial conversion process.

[0937] 2. The camera module is activated and the device (smartphone) captures video data in real time. The video data is acquired at a constant frame rate and generated in a stream format.

[0938] 3. The acquired video data is analyzed in real time using a facial recognition library such as OpenCV. The server detects faces in the video and compares them with pre-registered user facial data, excluding any matches from the conversion target.

[0939] 4. The detected face is transformed into another face in real time using a generative adversarial network (GAN) model such as StyleGAN2. This transformation process is fast and preserves the original face's natural angle and facial expression.

[0940] 5. Based on the options selected by the user, the server changes non-facial features (such as hairstyle and clothing) and converts the user to specific attributes. For example, it is possible to convert the user to an anime character or a foreign character.

[0941] 6. The converted video data is then distributed to viewers in real time via a video streaming library such as FFmpeg, allowing viewers to enjoy high-quality, seamless video.

[0942] Specific examples

[0943] For example, if a user takes a photo inside a physical store, the system can detect the faces of other customers and convert them into different faces before distributing the video, thereby avoiding portrait rights issues. This allows users to post photos and videos to social media with peace of mind.

[0944] Prompt Sentence Examples

[0945] "Please create a camera app that protects portrait rights by converting the faces of other customers in real time into videos taken in a store."

[0946] In this way, the present invention simultaneously realizes protection of portrait rights and provision of natural-looking images in brick-and-mortar stores and public places.

[0947] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0948] Step 1:

[0949] The user launches the smartphone app and performs the initial setup. Here, the user scans their face data with the camera and registers it in the system. The user's facial image data is used as input, and this is the output that is registered in the facial recognition database. This data is later used as an exclusion target during the face conversion process.

[0950] Step 2:

[0951] The device activates the camera module and captures video data in real time. The input is video data acquired from the camera, and the output is video data in a stream format generated at a constant frame rate. This video data is immediately sent to the server.

[0952] Step 3:

[0953] The server analyzes the received video data using an open-source face recognition library (e.g., OpenCV 4.5.3). The input is stream-format video data, and the output is data indicating the position of people's faces in the video. The server detects faces in the video based on this data.

[0954] Step 4:

[0955] The server compares the detected person's face with the user's face data registered in advance. The input is face position data and the user's face recognition data, and the output is a list of faces to be excluded from conversion. This prevents conversion even if the user's face data is included in the video.

[0956] Step 5:

[0957] The server uses a generative adversarial network (GAN) model (e.g., StyleGAN2) to convert the detected face into a different face in real time. The input is the recognized face position data and facial image data in the video, and the output is the converted facial image data. This data processing allows the conversion to be performed while maintaining the natural angle and expression of the original face.

[0958] Step 6:

[0959] The server changes parts of the character other than the face (such as hairstyle or clothing) based on the options selected by the user. The input is the user's options and video data, and the output is the video data after the changes. This data calculation results in a transformation to make the character look like an anime character or a foreign character, depending on the settings.

[0960] Step 7:

[0961] The server then uses a real-time video streaming library (e.g., FFmpeg 4.4) to deliver the converted video data to viewers. The input is the converted video data, and the output is the converted video delivered in real time. This allows viewers to enjoy high-quality video without delay.

[0962] These processing steps enable the system to safely broadcast live footage from within a physical store, providing natural-looking footage without violating the portrait rights of other customers.

[0963] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0964] The system of the present invention has the function of detecting human faces in video data in real time, converting the detected human faces into other faces using an AI model, and distributing the converted video data in real time. It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for changing hairstyles, clothing, and converting specific attributes. It also combines an emotion engine that recognizes user emotions, and can perform conversion processing and attribute changes within the video data based on the recognized emotions.

[0965] The implementation of this system is described in detail below.

[0966] The user launches the live streaming app and performs the initial setup. The streamer then points their face at the camera and scans it, which is then registered as facial recognition data in the system. This data is then used to exclude the streamer from face conversion later.

