System

Abstract

A system is provided.SOLUTION: A system, comprising: means for obtaining a facial image from a user; means for preprocessing the obtained facial image; means for analyzing the preprocessed facial image to predict an emotion of the user; means for receiving a text input from the user; means for generating a response based on the user emotion and the text input; and means for providing the generated response to the user.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] Currently, traditional customer service systems used by large retailers, technology enthusiasts, online education companies, and others struggle to fully understand customers' emotions and needs and provide personalized interactions. This poses a challenge for improving customer satisfaction. Furthermore, traditional systems are limited to specific input formats (e.g., text input) and cannot accommodate multimodal inputs (e.g., voice, images, etc.), making natural and smooth communication difficult. [Means for solving the problem]

[0005] The present invention solves the above problem by providing a system including: means for acquiring a facial image from a user; means for pre-processing the acquired facial image; means for analyzing the pre-processed facial image and predicting the user's emotion; means for receiving text input from the user; means for generating a response based on the user's emotion and the text input; and means for providing the generated response to the user.

[0006] This system can analyze customer sentiment in real time and generate personalized dialogue. Furthermore, by supporting multimodal inputs such as text, voice, and images, it can achieve more natural and smooth communication. As a result, it is possible to improve customer satisfaction and the efficiency of corporate service provision.

[0007] A "face image" is image data of a photograph of the user's face.

[0008] "Preprocessing" is the process of converting acquired data into a format suitable for analysis and recognition.

[0009] "Emotion prediction" means estimating a user's emotional state from information such as facial images and text input.

[0010] "Text input from the user" refers to a message or question entered by the user in text.

[0011] A "natural language processing model" is an artificial intelligence algorithm for understanding and generating human language.

[0012] "Generating a response" means creating an appropriate reply based on the input information and predicted emotions.

[0013] "Multimodal input" means the ability to receive input data in multiple formats, such as text, audio, and images.

[0014] "Providing to the user" means displaying or sending the generated response to the user. [Brief explanation of the drawings]

[0015] [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

[0016] 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.

[0017] First, the terms used in the following description will be explained.

[0018] 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).

[0019] 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.

[0020] 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.

[0021] 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.

[0022] 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."

[0023] [First embodiment]

[0024] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0025] 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.

[0026] 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).

[0027] 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.

[0028] 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.

[0029] 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.

[0030] 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.

[0031] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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."

[0036] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. Below, each element of the system and the processing flow of the program are explained in natural language.

[0037] The system begins by acquiring a facial image from the user's device. The device captures the user's facial image with a camera and sends it to the server. The server preprocesses the received facial image and inputs it into an emotion recognition model. The model analyzes the facial image and predicts the user's emotion (e.g., joy, sadness, anger, etc.).

[0038] Next, the user inputs text. Using a chat window on the device, the user enters a question or request and sends it to the server. The server receives the text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model. The natural language processing model generates a response to the user based on this information.

[0039] The generated response is then sent back to the device and displayed to the user. This allows the user to receive a personalized response that matches their emotions at that time. For example, if the user enters the text "I'm tired today," the server will recognize the emotion "tired" based on image analysis, and generate a response such as "Thank you for your hard work. Do you know how to relax?" based on that emotion and the text, and send it to the device.

[0040] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[0041] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[0042] The processing flow will be explained below.

[0043] Step 1:

[0044] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[0045] Step 2:

[0046] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image normalization and feature point extraction.

[0047] Step 3:

[0048] The server inputs the preprocessed facial image into an emotion recognition model. The emotion recognition model is composed of trained neural networks and other components, and predicts the user's emotions from the facial image. The predicted emotions are expressed with labels such as "happiness," "sadness," and "anger."

[0049] Step 4:

[0050] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[0051] Step 5:

[0052] The server receives text input from the user, then combines the received text with the previously predicted sentiment information to create a request that is input into a natural language processing model.

[0053] Step 6:

[0054] The server loads a natural language processing model and inputs the text and sentiment information together. The natural language processing model generates an optimal response based on the user's message and sentiment. For example, this response might be, "You're tired. Do you know how to relax?"

[0055] Step 7:

[0056] The server sends the generated response to the user's terminal.

[0057] Step 8:

[0058] The user can then check the response displayed on the device, which allows the user to receive a personalized response tailored to their emotions.

[0059] Step 9:

[0060] When the user performs voice input, the user uses the microphone of the terminal to record a voice message and transmits it to the server.

[0061] Step 10:

[0062] The server receives the voice data and loads an automatic speech recognition (ASR) model, which converts the voice data into text.

[0063] Step 11:

[0064] The server then uses the converted text and emotional information to re-feed data into the natural language processing model to generate an appropriate response.

[0065] Step 12:

[0066] The server sends the generated response to the user's terminal, where the user confirms it.

[0067] Through this series of steps, InsightBot Advanced uses information from multimodal inputs to understand sentiment and deliver personalized interactions.

[0068] Example 1

[0069] 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."

[0070] Conventional chat systems and automated response systems do not take into account the user's emotions, making it difficult to provide personalized responses, and many users end up with dissatisfied results. Furthermore, because they cannot handle multiple input modalities, such as voice and images in addition to text, they lack the ability to respond in a variety of usage scenarios.

[0071] 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.

[0072] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for using a natural language processing model to generate a response based on the user's emotion and the text input, and means for providing the generated response to the user, thereby enabling a personalized response based on the user's emotion to be provided in real time.

[0073] "User" refers to a person who performs an operation using a terminal.

[0074] "Facial image" refers to digital image data that captures an image of a user's face.

[0075] "Terminal" refers to an electronic device used by a user, such as a computer, smartphone, or tablet.

[0076] "Server" refers to a central system that receives and processes data sent by users.

[0077] "Preprocessing" refers to the initial processing of acquired data to convert it into a form that is easier to analyze.

[0078] An "emotion recognition model" refers to a machine learning model for analyzing a user's emotions from facial images, etc.

[0079] "Text input" refers to character string information that a user inputs through a terminal.

[0080] A "natural language processing model" refers to an algorithm or machine learning model that understands and generates natural language.

[0081] "Response" refers to an answer or message generated based on the user's input and predicted emotions.

[0082] "Voice input" refers to a method in which a user inputs instructions or questions into a terminal by voice.

[0083] "Speech recognition" refers to the technology of converting voice data into text data.

[0084] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. The following describes in detail each element of the system and the processing flow of the program.

[0085] This system begins by acquiring a facial image from the user's device. The user captures their own facial image using the device's camera. The device then sends the captured facial image to the server. The device can be a smartphone, tablet, or PC.

[0086] The server preprocesses the received facial images. This includes image resizing, noise reduction, and face detection using the OpenCV library. The preprocessed facial images are then input into an emotion recognition model built with TensorFlow. The model analyzes the facial images and predicts the user's emotions (e.g., joy, sadness, anger, etc.).

[0087] The user then enters text using a chat window on the terminal. For example, the user might enter "I'm tired today." The terminal transmits this text entry to the server.

[0088] The server receives the user's text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model, such as GPT-3. The following prompt sentence is generated and input into the model:

[0089] "Generate an appropriate response based on the user's sentiment and the text below:

[0090] Input text: I'm tired today

[0091] User Emotions: Tired

[0092] The natural language processing model uses this information to generate a response to the user, such as, "Thank you for your hard work. Do you know how to relax?"

[0093] The generated response is then sent back to the device and displayed to the user, allowing the user to receive a personalized response tailored to their emotions at that moment.

[0094] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[0095] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[0096] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0097] Step 1:

[0098] The user captures a facial image using the device's camera.

[0099] Specifically, the user activates the device's camera and takes a picture of their face. The input is the user's face image, which becomes the data used in the next step.

[0100] Step 2:

[0101] The device transmits the captured facial image to the server.

[0102] The device sends the captured facial image to the server via the Internet. Specifically, the image data is sent to the server via an HTTP POST request. The input is the facial image captured in step 1, and the output is the image data sent to the server.

[0103] Step 3:

[0104] The server preprocesses the received facial images.

[0105] The server preprocesses the received face image by using OpenCV to resize the image, remove noise, and detect faces. The input is the face image sent in step 2, and the output is the preprocessed face image.

[0106] Step 4:

[0107] The server inputs the preprocessed facial image into an emotion recognition model to predict the user's emotions.

[0108] The server inputs the preprocessed facial image into the emotion recognition model. Specifically, it uses TensorFlow to input the facial image into the emotion recognition model and predicts the user's emotion. The input is the preprocessed facial image, and the output is the predicted user emotion.

[0109] Step 5:

[0110] The user enters text in a chat window on the device.

[0111] In concrete terms, a user uses a chat window on a terminal to input text such as "I'm tired today." The input is the text input by the user, and the output is the text displayed on the terminal.

[0112] Step 6:

[0113] The terminal sends the user's text input to the server.

[0114] The terminal sends the user's text input to the server. Specifically, it sends the text data to the server via an HTTP POST request. The input is the text entered in step 5, and the output is the text data sent to the server.

[0115] Step 7:

[0116] The server inputs the user's text input and emotional information into a natural language processing model.

[0117] The server combines the user's text input with emotion information and inputs it into the natural language processing model. Specifically, it generates the following prompt sentence and inputs it into the natural language processing model:

[0118] "Generate an appropriate response based on the user's sentiment and the text below:

[0119] Input text: I'm tired today

[0120] User Emotions: Tired

[0121] The input is the user's text and emotional information, and the output is a prompt sent to a natural language processing model.

[0122] Step 8:

[0123] A natural language processing model generates an appropriate response to the user.

[0124] The natural language processing model generates a response based on the prompt. Specifically, the generative AI model generates a response such as, "Thank you for your hard work. Do you know how to relax?" The input is the prompt, and the output is the generated response.

[0125] Step 9:

[0126] The server sends the generated response to the terminal.

[0127] The server sends the generated response to the terminal. Specifically, it sends the response data to the terminal using an HTTP POST request. The input is the generated response, and the output is the response data sent to the terminal.

[0128] Step 10:

[0129] The terminal displays the response received from the server to the user.

[0130] The terminal displays the response received from the server in a chat window. Specifically, it displays the message "Thank you for your hard work. Do you know how to relax?" on the terminal screen. The input is the response data from the server, and the output is the display on the terminal.

[0131] (Application example 1)

[0132] 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."

[0133] Conventional customer service support systems lack the technology to analyze users' emotions in real time and provide appropriate responses based on those emotions, which limits the means for improving customer satisfaction. Furthermore, systems that assume only text input cannot realize interaction through voice input or image recognition, making it difficult to communicate naturally with users.

[0134] 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.

[0135] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for generating a response based on the user's emotion and the text input, means for providing the generated response to the user, means for capturing a facial image of the user with a camera mounted on the smart glasses, means for displaying personalized information on the smart glasses based on the emotion, and means for recognizing the user's voice input and generating a response based on the text, thereby realizing appropriate and natural dialogue that takes the user's emotion into consideration and improving customer satisfaction in physical stores.

[0136] "User" refers to a person who uses the system.

[0137] "Facial image" refers to image data of a user's face captured by a camera.

[0138] "Preprocessing" refers to image processing to make the acquired facial image easier to analyze.

[0139] "Emotion prediction" refers to the process of inferring a user's emotional state based on preprocessed facial images.

[0140] "Text input" refers to the act of a user inputting textual information into a system and the results of that action.

[0141] "Response generation" refers to the process of creating an appropriate response based on the user's sentiment and text input.

[0142] "Smart glasses" refers to a wearable device that incorporates a camera, display, etc. and can display information within the user's field of vision.

[0143] "Voice input" refers to the act of a user inputting voice information into a system and the results thereof.

[0144] A "natural language processing model" refers to an algorithm or model for understanding and generating human language.

[0145] "Personalized information" refers to information that is individually optimized according to the user's emotions and behavior.

[0146] "Capture" refers to the act of obtaining an image with a camera.

[0147] "Recognition" refers to analyzing input data and understanding its content and characteristics.

[0148] A system for implementing the present invention analyzes a user's emotions in real time and provides appropriate responses based on the emotions and the user's input. The system interacts with the user using smart glasses.

[0149] Hardware Configuration

[0150] The system uses the following hardware:

[0151] 1. Smart glasses: Equipped with a camera, display, and microphone.

[0152] 2. Server: A computer with powerful computing power and storage.

[0153] Software Configuration

[0154] The system uses the following software:

[0155] 1. Emotion recognition model: Model trained using Keras.

[0156] 2. Natural Language Processing model: A generative AI model (GPT-3) that uses the Transformers library.

[0157] 3. Speech Recognition Tools: Speech recognition software to convert the user's speech into text.

[0158] 4. gTTS and PyAudio: Software for converting response text into speech and playing it back.

[0159] Data processing and calculation

[0160] 1. Capture a user's face image:

[0161] The camera on the smart glasses captures the user's face.

[0162] The captured image data is transmitted to a server.

[0163] 2. Emotional Prediction:

[0164] The server preprocesses the received facial images, performs grayscale conversion and resizing.

[0165] The preprocessed image data is input into an emotion recognition model to predict the user's emotional state.

[0166] 3. Parsing user text input:

[0167] The smart glasses recognize voice input from the user and convert the voice data into text.

[0168] The converted text data and predicted emotion data are sent to the server.

[0169] 4. Response Generation:

[0170] The server uses a natural language processing model to generate an appropriate response based on the user's emotions and text input.

[0171] The generated response text is sent to the smart glasses and is displayed on the display and output as voice.

[0172] Specific examples

[0173] In a physical store, a customer is wearing smart glasses and asks, "Write a review about this product." The system then:

[0174] The camera in the smart glasses captures the customer's face and sends it to the server.

[0175] Predicting the emotion of "interested" using an emotion recognition model.

[0176] The questions are converted into text using voice recognition.

[0177] The following prompt sentences were generated using a natural language processing model (GPT-3) and responses were created:

[0178] text

[0179] User sentiment: Interested

[0180] Q: What are the reviews for this product?

[0181] Generates the response text "This product has high ratings and is well-received by many users."

[0182] Displayed on smart glasses and played back as audio at the same time.

[0183] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0184] Step 1:

[0185] The server controls the camera mounted on the smart glasses to capture the user's facial image. The server receives the camera image data as input and sends it to the server. The server generates the captured facial image data as output.

[0186] Step 2:

[0187] The server preprocesses the received facial image data by converting the image to grayscale and resizing it to the size required by the emotion recognition model. The server receives the facial image data as input and generates preprocessed image data as output.

[0188] Step 3:

[0189] The server inputs the preprocessed facial images into an emotion recognition model to predict the user's emotional state. This emotion recognition model is trained using Keras. It receives the preprocessed image data as input and generates the user's emotional data as output.

[0190] Step 4:

[0191] The user performs voice input through the smart glasses. The microphone in the smart glasses captures the voice data and transmits it to the terminal. The voice data is received as input and generated as output, which is transmitted to the terminal.

[0192] Step 5:

[0193] The server uses speech recognition software to convert the voice data into text. Specifically, it analyzes the voice data and generates text-based question data. The server receives the voice data as input and generates text data as output.

[0194] Step 6:

[0195] The server inputs the user's emotion data and text data into a generative AI model (GPT-3) to generate an appropriate response. The prompt uses a format such as "User emotion: Interested, Question: What are the reviews for this product?". Emotion data and text data are received as input, and response text data is generated as output.

[0196] Step 7:

[0197] The server sends the response text data to the smart glasses and displays it on the display. It also converts the response text into speech using gTTS and plays the speech to the user using PyAudio. The server receives the response text data as input and generates the text to be displayed on the display and the speech to be played as output.

[0198] 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.

[0199] The system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions. The following describes the system components and the program processing flow in natural language.

[0200] The system consists of several main components. First, the user's device plays an important role. This device is equipped with a camera, microphone, and input interface to capture the user's facial image, voice, and text input. The device then transmits this information to the server.

[0201] The server then performs several processing steps. First, the server receives the facial image sent by the user and performs preprocessing. This includes image normalization and feature point extraction. The preprocessed facial image is then passed to an emotion recognition engine. This engine analyzes the facial feature points and predicts the user's emotion. Examples of emotions include "happiness," "sadness," and "anger."

