System

Abstract

To provide a high-precision image recognition method in real time by improving image resolution and power consumption efficiency.SOLUTION: Noise is removed from the data stream received from the retinal sensor and the noise-removed data is normalized. Then, image analysis is performed using the normalized data, and the analysis result is transmitted to the terminal. In one example, the deep learning model is used to perform high-accuracy image recognition, and the analysis result can be notified to the user. This enables efficient and high-precision real-time image recognition, and overcomes the limitations of conventional frame-by-frame processing.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] Conventional image recognition systems use time-frame sensors to process data at a rate of vertical x horizontal pixels x number of frames per second, which limits the improvement of resolution and power consumption efficiency. Furthermore, while high-precision image recognition in real time is required, the difficulty of rapid response with frame-by-frame processing is an issue. Therefore, a more efficient and highly accurate image recognition method is needed. [Means for solving the problem]

[0005] The present invention provides a system that receives a continuous data stream from a retinal sensor and processes the data in real time. Specifically, the system removes noise from the received data stream and normalizes the noise-removed data. The normalized data is then used to perform image analysis and transmit the analysis results to a device. In one example, a deep learning model can be used to perform highly accurate image recognition and notify the user of the analysis results. This enables efficient, highly accurate, real-time image recognition, overcoming the limitations of conventional frame-by-frame processing.

[0006] A "retinal sensor" is an image recognition sensor that outputs data continuously in a stream format, unlike a time-frame sensor.

[0007] "Continuous data stream" refers to a format in which data is transmitted continuously in real time, rather than all at once.

[0008] "Noise reduction" is the process of filtering unwanted noise components from received data.

[0009] "Normalization" is the process of scaling the range of data to the range 0 to 1.

[0010] "Image analysis" refers to algorithms or models that take pre-processed data as input and perform object detection and classification.

[0011] A "terminal" is a device that receives the analysis results and notifies or displays them to the user.

[0012] A "deep learning model" is a type of machine learning that uses multi-layer neural networks to learn and make inferences.

[0013] A "buffer" is a memory area that temporarily stores data.

[0014] "Real-time" means there is almost no delay in processing and analyzing data.

[0015] A "communication channel" is a path or means for sending and receiving data.

[0016] "Communication" is the means or process by which data is conveyed from sender to receiver.

[0017] "User" refers to a person or thing that uses the system and receives the analysis results. [Brief explanation of the drawings]

[0018] [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

[0019] 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.

[0020] First, the terms used in the following description will be explained.

[0021] 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).

[0022] 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.

[0023] 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.

[0024] 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.

[0025] 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."

[0026] [First embodiment]

[0027] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0028] 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.

[0029] 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).

[0030] 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.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] 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.

[0036] 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.

[0037] 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.

[0038] 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."

[0039] The present invention is a system that uses a retinal sensor to continuously acquire data and perform image analysis in real time. Specifically, the system includes a server, a terminal, and a means for communicating between them.

[0040] System configuration

[0041] 1. Data Acquisition

[0042] The server receives the data stream from the retinal sensor. The data output by the sensor is sent in real time, divided into multiple data packets rather than frame by frame. The server receives this data and stores it in a buffer for processing.

[0043] Specific examples

[0044] The server processes data received from the retinal sensor every second and stores it in a buffer.

[0045] 2. Noise Reduction

[0046] The server retrieves the buffered data and applies a noise reduction filter to it, eliminating environmental and sensor noise components to obtain clean data.

[0047] Specific examples

[0048] The server applies a noise reduction filter to the received data to smooth out highly variable pixel values.

[0049] 3. Data normalization

[0050] The server scales the denoised data to a range between 0 and 1. This normalization process puts the data in a form suitable for analytical models.

[0051] Specific examples

[0052] The server scales the denoised data based on the maximum and minimum values.

[0053] 4. Image Analysis

[0054] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model consists of a multi-layered neural network and uses pre-trained parameters to achieve high accuracy in recognition.

[0055] Specific examples

[0056] The server inputs the normalized data into a deep learning model to detect faces and recognize facial expressions.

[0057] 5. Send results

[0058] The server sends the analysis results to the terminal. The terminal receives them and notifies or displays them to the user. The notification content and display format are dynamically changed depending on the analysis results.

[0059] Specific examples

[0060] The server sends the analysis results to the device, which then displays a notification on the user's smartphone saying "A face has been detected."

[0061] Usage example

[0062] Usage scenarios

[0063] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The monitoring personnel (users) receive alerts from their devices and can take prompt action.

[0064] As described above, the present invention provides a system that realizes continuous data stream processing using a retinal sensor and supports highly efficient and highly accurate image recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[0065] The processing flow will be explained below.

[0066] Step 1:

[0067] The server receives a continuous data stream from the retinal sensor. The sensor sends data packets at regular intervals, which the server buffers in real time, ready for the next processing step.