[0967] Next, the device activates the camera for live streaming and captures video data in real time, which is generated in a stream format and sent to the system at a constant frame rate.

[0968] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. The facial detection algorithm uses highly accurate technology to identify faces in each frame. It then references the facial data of streamers registered in advance and excludes faces that match from the conversion target.

[0969] The server then applies an AI model (e.g., a generative artificial network, or GAN) to the detected face to transform it into a different face in real time. The AI ​​model generates natural-looking facial features while preserving the original facial angle and expression.

[0970] Furthermore, an emotion engine is introduced to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The analysis results are classified into emotion categories such as joy, sadness, and anger.

[0971] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, additional processing can be performed to make the face in the video brighter or to change it into a smiling face. The attributes of the face and clothing generated by the AI ​​model can also be changed according to the emotion. For example, if the user selects an anime-style setting and is sad, detailed expressions such as the character shedding tears can be added.

[0972] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0973] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[0974] As described above, this system performs video processing that reflects the user's emotions, making it possible to provide viewers with consistently high-quality, natural-looking video.

[0975] The processing flow will be explained below.

[0976] Step 1:

[0977] The user launches the live streaming app and performs the initial setup. The user then points their face at the camera, scans it, and registers facial recognition data into the system. This data is used to identify the face as the streamer.

[0978] Step 2:

[0979] The device activates the camera for live streaming and captures video data in real time, which is generated in a stream format at a constant frame rate.

[0980] Step 3:

[0981] The device transmits the captured video data to the server in real time, with low latency and maintaining a stable connection with the server via the network.

[0982] Step 4:

[0983] The server analyzes the received video data and uses a facial recognition algorithm to detect faces in the video. High-precision facial recognition technology is used to identify faces in each frame of the video data.

[0984] Step 5:

[0985] The server compares the detected face data with the face data of the streamer registered in advance and excludes any matches from the conversion process. This means that the streamer's face is excluded from the conversion process.

[0986] Step 6:

[0987] The server then applies an AI model to the remaining faces, transforming them into a different face in real time. The AI ​​model (e.g., a generative adversarial network: GAN) generates natural-looking facial features while preserving the original facial angle and expression.

[0988] Step 7:

[0989] The device uses an emotion engine to recognize the user's emotions and analyzes them in real time through cameras and sensors, categorizing the results into emotion categories such as joy, sadness, and anger.

[0990] Step 8:

[0991] The server adjusts the transformation process in the video data based on the user's emotions recognized by the emotion engine. For example, if the server recognizes that the user is happy, it may transform the face of a person in the video into a brighter expression.

[0992] Step 9:

[0993] The server changes the facial and clothing attributes generated by the AI ​​model based on the user's emotions recognized by the emotion engine. For example, if the user selects the "anime character" setting, character expressions corresponding to the emotion will be added.

[0994] Step 10:

[0995] The server then delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay.

[0996] Step 11:

[0997] If a user wants to change settings while live streaming, they can change options through the app interface, and the changes will be immediately reflected on the server, where the video will be re-processed based on the new settings.

[0998] Step 12:

[0999] The server processes the video data in real time, converting and delivering it consistently until the end of the broadcast, allowing viewers to receive high-quality, natural video continuously.

[1000] Example 2

[1001] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1002] In real-time video streaming, there is a demand for providing viewers with higher quality and more natural-looking images. However, in technology for converting faces in streamed video, there is a lack of methods for excluding the streamer's own face from the conversion target, or for processing video that reflects the user's emotions. Therefore, there is a need for the development of a system that can appropriately convert the faces of people other than the streamer during live streaming, and further recognize the user's emotions and reflect them in the video.

[1003] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1004] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for distributing the converted video data in real time, means for identifying a user's emotion using an emotion engine, and means for adjusting the conversion process in the video data based on the identified emotion. This makes it possible to naturally convert the faces of people other than the broadcaster in real time during broadcasting, and further to reflect the user's emotion in the video.