[0202] Next, the user inputs text. Using the input interface on the device, the user inputs a message such as "I'm tired today" and sends it to the server. The server receives this text input and inputs it into a natural language processing model along with the previously predicted emotional information. The natural language processing model generates a response based on the emotional information and the text input. For example, this response could be a message such as "Thank you for your hard work. Do you know how to relax?"

[0203] A more refined approach involves voice input, where the user records their voice using the device's microphone and sends it to the server. The server receives the voice data and uses an automatic speech recognition (ASR) engine to convert the speech to text. The text is then combined with emotion recognition information and similarly fed into a natural language processing model to generate an appropriate response.

[0204] A unique feature of this system is that it also includes the ability to recognize a user's emotions using biometric information other than facial images (e.g., voice and gestures). For example, by analyzing voice tone and gesture movements, it is possible to more accurately predict emotions based on this data. Furthermore, by referencing past conversation history, it is possible to continuously track emotions and provide a deeper understanding.

[0205] For example, if a user inputs "I'm stressed today," the server will predict emotions such as "stress" or "anger" from the facial recognition results. Based on this, the system will generate a response such as "Do you have any recommendations for relaxation methods to reduce stress?" and provide it to the user.

[0206] As described above, by combining an emotion engine, the present invention makes it possible to analyze user emotions from multiple angles and provide more personalized interactions, thereby achieving significant improvements in customer experience and satisfaction.

[0207] The processing flow will be explained below.

[0208] A system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions.

[0209] Step 1:

[0210] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[0211] Step 2:

[0212] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image resizing, grayscale conversion, and facial feature point extraction.

[0213] Step 3:

[0214] The server passes the preprocessed facial image to an emotion recognition engine, which analyzes facial feature points and predicts the user's emotions. For example, it identifies emotions such as "happiness," "sadness," and "anger."

[0215] Step 4:

[0216] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[0217] Step 5:

[0218] The server processes the text input received from the user, integrates the received text input with the predicted emotional information, and creates a context to be input into the natural language processing model.

[0219] Step 6:

[0220] The server loads a natural language processing model and inputs text input along with emotional information. The natural language processing model generates an optimal response based on the user's input and emotional state. For example, it creates a response like, "You're tired. Do you know how to relax?"

[0221] Step 7:

[0222] The server sends the generated response to the user's terminal.

[0223] Step 8:

[0224] The user checks the response displayed on the device, allowing the user to receive a personalized response that matches their emotions.

[0225] Step 9:

[0226] When the user provides voice input, the terminal's microphone is used to record a voice message and send it to the server.

[0227] Step 10:

[0228] The server receives the voice data and loads an automatic speech recognition (ASR) engine, which converts the voice data into text.

[0229] Step 11:

[0230] The server then inputs the converted text and emotional information into a natural language processing model to generate an appropriate response, such as "What are your plans for today?"

[0231] Step 12:

[0232] The server sends the generated response to the user's terminal, and the user checks the response on the terminal.

[0233] Through this series of steps, the system can understand user emotions using multimodal input data and provide personalized interactions. Furthermore, by using an emotion engine, it can also utilize biometric information such as voice and gestures to more accurately recognize user emotions. This significantly improves customer experience and satisfaction.

[0234] Example 2

[0235] 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."

[0236] Conventional emotion analysis systems often have limited information for accurately predicting user emotions, using only facial images and text data, limiting their analytical accuracy. Furthermore, even systems that analyze emotions using user voice data have difficulty integrating the analysis of voice data with other data. Given these circumstances, the challenge is to analyze user emotions from multiple angles and generate more accurate and personalized responses.

[0237] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring image data from a user, means for preprocessing the acquired image data, means for analyzing the preprocessed image data and predicting the user's emotions, means for receiving text data from the user, means for generating a response based on the user's emotions and the text data, means for providing the generated response to the user, means for acquiring voice data and converting it into text data, and means for integrating and analyzing the emotion data and the voice data. This enables an integrated analysis of multiple data sources for a user, enabling more accurate and personalized responses.

[0238] A "user" is an entity that uses the system and provides data.

[0239] "Image data" is visual information captured using a camera or other imaging device.

[0240] "Preprocessing" refers to the initial processing operations performed to convert raw data into a format that is easier to analyze and process.

[0241] A "means for predicting emotions" is an algorithm or engine for inferring a user's emotional state based on acquired data.

[0242] "Text data" refers to information entered by a user in the form of text, typically using a keyboard or touch screen.

[0243] A "natural language processing model" is an algorithm or machine learning model for analyzing text data and understanding its meaning.

[0244] "Audio data" refers to audio information captured by a device such as a microphone, which needs to be converted into a text format.

[0245] An "automatic speech recognition (ASR) engine" is a technology for analyzing voice data and converting it into text data.

[0246] "Emotion data" refers to the emotional state predicted by the emotion recognition means, typically expressed numerically or categorically.

[0247] The "means for generating a response" is an algorithm or model for creating an appropriate response to provide to a user based on the emotional data and text data.

[0248] The "means for providing to the user" refers to an interface or device for transmitting the generated response to the user.

[0249] MODE FOR CARRYING OUT THE INVENTION

[0250] The system embodying the present invention analyzes various data of a user and generates a response based on the user's emotions. This system includes a user terminal, a server, and various software engines. The components and operation of the system are described in detail below.

[0251] User terminal

[0252] The user terminal includes the following hardware:

[0253] Camera: Used to capture the user's facial image.

[0254] Microphone: Used to record the user's voice input.

[0255] Input Interface: This includes keyboards and touchscreens.

[0256] The user terminal serves to transmit facial images, voice, and text input to the server in real time.

[0257] server

[0258] The server processes and analyzes the received data in cooperation with the following software engines:

[0259] Pre-processing engine: Normalizes the received face image and extracts facial feature points.

[0260] Emotion recognition engine: Analyzes preprocessed facial feature points to predict the user's emotions. For example, this engine can detect emotions such as "happiness," "sadness," and "anger."

[0261] Automatic Speech Recognition (ASR) engine: Converts user voice data into text.

[0262] Natural language processing (NLP) models: Generate appropriate responses based on text and sentiment data.

[0263] Data Handling

[0264] 1. Facial image capture and processing

[0265] The user faces the camera and the user terminal captures an image of the user's face.

[0266] The server receives the facial image, and the pre-processing engine normalizes the image and extracts feature points, such as identifying the positions of the eyes, mouth, and nose.

[0267] 2. Emotion analysis

[0268] The server's emotion recognition engine predicts the user's emotions based on the extracted feature points. For example, it analyzes the "muscle movements around the eyes" and the "angle of the corners of the mouth" to determine "happiness" or "sadness."

[0269] The results are stored as "Happiness: 80%", "Sadness: 15%", "Anger: 5%".

[0270] 3. Handling Text Input

[0271] The user uses the keyboard or touch screen on the device to input a message such as "I'm tired today" and transmits it to the server.

[0272] The server receives this text, combines it with emotional data, and inputs it into the NLP model.

[0273] 4. Processing voice input

[0274] The user uses a microphone to input speech, for example, "I'm feeling stressed today."

[0275] The server's ASR engine converts the voice data into text, then combines the converted text with emotional data and inputs it into the NLP model.

[0276] 5. Generating and Serving the Response

[0277] The NLP model uses emotional and textual data to generate a personalized response for the user, such as "Thank you for your hard work. Do you know how to relax?"

[0278] The generated response is sent from the server to the terminal and displayed to the user through the terminal's interface.

[0279] Examples and prompts

[0280] For example, if a user types "I'm feeling stressed today":

[0281] The user's device takes a facial image and sends it to the server.

[0282] The server preprocesses the images and uses an emotion recognition engine to predict emotions such as "stress" or "anger."

[0283] The user types the text "I'm stressed today" and sends it to the server.

[0284] The server inputs this information into an NLP model, generates a response such as "What relaxation techniques would you recommend to reduce stress?" and displays it on the device.

[0285] This system enables multifaceted analysis of user sentiment, enabling more accurate and personalized responses, resulting in a significant improvement in customer experience and satisfaction.

[0286] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0287] Processing steps of this system's program

[0288] Step 1: Capture user data

[0289] 1. Specific actions

[0290] The user faces the device's camera, which captures the face image and simultaneously records the voice through the microphone.

[0291] The user enters text using a keyboard or a touch screen.

[0292] 2. Input and Output

[0293] Input: User's face image, voice data, text input.

[0294] Output: The captured face image, voice data, and text data are sent from the device to the server.

[0295] Step 2: Preprocessing the face image

[0296] 1. Specific actions

[0297] The server receives the facial image sent from the terminal.

[0298] The pre-processing engine normalizes the image and extracts facial feature points (eyes, mouth, nose, etc.).

[0299] 2. Input and Output

[0300] Input: A face image sent from the device.

[0301] Output: Normalized and feature-point extracted face image data.

[0302] Step 3: Sentiment Analysis

[0303] 1. Specific actions

[0304] An emotion recognition engine analyzes the feature points of the pre-processed facial image data.

[0305] Algorithms are applied to predict emotions such as "happiness," "sadness," and "anger."

[0306] 2. Input and Output

[0307] Input: Feature points from preprocessed face image data.

[0308] Output: Predicted emotion data (e.g., "Happy: 70%", "Sad: 20%", "Anger: 10%").

[0309] Step 4: Integrating text data

[0310] 1. Specific actions

[0311] The server receives the text data sent by the user.

[0312] Integrate emotion data and text data.

[0313] 2. Input and Output

[0314] Input: Text data sent by the user, predicted emotion data.

[0315] Output: Integrated text and sentiment data.

[0316] Step 5: Process the audio data

[0317] 1. Specific actions

[0318] The server receives the voice data and converts the speech to text using an automatic speech recognition (ASR) engine.

[0319] The converted text data and emotion data are integrated.

[0320] 2. Input and Output

[0321] Input: Audio data sent by the user.

[0322] Output: Translated speech data, integrated text data and emotion data.

[0323] Step 6: Response Generation

[0324] 1. Specific actions

[0325] The server inputs the integrated text data and sentiment data into a natural language processing (NLP) model.

[0326] The NLP model generates the appropriate response.

[0327] 2. Input and Output

[0328] Input: Integrated text and sentiment data.

[0329] Output: The generated response (e.g., "Good work! Do you know how to relax?").

[0330] Step 7: Providing a response

[0331] 1. Specific actions

[0332] The server sends the generated response to the terminal.

[0333] The terminal displays the received response to the user.

[0334] 2. Input and Output

[0335] Input: The generated response.

[0336] Output: The response that is displayed to the user.

[0337] Through the above processing steps, various types of user data can be analyzed from multiple angles, and a personalized response can be generated and provided based on the results.

[0338] (Application example 2)

[0339] 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."

[0340] Detecting suspicious individuals and early detection of potential threats is a challenge for modern security systems. It is particularly difficult to quickly and accurately analyze the emotions of individuals in a crowd and provide appropriate responses. Therefore, there is a need for a system that allows security staff to analyze emotions in real time on-site and issue appropriate alerts and responses based on the results.

[0341] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring a facial image from a user's terminal, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for acquiring voice from the user's terminal, means for converting the acquired voice into text, means for generating a response based on the user's emotion and text input, and means for providing the generated response to the user's terminal. This enables security staff to analyze individual emotions in real time on site and detect and respond to potential threats in advance.

[0342] A "user terminal" is a device equipped with input devices such as a camera and microphone, which captures facial images and voices from the user and transmits them to a server.

[0343] The "means for acquiring a facial image" refers to a function for capturing a facial image of a user using a camera mounted on the user's terminal.

[0344] The "means for preprocessing a facial image" refers to a function for performing preprocessing such as image normalization and feature point detection on an acquired facial image.

[0345] The "means for analyzing a facial image and predicting a user's emotions" refers to a function that analyzes facial feature points based on a preprocessed facial image and predicts a user's emotions.

[0346] The "means for acquiring voice" refers to a function for capturing the user's voice using a microphone installed in the user's terminal.

[0347] The "means for converting voice to text" refers to a function for converting acquired voice data into text data.

[0348] The "means for generating a response" refers to a function that uses a natural language processing model to generate an appropriate response based on the user's sentiment and text input.

[0349] The "means for providing the generated response to the user" refers to a function for displaying the generated response on the user's terminal or reproducing it as audio.

[0350] A "generative AI model" is an artificial intelligence model that generates natural-sounding language responses based on input data.

[0351] This invention is a system that uses multimodal data (facial images, voice) acquired from a user's device to analyze emotions in real time and generate and provide responses based on that data. The purpose of this system is to analyze a user's emotions from multiple angles and realize more accurate and appropriate responses.

[0352] Hardware and software used

[0353] 1. Hardware used:

[0354] Smart glasses: Equipped with a camera and microphone to capture the user's facial image and voice.

[0355] Server: Preprocesses data, analyzes it, and generates responses.

[0356] 2. Software used:

[0357] OpenCV (cv2): Image capture and preprocessing.

[0358] dlib: Face detection and feature extraction.

[0359] speech_recognition: Speech capture and text conversion.

[0360] tensorflow: Emotion recognition model.

[0361] transformers(Hugging Face): Natural language processing model.

[0362] Process Overview

[0363] Facial image capture and analysis

[0364] The user wears the smart glasses and captures facial images in real time through the camera. The images are sent to the server and preprocessed using OpenCV. This includes image normalization and facial feature point extraction using dlib. Then, emotions are analyzed from the images using a TensorFlow emotion recognition model.

[0365] Audio capture and conversion

[0366] Similarly, the user's voice is captured through a microphone built into the smart glasses, which is then sent to the server and converted to text using speech_recognition.

[0367] Generating and serving the response

[0368] The server generates an appropriate response using a generative AI model based on the analyzed emotion results and the text converted from speech, and this response is fed back to the user through the smart glasses.

[0369] Specific examples

[0370] For example, if security staff wear smart glasses at the entrance to an event venue, they can capture the faces and voices of attendees and analyze their emotions. If anger or stress is detected, the system can generate an alert that the attendee is likely overly excited and notify security staff.

[0371] Prompt Sentence Examples

[0372] "This user's emotion is anger. Then respond appropriately to the following text: 'I'm frustrated with the traffic today.'"

[0373] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0374] Step 1:

[0375] The user's device captures a facial image using the camera in the smart glasses. Specifically, the camera acquires an image and stores it as digital data. The input is the facial image captured by the camera. This data is sent to the server.

[0376] Step 2:

[0377] The server preprocesses the received facial image data. Specifically, it normalizes the image using OpenCV and then extracts facial feature points using dlib. The input is the captured facial image, and the output is the preprocessed image data and coordinate information of the feature points.

[0378] Step 3:

[0379] The server inputs the preprocessed facial image data into an emotion recognition model to predict the user's emotion. Specifically, the data is input into the TensorFlow emotion recognition model and an estimated emotion label is obtained. The input is the coordinate information of the feature points, and the output is the emotion label (e.g., "joy," "sadness," or "anger").

[0380] Step 4:

[0381] The user's device captures voice data using the microphone in the smart glasses. Specifically, the microphone records the voice and converts it into digital data. The input is the user's voice, and the output is voice data. This voice data is then sent to the server.

[0382] Step 5:

[0383] The server converts the received voice data into text data. Specifically, it uses the speech_recognition library to perform voice recognition and converts it into text. The input is voice data, and the output is text data.

[0384] Step 6:

[0385] The server generates a response based on the emotion analysis results and the text data converted from the speech. Specifically, it uses Hugging Face's transformers library to input a prompt into a generative AI model and generate an appropriate response. The input is emotion labels and text data, and the output is a response in natural language.

[0386] Step 7:

[0387] The server then sends the generated response to the user's device. Specifically, the server sends the generated response to the user's smart glasses via a network. The input is a natural language response, and the output is displayed or played on the smart glasses' display or audio output.

[0388] Step 8:

[0389] The user's device provides the generated response to the user using the smart glasses' display or audio output. Specifically, the response is displayed as text on the display or played as audio output. The input is the response received from the server, and the output is a display or audio notification to the user.