[0068] Step 2:

[0069] The server retrieves the buffered data and performs a denoising process, applying a filtering algorithm to remove external environmental noise and sensor noise components. The denoised data is clean and suitable for subsequent normalization.

[0070] Step 3:

[0071] The server normalizes the denoised data, scaling it based on the minimum and maximum values ​​of the data to convert it into a range between 0 and 1. This normalization process optimizes the input data for the analytical model.

[0072] Step 4:

[0073] The server uses the normalized data to perform image analysis. The data is then fed into a deep learning model to perform object detection and classification. Specifically, a neural network analyzes the data using multiple layers of parameters to perform tasks such as face detection and facial expression recognition.

[0074] Step 5:

[0075] The server sends the analysis results to the terminal, which then packages the analysis results in a predefined format and sends them to the terminal through a communication channel, allowing the terminal to receive the analysis results in real time.

[0076] Step 6:

[0077] The device will notify the user of the analysis results it has received in various forms, including visual alerts, audio alerts, and messages, so that the user can take further action.

[0078] Step 7:

[0079] The user can check the display on the device and take action as needed. For example, in the case of a surveillance system, the user can take action such as zooming the camera to the area where the face was detected.

[0080] By going through this series of steps, this system achieves continuous, real-time data processing using a retinal sensor, enabling highly accurate image recognition.

[0081] Example 1

[0082] 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."

[0083] Conventional image analysis systems process data on a frame-by-frame basis, but this method makes real-time analysis difficult and can result in excessive delays. Furthermore, environmental noise and sensor noise components can reduce the accuracy of data analysis. The present invention aims to solve these problems and achieve high-precision image analysis in real time. Furthermore, the present invention aims to shorten response times in surveillance and security systems by quickly notifying users of analysis results.

[0084] 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.

[0085] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data stream in a buffer, means for removing noise from the data stored in the buffer, means for normalizing the data from which the noise has been removed, means for performing image analysis using a deep learning model using the normalized data, and means for transmitting the analysis results to the terminal, thereby enabling highly accurate data analysis in real time and prompt notification of the results.

[0086] A "retinal sensor" is a sensor device that receives light, converts it into an electrical signal, and outputs visual information as digital data.

[0087] A "data stream" is a continuously transmitted stream of signals or data.

[0088] A "buffer" is a memory area or storage that temporarily stores received data.

[0089] "Noise" and "noise removal filters" are algorithms or devices used to remove unwanted signals or noise components from data.

[0090] "Data normalization" is the process of converting data into a certain range, usually scaling it to the range 0 to 1.

[0091] A "deep learning model" is a machine learning model that uses a multi-layer neural network to analyze data.

[0092] "Image analysis" is the process of extracting information such as object detection and classification from image data.

[0093] "Analysis results" refers to information or data derived from deep learning models or other analytical means.

[0094] A "terminal" is a device that communicates with a server via a network and displays or notifies the results of analysis.

[0095] The present invention provides a system that uses a retinal sensor to continuously acquire data and perform real-time image analysis. The system includes a server, a terminal, and a means for communicating between them.

[0096] Data Acquisition

[0097] The server receives the data stream sent from the retinal sensor in real time. Specifically, the data continuously output by the sensor is divided into multiple data packets and received via the network. The server stores these data packets in a buffer. For example, the server receives approximately 1,000 packets of data per second and stores each packet sequentially in the buffer.

[0098] Noise Reduction

[0099] The server removes noise from the data stored in the buffer. By applying a noise reduction filter, environmental noise and sensor noise components are reduced. The server's processing unit passes 50% of the data in the buffer through the noise reduction filter every 10 ms. The filtered data has 80% less noise compared to the original data.

[0100] Data normalization

[0101] The server scales the noise-removed data to a range of 0 to 1. This normalization process makes the data suitable for analytical models. The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of the data normalization.

[0102] Image analysis

[0103] The server uses the normalized data to perform image analysis using a deep learning model. This model is composed of a multi-layer neural network and performs highly accurate analysis using parameters based on the training data. The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step. During this process, the model maintains an accuracy of over 95%.

[0104] Send results

[0105] The server sends the analysis results to the device. The device receives the analysis results and notifies or displays them appropriately to the user. The notification content and display format change dynamically depending on the analysis results. For example, the device receives the analysis results from the server every second and displays a pop-up notification on the user's smartphone saying "A face has been detected." The user can immediately check the notification and take additional action if necessary.

[0106] Usage example

[0107] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The person in charge of monitoring (user) receives an alert from the device and can respond promptly.

[0108] Example of input prompt for generative AI model

[0109] Please explain the process of using a retinal sensor in a surveillance system to acquire data in real time and perform image analysis. Please describe in detail each step from data acquisition to sending the results, dividing it into the roles of the server, device, and user.