[1005] "Real-time" refers to the ability of a system to operate such that data is processed and responded to almost immediately, implying very little processing delay.

[1006] "Video data" means data that describes visual information captured by a camera or other imaging device.

[1007] "Human face" refers to the facial features of an individual human being within video data, and specifically refers to the area that can be identified by facial recognition technology.

[1008] "Means for detecting" refers to a combination of hardware and software capable of identifying and extracting a particular target.

[1009] "Generative AI models" refer to algorithms that use artificial intelligence to generate new data and features, and specifically include generative artificial network (GAN).

[1010] "Means for converting" refers to software or hardware that has the functionality to change data of one format into another format.

[1011] "Distribution means" refers to a system capable of transmitting data to a specific receiver and making it available in a usable form.

[1012] "Emotion engine" refers to a combination of software and hardware for recognizing and analyzing user emotions in real time.

[1013] "User emotion" refers to the user's psychological state as recognized from the video and audio data being distributed, and includes emotion categories such as joy, sadness, and anger.

[1014] "Adjusting means" refers to hardware and software that has the functionality to change the system's behavior or output based on specific conditions.

[1015] The system of the present invention has the function of detecting human faces in video data in real time during live streaming, converting the detected faces into other faces using a generative AI model, and distributing the converted video data in real time.It also has a means for excluding pre-registered faces of specific people from the conversion targets, and a means for recognizing user emotions and changing the attributes of the video based on the emotions.

[1016] The user launches the live streaming app and performs the initial setup. For example, the streamer points the camera at their face and scans it, registering it as facial recognition data in the system. This data is later used to exclude the streamer from face conversion. The user scans their face using a smartphone or computer, following the app's instructions.

[1017] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the system at a certain frame rate (e.g., 30 fps). At this time, the device uses a high-resolution camera to capture clear video.

[1018] The server analyzes the received video data and uses a facial recognition algorithm to detect faces of people appearing in the video. The facial detection algorithm uses open-source facial recognition libraries (e.g., dlib or OpenCV) to perform highly accurate face detection. The server performs face detection for each series of frames and excludes matching faces from conversion by referencing the facial data of broadcasters registered in advance.

[1019] The server then applies a generative AI model to the detected face, transforming it into a different face in real time. This model uses generative artificial intelligence (GAN) or deep learning. For example, GANs preserve the original facial features while generating new facial features. During this process, the server carefully transforms the face to maintain the original facial angle and facial expression.

[1020] Furthermore, an emotion engine is implemented to recognize the user's emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time. The emotion engine uses facial expression recognition technology and voice analysis technology to classify the user's emotions into joy, sadness, anger, etc. The analysis results are sent to the server in real time.

[1021] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process within the video data. For example, if the user is happy, the server adds a process to brighten the face in the video and transform it into a smiling face. The attributes of the face and clothing generated by the generative AI model are also changed according to the user's emotions. For example, if the user selects an anime-style setting and is sad, the character will shed tears.

[1022] Finally, the server delivers the converted video data to viewers in real time, allowing them to enjoy natural, high-quality video without delay via live streaming platforms (such as YouTube Live or Twitch).

[1023] For example, if a user is live streaming at a tourist spot and the emotion engine detects "joy," the faces of all people except the streamer are converted to natural, bright smiles, allowing viewers to enjoy the video with a fun atmosphere. Furthermore, the streamer's own face is excluded from the conversion because it was registered in advance.

[1024] Example prompt sentence:

[1025] Start a live stream at a tourist spot, and have it detect people's faces in real time and convert their expressions into happy expressions. The streamer's own face will not be converted, ensuring a fun-filled stream for your viewers.

[1026] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1027] Step 1:

[1028] The user launches the live streaming app and performs the initial setup. The user then points the camera at their face to scan it and registers it as facial recognition data in the system. The facial image captured by the camera is used as input, and a facial recognition algorithm is applied to obtain the broadcaster's facial data. As output, the broadcaster's facial data is stored in the system's database.