[0390] 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.

[0391] 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.

[0392] 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.

[0393] [Second embodiment]

[0394] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0395] 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.

[0396] 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).

[0397] 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.

[0398] 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.

[0399] 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).

[0400] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for 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.

[0401] 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.

[0402] 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.

[0403] 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.

[0404] In the smart glasses 214, the 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.

[0405] 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."

[0406] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. Below, each element of the system and the processing flow of the program are explained in natural language.

[0407] The system begins by acquiring a facial image from the user's device. The device captures the user's facial image with a camera and sends it to the server. The server preprocesses the received facial image and inputs it into an emotion recognition model. The model analyzes the facial image and predicts the user's emotion (e.g., joy, sadness, anger, etc.).

[0408] Next, the user inputs text. Using a chat window on the device, the user enters a question or request and sends it to the server. The server receives the text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model. The natural language processing model generates a response to the user based on this information.

[0409] The generated response is then sent back to the device and displayed to the user. This allows the user to receive a personalized response that matches their emotions at that time. For example, if the user enters the text "I'm tired today," the server will recognize the emotion "tired" based on image analysis, and generate a response such as "Thank you for your hard work. Do you know how to relax?" based on that emotion and the text, and send it to the device.

[0410] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[0411] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[0412] The processing flow will be explained below.

[0413] Step 1:

[0414] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[0415] Step 2:

[0416] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image normalization and feature point extraction.

[0417] Step 3:

[0418] The server inputs the preprocessed facial image into an emotion recognition model. The emotion recognition model is composed of trained neural networks and other components, and predicts the user's emotions from the facial image. The predicted emotions are expressed with labels such as "happiness," "sadness," and "anger."

[0419] Step 4:

[0420] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[0421] Step 5:

[0422] The server receives text input from the user, then combines the received text with the previously predicted sentiment information to create a request that is input into a natural language processing model.

[0423] Step 6:

[0424] The server loads a natural language processing model and inputs the text and sentiment information together. The natural language processing model generates an optimal response based on the user's message and sentiment. For example, this response might be, "You're tired. Do you know how to relax?"

[0425] Step 7:

[0426] The server sends the generated response to the user's terminal.

[0427] Step 8:

[0428] The user can then check the response displayed on the device, which allows the user to receive a personalized response tailored to their emotions.

[0429] Step 9:

[0430] When the user performs voice input, the user uses the microphone of the terminal to record a voice message and transmits it to the server.

[0431] Step 10:

[0432] The server receives the voice data and loads an automatic speech recognition (ASR) model, which converts the voice data into text.

[0433] Step 11:

[0434] The server then uses the converted text and emotional information to re-feed data into the natural language processing model to generate an appropriate response.

[0435] Step 12:

[0436] The server sends the generated response to the user's terminal, where the user confirms it.

[0437] Through this series of steps, InsightBot Advanced uses information from multimodal inputs to understand sentiment and deliver personalized interactions.

[0438] Example 1

[0439] 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."

[0440] Conventional chat systems and automated response systems do not take into account the user's emotions, making it difficult to provide personalized responses, and many users end up with dissatisfied results. Furthermore, because they cannot handle multiple input modalities, such as voice and images in addition to text, they lack the ability to respond in a variety of usage scenarios.

[0441] 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.

[0442] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for using a natural language processing model to generate a response based on the user's emotion and the text input, and means for providing the generated response to the user, thereby enabling a personalized response based on the user's emotion to be provided in real time.

[0443] "User" refers to a person who performs an operation using a terminal.

[0444] "Facial image" refers to digital image data that captures an image of a user's face.

[0445] "Terminal" refers to an electronic device used by a user, such as a computer, smartphone, or tablet.

[0446] "Server" refers to a central system that receives and processes data sent by users.

[0447] "Preprocessing" refers to the initial processing of acquired data to convert it into a form that is easier to analyze.

[0448] An "emotion recognition model" refers to a machine learning model for analyzing a user's emotions from facial images, etc.

[0449] "Text input" refers to character string information that a user inputs through a terminal.

[0450] A "natural language processing model" refers to an algorithm or machine learning model that understands and generates natural language.

[0451] "Response" refers to an answer or message generated based on the user's input and predicted emotions.

[0452] "Voice input" refers to a method in which a user inputs instructions or questions into a terminal by voice.

[0453] "Speech recognition" refers to the technology of converting voice data into text data.

[0454] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. The following describes in detail each element of the system and the processing flow of the program.

[0455] This system begins by acquiring a facial image from the user's device. The user captures their own facial image using the device's camera. The device then sends the captured facial image to the server. The device can be a smartphone, tablet, or PC.

[0456] The server preprocesses the received facial images. This includes image resizing, noise reduction, and face detection using the OpenCV library. The preprocessed facial images are then input into an emotion recognition model built with TensorFlow. The model analyzes the facial images and predicts the user's emotions (e.g., joy, sadness, anger, etc.).

[0457] The user then enters text using a chat window on the terminal. For example, the user might enter "I'm tired today." The terminal transmits this text entry to the server.

[0458] The server receives the user's text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model, such as GPT-3. The following prompt sentence is generated and input into the model:

[0459] "Generate an appropriate response based on the user's sentiment and the text below:

[0460] Input text: I'm tired today

[0461] User Emotions: Tired

[0462] The natural language processing model uses this information to generate a response to the user, such as, "Thank you for your hard work. Do you know how to relax?"

[0463] The generated response is then sent back to the device and displayed to the user, allowing the user to receive a personalized response tailored to their emotions at that moment.

[0464] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[0465] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[0466] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0467] Step 1:

[0468] The user captures a facial image using the device's camera.

[0469] Specifically, the user activates the device's camera and takes a picture of their face. The input is the user's face image, which becomes the data used in the next step.

[0470] Step 2:

[0471] The device transmits the captured facial image to the server.

[0472] The device sends the captured facial image to the server via the Internet. Specifically, the image data is sent to the server via an HTTP POST request. The input is the facial image captured in step 1, and the output is the image data sent to the server.

[0473] Step 3:

[0474] The server preprocesses the received facial images.

[0475] The server preprocesses the received face image by using OpenCV to resize the image, remove noise, and detect faces. The input is the face image sent in step 2, and the output is the preprocessed face image.

[0476] Step 4:

[0477] The server inputs the preprocessed facial image into an emotion recognition model to predict the user's emotions.

[0478] The server inputs the preprocessed facial image into the emotion recognition model. Specifically, it uses TensorFlow to input the facial image into the emotion recognition model and predicts the user's emotion. The input is the preprocessed facial image, and the output is the predicted user emotion.

[0479] Step 5:

[0480] The user enters text in a chat window on the device.

[0481] In concrete terms, a user uses a chat window on a terminal to input text such as "I'm tired today." The input is the text input by the user, and the output is the text displayed on the terminal.

[0482] Step 6:

[0483] The terminal sends the user's text input to the server.

[0484] The terminal sends the user's text input to the server. Specifically, it sends the text data to the server via an HTTP POST request. The input is the text entered in step 5, and the output is the text data sent to the server.

[0485] Step 7:

[0486] The server inputs the user's text input and emotional information into a natural language processing model.

[0487] The server combines the user's text input with emotion information and inputs it into the natural language processing model. Specifically, it generates the following prompt sentence and inputs it into the natural language processing model:

[0488] "Generate an appropriate response based on the user's sentiment and the text below:

[0489] Input text: I'm tired today

[0490] User Emotions: Tired

[0491] The input is the user's text and emotional information, and the output is a prompt sent to a natural language processing model.

[0492] Step 8:

[0493] A natural language processing model generates an appropriate response to the user.

[0494] The natural language processing model generates a response based on the prompt. Specifically, the generative AI model generates a response such as, "Thank you for your hard work. Do you know how to relax?" The input is the prompt, and the output is the generated response.

[0495] Step 9:

[0496] The server sends the generated response to the terminal.

[0497] The server sends the generated response to the terminal. Specifically, it sends the response data to the terminal using an HTTP POST request. The input is the generated response, and the output is the response data sent to the terminal.

[0498] Step 10:

[0499] The terminal displays the response received from the server to the user.

[0500] The terminal displays the response received from the server in a chat window. Specifically, it displays the message "Thank you for your hard work. Do you know how to relax?" on the terminal screen. The input is the response data from the server, and the output is the display on the terminal.

[0501] (Application example 1)

[0502] 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."

[0503] Conventional customer service support systems lack the technology to analyze users' emotions in real time and provide appropriate responses based on those emotions, which limits the means for improving customer satisfaction. Furthermore, systems that assume only text input cannot realize interaction through voice input or image recognition, making it difficult to communicate naturally with users.

[0504] 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.

[0505] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for generating a response based on the user's emotion and the text input, means for providing the generated response to the user, means for capturing a facial image of the user with a camera mounted on the smart glasses, means for displaying personalized information on the smart glasses based on the emotion, and means for recognizing the user's voice input and generating a response based on the text, thereby realizing appropriate and natural dialogue that takes the user's emotion into consideration and improving customer satisfaction in physical stores.

[0506] "User" refers to a person who uses the system.

[0507] "Facial image" refers to image data of a user's face captured by a camera.

[0508] "Preprocessing" refers to image processing to make the acquired facial image easier to analyze.

[0509] "Emotion prediction" refers to the process of inferring a user's emotional state based on preprocessed facial images.

[0510] "Text input" refers to the act of a user inputting textual information into a system and the results of that action.

[0511] "Response generation" refers to the process of creating an appropriate response based on the user's sentiment and text input.

[0512] "Smart glasses" refers to a wearable device that incorporates a camera, display, etc. and can display information within the user's field of vision.

[0513] "Voice input" refers to the act of a user inputting voice information into a system and the results thereof.

[0514] A "natural language processing model" refers to an algorithm or model for understanding and generating human language.

[0515] "Personalized information" refers to information that is individually optimized according to the user's emotions and behavior.

[0516] "Capture" refers to the act of obtaining an image with a camera.

[0517] "Recognition" refers to analyzing input data and understanding its content and characteristics.

[0518] A system for implementing the present invention analyzes a user's emotions in real time and provides appropriate responses based on the emotions and the user's input. The system interacts with the user using smart glasses.

[0519] Hardware Configuration

[0520] The system uses the following hardware:

[0521] 1. Smart glasses: Equipped with a camera, display, and microphone.

[0522] 2. Server: A computer with powerful computing power and storage.

[0523] Software Configuration

[0524] The system uses the following software:

[0525] 1. Emotion recognition model: Model trained using Keras.

[0526] 2. Natural Language Processing model: A generative AI model (GPT-3) that uses the Transformers library.

[0527] 3. Speech Recognition Tools: Speech recognition software to convert the user's speech into text.

[0528] 4. gTTS and PyAudio: Software for converting response text into speech and playing it back.

[0529] Data processing and calculation

[0530] 1. Capture a user's face image:

[0531] The camera on the smart glasses captures the user's face.

[0532] The captured image data is transmitted to a server.

[0533] 2. Emotional Prediction:

[0534] The server preprocesses the received facial images, performs grayscale conversion and resizing.

[0535] The preprocessed image data is input into an emotion recognition model to predict the user's emotional state.

[0536] 3. Parsing user text input:

[0537] The smart glasses recognize voice input from the user and convert the voice data into text.

[0538] The converted text data and predicted emotion data are sent to the server.

[0539] 4. Response Generation:

[0540] The server uses a natural language processing model to generate an appropriate response based on the user's emotions and text input.

[0541] The generated response text is sent to the smart glasses and is displayed on the display and output as voice.

[0542] Specific examples

[0543] In a physical store, a customer is wearing smart glasses and asks, "Write a review about this product." The system then:

[0544] The camera in the smart glasses captures the customer's face and sends it to the server.

[0545] Predicting the emotion of "interested" using an emotion recognition model.

[0546] The questions are converted into text using voice recognition.

[0547] The following prompt sentences were generated using a natural language processing model (GPT-3) and responses were created:

[0548] text

[0549] User sentiment: Interested

[0550] Q: What are the reviews for this product?

[0551] Generates the response text "This product has high ratings and is well-received by many users."

[0552] Displayed on smart glasses and played back as audio at the same time.

[0553] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0554] Step 1:

[0555] The server controls the camera mounted on the smart glasses to capture the user's facial image. The server receives the camera image data as input and sends it to the server. The server generates the captured facial image data as output.

[0556] Step 2:

[0557] The server preprocesses the received facial image data by converting the image to grayscale and resizing it to the size required by the emotion recognition model. The server receives the facial image data as input and generates preprocessed image data as output.

[0558] Step 3:

[0559] The server inputs the preprocessed facial images into an emotion recognition model to predict the user's emotional state. This emotion recognition model is trained using Keras. It receives the preprocessed image data as input and generates the user's emotional data as output.

[0560] Step 4:

[0561] The user performs voice input through the smart glasses. The microphone in the smart glasses captures the voice data and transmits it to the terminal. The voice data is received as input and generated as output, which is transmitted to the terminal.

[0562] Step 5:

[0563] The server uses speech recognition software to convert the voice data into text. Specifically, it analyzes the voice data and generates text-based question data. The server receives the voice data as input and generates text data as output.

[0564] Step 6:

[0565] The server inputs the user's emotion data and text data into a generative AI model (GPT-3) to generate an appropriate response. The prompt uses a format such as "User emotion: Interested, Question: What are the reviews for this product?". Emotion data and text data are received as input, and response text data is generated as output.

[0566] Step 7:

[0567] The server sends the response text data to the smart glasses and displays it on the display. It also converts the response text into speech using gTTS and plays the speech to the user using PyAudio. The server receives the response text data as input and generates the text to be displayed on the display and the speech to be played as output.

[0568] 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.

[0569] The system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions. The following describes the system components and the program processing flow in natural language.

[0570] The system consists of several main components. First, the user's device plays an important role. This device is equipped with a camera, microphone, and input interface to capture the user's facial image, voice, and text input. The device then transmits this information to the server.

[0571] The server then performs several processing steps. First, the server receives the facial image sent by the user and performs preprocessing. This includes image normalization and feature point extraction. The preprocessed facial image is then passed to an emotion recognition engine. This engine analyzes the facial feature points and predicts the user's emotion. Examples of emotions include "happiness," "sadness," and "anger."

[0572] Next, the user inputs text. Using the input interface on the device, the user inputs a message such as "I'm tired today" and sends it to the server. The server receives this text input and inputs it into a natural language processing model along with the previously predicted emotional information. The natural language processing model generates a response based on the emotional information and the text input. For example, this response could be a message such as "Thank you for your hard work. Do you know how to relax?"

[0573] A more refined approach involves voice input, where the user records their voice using the device's microphone and sends it to the server. The server receives the voice data and uses an automatic speech recognition (ASR) engine to convert the speech to text. The text is then combined with emotion recognition information and similarly fed into a natural language processing model to generate an appropriate response.

[0574] A unique feature of this system is that it also includes the ability to recognize a user's emotions using biometric information other than facial images (e.g., voice and gestures). For example, by analyzing voice tone and gesture movements, it is possible to more accurately predict emotions based on this data. Furthermore, by referencing past conversation history, it is possible to continuously track emotions and provide a deeper understanding.

[0575] For example, if a user inputs "I'm stressed today," the server will predict emotions such as "stress" or "anger" from the facial recognition results. Based on this, the system will generate a response such as "Do you have any recommendations for relaxation methods to reduce stress?" and provide it to the user.

[0576] As described above, by combining an emotion engine, the present invention makes it possible to analyze user emotions from multiple angles and provide more personalized interactions, thereby achieving significant improvements in customer experience and satisfaction.

[0577] The processing flow will be explained below.

[0578] A system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions.

[0579] Step 1:

[0580] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[0581] Step 2:

[0582] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image resizing, grayscale conversion, and facial feature point extraction.

[0583] Step 3:

[0584] The server passes the preprocessed facial image to an emotion recognition engine, which analyzes facial feature points and predicts the user's emotions. For example, it identifies emotions such as "happiness," "sadness," and "anger."