[0110] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0111] Step 1:

[0112] The server receives a continuous data stream from the retinal sensor in real time by dividing the data continuously output by the sensor into multiple data packets and receiving them over the network.

[0113] Input: Data packet from retinal sensor

[0114] Output: Buffered data packets

[0115] Specific operation: The server receives approximately 1,000 packets of data per second and stores each packet sequentially in a buffer.

[0116] Step 2:

[0117] The server applies a noise reduction filter to remove noise from the buffered data, which reduces environmental noise and sensor noise components.

[0118] Input: raw data stored in a buffer

[0119] Output: Clean data with noise removed

[0120] How it works: Every 10 ms, the server's processing unit passes 50% of the data in the buffer through a noise reduction filter. The filtered data has 80% less noise than the original data.

[0121] Step 3:

[0122] The server scales the denoised data to a range between 0 and 1. The normalization process involves scaling each data point based on the maximum and minimum values ​​of the data.

[0123] Input: Clean data with noise removed

[0124] Output: Normalized data

[0125] Specific operation: The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of data normalization.

[0126] Step 4:

[0127] The server uses the normalized data to perform image analysis using a deep learning model, which is composed of a multi-layer neural network and uses parameters based on the training data to perform highly accurate analysis.

[0128] Input: Normalized data

[0129] Output: Image analysis results

[0130] Specific operation: The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step, maintaining the model accuracy of over 95% during this process.

[0131] Step 5:

[0132] The server sends the results of the image analysis to the device. The device receives the results and notifies or displays the appropriate information to the user. The notification content and display format are dynamically changed depending on the analysis results.

[0133] Input: Image analysis results

[0134] Output: Information reported to the user

[0135] Specific operation: The device receives the analysis results from the server every second and displays a "Face detected" notification on the user's smartphone. The user can immediately check the notification and take additional action if necessary.

[0136] (Application example 1)

[0137] 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."

[0138] Conventional real-time image analysis systems often lack the efficiency and accuracy required for continuous data processing and noise removal, resulting in delays and false positives when detecting suspicious individuals, particularly in the security field. Furthermore, when notifying a device of analysis results, the system lacks the functionality to respond immediately, requiring a fast and accurate response.

[0139] 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.

[0140] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis using the normalized data, means for transmitting the analysis results to the terminal, means for generating a security notification based on the analysis results and displaying it on the terminal, and means for detecting suspicious activity based on the image analysis results in real time, thereby enabling highly efficient and highly accurate security monitoring.

[0141] A "retinal sensor" is a sensor that continuously acquires video data, unlike conventional frame-by-frame imaging methods, and generates a data stream that can be analyzed in real time.

[0142] A "continuous data stream" is a data format in which data is transmitted continuously, without interruption, rather than in specific frames.

[0143] "Noise removal" is the process of removing unnecessary environmental noise and sensor noise components from acquired data to obtain clean data.

[0144] "Data normalization" is the process of scaling noise-removed data to a range of 0 to 1 and converting it into a format suitable for analytical models.

[0145] "Image analysis" is the process of using data to detect and classify objects using algorithms such as deep learning models.

[0146] A "security notification" is a notification that notifies the user when suspicious activity or anomalies are detected based on the analysis results.

[0147] A "terminal" refers to an electronic device such as a smartphone or smart glasses that receives the analysis results and notifies the user.

[0148] "Real-time" refers to data acquisition and processing occurring immediately, with little or no delay.

[0149] "Detecting suspicious activity" is the process of identifying suspicious behavior or individuals based on the results of image analysis and providing information for responding.

[0150] The system for realizing this invention is composed of a retinal sensor, a server, and a terminal. The specific operation of each component and its program are described in detail below.

[0151] Hardware and Software Use Cases

[0152] This system uses the following hardware and software:

[0153] Hardware

[0154] Retinal sensor: A sensor that continuously acquires video data and generates a data stream that can be analyzed in real time.

[0155] Smartphone or smart glasses: A mobile device carried by the user.

[0156] Server: Receives data, processes it, and sends the analysis results.

[0157] software

[0158] Python: An entire programming language.

[0159] OpenCV: An image processing library, used for denoising and preprocessing data.

[0160] TENSORFLOW® / Keras: A deep learning library used for image analysis models.

[0161] Flask: A web framework used for communication between the server and the terminal and for notifying results.

[0162] Data processing and calculation flow

[0163] The server starts by receiving a continuous data stream from the retinal sensor. The received data is first stored in a buffer. Next, noise is removed from the data using OpenCV to obtain clean data. The noise-removed data is normalized to a range of 0 to 1, and image analysis is performed using a deep learning model. If suspicious activity is detected as a result of the analysis, the results are sent to the device. This series of processes is performed in real time.

[0164] The device receives the analysis results from the server and generates and displays security notifications to the user as needed, such as a "Suspicious person detected" message on a smartphone or smart glasses.