[1029] Step 2:

[1030] The device activates the camera for live streaming and captures video data in real time. As input, it uses the video stream from the camera and captures data at a constant frame rate (e.g., 30 fps). As output, the captured video data is generated in a stream format and sent to the server.

[1031] Step 3:

[1032] The server analyzes the received video data and uses a facial recognition algorithm to detect human faces in the video. As input, it uses the received video frames and applies a facial recognition algorithm (e.g., dlib or OpenCV). As output, it generates location information and feature point data of the detected faces.

[1033] Step 4:

[1034] The server references the face data of the streamer registered in advance and compares it with the detected face. As input, it uses the face data of the streamer registered in step 1 and the detected face data obtained in step 3. As output, it generates a list in which matching faces are excluded from conversion.

[1035] Step 5:

[1036] The server applies a generative AI model to the detected face and transforms it into a different face in real time. It uses facial position information and feature point data as input and uses a generative artificial network (GAN). The output is image data of the transformed face. During this process, the original facial angle and facial expression are maintained.

[1037] Step 6:

[1038] The emotion engine analyzes the user's emotions in real time using a camera or sensor on the device that can identify the user's emotions. As input, it uses data obtained from the camera or sensor (facial expressions and voice data) and applies emotion analysis algorithms. As output, it generates an analysis result of emotion categories such as joy, sadness, anger, etc. and sends it to the server.

[1039] Step 7:

[1040] Based on the user's emotions recognized by the emotion engine, the server adjusts the transformation process in the video data. The inputs are the emotion analysis results and the transformed facial image data. The output is video data that has been adjusted according to the emotion. For example, if the user is happy, the face in the video is made brighter and transformed into a smiling face.

[1041] Step 8:

[1042] The server delivers the converted video data to viewers in real time. The server uses the adjusted video data as input and sends the data via a live streaming platform. The output is high-quality, latency-free video delivered to viewers, allowing them to enjoy a natural, real-time experience.

[1043] (Application example 2)

[1044] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1045] Providing a consistently visually pleasing experience to viewers is challenging in current live streaming and video distribution services. In particular, there is a need for technology that can flexibly address the needs of privacy protection and emotion-based video conversion. Furthermore, there is no system that can exclude specific people's faces from conversion and dynamically process video based on emotions in real time.

[1046] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1047] In this invention, the server includes means for detecting a person's face in video data in real time, means for converting the detected person's face into another face using a generative AI model, means for delivering the converted video data in real time, and means for adjusting the conversion processing in the video data based on the recognized emotion using an emotion engine that recognizes emotion. This makes it possible to provide viewers with consistently high-quality, natural video and realize video processing that corresponds to the user's emotion in real time.

[1048] "Real-time" refers to a state in which results are available immediately at the moment a process or operation is performed.

[1049] "Video Data" refers to video and image information captured by a camera or other device.

[1050] "Means for detecting faces" refers to technologies and algorithms that identify and locate human faces within video data.

[1051] "Generative AI models" refer to artificial intelligence models that learn from specific input data and generate new data. This includes technologies such as generative artificial network (GAN).

[1052] "Face conversion method" refers to the technology or algorithm used to replace a detected person's face with a different face.

[1053] "Means of distribution" refers to the technology and infrastructure used to transmit and display the converted video data to viewers in real time via the Internet.

[1054] An "emotion engine" refers to the technology or algorithms used to identify and categorize people's emotions.

[1055] "Means for adjusting the transformation process" refers to technologies and algorithms that dynamically change and optimize the way video data is transformed based on emotion or other criteria.

[1056] "Means for excluding the faces of specific people from conversion targets" refers to technologies or algorithms that detect the faces of specific people registered in advance and exclude them from real-time conversion targets.

[1057] "Means for changing hairstyles and clothing, and converting to specific attributes" refers to technologies and algorithms that change the hairstyles and clothing of people in video data to different ones and convert them to specific attributes (for example, to look like an anime character).