[0585] Step 4:

[0586] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[0587] Step 5:

[0588] The server processes the text input received from the user, integrates the received text input with the predicted emotional information, and creates a context to be input into the natural language processing model.

[0589] Step 6:

[0590] The server loads a natural language processing model and inputs text input along with emotional information. The natural language processing model generates an optimal response based on the user's input and emotional state. For example, it creates a response like, "You're tired. Do you know how to relax?"

[0591] Step 7:

[0592] The server sends the generated response to the user's terminal.

[0593] Step 8:

[0594] The user checks the response displayed on the device, allowing the user to receive a personalized response that matches their emotions.

[0595] Step 9:

[0596] When the user provides voice input, the terminal's microphone is used to record a voice message and send it to the server.

[0597] Step 10:

[0598] The server receives the voice data and loads an automatic speech recognition (ASR) engine, which converts the voice data into text.

[0599] Step 11:

[0600] The server then inputs the converted text and emotional information into a natural language processing model to generate an appropriate response, such as "What are your plans for today?"

[0601] Step 12:

[0602] The server sends the generated response to the user's terminal, and the user checks the response on the terminal.

[0603] Through this series of steps, the system can understand user emotions using multimodal input data and provide personalized interactions. Furthermore, by using an emotion engine, it can also utilize biometric information such as voice and gestures to more accurately recognize user emotions. This significantly improves customer experience and satisfaction.

[0604] Example 2

[0605] 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."

[0606] Conventional emotion analysis systems often have limited information for accurately predicting user emotions, using only facial images and text data, limiting their analytical accuracy. Furthermore, even systems that analyze emotions using user voice data have difficulty integrating the analysis of voice data with other data. Given these circumstances, the challenge is to analyze user emotions from multiple angles and generate more accurate and personalized responses.

[0607] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring image data from a user, means for preprocessing the acquired image data, means for analyzing the preprocessed image data and predicting the user's emotions, means for receiving text data from the user, means for generating a response based on the user's emotions and the text data, means for providing the generated response to the user, means for acquiring voice data and converting it into text data, and means for integrating and analyzing the emotion data and the voice data. This enables an integrated analysis of multiple data sources for a user, enabling more accurate and personalized responses.

[0608] A "user" is an entity that uses the system and provides data.

[0609] "Image data" is visual information captured using a camera or other imaging device.

[0610] "Preprocessing" refers to the initial processing operations performed to convert raw data into a format that is easier to analyze and process.

[0611] A "means for predicting emotions" is an algorithm or engine for inferring a user's emotional state based on acquired data.

[0612] "Text data" refers to information entered by a user in the form of text, typically using a keyboard or touch screen.

[0613] A "natural language processing model" is an algorithm or machine learning model for analyzing text data and understanding its meaning.

[0614] "Audio data" refers to audio information captured by a device such as a microphone, which needs to be converted into a text format.

[0615] An "automatic speech recognition (ASR) engine" is a technology for analyzing voice data and converting it into text data.

[0616] "Emotion data" refers to the emotional state predicted by the emotion recognition means, typically expressed numerically or categorically.

[0617] The "means for generating a response" is an algorithm or model for creating an appropriate response to provide to a user based on the emotional data and text data.

[0618] The "means for providing to the user" refers to an interface or device for transmitting the generated response to the user.

[0619] MODE FOR CARRYING OUT THE INVENTION

[0620] The system embodying the present invention analyzes various data of a user and generates a response based on the user's emotions. This system includes a user terminal, a server, and various software engines. The components and operation of the system are described in detail below.

[0621] User terminal

[0622] The user terminal includes the following hardware:

[0623] Camera: Used to capture the user's facial image.

[0624] Microphone: Used to record the user's voice input.

[0625] Input Interface: This includes keyboards and touchscreens.

[0626] The user terminal serves to transmit facial images, voice, and text input to the server in real time.

[0627] server

[0628] The server processes and analyzes the received data in cooperation with the following software engines:

[0629] Pre-processing engine: Normalizes the received face image and extracts facial feature points.

[0630] Emotion recognition engine: Analyzes preprocessed facial feature points to predict the user's emotions. For example, this engine can detect emotions such as "happiness," "sadness," and "anger."

[0631] Automatic Speech Recognition (ASR) engine: Converts user voice data into text.

[0632] Natural language processing (NLP) models: Generate appropriate responses based on text and sentiment data.

[0633] Data Handling

[0634] 1. Facial image capture and processing

[0635] The user faces the camera and the user terminal captures an image of the user's face.

[0636] The server receives the facial image, and the pre-processing engine normalizes the image and extracts feature points, such as identifying the positions of the eyes, mouth, and nose.

[0637] 2. Emotion analysis

[0638] The server's emotion recognition engine predicts the user's emotions based on the extracted feature points. For example, it analyzes the "muscle movements around the eyes" and the "angle of the corners of the mouth" to determine "happiness" or "sadness."

[0639] The results are stored as "Happiness: 80%", "Sadness: 15%", "Anger: 5%".

[0640] 3. Handling Text Input

[0641] The user uses the keyboard or touch screen on the device to input a message such as "I'm tired today" and transmits it to the server.

[0642] The server receives this text, combines it with emotional data, and inputs it into the NLP model.

[0643] 4. Processing voice input

[0644] The user uses a microphone to input speech, for example, "I'm feeling stressed today."

[0645] The server's ASR engine converts the voice data into text, then combines the converted text with emotional data and inputs it into the NLP model.

[0646] 5. Generating and Serving the Response

[0647] The NLP model uses emotional and textual data to generate a personalized response for the user, such as "Thank you for your hard work. Do you know how to relax?"

[0648] The generated response is sent from the server to the terminal and displayed to the user through the terminal's interface.

[0649] Examples and prompts

[0650] For example, if a user types "I'm feeling stressed today":

[0651] The user's device takes a facial image and sends it to the server.

[0652] The server preprocesses the images and uses an emotion recognition engine to predict emotions such as "stress" or "anger."

[0653] The user types the text "I'm stressed today" and sends it to the server.

[0654] The server inputs this information into an NLP model, generates a response such as "What relaxation techniques would you recommend to reduce stress?" and displays it on the device.

[0655] This system enables multifaceted analysis of user sentiment, enabling more accurate and personalized responses, resulting in a significant improvement in customer experience and satisfaction.

[0656] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0657] Processing steps of this system's program

[0658] Step 1: Capture user data

[0659] 1. Specific actions

[0660] The user faces the device's camera, which captures the face image and simultaneously records the voice through the microphone.

[0661] The user enters text using a keyboard or a touch screen.

[0662] 2. Input and Output

[0663] Input: User's face image, voice data, text input.

[0664] Output: The captured face image, voice data, and text data are sent from the device to the server.

[0665] Step 2: Preprocessing the face image

[0666] 1. Specific actions

[0667] The server receives the facial image sent from the terminal.

[0668] The pre-processing engine normalizes the image and extracts facial feature points (eyes, mouth, nose, etc.).

[0669] 2. Input and Output

[0670] Input: A face image sent from the device.

[0671] Output: Normalized and feature-point extracted face image data.

[0672] Step 3: Sentiment Analysis

[0673] 1. Specific actions

[0674] An emotion recognition engine analyzes the feature points of the pre-processed facial image data.

[0675] Algorithms are applied to predict emotions such as "happiness," "sadness," and "anger."

[0676] 2. Input and Output

[0677] Input: Feature points from preprocessed face image data.

[0678] Output: Predicted emotion data (e.g., "Happy: 70%", "Sad: 20%", "Anger: 10%").

[0679] Step 4: Integrating text data

[0680] 1. Specific actions

[0681] The server receives the text data sent by the user.

[0682] Integrate emotion data and text data.

[0683] 2. Input and Output

[0684] Input: Text data sent by the user, predicted emotion data.

[0685] Output: Integrated text and sentiment data.

[0686] Step 5: Process the audio data

[0687] 1. Specific actions

[0688] The server receives the voice data and converts the speech to text using an automatic speech recognition (ASR) engine.

[0689] The converted text data and emotion data are integrated.

[0690] 2. Input and Output

[0691] Input: Audio data sent by the user.

[0692] Output: Translated speech data, integrated text data and emotion data.

[0693] Step 6: Response Generation

[0694] 1. Specific actions

[0695] The server inputs the integrated text data and sentiment data into a natural language processing (NLP) model.

[0696] The NLP model generates the appropriate response.

[0697] 2. Input and Output

[0698] Input: Integrated text and sentiment data.

[0699] Output: The generated response (e.g., "Good work! Do you know how to relax?").

[0700] Step 7: Providing a response

[0701] 1. Specific actions

[0702] The server sends the generated response to the terminal.

[0703] The terminal displays the received response to the user.

[0704] 2. Input and Output

[0705] Input: The generated response.

[0706] Output: The response that is displayed to the user.

[0707] Through the above processing steps, various types of user data can be analyzed from multiple angles, and a personalized response can be generated and provided based on the results.

[0708] (Application example 2)

[0709] 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."

[0710] Detecting suspicious individuals and early detection of potential threats is a challenge for modern security systems. It is particularly difficult to quickly and accurately analyze the emotions of individuals in a crowd and provide appropriate responses. Therefore, there is a need for a system that allows security staff to analyze emotions in real time on-site and issue appropriate alerts and responses based on the results.

[0711] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring a facial image from a user's terminal, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for acquiring voice from the user's terminal, means for converting the acquired voice into text, means for generating a response based on the user's emotion and text input, and means for providing the generated response to the user's terminal. This enables security staff to analyze individual emotions in real time on site and detect and respond to potential threats in advance.

[0712] A "user terminal" is a device equipped with input devices such as a camera and microphone, which captures facial images and voices from the user and transmits them to a server.

[0713] The "means for acquiring a facial image" refers to a function for capturing a facial image of a user using a camera mounted on the user's terminal.

[0714] The "means for preprocessing a facial image" refers to a function for performing preprocessing such as image normalization and feature point detection on an acquired facial image.

[0715] The "means for analyzing a facial image and predicting a user's emotions" refers to a function that analyzes facial feature points based on a preprocessed facial image and predicts a user's emotions.

[0716] The "means for acquiring voice" refers to a function for capturing the user's voice using a microphone installed in the user's terminal.

[0717] The "means for converting voice to text" refers to a function for converting acquired voice data into text data.

[0718] The "means for generating a response" refers to a function that uses a natural language processing model to generate an appropriate response based on the user's sentiment and text input.

[0719] The "means for providing the generated response to the user" refers to a function for displaying the generated response on the user's terminal or reproducing it as audio.

[0720] A "generative AI model" is an artificial intelligence model that generates natural-sounding language responses based on input data.

[0721] This invention is a system that uses multimodal data (facial images, voice) acquired from a user's device to analyze emotions in real time and generate and provide responses based on that data. The purpose of this system is to analyze a user's emotions from multiple angles and realize more accurate and appropriate responses.

[0722] Hardware and software used

[0723] 1. Hardware used:

[0724] Smart glasses: Equipped with a camera and microphone to capture the user's facial image and voice.

[0725] Server: Preprocesses data, analyzes it, and generates responses.

[0726] 2. Software used:

[0727] OpenCV (cv2): Image capture and preprocessing.

[0728] dlib: Face detection and feature extraction.

[0729] speech_recognition: Speech capture and text conversion.

[0730] tensorflow: Emotion recognition model.

[0731] transformers(Hugging Face): Natural language processing model.

[0732] Process Overview

[0733] Facial image capture and analysis

[0734] The user wears the smart glasses and captures facial images in real time through the camera. The images are sent to the server and preprocessed using OpenCV. This includes image normalization and facial feature point extraction using dlib. Then, emotions are analyzed from the images using a TensorFlow emotion recognition model.

[0735] Audio capture and conversion

[0736] Similarly, the user's voice is captured through a microphone built into the smart glasses, which is then sent to the server and converted to text using speech_recognition.

[0737] Generating and serving the response

[0738] The server generates an appropriate response using a generative AI model based on the analyzed emotion results and the text converted from speech, and this response is fed back to the user through the smart glasses.

[0739] Specific examples

[0740] For example, if security staff wear smart glasses at the entrance to an event venue, they can capture the faces and voices of attendees and analyze their emotions. If anger or stress is detected, the system can generate an alert that the attendee is likely overly excited and notify security staff.

[0741] Prompt Sentence Examples

[0742] "This user's emotion is anger. Then respond appropriately to the following text: 'I'm frustrated with the traffic today.'"

[0743] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0744] Step 1:

[0745] The user's device captures a facial image using the camera in the smart glasses. Specifically, the camera acquires an image and stores it as digital data. The input is the facial image captured by the camera. This data is sent to the server.

[0746] Step 2:

[0747] The server preprocesses the received facial image data. Specifically, it normalizes the image using OpenCV and then extracts facial feature points using dlib. The input is the captured facial image, and the output is the preprocessed image data and coordinate information of the feature points.

[0748] Step 3:

[0749] The server inputs the preprocessed facial image data into an emotion recognition model to predict the user's emotion. Specifically, the data is input into the TensorFlow emotion recognition model and an estimated emotion label is obtained. The input is the coordinate information of the feature points, and the output is the emotion label (e.g., "joy," "sadness," or "anger").

[0750] Step 4:

[0751] The user's device captures voice data using the microphone in the smart glasses. Specifically, the microphone records the voice and converts it into digital data. The input is the user's voice, and the output is voice data. This voice data is then sent to the server.

[0752] Step 5:

[0753] The server converts the received voice data into text data. Specifically, it uses the speech_recognition library to perform voice recognition and converts it into text. The input is voice data, and the output is text data.

[0754] Step 6:

[0755] The server generates a response based on the emotion analysis results and the text data converted from the speech. Specifically, it uses Hugging Face's transformers library to input a prompt into a generative AI model and generate an appropriate response. The input is emotion labels and text data, and the output is a response in natural language.

[0756] Step 7:

[0757] The server then sends the generated response to the user's device. Specifically, the server sends the generated response to the user's smart glasses via a network. The input is a natural language response, and the output is displayed or played on the smart glasses' display or audio output.

[0758] Step 8:

[0759] The user's device provides the generated response to the user using the smart glasses' display or audio output. Specifically, the response is displayed as text on the display or played as audio output. The input is the response received from the server, and the output is a display or audio notification to the user.

[0760] 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.

[0761] 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.

[0762] 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.

[0763] [Third embodiment]

[0764] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0765] 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.

[0766] 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).

[0767] 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.

[0768] 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.

[0769] 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).

[0770] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for 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.

[0771] 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.

[0772] 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.

[0773] 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.

[0774] 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.

[0775] 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."

[0776] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. Below, each element of the system and the processing flow of the program are explained in natural language.

[0777] The system begins by acquiring a facial image from the user's device. The device captures the user's facial image with a camera and sends it to the server. The server preprocesses the received facial image and inputs it into an emotion recognition model. The model analyzes the facial image and predicts the user's emotion (e.g., joy, sadness, anger, etc.).

[0778] Next, the user inputs text. Using a chat window on the device, the user enters a question or request and sends it to the server. The server receives the text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model. The natural language processing model generates a response to the user based on this information.

[0779] The generated response is then sent back to the device and displayed to the user. This allows the user to receive a personalized response that matches their emotions at that time. For example, if the user enters the text "I'm tired today," the server will recognize the emotion "tired" based on image analysis, and generate a response such as "Thank you for your hard work. Do you know how to relax?" based on that emotion and the text, and send it to the device.

[0780] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[0781] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[0782] The processing flow will be explained below.

[0783] Step 1:

[0784] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[0785] Step 2:

[0786] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image normalization and feature point extraction.

[0787] Step 3:

[0788] The server inputs the preprocessed facial image into an emotion recognition model. The emotion recognition model is composed of trained neural networks and other components, and predicts the user's emotions from the facial image. The predicted emotions are expressed with labels such as "happiness," "sadness," and "anger."

[0789] Step 4:

[0790] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[0791] Step 5:

[0792] The server receives text input from the user, then combines the received text with the previously predicted sentiment information to create a request that is input into a natural language processing model.

[0793] Step 6:

[0794] The server loads a natural language processing model and inputs the text and sentiment information together. The natural language processing model generates an optimal response based on the user's message and sentiment. For example, this response might be, "You're tired. Do you know how to relax?"