[0165] Specific examples of processing

[0166] This system is used in a scenario where office entrances are monitored. When a retinal sensor installed in the smart glasses detects a suspicious person or movement, the data is sent to a server in real time. The server processes the data and notifies the user of the analysis results on their smartphone. This allows the user to take prompt action even when they are in a remote location.

[0167] Prompt Sentence Examples

[0168] "Implement a security application that uses a retinal sensor to monitor the surroundings in real time with high accuracy and detect suspicious individuals. The application should be installable on smartphones or smart glasses, using Python, OpenCV, TensorFlow, and Flask."

[0169] In this way, the present invention realizes continuous data stream processing using a retinal sensor, and can provide highly efficient and highly accurate security monitoring.

[0170] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0171] Step 1:

[0172] A continuous data stream is acquired from the retinal sensor.

[0173] Input: Raw data stream from the retinal sensor.

[0174] Data processing / calculation: The image data captured by the sensor is acquired continuously, rather than frame by frame.

[0175] Output: A continuous data stream.

[0176] Step 2:

[0177] The server receives the data stream and stores it in a buffer.

[0178] Input: A continuous data stream.

[0179] Data processing / calculation: The received data stream is temporarily stored in a buffer.

[0180] Output: The raw data stored in the buffer.

[0181] Step 3:

[0182] The server applies a noise reduction filter to the buffered data.

[0183] Input: Raw data stored in a buffer.

[0184] Data processing / calculation: Using OpenCV, unnecessary information such as environmental noise and sensor noise is filtered out.

[0185] Specific operation: Convert to grayscale and apply a filter such as GaussianBlur.

[0186] Output: Denoised data.

[0187] Step 4:

[0188] The server normalizes the denoised data.

[0189] Input: Denoised data.

[0190] Data processing / calculation: Scale the data to the range 0 to 1.

[0191] Specific operation: Normalization is performed using the maximum and minimum values ​​of the data.

[0192] Output: Normalized data.

[0193] Step 5:

[0194] The server inputs the normalized data into a deep learning model to perform image analysis.

[0195] Input: Normalized data.

[0196] Data processing / calculation: Image analysis is performed using a trained model using TensorFlow / Keras to detect suspicious individuals and abnormal behavior.

[0197] Specific operation: Analysis is performed using a neural network.

[0198] Output: Analysis results (such as suspicious person detection).

[0199] Step 6:

[0200] The server transmits the analysis results to the terminal.

[0201] Input: Image analysis results.

[0202] Data processing / calculation: Analysis results are sent to the terminal.

[0203] Specific operation: Send the result via HTTP request using Flask.

[0204] Output: The analysis results are sent to the terminal.

[0205] Step 7:

[0206] The device displays a notification to the user based on the received analysis results.

[0207] Input: Analysis results sent from the server.

[0208] Data processing / calculation: The received results are notified to the user in an easy-to-understand format.

[0209] Specific operation: Display an alert on your smartphone or smart glasses.

[0210] Output: A notification is displayed to the user.

[0211] In this way, the server, terminal, and user play their respective roles at each step, achieving highly accurate real-time image analysis.

[0212] 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.

[0213] The present invention is a system for seamlessly processing data acquired from a retinal sensor to perform image analysis and emotion recognition. Specifically, it includes a server, a terminal, and an emotion engine.

[0214] System configuration

[0215] 1. Data Acquisition

[0216] The server receives a continuous data stream from the retinal sensor, which stores the data in a buffer in real time.

[0217] Specific examples

[0218] The server continuously receives data from the retinal sensor every second and stores it in a buffer.

[0219] 2. Noise Reduction

[0220] The server retrieves the data stored in the buffer and applies a noise reduction filter to remove the noise components, resulting in clean data.

[0221] Specific examples

[0222] The server applies a noise reduction filter to the received data to remove environmental noise and sensor noise.

[0223] 3. Data normalization

[0224] The server normalizes the denoised data, scaling it to the range 0-1 and converting it into a form suitable for analytical models.

[0225] Specific examples

[0226] The server scales the denoised data based on the maximum and minimum values.

[0227] 4. Image Analysis

[0228] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model is composed of a multi-layered neural network, achieving highly accurate recognition.

[0229] Specific examples

[0230] The server inputs the normalized data into a deep learning model to detect faces and classify objects.

[0231] 5. Emotion recognition

[0232] The emotion engine uses the results of image analysis and additional sensor data to recognize the user's emotions. It uses deep learning models to analyze emotions from the user's facial expressions and voice.

[0233] Specific examples

[0234] The emotion engine uses deep learning models to analyze the user's facial expressions to determine joy, anger, sadness, and happiness.

[0235] 6. Send results

[0236] The server sends the results of image analysis and emotion recognition to the device. The analysis results are packaged in a predefined format and sent to the device in real time. The device then receives the analysis results and notifies the user.