[1058] The present invention provides a system that detects human faces in video data in real time, converts the detected faces into different faces using a generative AI model, and distributes the video in real time. Specific embodiments are described below.

[1059] First, the user launches the live streaming application on their device, which can be a smartphone, smart glasses, or a head-mounted display. The application has a function that allows the streamer to point the camera at their face, scan it, and register it as facial recognition data in the system. This data is used to exclude the streamer's face from conversion targets.

[1060] Next, the device activates the camera and captures video data in real time. The captured video data is sent to the server as a stream at a fixed frame rate. The server analyzes the received video data and uses dlib's facial recognition algorithm to detect human faces in the video.

[1061] The server applies a generative AI model (e.g., a generative adversarial network: GAN) to the detected face and converts it into a different face in real time. This generative AI model generates natural-looking facial features and can preserve the original facial angle and expression. It also references the facial data of pre-registered streamers and excludes matching faces from the conversion target.

[1062] Furthermore, the server introduces an emotion engine to recognize emotions. If the device is equipped with a camera or sensor that can identify the user's emotions, the emotion engine will analyze the user's emotions in real time and adjust the conversion process within the video data based on the analysis results. For example, if the user is happy, the server can convert the face in the video into a bright, smiling face.

[1063] The server then delivers the converted video data to viewers in real time, allowing viewers to instantly enjoy high-quality, natural-looking video and a visual experience that reflects the broadcaster's intentions and emotions.

[1064] As a concrete example, consider a case where a user is live streaming at a tourist spot and the emotion engine detects "joy" during the stream. In this case, the faces of all other people in the stream will be converted to bright, smiling faces. An example of a prompt sentence is as follows:

[1065] Example prompt sentence:

[1066] "Identify the faces of people in a photo, recognize the emotions on those faces, and convert them into appropriate face images based on that. 1. Face recognition: using dlib 2. Emotion recognition: using EmotionRecognizer 3. Emotion-based face transformation: using GAN to generate natural facial expressions"

[1067] In this way, the system is able to respond to the user's emotions and provide high quality, visually pleasing video in real time.

[1068] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1069] Step 1:

[1070] The user launches the live streaming application on their device. They point the camera at their own face, scan it, and register it as facial recognition data in the system. This data is sent from the device to the server and used to exclude the broadcaster's face from conversion. The input is facial image data from the camera, and the output is facial recognition data.

[1071] Step 2:

[1072] The device activates the camera for live streaming and captures video data in real time. The video data is generated in a stream format and sent to the server at a constant frame rate. The input is video data from the camera, and the output is video data in stream format.

[1073] Step 3:

[1074] The server prepares to analyze the received video data. It uses dlib's facial recognition algorithm to detect faces in the video. It references the face data of pre-registered streamers and excludes matching faces from the conversion target. The input is stream-format video data, and the output is the face coordinates and the streamer's face data detection results.

[1075] Step 4:

[1076] The server applies a generative AI model (e.g., a generative artificial network (GAN)) to the detected face and converts it into a different face in real time, preserving the original facial angle and expression. The input is the face coordinates and facial image data, and the output is the converted facial image data.

[1077] Step 5:

[1078] The emotion engine analyzes the user's emotions in real time using cameras and sensors that can identify the user's emotions. The analysis results are classified into emotion categories such as joy, sadness, and anger. The input is data from the cameras and sensors, and the output is emotion category data.

[1079] Step 6:

[1080] The server adjusts the transformation process in the video data based on the emotion category. For example, if the user is happy, additional processing is performed to transform the face in the video into a brighter, smiling face. The input is emotion category data and transformed face image data, and the output is adjusted face image data.

[1081] Step 7:

[1082] The server delivers the converted video data to viewers in real time, allowing viewers to enjoy natural, high-quality video without delay. The input is adjusted video data, and the output is video data delivered in real time.

[1083] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1084] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1085] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1086] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1087] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1088] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1089] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1090] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1091] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1092] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1093] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1094] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1095] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1096] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1097] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1098] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1099] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1100] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1101] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1102] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1103] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1104] The following is further disclosed regarding the above embodiment.