[0795] Step 7:

[0796] The server sends the generated response to the user's terminal.

[0797] Step 8:

[0798] The user can then check the response displayed on the device, which allows the user to receive a personalized response tailored to their emotions.

[0799] Step 9:

[0800] When the user performs voice input, the user uses the microphone of the terminal to record a voice message and transmits it to the server.

[0801] Step 10:

[0802] The server receives the voice data and loads an automatic speech recognition (ASR) model, which converts the voice data into text.

[0803] Step 11:

[0804] The server then uses the converted text and emotional information to re-feed data into the natural language processing model to generate an appropriate response.

[0805] Step 12:

[0806] The server sends the generated response to the user's terminal, where the user confirms it.

[0807] Through this series of steps, InsightBot Advanced uses information from multimodal inputs to understand sentiment and deliver personalized interactions.

[0808] Example 1

[0809] 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."

[0810] Conventional chat systems and automated response systems do not take into account the user's emotions, making it difficult to provide personalized responses, and many users end up with dissatisfied results. Furthermore, because they cannot handle multiple input modalities, such as voice and images in addition to text, they lack the ability to respond in a variety of usage scenarios.

[0811] 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.

[0812] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for using a natural language processing model to generate a response based on the user's emotion and the text input, and means for providing the generated response to the user, thereby enabling a personalized response based on the user's emotion to be provided in real time.

[0813] "User" refers to a person who performs an operation using a terminal.

[0814] "Facial image" refers to digital image data that captures an image of a user's face.

[0815] "Terminal" refers to an electronic device used by a user, such as a computer, smartphone, or tablet.

[0816] "Server" refers to a central system that receives and processes data sent by users.

[0817] "Preprocessing" refers to the initial processing of acquired data to convert it into a form that is easier to analyze.

[0818] An "emotion recognition model" refers to a machine learning model for analyzing a user's emotions from facial images, etc.

[0819] "Text input" refers to character string information that a user inputs through a terminal.

[0820] A "natural language processing model" refers to an algorithm or machine learning model that understands and generates natural language.

[0821] "Response" refers to an answer or message generated based on the user's input and predicted emotions.

[0822] "Voice input" refers to a method in which a user inputs instructions or questions into a terminal by voice.

[0823] "Speech recognition" refers to the technology of converting voice data into text data.

[0824] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. The following describes in detail each element of the system and the processing flow of the program.

[0825] This system begins by acquiring a facial image from the user's device. The user captures their own facial image using the device's camera. The device then sends the captured facial image to the server. The device can be a smartphone, tablet, or PC.

[0826] The server preprocesses the received facial images. This includes image resizing, noise reduction, and face detection using the OpenCV library. The preprocessed facial images are then input into an emotion recognition model built with TensorFlow. The model analyzes the facial images and predicts the user's emotions (e.g., joy, sadness, anger, etc.).

[0827] The user then enters text using a chat window on the terminal. For example, the user might enter "I'm tired today." The terminal transmits this text entry to the server.

[0828] The server receives the user's text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model, such as GPT-3. The following prompt sentence is generated and input into the model:

[0829] "Generate an appropriate response based on the user's sentiment and the text below:

[0830] Input text: I'm tired today

[0831] User Emotions: Tired

[0832] The natural language processing model uses this information to generate a response to the user, such as, "Thank you for your hard work. Do you know how to relax?"

[0833] The generated response is then sent back to the device and displayed to the user, allowing the user to receive a personalized response tailored to their emotions at that moment.

[0834] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[0835] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[0836] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0837] Step 1:

[0838] The user captures a facial image using the device's camera.

[0839] Specifically, the user activates the device's camera and takes a picture of their face. The input is the user's face image, which becomes the data used in the next step.

[0840] Step 2:

[0841] The device transmits the captured facial image to the server.

[0842] The device sends the captured facial image to the server via the Internet. Specifically, the image data is sent to the server via an HTTP POST request. The input is the facial image captured in step 1, and the output is the image data sent to the server.

[0843] Step 3:

[0844] The server preprocesses the received facial images.

[0845] The server preprocesses the received face image by using OpenCV to resize the image, remove noise, and detect faces. The input is the face image sent in step 2, and the output is the preprocessed face image.

[0846] Step 4:

[0847] The server inputs the preprocessed facial image into an emotion recognition model to predict the user's emotions.

[0848] The server inputs the preprocessed facial image into the emotion recognition model. Specifically, it uses TensorFlow to input the facial image into the emotion recognition model and predicts the user's emotion. The input is the preprocessed facial image, and the output is the predicted user emotion.

[0849] Step 5:

[0850] The user enters text in a chat window on the device.

[0851] In concrete terms, a user uses a chat window on a terminal to input text such as "I'm tired today." The input is the text input by the user, and the output is the text displayed on the terminal.

[0852] Step 6:

[0853] The terminal sends the user's text input to the server.

[0854] The terminal sends the user's text input to the server. Specifically, it sends the text data to the server via an HTTP POST request. The input is the text entered in step 5, and the output is the text data sent to the server.

[0855] Step 7:

[0856] The server inputs the user's text input and emotional information into a natural language processing model.

[0857] The server combines the user's text input with emotion information and inputs it into the natural language processing model. Specifically, it generates the following prompt sentence and inputs it into the natural language processing model:

[0858] "Generate an appropriate response based on the user's sentiment and the text below:

[0859] Input text: I'm tired today

[0860] User Emotions: Tired

[0861] The input is the user's text and emotional information, and the output is a prompt sent to a natural language processing model.

[0862] Step 8:

[0863] A natural language processing model generates an appropriate response to the user.

[0864] The natural language processing model generates a response based on the prompt. Specifically, the generative AI model generates a response such as, "Thank you for your hard work. Do you know how to relax?" The input is the prompt, and the output is the generated response.

[0865] Step 9:

[0866] The server sends the generated response to the terminal.

[0867] The server sends the generated response to the terminal. Specifically, it sends the response data to the terminal using an HTTP POST request. The input is the generated response, and the output is the response data sent to the terminal.

[0868] Step 10:

[0869] The terminal displays the response received from the server to the user.

[0870] The terminal displays the response received from the server in a chat window. Specifically, it displays the message "Thank you for your hard work. Do you know how to relax?" on the terminal screen. The input is the response data from the server, and the output is the display on the terminal.

[0871] (Application example 1)

[0872] 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."

[0873] Conventional customer service support systems lack the technology to analyze users' emotions in real time and provide appropriate responses based on those emotions, which limits the means for improving customer satisfaction. Furthermore, systems that assume only text input cannot realize interaction through voice input or image recognition, making it difficult to communicate naturally with users.

[0874] 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.

[0875] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for generating a response based on the user's emotion and the text input, means for providing the generated response to the user, means for capturing a facial image of the user with a camera mounted on the smart glasses, means for displaying personalized information on the smart glasses based on the emotion, and means for recognizing the user's voice input and generating a response based on the text, thereby realizing appropriate and natural dialogue that takes the user's emotion into consideration and improving customer satisfaction in physical stores.

[0876] "User" refers to a person who uses the system.

[0877] "Facial image" refers to image data of a user's face captured by a camera.

[0878] "Preprocessing" refers to image processing to make the acquired facial image easier to analyze.

[0879] "Emotion prediction" refers to the process of inferring a user's emotional state based on preprocessed facial images.

[0880] "Text input" refers to the act of a user inputting textual information into a system and the results of that action.

[0881] "Response generation" refers to the process of creating an appropriate response based on the user's sentiment and text input.

[0882] "Smart glasses" refers to a wearable device that incorporates a camera, display, etc. and can display information within the user's field of vision.

[0883] "Voice input" refers to the act of a user inputting voice information into a system and the results thereof.

[0884] A "natural language processing model" refers to an algorithm or model for understanding and generating human language.

[0885] "Personalized information" refers to information that is individually optimized according to the user's emotions and behavior.

[0886] "Capture" refers to the act of obtaining an image with a camera.

[0887] "Recognition" refers to analyzing input data and understanding its content and characteristics.

[0888] A system for implementing the present invention analyzes a user's emotions in real time and provides appropriate responses based on the emotions and the user's input. The system interacts with the user using smart glasses.

[0889] Hardware Configuration

[0890] The system uses the following hardware:

[0891] 1. Smart glasses: Equipped with a camera, display, and microphone.

[0892] 2. Server: A computer with powerful computing power and storage.

[0893] Software Configuration

[0894] The system uses the following software:

[0895] 1. Emotion recognition model: Model trained using Keras.

[0896] 2. Natural Language Processing model: A generative AI model (GPT-3) that uses the Transformers library.

[0897] 3. Speech Recognition Tools: Speech recognition software to convert the user's speech into text.

[0898] 4. gTTS and PyAudio: Software for converting response text into speech and playing it back.

[0899] Data processing and calculation

[0900] 1. Capture a user's face image:

[0901] The camera on the smart glasses captures the user's face.

[0902] The captured image data is transmitted to a server.

[0903] 2. Emotional Prediction:

[0904] The server preprocesses the received facial images, performs grayscale conversion and resizing.

[0905] The preprocessed image data is input into an emotion recognition model to predict the user's emotional state.

[0906] 3. Parsing user text input:

[0907] The smart glasses recognize voice input from the user and convert the voice data into text.

[0908] The converted text data and predicted emotion data are sent to the server.

[0909] 4. Response Generation:

[0910] The server uses a natural language processing model to generate an appropriate response based on the user's emotions and text input.

[0911] The generated response text is sent to the smart glasses and is displayed on the display and output as voice.

[0912] Specific examples

[0913] In a physical store, a customer is wearing smart glasses and asks, "Write a review about this product." The system then:

[0914] The camera in the smart glasses captures the customer's face and sends it to the server.

[0915] Predicting the emotion of "interested" using an emotion recognition model.

[0916] The questions are converted into text using voice recognition.

[0917] The following prompt sentences were generated using a natural language processing model (GPT-3) and responses were created:

[0918] text

[0919] User sentiment: Interested

[0920] Q: What are the reviews for this product?

[0921] Generates the response text "This product has high ratings and is well-received by many users."

[0922] Displayed on smart glasses and played back as audio at the same time.

[0923] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0924] Step 1:

[0925] The server controls the camera mounted on the smart glasses to capture the user's facial image. The server receives the camera image data as input and sends it to the server. The server generates the captured facial image data as output.

[0926] Step 2:

[0927] The server preprocesses the received facial image data by converting the image to grayscale and resizing it to the size required by the emotion recognition model. The server receives the facial image data as input and generates preprocessed image data as output.

[0928] Step 3:

[0929] The server inputs the preprocessed facial images into an emotion recognition model to predict the user's emotional state. This emotion recognition model is trained using Keras. It receives the preprocessed image data as input and generates the user's emotional data as output.

[0930] Step 4:

[0931] The user performs voice input through the smart glasses. The microphone in the smart glasses captures the voice data and transmits it to the terminal. The voice data is received as input and generated as output, which is transmitted to the terminal.

[0932] Step 5:

[0933] The server uses speech recognition software to convert the voice data into text. Specifically, it analyzes the voice data and generates text-based question data. The server receives the voice data as input and generates text data as output.

[0934] Step 6:

[0935] The server inputs the user's emotion data and text data into a generative AI model (GPT-3) to generate an appropriate response. The prompt uses a format such as "User emotion: Interested, Question: What are the reviews for this product?". Emotion data and text data are received as input, and response text data is generated as output.

[0936] Step 7:

[0937] The server sends the response text data to the smart glasses and displays it on the display. It also converts the response text into speech using gTTS and plays the speech to the user using PyAudio. The server receives the response text data as input and generates the text to be displayed on the display and the speech to be played as output.

[0938] 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.

[0939] The system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions. The following describes the system components and the program processing flow in natural language.

[0940] The system consists of several main components. First, the user's device plays an important role. This device is equipped with a camera, microphone, and input interface to capture the user's facial image, voice, and text input. The device then transmits this information to the server.

[0941] The server then performs several processing steps. First, the server receives the facial image sent by the user and performs preprocessing. This includes image normalization and feature point extraction. The preprocessed facial image is then passed to an emotion recognition engine. This engine analyzes the facial feature points and predicts the user's emotion. Examples of emotions include "happiness," "sadness," and "anger."

[0942] Next, the user inputs text. Using the input interface on the device, the user inputs a message such as "I'm tired today" and sends it to the server. The server receives this text input and inputs it into a natural language processing model along with the previously predicted emotional information. The natural language processing model generates a response based on the emotional information and the text input. For example, this response could be a message such as "Thank you for your hard work. Do you know how to relax?"

[0943] A more refined approach involves voice input, where the user records their voice using the device's microphone and sends it to the server. The server receives the voice data and uses an automatic speech recognition (ASR) engine to convert the speech to text. The text is then combined with emotion recognition information and similarly fed into a natural language processing model to generate an appropriate response.

[0944] A unique feature of this system is that it also includes the ability to recognize a user's emotions using biometric information other than facial images (e.g., voice and gestures). For example, by analyzing voice tone and gesture movements, it is possible to more accurately predict emotions based on this data. Furthermore, by referencing past conversation history, it is possible to continuously track emotions and provide a deeper understanding.

[0945] For example, if a user inputs "I'm stressed today," the server will predict emotions such as "stress" or "anger" from the facial recognition results. Based on this, the system will generate a response such as "Do you have any recommendations for relaxation methods to reduce stress?" and provide it to the user.

[0946] As described above, by combining an emotion engine, the present invention makes it possible to analyze user emotions from multiple angles and provide more personalized interactions, thereby achieving significant improvements in customer experience and satisfaction.

[0947] The processing flow will be explained below.

[0948] A system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions.

[0949] Step 1:

[0950] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[0951] Step 2:

[0952] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image resizing, grayscale conversion, and facial feature point extraction.

[0953] Step 3:

[0954] The server passes the preprocessed facial image to an emotion recognition engine, which analyzes facial feature points and predicts the user's emotions. For example, it identifies emotions such as "happiness," "sadness," and "anger."

[0955] Step 4:

[0956] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[0957] Step 5:

[0958] The server processes the text input received from the user, integrates the received text input with the predicted emotional information, and creates a context to be input into the natural language processing model.

[0959] Step 6:

[0960] The server loads a natural language processing model and inputs text input along with emotional information. The natural language processing model generates an optimal response based on the user's input and emotional state. For example, it creates a response like, "You're tired. Do you know how to relax?"

[0961] Step 7:

[0962] The server sends the generated response to the user's terminal.

[0963] Step 8:

[0964] The user checks the response displayed on the device, allowing the user to receive a personalized response that matches their emotions.

[0965] Step 9:

[0966] When the user provides voice input, the terminal's microphone is used to record a voice message and send it to the server.

[0967] Step 10:

[0968] The server receives the voice data and loads an automatic speech recognition (ASR) engine, which converts the voice data into text.

[0969] Step 11:

[0970] The server then inputs the converted text and emotional information into a natural language processing model to generate an appropriate response, such as "What are your plans for today?"

[0971] Step 12:

[0972] The server sends the generated response to the user's terminal, and the user checks the response on the terminal.

[0973] Through this series of steps, the system can understand user emotions using multimodal input data and provide personalized interactions. Furthermore, by using an emotion engine, it can also utilize biometric information such as voice and gestures to more accurately recognize user emotions. This significantly improves customer experience and satisfaction.

[0974] Example 2

[0975] 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."

[0976] Conventional emotion analysis systems often have limited information for accurately predicting user emotions, using only facial images and text data, limiting their analytical accuracy. Furthermore, even systems that analyze emotions using user voice data have difficulty integrating the analysis of voice data with other data. Given these circumstances, the challenge is to analyze user emotions from multiple angles and generate more accurate and personalized responses.

[0977] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring image data from a user, means for preprocessing the acquired image data, means for analyzing the preprocessed image data and predicting the user's emotions, means for receiving text data from the user, means for generating a response based on the user's emotions and the text data, means for providing the generated response to the user, means for acquiring voice data and converting it into text data, and means for integrating and analyzing the emotion data and the voice data. This enables an integrated analysis of multiple data sources for a user, enabling more accurate and personalized responses.