[0237] Specific examples

[0238] The server sends the results of image analysis and emotion recognition to the device, which then displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0239] Usage example

[0240] Usage scenarios

[0241] The online education system analyzes students' facial expressions and voices in real time to understand their emotions. This allows instructors to understand the students' emotional state and adjust the educational content in real time. For example, if a student appears confused during a lecture, the data can be analyzed by another engine and a message will be sent to the instructor saying, "The student is confused." This allows the instructor to respond by providing additional explanations.

[0242] As described above, the present invention provides a system that uses a retinal sensor to realize real-time processing of continuous data streams, image analysis, and emotion recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[0243] The processing flow will be explained below.

[0244] Step 1:

[0245] The server receives a continuous data stream from the retinal sensor. The data generated by the sensor is sent in real time to the server, which stores it in a buffer where it is ready for the next processing step.

[0246] Step 2:

[0247] The server retrieves the buffered data and performs a denoising process, applying filtering algorithms to remove environmental and sensor noise. This clean data provides the basis for more accurate analysis.

[0248] Step 3:

[0249] The server then performs a normalization process on the denoised data, scaling it to a range of 0 to 1 and converting it into a form that is compatible with the analytical model. This normalization process allows the analytical model to function efficiently.

[0250] Step 4:

[0251] The server uses the normalized data to perform image analysis. It then uses deep learning models to analyze the data and perform object detection and classification. For example, the server uses deep learning models to detect human faces and classify objects.

[0252] Step 5:

[0253] The emotion engine uses image analysis results and additional sensor data to recognize the user's emotions. Deep learning models are applied to emotion recognition, analyzing the user's facial expressions and voice data. The analysis results are classified into emotion categories such as "happiness," "sadness," and "surprise."

[0254] Step 6:

[0255] The server sends the results of image analysis and emotion recognition to the device, which then packages the results in a predefined format and sends them to the device. The device processes the results and notifies the user.

[0256] Step 7:

[0257] The device displays the analysis results received from the server and notifies the user in a variety of notification formats, including visual alerts, audio notifications, and messages. The user can then take action based on the notification.

[0258] Step 8:

[0259] The user checks the display on the device and takes action as necessary. For example, in the case of a surveillance system, after receiving a notification, the user may take action such as zooming in on the area where suspicious activity was detected.

[0260] Through these steps, this system achieves real-time data processing, image analysis, and emotion recognition using a retinal sensor, enabling it to provide users with fast and accurate information. This overcomes the limitations of conventional frame-by-frame processing methods and enables more efficient operation.

[0261] Example 2

[0262] 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."

[0263] Conventional image analysis and emotion recognition systems face the challenge of processing the entire process, from data acquisition to analysis, recognition, and notification, in real time. This can lead to issues with reduced analysis accuracy, particularly when dealing with raw data containing large amounts of noise. Furthermore, there is a lack of effective means for quickly notifying users of analysis results. To address these issues, a system capable of high-precision data processing and analysis in real time is required.

[0264] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0265] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data in a buffer, means for applying a noise reduction filter to the data stored in the buffer, means for normalizing the noise-reduced data, means for performing image analysis using a deep learning model with the normalized data, means for performing emotion recognition using the image analysis results and additional sensor data, means for transmitting the analysis results and emotion recognition results to a terminal, and means for notifying the user of the results on the terminal. This enables noise reduction and data normalization to be performed in real time, achieving highly accurate image analysis and emotion recognition, and quickly notifying the user of the analysis results.

[0266] A "retinal sensor" is a sensor that detects light and outputs visual information as an electrical signal.

[0267] A "continuous data stream" is a continuous flow of data that is sent without interruption.

[0268] A "buffer" is a temporary storage area for data, and is an area used to adjust the difference between the data reception speed and the data processing speed.

[0269] A "noise removal filter" is an algorithm or hardware for removing unwanted noise components contained in data.

[0270] "Normalization" is the process of scaling data to a certain range (usually between 0 and 1).

[0271] A "deep learning model" is a machine learning model that uses a multi-layer neural network to learn the characteristics of data and perform analysis and prediction.

[0272] "Image analysis" is the process of detecting and classifying objects from image data, and is primarily performed using deep learning models.

[0273] "Emotion recognition" is the process of identifying a person's emotional state from image and audio data.

[0274] A "terminal" is an equipment or device that is directly operated by a user, and examples include smartphones and tablets.

[0275] MODE FOR CARRYING OUT THE INVENTION

[0276] The present invention provides a system for image analysis and emotion recognition that processes a continuous data stream acquired from a retinal sensor in real time. The system mainly includes a server, a terminal, and an emotion engine.

[0277] 1. Data Acquisition

[0278] The server receives the data sent from the retina sensor in real time and stores it in a buffer in memory, preventing data interruptions and ensuring a continuous data stream.

[0279] Specific examples

[0280] The server receives data from the retina sensor every second and stores it in a buffer.