[1105] (Claim 1)

[1106] means for detecting human faces in video data in real time;

[1107] A method for converting the detected person's face into another face using an AI model;

[1108] a means for delivering the converted video data in real time;

[1109] A system including:

[1110] (Claim 2)

[1111] 2. The system according to claim 1, further comprising means for excluding faces of specific people registered in advance from the conversion target.

[1112] (Claim 3)

[1113] 10. The system of claim 1, further comprising means for changing hairstyle and clothing, and converting into specific attributes.

[1114] "Example 1"

[1115] (Claim 1)

[1116] A method for users to launch the live streaming app, scan their face, and register facial recognition data;

[1117] A means for the device to activate the camera, capture video data in real time, and generate it in a stream format at a constant frame rate;

[1118] a means for analyzing the video data received by the server and detecting the faces of people appearing in the video using a facial recognition algorithm;

[1119] A means for converting the detected person's face into another face using a generative AI model;

[1120] A means to change non-facial features or convert specific attributes based on user-selected options;

[1121] a means for delivering the converted video data in real time;

[1122] A system including:

[1123] (Claim 2)

[1124] 2. The system according to claim 1, further comprising means for excluding faces of specific people registered in advance from the conversion target.

[1125] (Claim 3)

[1126] 2. The system according to claim 1, further comprising means for changing hairstyle and clothing and converting into specific attributes based on option settings selected by the user.

[1127] "Application Example 1"

[1128] (Claim 1)

[1129] means for detecting human faces in video data in real time;

[1130] A method for converting the detected person's face into another face using an AI model;

[1131] a means for delivering the converted video data in real time;

[1132] A means for excluding faces of specific people registered in advance from the conversion target;

[1133] A means to change hairstyles and clothing, and to convert to specific attributes,

[1134] A method to process photos and videos taken by customers in the store in real time and change the faces of other customers into different faces,

[1135] A system including:

[1136] (Claim 2)

[1137] 10. The system of claim 1, further comprising means for a user to perform initial setup and register his or her facial data.

[1138] (Claim 3)

[1139] 10. The system of claim 1, further comprising means for performing additional transformations based on options selected by the user.

[1140] "Example 2: Combining Emotion Engines"

[1141] (Claim 1)

[1142] means for detecting human faces in video data in real time;

[1143] A means for converting the detected person's face into another face using a generative AI model;

[1144] a means for delivering the converted video data in real time;

[1145] means for identifying a user's emotion using an emotion engine;

[1146] means for adjusting transformations within the video data based on the identified emotions;

[1147] A system including:

[1148] (Claim 2)

[1149] 2. The system according to claim 1, further comprising means for excluding faces of specific people registered in advance from the conversion target.

[1150] (Claim 3)

[1151] 10. The system of claim 1, further comprising means for changing hairstyle and clothing to convert into specific attributes.

[1152] "Application example 2 when combining emotion engines"

[1153] (Claim 1)

[1154] means for detecting human faces in video data in real time;

[1155] A means for converting the detected person's face into another face using a generative AI model;

[1156] a means for delivering the converted video data in real time;

[1157] means for using an emotion engine to recognize emotions and adjusting transformations within the video data based on the recognized emotions;

[1158] …

[1159] A system including:

[1160] (Claim 2)

[1161] 2. The system according to claim 1, further comprising means for excluding faces of specific people registered in advance from the conversion target.

[1162] (Claim 3)

[1163] 10. The system of claim 1, further comprising means for changing hairstyle and clothing, and converting into specific attributes. [Explanation of symbols]

[1164] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for detecting human faces in video data in real time; A method for converting the detected person's face into another face using an AI model; a means for delivering the converted video data in real time; A system including:

2. The system according to claim 1 , further comprising means for excluding faces of specific people registered in advance from the conversion target.

3. The system of claim 1 further comprising means for changing hairstyle and clothing, and converting into specific attributes.

Images