[0978] A "user" is an entity that uses the system and provides data.

[0979] "Image data" is visual information captured using a camera or other imaging device.

[0980] "Preprocessing" refers to the initial processing operations performed to convert raw data into a format that is easier to analyze and process.

[0981] A "means for predicting emotions" is an algorithm or engine for inferring a user's emotional state based on acquired data.

[0982] "Text data" refers to information entered by a user in the form of text, typically using a keyboard or touch screen.

[0983] A "natural language processing model" is an algorithm or machine learning model for analyzing text data and understanding its meaning.

[0984] "Audio data" refers to audio information captured by a device such as a microphone, which needs to be converted into a text format.

[0985] An "automatic speech recognition (ASR) engine" is a technology for analyzing voice data and converting it into text data.

[0986] "Emotion data" refers to the emotional state predicted by the emotion recognition means, typically expressed numerically or categorically.

[0987] The "means for generating a response" is an algorithm or model for creating an appropriate response to provide to a user based on the emotional data and text data.

[0988] The "means for providing to the user" refers to an interface or device for transmitting the generated response to the user.

[0989] MODE FOR CARRYING OUT THE INVENTION

[0990] The system embodying the present invention analyzes various data of a user and generates a response based on the user's emotions. This system includes a user terminal, a server, and various software engines. The components and operation of the system are described in detail below.

[0991] User terminal

[0992] The user terminal includes the following hardware:

[0993] Camera: Used to capture the user's facial image.

[0994] Microphone: Used to record the user's voice input.

[0995] Input Interface: This includes keyboards and touchscreens.

[0996] The user terminal serves to transmit facial images, voice, and text input to the server in real time.

[0997] server

[0998] The server processes and analyzes the received data in cooperation with the following software engines:

[0999] Pre-processing engine: Normalizes the received face image and extracts facial feature points.

[1000] Emotion recognition engine: Analyzes preprocessed facial feature points to predict the user's emotions. For example, this engine can detect emotions such as "happiness," "sadness," and "anger."

[1001] Automatic Speech Recognition (ASR) engine: Converts user voice data into text.

[1002] Natural language processing (NLP) models: Generate appropriate responses based on text and sentiment data.

[1003] Data Handling

[1004] 1. Facial image capture and processing

[1005] The user faces the camera and the user terminal captures an image of the user's face.

[1006] The server receives the facial image, and the pre-processing engine normalizes the image and extracts feature points, such as identifying the positions of the eyes, mouth, and nose.

[1007] 2. Emotion analysis

[1008] The server's emotion recognition engine predicts the user's emotions based on the extracted feature points. For example, it analyzes the "muscle movements around the eyes" and the "angle of the corners of the mouth" to determine "happiness" or "sadness."

[1009] The results are stored as "Happiness: 80%", "Sadness: 15%", "Anger: 5%".

[1010] 3. Handling Text Input

[1011] The user uses the keyboard or touch screen on the device to input a message such as "I'm tired today" and transmits it to the server.

[1012] The server receives this text, combines it with emotional data, and inputs it into the NLP model.

[1013] 4. Processing voice input

[1014] The user uses a microphone to input speech, for example, "I'm feeling stressed today."

[1015] The server's ASR engine converts the voice data into text, then combines the converted text with emotional data and inputs it into the NLP model.

[1016] 5. Generating and Serving the Response

[1017] The NLP model uses emotional and textual data to generate a personalized response for the user, such as "Thank you for your hard work. Do you know how to relax?"

[1018] The generated response is sent from the server to the terminal and displayed to the user through the terminal's interface.

[1019] Examples and prompts

[1020] For example, if a user types "I'm feeling stressed today":

[1021] The user's device takes a facial image and sends it to the server.

[1022] The server preprocesses the images and uses an emotion recognition engine to predict emotions such as "stress" or "anger."

[1023] The user types the text "I'm stressed today" and sends it to the server.

[1024] The server inputs this information into an NLP model, generates a response such as "What relaxation techniques would you recommend to reduce stress?" and displays it on the device.

[1025] This system enables multifaceted analysis of user sentiment, enabling more accurate and personalized responses, resulting in a significant improvement in customer experience and satisfaction.

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

[1027] Processing steps of this system's program

[1028] Step 1: Capture user data

[1029] 1. Specific actions

[1030] The user faces the device's camera, which captures the face image and simultaneously records the voice through the microphone.

[1031] The user enters text using a keyboard or a touch screen.

[1032] 2. Input and Output

[1033] Input: User's face image, voice data, text input.

[1034] Output: The captured face image, voice data, and text data are sent from the device to the server.

[1035] Step 2: Preprocessing the face image

[1036] 1. Specific actions

[1037] The server receives the facial image sent from the terminal.

[1038] The pre-processing engine normalizes the image and extracts facial feature points (eyes, mouth, nose, etc.).

[1039] 2. Input and Output

[1040] Input: A face image sent from the device.

[1041] Output: Normalized and feature-point extracted face image data.

[1042] Step 3: Sentiment Analysis

[1043] 1. Specific actions

[1044] An emotion recognition engine analyzes the feature points of the pre-processed facial image data.

[1045] Algorithms are applied to predict emotions such as "happiness," "sadness," and "anger."

[1046] 2. Input and Output

[1047] Input: Feature points from preprocessed face image data.

[1048] Output: Predicted emotion data (e.g., "Happy: 70%", "Sad: 20%", "Anger: 10%").

[1049] Step 4: Integrating text data

[1050] 1. Specific actions

[1051] The server receives the text data sent by the user.

[1052] Integrate emotion data and text data.

[1053] 2. Input and Output

[1054] Input: Text data sent by the user, predicted emotion data.

[1055] Output: Integrated text and sentiment data.

[1056] Step 5: Process the audio data

[1057] 1. Specific actions

[1058] The server receives the voice data and converts the speech to text using an automatic speech recognition (ASR) engine.

[1059] The converted text data and emotion data are integrated.

[1060] 2. Input and Output

[1061] Input: Audio data sent by the user.

[1062] Output: Translated speech data, integrated text data and emotion data.

[1063] Step 6: Response Generation

[1064] 1. Specific actions

[1065] The server inputs the integrated text data and sentiment data into a natural language processing (NLP) model.

[1066] The NLP model generates the appropriate response.

[1067] 2. Input and Output

[1068] Input: Integrated text and sentiment data.

[1069] Output: The generated response (e.g., "Good work! Do you know how to relax?").

[1070] Step 7: Providing a response

[1071] 1. Specific actions

[1072] The server sends the generated response to the terminal.

[1073] The terminal displays the received response to the user.

[1074] 2. Input and Output

[1075] Input: The generated response.

[1076] Output: The response that is displayed to the user.

[1077] Through the above processing steps, various types of user data can be analyzed from multiple angles, and a personalized response can be generated and provided based on the results.

[1078] (Application example 2)

[1079] 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."

[1080] Detecting suspicious individuals and early detection of potential threats is a challenge for modern security systems. It is particularly difficult to quickly and accurately analyze the emotions of individuals in a crowd and provide appropriate responses. Therefore, there is a need for a system that allows security staff to analyze emotions in real time on-site and issue appropriate alerts and responses based on the results.

[1081] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring a facial image from a user's terminal, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for acquiring voice from the user's terminal, means for converting the acquired voice into text, means for generating a response based on the user's emotion and text input, and means for providing the generated response to the user's terminal. This enables security staff to analyze individual emotions in real time on site and detect and respond to potential threats in advance.

[1082] A "user terminal" is a device equipped with input devices such as a camera and microphone, which captures facial images and voices from the user and transmits them to a server.

[1083] The "means for acquiring a facial image" refers to a function for capturing a facial image of a user using a camera mounted on the user's terminal.

[1084] The "means for preprocessing a facial image" refers to a function for performing preprocessing such as image normalization and feature point detection on an acquired facial image.

[1085] The "means for analyzing a facial image and predicting a user's emotions" refers to a function that analyzes facial feature points based on a preprocessed facial image and predicts a user's emotions.

[1086] The "means for acquiring voice" refers to a function for capturing the user's voice using a microphone installed in the user's terminal.

[1087] The "means for converting voice to text" refers to a function for converting acquired voice data into text data.

[1088] The "means for generating a response" refers to a function that uses a natural language processing model to generate an appropriate response based on the user's sentiment and text input.

[1089] The "means for providing the generated response to the user" refers to a function for displaying the generated response on the user's terminal or reproducing it as audio.

[1090] A "generative AI model" is an artificial intelligence model that generates natural-sounding language responses based on input data.

[1091] This invention is a system that uses multimodal data (facial images, voice) acquired from a user's device to analyze emotions in real time and generate and provide responses based on that data. The purpose of this system is to analyze a user's emotions from multiple angles and realize more accurate and appropriate responses.

[1092] Hardware and software used

[1093] 1. Hardware used:

[1094] Smart glasses: Equipped with a camera and microphone to capture the user's facial image and voice.

[1095] Server: Preprocesses data, analyzes it, and generates responses.

[1096] 2. Software used:

[1097] OpenCV (cv2): Image capture and preprocessing.

[1098] dlib: Face detection and feature extraction.

[1099] speech_recognition: Speech capture and text conversion.

[1100] tensorflow: Emotion recognition model.

[1101] transformers(Hugging Face): Natural language processing model.

[1102] Process Overview

[1103] Facial image capture and analysis

[1104] The user wears the smart glasses and captures facial images in real time through the camera. The images are sent to the server and preprocessed using OpenCV. This includes image normalization and facial feature point extraction using dlib. Then, emotions are analyzed from the images using a TensorFlow emotion recognition model.

[1105] Audio capture and conversion

[1106] Similarly, the user's voice is captured through a microphone built into the smart glasses, which is then sent to the server and converted to text using speech_recognition.

[1107] Generating and serving the response

[1108] The server generates an appropriate response using a generative AI model based on the analyzed emotion results and the text converted from speech, and this response is fed back to the user through the smart glasses.

[1109] Specific examples

[1110] For example, if security staff wear smart glasses at the entrance to an event venue, they can capture the faces and voices of attendees and analyze their emotions. If anger or stress is detected, the system can generate an alert that the attendee is likely overly excited and notify security staff.

[1111] Prompt Sentence Examples

[1112] "This user's emotion is anger. Then respond appropriately to the following text: 'I'm frustrated with the traffic today.'"

[1113] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1114] Step 1:

[1115] The user's device captures a facial image using the camera in the smart glasses. Specifically, the camera acquires an image and stores it as digital data. The input is the facial image captured by the camera. This data is sent to the server.

[1116] Step 2:

[1117] The server preprocesses the received facial image data. Specifically, it normalizes the image using OpenCV and then extracts facial feature points using dlib. The input is the captured facial image, and the output is the preprocessed image data and coordinate information of the feature points.

[1118] Step 3:

[1119] The server inputs the preprocessed facial image data into an emotion recognition model to predict the user's emotion. Specifically, the data is input into the TensorFlow emotion recognition model and an estimated emotion label is obtained. The input is the coordinate information of the feature points, and the output is the emotion label (e.g., "joy," "sadness," or "anger").

[1120] Step 4:

[1121] The user's device captures voice data using the microphone in the smart glasses. Specifically, the microphone records the voice and converts it into digital data. The input is the user's voice, and the output is voice data. This voice data is then sent to the server.

[1122] Step 5:

[1123] The server converts the received voice data into text data. Specifically, it uses the speech_recognition library to perform voice recognition and converts it into text. The input is voice data, and the output is text data.

[1124] Step 6:

[1125] The server generates a response based on the emotion analysis results and the text data converted from the speech. Specifically, it uses Hugging Face's transformers library to input a prompt into a generative AI model and generate an appropriate response. The input is emotion labels and text data, and the output is a response in natural language.

[1126] Step 7:

[1127] The server then sends the generated response to the user's device. Specifically, the server sends the generated response to the user's smart glasses via a network. The input is a natural language response, and the output is displayed or played on the smart glasses' display or audio output.

[1128] Step 8:

[1129] The user's device provides the generated response to the user using the smart glasses' display or audio output. Specifically, the response is displayed as text on the display or played as audio output. The input is the response received from the server, and the output is a display or audio notification to the user.

[1130] 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.

[1131] 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.

[1132] 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.

[1133] [Fourth embodiment]

[1134] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1135] 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.

[1136] 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).

[1137] 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.

[1138] 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.

[1139] 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).

[1140] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for 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.

[1141] 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.

[1142] 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.

[1143] 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.

[1144] 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.

[1145] 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.

[1146] 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."

[1147] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. Below, each element of the system and the processing flow of the program are explained in natural language.

[1148] The system begins by acquiring a facial image from the user's device. The device captures the user's facial image with a camera and sends it to the server. The server preprocesses the received facial image and inputs it into an emotion recognition model. The model analyzes the facial image and predicts the user's emotion (e.g., joy, sadness, anger, etc.).

[1149] Next, the user inputs text. Using a chat window on the device, the user enters a question or request and sends it to the server. The server receives the text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model. The natural language processing model generates a response to the user based on this information.

[1150] The generated response is then sent back to the device and displayed to the user. This allows the user to receive a personalized response that matches their emotions at that time. For example, if the user enters the text "I'm tired today," the server will recognize the emotion "tired" based on image analysis, and generate a response such as "Thank you for your hard work. Do you know how to relax?" based on that emotion and the text, and send it to the device.

[1151] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[1152] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[1153] The processing flow will be explained below.

[1154] Step 1:

[1155] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[1156] Step 2:

[1157] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image normalization and feature point extraction.

[1158] Step 3:

[1159] The server inputs the preprocessed facial image into an emotion recognition model. The emotion recognition model is composed of trained neural networks and other components, and predicts the user's emotions from the facial image. The predicted emotions are expressed with labels such as "happiness," "sadness," and "anger."

[1160] Step 4:

[1161] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[1162] Step 5:

[1163] The server receives text input from the user, then combines the received text with the previously predicted sentiment information to create a request that is input into a natural language processing model.

[1164] Step 6:

[1165] The server loads a natural language processing model and inputs the text and sentiment information together. The natural language processing model generates an optimal response based on the user's message and sentiment. For example, this response might be, "You're tired. Do you know how to relax?"

[1166] Step 7:

[1167] The server sends the generated response to the user's terminal.

[1168] Step 8:

[1169] The user can then check the response displayed on the device, which allows the user to receive a personalized response tailored to their emotions.

[1170] Step 9:

[1171] When the user performs voice input, the user uses the microphone of the terminal to record a voice message and transmits it to the server.

[1172] Step 10:

[1173] The server receives the voice data and loads an automatic speech recognition (ASR) model, which converts the voice data into text.

[1174] Step 11:

[1175] The server then uses the converted text and emotional information to re-feed data into the natural language processing model to generate an appropriate response.

[1176] Step 12:

[1177] The server sends the generated response to the user's terminal, where the user confirms it.

[1178] Through this series of steps, InsightBot Advanced uses information from multimodal inputs to understand sentiment and deliver personalized interactions.

[1179] Example 1

[1180] 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."

[1181] Conventional chat systems and automated response systems do not take into account the user's emotions, making it difficult to provide personalized responses, and many users end up with dissatisfied results. Furthermore, because they cannot handle multiple input modalities, such as voice and images in addition to text, they lack the ability to respond in a variety of usage scenarios.

[1182] 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.

[1183] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for using a natural language processing model to generate a response based on the user's emotion and the text input, and means for providing the generated response to the user, thereby enabling a personalized response based on the user's emotion to be provided in real time.

[1184] "User" refers to a person who performs an operation using a terminal.

[1185] "Facial image" refers to digital image data that captures an image of a user's face.

[1186] "Terminal" refers to an electronic device used by a user, such as a computer, smartphone, or tablet.

[1187] "Server" refers to a central system that receives and processes data sent by users.

[1188] "Preprocessing" refers to the initial processing of acquired data to convert it into a form that is easier to analyze.

[1189] An "emotion recognition model" refers to a machine learning model for analyzing a user's emotions from facial images, etc.