[0281] 2. Noise Reduction

[0282] The server applies a noise reduction filter to the buffered data, removing sensor noise and environmental noise to obtain clean data.

[0283] Specific examples

[0284] The server applies a low-pass filter to the data in the buffer to remove unwanted high frequency components.

[0285] 3. Data normalization

[0286] The server normalizes the denoised data, specifically scaling the data values ​​to the range 0 to 1.

[0287] Specific examples

[0288] The server takes the maximum value of the data as 1 and the minimum value as 0 and scales each data point to fit into this range.

[0289] 4. Image Analysis

[0290] The server uses the normalized data to perform highly accurate image analysis using a deep learning model, which is built using a multi-layer neural network.

[0291] Specific examples

[0292] The server inputs the data into a neural network model built using a deep learning library (e.g., TensorFlow, PyTorch) to detect human faces and classify objects.

[0293] 5. Emotion recognition

[0294] The emotion engine uses a deep learning model to analyze the user's emotions based on the results of image analysis and additional sensor data, allowing it to identify emotions from the user's facial expressions and voice.

[0295] Specific examples

[0296] The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial expressions and voice data to identify emotions.

[0297] 6. Send results

[0298] The server sends the results of image analysis and emotion recognition to the device, which receives them and notifies the user of the results.

[0299] Specific examples

[0300] The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0301] Prompt Sentence Examples

[0302] Below are some examples of prompts to input to the generative AI model.

[0303] Example prompts

[0304] Please explain in detail how the system works, which continuously receives data from a retinal sensor and performs noise removal, data normalization, image analysis, and emotion recognition in real time.

[0305] The above is an embodiment of the system. This invention enables highly accurate data processing and analysis in real time, and allows analysis results to be quickly notified to the user.

[0306] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0307] Step 1:

[0308] The server receives a continuous data stream from the retinal sensor, which is buffered in real time.

[0309] Input: Raw data from the retinal sensor

[0310] Output: raw data stored in a buffer

[0311] Specific operation: The server receives data every second and immediately stores the data in a buffer to prevent data interruptions.

[0312] Step 2:

[0313] The server applies a noise reduction filter to the buffered data, which removes unwanted noise components.

[0314] Input: raw data stored in a buffer

[0315] Output: Clean data after noise removal

[0316] What it does: The server applies a low-pass filter to remove environmental noise and sensor noise, for example, a Butterworth filter to reduce noise in a specific frequency band.

[0317] Step 3:

[0318] The server normalizes the denoised data to a range of 0 to 1, converting it into a data format suitable for the analytical model.

[0319] Input: Clean data after noise removal

[0320] Output: Normalized data

[0321] What it does: The server scales each value of the data based on the maximum and minimum values. For example, if the original data ranges from 0 to 255, it divides each value by 255 to scale it to the range of 0 to 1.

[0322] Step 4:

[0323] The server runs a deep learning model on the normalized data to analyze the images, resulting in object detection and classification.

[0324] Input: Normalized data

[0325] Output: Image analysis results (e.g., object detection and classification results)

[0326] What it does: The server uses deep learning libraries (e.g., TensorFlow, PyTorch) to apply neural network models to perform image analysis, such as detecting human faces and classifying their attributes.

[0327] Step 5:

[0328] The emotion engine performs emotion recognition based on the results of image analysis and additional sensor data, making it possible to identify the user's emotions.

[0329] Input: Image analysis results, additional sensor data

[0330] Output: Emotion recognition results (e.g., joy, anger, sadness, and happiness)

[0331] Specific operation: The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial and voice data and identify emotions. For example, it determines that the user is "happy" based on their facial expression.

[0332] Step 6:

[0333] The server transmits the results of the image analysis and emotion recognition to the terminal, and the terminal notifies the user of the results.

[0334] Input: Image analysis results, emotion recognition results

[0335] Output: The result reported to the user

[0336] Specific operation: The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0337] (Application example 2)

[0338] 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."

[0339] Conventional factory robot systems have been unable to monitor worker emotions and fatigue levels in real time and provide appropriate feedback according to the situation. This has made it difficult to ensure worker safety and has left challenges in creating an efficient work environment. The present invention aims to solve this problem by providing a system that enables worker emotion recognition and fatigue level monitoring.

[0340] 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 receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis and emotion recognition using the normalized data, means for transmitting the analysis results and emotion recognition results to the terminal and providing feedback according to the user's situation based on the analysis results, means for continuously buffering preprocessed data, means for analyzing the data received from the retinal sensor in real time and monitoring the worker's emotion and fatigue level, and means for processing the received data stream in real time and notifying the worker to take a break based on the analysis results. This makes it possible to monitor the worker's emotion and fatigue level in real time and create a safe and efficient work environment.

[0341] A "retinal sensor" is a device that senses light, generates signals, and outputs a continuous data stream, mimicking the function of the human retina.