[1190] "Text input" refers to character string information that a user inputs through a terminal.

[1191] A "natural language processing model" refers to an algorithm or machine learning model that understands and generates natural language.

[1192] "Response" refers to an answer or message generated based on the user's input and predicted emotions.

[1193] "Voice input" refers to a method in which a user inputs instructions or questions into a terminal by voice.

[1194] "Speech recognition" refers to the technology of converting voice data into text data.

[1195] The system for implementing the present invention analyzes a user's emotions in real time and provides an appropriate response based on the emotions and input from the user. The following describes in detail each element of the system and the processing flow of the program.

[1196] This system begins by acquiring a facial image from the user's device. The user captures their own facial image using the device's camera. The device then sends the captured facial image to the server. The device can be a smartphone, tablet, or PC.

[1197] The server preprocesses the received facial images. This includes image resizing, noise reduction, and face detection using the OpenCV library. The preprocessed facial images are then input into an emotion recognition model built with TensorFlow. The model analyzes the facial images and predicts the user's emotions (e.g., joy, sadness, anger, etc.).

[1198] The user then enters text using a chat window on the terminal. For example, the user might enter "I'm tired today." The terminal transmits this text entry to the server.

[1199] The server receives the user's text input, combines it with the previously predicted emotional information, and inputs it into a natural language processing model, such as GPT-3. The following prompt sentence is generated and input into the model:

[1200] "Generate an appropriate response based on the user's sentiment and the text below:

[1201] Input text: I'm tired today

[1202] User Emotions: Tired

[1203] The natural language processing model uses this information to generate a response to the user, such as, "Thank you for your hard work. Do you know how to relax?"

[1204] The generated response is then sent back to the device and displayed to the user, allowing the user to receive a personalized response tailored to their emotions at that moment.

[1205] Additionally, when a user speaks, the device captures the voice data and sends it to the server, which then uses a speech recognition model to convert the speech into text and generate a response based on that text. This also takes emotional information into account, resulting in a more relevant and personalized interaction.

[1206] As described above, this invention enables natural and smooth communication that takes into account the user's emotions, thereby improving customer satisfaction. Furthermore, by supporting multimodal inputs such as text, voice, and images, the system can be used in a wider range of scenarios and functions effectively in a variety of environments.

[1207] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1208] Step 1:

[1209] The user captures a facial image using the device's camera.

[1210] Specifically, the user activates the device's camera and takes a picture of their face. The input is the user's face image, which becomes the data used in the next step.

[1211] Step 2:

[1212] The device transmits the captured facial image to the server.

[1213] The device sends the captured facial image to the server via the Internet. Specifically, the image data is sent to the server via an HTTP POST request. The input is the facial image captured in step 1, and the output is the image data sent to the server.

[1214] Step 3:

[1215] The server preprocesses the received facial images.

[1216] The server preprocesses the received face image by using OpenCV to resize the image, remove noise, and detect faces. The input is the face image sent in step 2, and the output is the preprocessed face image.

[1217] Step 4:

[1218] The server inputs the preprocessed facial image into an emotion recognition model to predict the user's emotions.

[1219] The server inputs the preprocessed facial image into the emotion recognition model. Specifically, it uses TensorFlow to input the facial image into the emotion recognition model and predicts the user's emotion. The input is the preprocessed facial image, and the output is the predicted user emotion.

[1220] Step 5:

[1221] The user enters text in a chat window on the device.

[1222] In concrete terms, a user uses a chat window on a terminal to input text such as "I'm tired today." The input is the text input by the user, and the output is the text displayed on the terminal.

[1223] Step 6:

[1224] The terminal sends the user's text input to the server.

[1225] The terminal sends the user's text input to the server. Specifically, it sends the text data to the server via an HTTP POST request. The input is the text entered in step 5, and the output is the text data sent to the server.

[1226] Step 7:

[1227] The server inputs the user's text input and emotional information into a natural language processing model.

[1228] The server combines the user's text input with emotion information and inputs it into the natural language processing model. Specifically, it generates the following prompt sentence and inputs it into the natural language processing model:

[1229] "Generate an appropriate response based on the user's sentiment and the text below:

[1230] Input text: I'm tired today

[1231] User Emotions: Tired

[1232] The input is the user's text and emotional information, and the output is a prompt sent to a natural language processing model.

[1233] Step 8:

[1234] A natural language processing model generates an appropriate response to the user.

[1235] The natural language processing model generates a response based on the prompt. Specifically, the generative AI model generates a response such as, "Thank you for your hard work. Do you know how to relax?" The input is the prompt, and the output is the generated response.

[1236] Step 9:

[1237] The server sends the generated response to the terminal.

[1238] The server sends the generated response to the terminal. Specifically, it sends the response data to the terminal using an HTTP POST request. The input is the generated response, and the output is the response data sent to the terminal.

[1239] Step 10:

[1240] The terminal displays the response received from the server to the user.

[1241] The terminal displays the response received from the server in a chat window. Specifically, it displays the message "Thank you for your hard work. Do you know how to relax?" on the terminal screen. The input is the response data from the server, and the output is the display on the terminal.

[1242] (Application example 1)

[1243] 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."

[1244] Conventional customer service support systems lack the technology to analyze users' emotions in real time and provide appropriate responses based on those emotions, which limits the means for improving customer satisfaction. Furthermore, systems that assume only text input cannot realize interaction through voice input or image recognition, making it difficult to communicate naturally with users.

[1245] 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.

[1246] In this invention, the server includes means for acquiring a facial image from a user, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for receiving a text input from the user, means for generating a response based on the user's emotion and the text input, means for providing the generated response to the user, means for capturing a facial image of the user with a camera mounted on the smart glasses, means for displaying personalized information on the smart glasses based on the emotion, and means for recognizing the user's voice input and generating a response based on the text, thereby realizing appropriate and natural dialogue that takes the user's emotion into consideration and improving customer satisfaction in physical stores.

[1247] "User" refers to a person who uses the system.

[1248] "Facial image" refers to image data of a user's face captured by a camera.

[1249] "Preprocessing" refers to image processing to make the acquired facial image easier to analyze.

[1250] "Emotion prediction" refers to the process of inferring a user's emotional state based on preprocessed facial images.

[1251] "Text input" refers to the act of a user inputting textual information into a system and the results of that action.

[1252] "Response generation" refers to the process of creating an appropriate response based on the user's sentiment and text input.

[1253] "Smart glasses" refers to a wearable device that incorporates a camera, display, etc. and can display information within the user's field of vision.

[1254] "Voice input" refers to the act of a user inputting voice information into a system and the results thereof.

[1255] A "natural language processing model" refers to an algorithm or model for understanding and generating human language.

[1256] "Personalized information" refers to information that is individually optimized according to the user's emotions and behavior.

[1257] "Capture" refers to the act of obtaining an image with a camera.

[1258] "Recognition" refers to analyzing input data and understanding its content and characteristics.

[1259] A system for implementing the present invention analyzes a user's emotions in real time and provides appropriate responses based on the emotions and the user's input. The system interacts with the user using smart glasses.

[1260] Hardware Configuration

[1261] The system uses the following hardware:

[1262] 1. Smart glasses: Equipped with a camera, display, and microphone.

[1263] 2. Server: A computer with powerful computing power and storage.

[1264] Software Configuration

[1265] The system uses the following software:

[1266] 1. Emotion recognition model: Model trained using Keras.

[1267] 2. Natural Language Processing model: A generative AI model (GPT-3) that uses the Transformers library.

[1268] 3. Speech Recognition Tools: Speech recognition software to convert the user's speech into text.

[1269] 4. gTTS and PyAudio: Software for converting response text into speech and playing it back.

[1270] Data processing and calculation

[1271] 1. Capture a user's face image:

[1272] The camera on the smart glasses captures the user's face.

[1273] The captured image data is transmitted to a server.

[1274] 2. Emotional Prediction:

[1275] The server preprocesses the received facial images, performs grayscale conversion and resizing.

[1276] The preprocessed image data is input into an emotion recognition model to predict the user's emotional state.

[1277] 3. Parsing user text input:

[1278] The smart glasses recognize voice input from the user and convert the voice data into text.

[1279] The converted text data and predicted emotion data are sent to the server.

[1280] 4. Response Generation:

[1281] The server uses a natural language processing model to generate an appropriate response based on the user's emotions and text input.

[1282] The generated response text is sent to the smart glasses and is displayed on the display and output as voice.

[1283] Specific examples

[1284] In a physical store, a customer is wearing smart glasses and asks, "Write a review about this product." The system then:

[1285] The camera in the smart glasses captures the customer's face and sends it to the server.

[1286] Predicting the emotion of "interested" using an emotion recognition model.

[1287] The questions are converted into text using voice recognition.

[1288] The following prompt sentences were generated using a natural language processing model (GPT-3) and responses were created:

[1289] text

[1290] User sentiment: Interested

[1291] Q: What are the reviews for this product?

[1292] Generates the response text "This product has high ratings and is well-received by many users."

[1293] Displayed on smart glasses and played back as audio at the same time.

[1294] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1295] Step 1:

[1296] The server controls the camera mounted on the smart glasses to capture the user's facial image. The server receives the camera image data as input and sends it to the server. The server generates the captured facial image data as output.

[1297] Step 2:

[1298] The server preprocesses the received facial image data by converting the image to grayscale and resizing it to the size required by the emotion recognition model. The server receives the facial image data as input and generates preprocessed image data as output.

[1299] Step 3:

[1300] The server inputs the preprocessed facial images into an emotion recognition model to predict the user's emotional state. This emotion recognition model is trained using Keras. It receives the preprocessed image data as input and generates the user's emotional data as output.

[1301] Step 4:

[1302] The user performs voice input through the smart glasses. The microphone in the smart glasses captures the voice data and transmits it to the terminal. The voice data is received as input and generated as output, which is transmitted to the terminal.

[1303] Step 5:

[1304] The server uses speech recognition software to convert the voice data into text. Specifically, it analyzes the voice data and generates text-based question data. The server receives the voice data as input and generates text data as output.

[1305] Step 6:

[1306] The server inputs the user's emotion data and text data into a generative AI model (GPT-3) to generate an appropriate response. The prompt uses a format such as "User emotion: Interested, Question: What are the reviews for this product?". Emotion data and text data are received as input, and response text data is generated as output.

[1307] Step 7:

[1308] The server sends the response text data to the smart glasses and displays it on the display. It also converts the response text into speech using gTTS and plays the speech to the user using PyAudio. The server receives the response text data as input and generates the text to be displayed on the display and the speech to be played as output.

[1309] 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.

[1310] The system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions. The following describes the system components and the program processing flow in natural language.

[1311] The system consists of several main components. First, the user's device plays an important role. This device is equipped with a camera, microphone, and input interface to capture the user's facial image, voice, and text input. The device then transmits this information to the server.

[1312] The server then performs several processing steps. First, the server receives the facial image sent by the user and performs preprocessing. This includes image normalization and feature point extraction. The preprocessed facial image is then passed to an emotion recognition engine. This engine analyzes the facial feature points and predicts the user's emotion. Examples of emotions include "happiness," "sadness," and "anger."

[1313] Next, the user inputs text. Using the input interface on the device, the user inputs a message such as "I'm tired today" and sends it to the server. The server receives this text input and inputs it into a natural language processing model along with the previously predicted emotional information. The natural language processing model generates a response based on the emotional information and the text input. For example, this response could be a message such as "Thank you for your hard work. Do you know how to relax?"

[1314] A more refined approach involves voice input, where the user records their voice using the device's microphone and sends it to the server. The server receives the voice data and uses an automatic speech recognition (ASR) engine to convert the speech to text. The text is then combined with emotion recognition information and similarly fed into a natural language processing model to generate an appropriate response.

[1315] A unique feature of this system is that it also includes the ability to recognize a user's emotions using biometric information other than facial images (e.g., voice and gestures). For example, by analyzing voice tone and gesture movements, it is possible to more accurately predict emotions based on this data. Furthermore, by referencing past conversation history, it is possible to continuously track emotions and provide a deeper understanding.

[1316] For example, if a user inputs "I'm stressed today," the server will predict emotions such as "stress" or "anger" from the facial recognition results. Based on this, the system will generate a response such as "Do you have any recommendations for relaxation methods to reduce stress?" and provide it to the user.

[1317] As described above, by combining an emotion engine, the present invention makes it possible to analyze user emotions from multiple angles and provide more personalized interactions, thereby achieving significant improvements in customer experience and satisfaction.

[1318] The processing flow will be explained below.

[1319] A system embodying the present invention analyzes a user's emotions in real time using multimodal data such as facial images, text input, and voice input, and generates a response based on those emotions.

[1320] Step 1:

[1321] The user captures a facial image using the device's camera, which is then immediately sent to the server.

[1322] Step 2:

[1323] The server receives the facial image sent by the user and then pre-processes the facial image, which includes image resizing, grayscale conversion, and facial feature point extraction.

[1324] Step 3:

[1325] The server passes the preprocessed facial image to an emotion recognition engine, which analyzes facial feature points and predicts the user's emotions. For example, it identifies emotions such as "happiness," "sadness," and "anger."

[1326] Step 4:

[1327] The user inputs a message into the text input field on the terminal, for example, "I'm tired today," and presses the send button to send it to the server.

[1328] Step 5:

[1329] The server processes the text input received from the user, integrates the received text input with the predicted emotional information, and creates a context to be input into the natural language processing model.

[1330] Step 6:

[1331] The server loads a natural language processing model and inputs text input along with emotional information. The natural language processing model generates an optimal response based on the user's input and emotional state. For example, it creates a response like, "You're tired. Do you know how to relax?"

[1332] Step 7:

[1333] The server sends the generated response to the user's terminal.

[1334] Step 8:

[1335] The user checks the response displayed on the device, allowing the user to receive a personalized response that matches their emotions.

[1336] Step 9:

[1337] When the user provides voice input, the terminal's microphone is used to record a voice message and send it to the server.

[1338] Step 10:

[1339] The server receives the voice data and loads an automatic speech recognition (ASR) engine, which converts the voice data into text.

[1340] Step 11:

[1341] The server then inputs the converted text and emotional information into a natural language processing model to generate an appropriate response, such as "What are your plans for today?"

[1342] Step 12:

[1343] The server sends the generated response to the user's terminal, and the user checks the response on the terminal.

[1344] Through this series of steps, the system can understand user emotions using multimodal input data and provide personalized interactions. Furthermore, by using an emotion engine, it can also utilize biometric information such as voice and gestures to more accurately recognize user emotions. This significantly improves customer experience and satisfaction.

[1345] Example 2

[1346] 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."

[1347] Conventional emotion analysis systems often have limited information for accurately predicting user emotions, using only facial images and text data, limiting their analytical accuracy. Furthermore, even systems that analyze emotions using user voice data have difficulty integrating the analysis of voice data with other data. Given these circumstances, the challenge is to analyze user emotions from multiple angles and generate more accurate and personalized responses.

[1348] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring image data from a user, means for preprocessing the acquired image data, means for analyzing the preprocessed image data and predicting the user's emotions, means for receiving text data from the user, means for generating a response based on the user's emotions and the text data, means for providing the generated response to the user, means for acquiring voice data and converting it into text data, and means for integrating and analyzing the emotion data and the voice data. This enables an integrated analysis of multiple data sources for a user, enabling more accurate and personalized responses.

[1349] A "user" is an entity that uses the system and provides data.

[1350] "Image data" is visual information captured using a camera or other imaging device.

[1351] "Preprocessing" refers to the initial processing operations performed to convert raw data into a format that is easier to analyze and process.

[1352] A "means for predicting emotions" is an algorithm or engine for inferring a user's emotional state based on acquired data.

[1353] "Text data" refers to information entered by a user in the form of text, typically using a keyboard or touch screen.

[1354] A "natural language processing model" is an algorithm or machine learning model for analyzing text data and understanding its meaning.

[1355] "Audio data" refers to audio information captured by a device such as a microphone, which needs to be converted into a text format.

[1356] An "automatic speech recognition (ASR) engine" is a technology for analyzing voice data and converting it into text data.

[1357] "Emotion data" refers to the emotional state predicted by the emotion recognition means, typically expressed numerically or categorically.

[1358] The "means for generating a response" is an algorithm or model for creating an appropriate response to provide to a user based on the emotional data and text data.

[1359] The "means for providing to the user" refers to an interface or device for transmitting the generated response to the user.

[1360] MODE FOR CARRYING OUT THE INVENTION

[1361] The system embodying the present invention analyzes various data of a user and generates a response based on the user's emotions. This system includes a user terminal, a server, and various software engines. The components and operation of the system are described in detail below.

[1362] User terminal

[1363] The user terminal includes the following hardware:

[1364] Camera: Used to capture the user's facial image.

[1365] Microphone: Used to record the user's voice input.

[1366] Input Interface: This includes keyboards and touchscreens.

[1367] The user terminal serves to transmit facial images, voice, and text input to the server in real time.

[1368] server

[1369] The server processes and analyzes the received data in cooperation with the following software engines:

[1370] Pre-processing engine: Normalizes the received face image and extracts facial feature points.

[1371] Emotion recognition engine: Analyzes preprocessed facial feature points to predict the user's emotions. For example, this engine can detect emotions such as "happiness," "sadness," and "anger."

[1372] Automatic Speech Recognition (ASR) engine: Converts user voice data into text.

[1373] Natural language processing (NLP) models: Generate appropriate responses based on text and sentiment data.

[1374] Data Handling

[1375] 1. Facial image capture and processing

[1376] The user faces the camera and the user terminal captures an image of the user's face.

[1377] The server receives the facial image, and the pre-processing engine normalizes the image and extracts feature points, such as identifying the positions of the eyes, mouth, and nose.

[1378] 2. Emotion analysis

[1379] The server's emotion recognition engine predicts the user's emotions based on the extracted feature points. For example, it analyzes the "muscle movements around the eyes" and the "angle of the corners of the mouth" to determine "happiness" or "sadness."

[1380] The results are stored as "Happiness: 80%", "Sadness: 15%", "Anger: 5%".

[1381] 3. Handling Text Input

[1382] The user uses the keyboard or touch screen on the device to input a message such as "I'm tired today" and transmits it to the server.

[1383] The server receives this text, combines it with emotional data, and inputs it into the NLP model.

[1384] 4. Processing voice input

[1385] The user uses a microphone to input speech, for example, "I'm feeling stressed today."

[1386] The server's ASR engine converts the voice data into text, then combines the converted text with emotional data and inputs it into the NLP model.

[1387] 5. Generating and Serving the Response

[1388] The NLP model uses emotional and textual data to generate a personalized response for the user, such as "Thank you for your hard work. Do you know how to relax?"

[1389] The generated response is sent from the server to the terminal and displayed to the user through the terminal's interface.

[1390] Examples and prompts

[1391] For example, if a user types "I'm feeling stressed today":

[1392] The user's device takes a facial image and sends it to the server.

[1393] The server preprocesses the images and uses an emotion recognition engine to predict emotions such as "stress" or "anger."

[1394] The user types the text "I'm stressed today" and sends it to the server.

[1395] The server inputs this information into an NLP model, generates a response such as "What relaxation techniques would you recommend to reduce stress?" and displays it on the device.

[1396] This system enables multifaceted analysis of user sentiment, enabling more accurate and personalized responses, resulting in a significant improvement in customer experience and satisfaction.

[1397] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1398] Processing steps of this system's program

[1399] Step 1: Capture user data

[1400] 1. Specific actions

[1401] The user faces the device's camera, which captures the face image and simultaneously records the voice through the microphone.

[1402] The user enters text using a keyboard or a touch screen.

[1403] 2. Input and Output

[1404] Input: User's face image, voice data, text input.

[1405] Output: The captured face image, voice data, and text data are sent from the device to the server.

[1406] Step 2: Preprocessing the face image

[1407] 1. Specific actions

[1408] The server receives the facial image sent from the terminal.

[1409] The pre-processing engine normalizes the image and extracts facial feature points (eyes, mouth, nose, etc.).

[1410] 2. Input and Output

[1411] Input: A face image sent from the device.

[1412] Output: Normalized and feature-point extracted face image data.

[1413] Step 3: Sentiment Analysis

[1414] 1. Specific actions

[1415] An emotion recognition engine analyzes the feature points of the pre-processed facial image data.

[1416] Algorithms are applied to predict emotions such as "happiness," "sadness," and "anger."

[1417] 2. Input and Output

[1418] Input: Feature points from preprocessed face image data.

[1419] Output: Predicted emotion data (e.g., "Happy: 70%", "Sad: 20%", "Anger: 10%").

[1420] Step 4: Integrating text data

[1421] 1. Specific actions

[1422] The server receives the text data sent by the user.

[1423] Integrate emotion data and text data.

[1424] 2. Input and Output

[1425] Input: Text data sent by the user, predicted emotion data.

[1426] Output: Integrated text and sentiment data.

[1427] Step 5: Process the audio data

[1428] 1. Specific actions

[1429] The server receives the voice data and converts the speech to text using an automatic speech recognition (ASR) engine.

[1430] The converted text data and emotion data are integrated.

[1431] 2. Input and Output

[1432] Input: Audio data sent by the user.

[1433] Output: Translated speech data, integrated text data and emotion data.

[1434] Step 6: Response Generation

[1435] 1. Specific actions

[1436] The server inputs the integrated text data and sentiment data into a natural language processing (NLP) model.

[1437] The NLP model generates the appropriate response.

[1438] 2. Input and Output

[1439] Input: Integrated text and sentiment data.

[1440] Output: The generated response (e.g., "Good work! Do you know how to relax?").

[1441] Step 7: Providing a response

[1442] 1. Specific actions

[1443] The server sends the generated response to the terminal.

[1444] The terminal displays the received response to the user.

[1445] 2. Input and Output

[1446] Input: The generated response.

[1447] Output: The response that is displayed to the user.

[1448] Through the above processing steps, various types of user data can be analyzed from multiple angles, and a personalized response can be generated and provided based on the results.

[1449] (Application example 2)

[1450] 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."

[1451] Detecting suspicious individuals and early detection of potential threats is a challenge for modern security systems. It is particularly difficult to quickly and accurately analyze the emotions of individuals in a crowd and provide appropriate responses. Therefore, there is a need for a system that allows security staff to analyze emotions in real time on-site and issue appropriate alerts and responses based on the results.

[1452] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring a facial image from a user's terminal, means for preprocessing the acquired facial image, means for analyzing the preprocessed facial image and predicting the user's emotion, means for acquiring voice from the user's terminal, means for converting the acquired voice into text, means for generating a response based on the user's emotion and text input, and means for providing the generated response to the user's terminal. This enables security staff to analyze individual emotions in real time on site and detect and respond to potential threats in advance.

[1453] A "user terminal" is a device equipped with input devices such as a camera and microphone, which captures facial images and voices from the user and transmits them to a server.

[1454] The "means for acquiring a facial image" refers to a function for capturing a facial image of a user using a camera mounted on the user's terminal.

[1455] The "means for preprocessing a facial image" refers to a function for performing preprocessing such as image normalization and feature point detection on an acquired facial image.

[1456] The "means for analyzing a facial image and predicting a user's emotions" refers to a function that analyzes facial feature points based on a preprocessed facial image and predicts a user's emotions.

[1457] The "means for acquiring voice" refers to a function for capturing the user's voice using a microphone installed in the user's terminal.

[1458] The "means for converting voice to text" refers to a function for converting acquired voice data into text data.

[1459] The "means for generating a response" refers to a function that uses a natural language processing model to generate an appropriate response based on the user's sentiment and text input.

[1460] The "means for providing the generated response to the user" refers to a function for displaying the generated response on the user's terminal or reproducing it as audio.

[1461] A "generative AI model" is an artificial intelligence model that generates natural-sounding language responses based on input data.

[1462] This invention is a system that uses multimodal data (facial images, voice) acquired from a user's device to analyze emotions in real time and generate and provide responses based on that data. The purpose of this system is to analyze a user's emotions from multiple angles and realize more accurate and appropriate responses.

[1463] Hardware and software used

[1464] 1. Hardware used:

[1465] Smart glasses: Equipped with a camera and microphone to capture the user's facial image and voice.

[1466] Server: Preprocesses data, analyzes it, and generates responses.

[1467] 2. Software used:

[1468] OpenCV (cv2): Image capture and preprocessing.

[1469] dlib: Face detection and feature extraction.

[1470] speech_recognition: Speech capture and text conversion.

[1471] tensorflow: Emotion recognition model.

[1472] transformers(Hugging Face): Natural language processing model.

[1473] Process Overview

[1474] Facial image capture and analysis

[1475] The user wears the smart glasses and captures facial images in real time through the camera. The images are sent to the server and preprocessed using OpenCV. This includes image normalization and facial feature point extraction using dlib. Then, emotions are analyzed from the images using a TensorFlow emotion recognition model.

[1476] Audio capture and conversion

[1477] Similarly, the user's voice is captured through a microphone built into the smart glasses, which is then sent to the server and converted to text using speech_recognition.

[1478] Generating and serving the response

[1479] The server generates an appropriate response using a generative AI model based on the analyzed emotion results and the text converted from speech, and this response is fed back to the user through the smart glasses.

[1480] Specific examples

[1481] For example, if security staff wear smart glasses at the entrance to an event venue, they can capture the faces and voices of attendees and analyze their emotions. If anger or stress is detected, the system can generate an alert that the attendee is likely overly excited and notify security staff.

[1482] Prompt Sentence Examples

[1483] "This user's emotion is anger. Then respond appropriately to the following text: 'I'm frustrated with the traffic today.'"

[1484] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1485] Step 1:

[1486] The user's device captures a facial image using the camera in the smart glasses. Specifically, the camera acquires an image and stores it as digital data. The input is the facial image captured by the camera. This data is sent to the server.

[1487] Step 2:

[1488] The server preprocesses the received facial image data. Specifically, it normalizes the image using OpenCV and then extracts facial feature points using dlib. The input is the captured facial image, and the output is the preprocessed image data and coordinate information of the feature points.

[1489] Step 3:

[1490] The server inputs the preprocessed facial image data into an emotion recognition model to predict the user's emotion. Specifically, the data is input into the TensorFlow emotion recognition model and an estimated emotion label is obtained. The input is the coordinate information of the feature points, and the output is the emotion label (e.g., "joy," "sadness," or "anger").

[1491] Step 4:

[1492] The user's device captures voice data using the microphone in the smart glasses. Specifically, the microphone records the voice and converts it into digital data. The input is the user's voice, and the output is voice data. This voice data is then sent to the server.

[1493] Step 5:

[1494] The server converts the received voice data into text data. Specifically, it uses the speech_recognition library to perform voice recognition and converts it into text. The input is voice data, and the output is text data.

[1495] Step 6:

[1496] The server generates a response based on the emotion analysis results and the text data converted from the speech. Specifically, it uses Hugging Face's transformers library to input a prompt into a generative AI model and generate an appropriate response. The input is emotion labels and text data, and the output is a response in natural language.

[1497] Step 7:

[1498] The server then sends the generated response to the user's device. Specifically, the server sends the generated response to the user's smart glasses via a network. The input is a natural language response, and the output is displayed or played on the smart glasses' display or audio output.

[1499] Step 8:

[1500] The user's device provides the generated response to the user using the smart glasses' display or audio output. Specifically, the response is displayed as text on the display or played as audio output. The input is the response received from the server, and the output is a display or audio notification to the user.

[1501] 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.

[1502] 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.

[1503] 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.

[1504] 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.

[1505] 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.

[1506] 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.

[1507] 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).

[1508] 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.

[1509] 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."

[1510] 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.

[1511] 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).

[1512] 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.

[1513] 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.

[1514] 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.

[1515] 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.

[1516] 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.

[1517] 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.

[1518] 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.

[1519] 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.

[1520] 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.

[1521] 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.

[1522] The following is further disclosed regarding the above embodiment.

[1523] (Claim 1)

[1524] means for acquiring a facial image from a user;

[1525] means for pre-processing the acquired facial images;

[1526] means for analyzing the preprocessed facial image and predicting the user's emotion;

[1527] means for receiving text input from a user;

[1528] means for generating a response based on the user's emotion and text input;

[1529] means for providing the generated response to a user;

[1530] A system including:

[1531] (Claim 2)

[1532] 2. The system of claim 1, wherein the means for preprocessing the facial image includes means for detecting facial feature points.

[1533] (Claim 3)

[1534] 10. The system of claim 1, wherein the means for generating a response uses a natural language processing model.

[1535] "Example 1"

[1536] (Claim 1)

[1537] means for acquiring a facial image from a user;

[1538] means for pre-processing the acquired facial images;

[1539] means for analyzing the preprocessed facial image and predicting the user's emotion;

[1540] means for receiving text input from a user;

[1541] means for using a natural language processing model to generate a response based on the user's sentiment and text input;

[1542] means for providing the generated response to a user;

[1543] A system including:

[1544] (Claim 2)

[1545] 2. The system of claim 1, wherein the means for preprocessing the facial image includes means for detecting facial feature points.

[1546] (Claim 3)

[1547] 10. The system of claim 1, wherein the means for receiving text input further comprises means for receiving voice input and performing voice recognition.

[1548] "Application Example 1"

[1549] (Claim 1)

[1550] means for acquiring a facial image from a user;

[1551] means for pre-processing the acquired facial images;

[1552] means for analyzing the preprocessed facial image and predicting the user's emotion;

[1553] means for receiving text input from a user;

[1554] means for generating a response based on the user's emotion and text input;

[1555] means for providing the generated response to a user;

[1556] means for capturing a facial image of a user with a camera mounted on the smart glasses;

[1557] means for displaying personalized information based on the emotion on the smart glasses;

[1558] means for recognizing a user's voice input and generating a response based on the text;

[1559] A system including:

[1560] (Claim 2)

[1561] 2. The system of claim 1, wherein the means for preprocessing the facial image includes means for detecting facial feature points.

[1562] (Claim 3)

[1563] 10. The system of claim 1, wherein the means for generating a response uses a natural language processing model.

[1564] "Example 2: Combining Emotion Engines"

[1565] (Claim 1)

[1566] means for obtaining image data from a user;

[1567] means for pre-processing the acquired image data;

[1568] means for analyzing the preprocessed image data and predicting a user's emotion;

[1569] means for receiving text data from a user;

[1570] means for generating a response based on the user's emotion and text data;

[1571] means for providing the generated response to a user;

[1572] A means for acquiring voice data and converting it into text data;

[1573] A means for integrating and analyzing emotion data and voice data;

[1574] A system including:

[1575] (Claim 2)

[1576] 2. The system of claim 1, wherein the means for preprocessing the image data includes means for normalizing the image and detecting feature points.

[1577] (Claim 3)

[1578] 10. The system of claim 1, wherein the means for generating a response uses a natural language processing model.

[1579] "Application example 2 when combining emotion engines"

[1580] (Claim 1)

[1581] A means for acquiring a face image from a user's terminal;

[1582] means for pre-processing the acquired facial images;

[1583] means for analyzing the preprocessed facial image and predicting the user's emotion;

[1584] means for acquiring audio from a user's terminal;

[1585] means for converting the acquired speech into text;

[1586] means for generating a response based on the user's emotion and text input;

[1587] means for providing the generated response to the user's terminal;

[1588] A system including:

[1589] (Claim 2)

[1590] 2. The system of claim 1, wherein the means for preprocessing the facial image includes means for detecting facial feature points.

[1591] (Claim 3)

[1592] 10. The system of claim 1, wherein the means for generating a response uses a generative AI model. [Explanation of symbols]

[1593] 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 acquiring a facial image from a user; means for pre-processing the acquired facial images; means for analyzing the preprocessed facial image and predicting the user's emotion; means for receiving text input from a user; means for generating a response based on the user's emotion and text input; means for providing the generated response to a user; A system including:

2. 2. The system of claim 1, wherein the means for preprocessing the facial image includes means for detecting facial feature points.

3. The system of claim 1 , wherein the means for generating a response uses a natural language processing model.

Images