[0342] A "continuous data stream" refers to a data flow in which data is transmitted continuously rather than intermittently.

[0343] "Noise" refers to unnecessary components contained in sensor data that prevent accurate information from being obtained.

[0344] A "means for removing noise" is a method or device for removing unnecessary components from received data to obtain clean data.

[0345] A "means for normalizing data" is a method or device that scales data to a particular range and converts it into a form suitable for analysis.

[0346] "Image analysis" refers to the method or process of processing acquired image data to detect and classify objects.

[0347] "Emotion recognition" is a method of analyzing and identifying a user's emotional state using images and additional sensor data.

[0348] "Analysis results" refers to the information obtained as a result of data analysis, which is used to prompt decision-making and action.

[0349] "Feedback" refers to information or instructions provided to the user based on the analysis results, to help them maintain appropriate behavior or state.

[0350] "Terminal" refers to a computer or smart device that is directly operated by a user and is used to display analysis results and receive notifications.

[0351] A "buffer" is a storage device that temporarily stores data, allowing for smooth reception and processing of the data.

[0352] "Real-time" refers to data acquisition, processing, and feedback occurring immediately, without delay.

[0353] In this invention, a system is constructed that monitors the emotions and fatigue levels of factory workers in real time to provide a safe and efficient working environment. A detailed embodiment of this system will be described below.

[0354] The server first receives the data stream from the retinal sensor, which captures information such as the worker's facial expressions and gaze in real time. The captured data is temporarily stored in a buffer. A noise reduction filter is then applied to the stored data to remove unnecessary noise components, resulting in clean data.

[0355] The noise-removed data is then normalized to a range of 0 to 1. A deep learning model is then used to analyze the image and recognize emotions. The image analysis process detects the worker's face and facial expressions, and emotion recognition is performed based on this information.

[0356] The results of emotion recognition are sent from the server to the device in real time. Based on the analysis results, the device provides the worker with context-appropriate feedback. This feedback is displayed on the worker's smart glasses. For example, if the device determines that the worker is tired, a notification will be displayed encouraging them to take a break.

[0357] Specific examples of hardware and software used

[0358] Hardware: retinal sensor, smart glasses, server

[0359] Software: TensorFlow / Keras (for deep learning), OpenCV (for image processing), Flask (to build the API)

[0360] Specific examples

[0361] If a factory worker is operating a machine without expressing any emotion for a long period of time, the retinal sensor will capture this behavior and the server will analyze it in real time. If this analysis determines that the worker is fatigued, the system will display a notification on the smart glasses telling the worker to take a break.

[0362] Prompt Sentence Examples

[0363] Create a Python program that detects fatigue when a factory worker operates a machine without expressing any emotion for a long period of time and displays a notification encouraging them to take a break.

[0364] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0365] Step 1:

[0366] The server receives a continuous data stream from the retinal sensor. The input is the raw data from the retinal sensor, and the output is the received data stream. This data stream contains information such as the worker's facial expressions and gaze.

[0367] Step 2:

[0368] The server stores the received data stream in a buffer. The input is the data stream received in step 1, and the output is the data stored in the buffer. In this step, data is temporarily stored.

[0369] Step 3:

[0370] The server applies a noise reduction filter to the buffered data. The input is the raw data stored in the buffer, and the output is the clean, denoised data. This filtering process removes environmental noise and sensor noise.

[0371] Step 4:

[0372] The server normalizes the denoised data. The input is the denoised data and the output is the normalised data. The normalisation process scales the data to the range 0 to 1.

[0373] Step 5:

[0374] The server uses the normalized data to perform image analysis. The input is the normalized data, and the output is the result of image analysis. This image analysis uses a deep learning model to detect and classify objects.

[0375] Step 6:

[0376] The server performs emotion recognition based on the results of image analysis. The input is the result of image analysis, and the output is the result of emotion recognition. The type of emotion (joy, anger, sadness, happiness, etc.) and fatigue level are analyzed using a deep learning model.

[0377] Step 7:

[0378] The server sends the results of image analysis and emotion recognition to the terminal. The input is the emotion recognition result, and the output is the analysis result and emotion recognition result sent to the receiving terminal. This data is sent in real time.

[0379] Step 8:

[0380] The terminal provides feedback to the user based on the analysis results received from the server. The input is the analysis results and emotion recognition results sent from the server, and the output is a notification displayed to the user. As a specific example, a notification saying "Please take a break" is displayed on smart glasses. This notification allows the user (worker) to receive appropriate feedback.

[0381] 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.

[0382] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0383] 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.

[0384] [Second embodiment]

[0385] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0386] 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.

[0387] 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).

[0388] 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.

[0389] 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.

[0390] 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).

[0391] 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.

[0392] 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.

[0393] 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.

[0394] 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.

[0395] 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.

[0396] 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."

[0397] The present invention is a system that uses a retinal sensor to continuously acquire data and perform image analysis in real time. Specifically, the system includes a server, a terminal, and a means for communicating between them.

[0398] System configuration

[0399] 1. Data Acquisition

[0400] The server receives the data stream from the retinal sensor. The data output by the sensor is sent in real time, divided into multiple data packets rather than frame by frame. The server receives this data and stores it in a buffer for processing.

[0401] Specific examples

[0402] The server processes data received from the retinal sensor every second and stores it in a buffer.

[0403] 2. Noise Reduction

[0404] The server retrieves the buffered data and applies a noise reduction filter to it, eliminating environmental and sensor noise components to obtain clean data.

[0405] Specific examples

[0406] The server applies a noise reduction filter to the received data to smooth out highly variable pixel values.

[0407] 3. Data normalization

[0408] The server scales the denoised data to a range between 0 and 1. This normalization process puts the data in a form suitable for analytical models.

[0409] Specific examples

[0410] The server scales the denoised data based on the maximum and minimum values.

[0411] 4. Image Analysis

[0412] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model consists of a multi-layered neural network and uses pre-trained parameters to achieve high accuracy in recognition.

[0413] Specific examples

[0414] The server inputs the normalized data into a deep learning model to detect faces and recognize facial expressions.

[0415] 5. Send results

[0416] The server sends the analysis results to the terminal. The terminal receives them and notifies or displays them to the user. The notification content and display format are dynamically changed depending on the analysis results.

[0417] Specific examples

[0418] The server sends the analysis results to the device, which then displays a notification on the user's smartphone saying "A face has been detected."

[0419] Usage example

[0420] Usage scenarios

[0421] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The monitoring personnel (users) receive alerts from their devices and can take prompt action.

[0422] As described above, the present invention provides a system that realizes continuous data stream processing using a retinal sensor and supports highly efficient and highly accurate image recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[0423] The processing flow will be explained below.

[0424] Step 1:

[0425] The server receives a continuous data stream from the retinal sensor. The sensor sends data packets at regular intervals, which the server buffers in real time, ready for the next processing step.

[0426] Step 2:

[0427] The server retrieves the buffered data and performs a denoising process, applying a filtering algorithm to remove external environmental noise and sensor noise components. The denoised data is clean and suitable for subsequent normalization.

[0428] Step 3:

[0429] The server normalizes the denoised data, scaling it based on the minimum and maximum values ​​of the data to convert it into a range between 0 and 1. This normalization process optimizes the input data for the analytical model.

[0430] Step 4:

[0431] The server uses the normalized data to perform image analysis. The data is then fed into a deep learning model to perform object detection and classification. Specifically, a neural network analyzes the data using multiple layers of parameters to perform tasks such as face detection and facial expression recognition.

[0432] Step 5:

[0433] The server sends the analysis results to the terminal, which then packages the analysis results in a predefined format and sends them to the terminal through a communication channel, allowing the terminal to receive the analysis results in real time.

[0434] Step 6:

[0435] The device will notify the user of the analysis results it has received in various forms, including visual alerts, audio alerts, and messages, so that the user can take further action.

[0436] Step 7:

[0437] The user can check the display on the device and take action as needed. For example, in the case of a surveillance system, the user can take action such as zooming the camera to the area where the face was detected.

[0438] By going through this series of steps, this system achieves continuous, real-time data processing using a retinal sensor, enabling highly accurate image recognition.

[0439] Example 1

[0440] 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."

[0441] Conventional image analysis systems process data on a frame-by-frame basis, but this method makes real-time analysis difficult and can result in excessive delays. Furthermore, environmental noise and sensor noise components can reduce the accuracy of data analysis. The present invention aims to solve these problems and achieve high-precision image analysis in real time. Furthermore, the present invention aims to shorten response times in surveillance and security systems by quickly notifying users of analysis results.

[0442] 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.

[0443] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data stream in a buffer, means for removing noise from the data stored in the buffer, means for normalizing the data from which the noise has been removed, means for performing image analysis using a deep learning model using the normalized data, and means for transmitting the analysis results to the terminal, thereby enabling highly accurate data analysis in real time and prompt notification of the results.

[0444] A "retinal sensor" is a sensor device that receives light, converts it into an electrical signal, and outputs visual information as digital data.

[0445] A "data stream" is a continuously transmitted stream of signals or data.

[0446] A "buffer" is a memory area or storage that temporarily stores received data.

[0447] "Noise" and "noise removal filters" are algorithms or devices used to remove unwanted signals or noise components from data.

[0448] "Data normalization" is the process of converting data into a certain range, usually scaling it to the range 0 to 1.

[0449] A "deep learning model" is a machine learning model that uses a multi-layer neural network to analyze data.

[0450] "Image analysis" is the process of extracting information such as object detection and classification from image data.

[0451] "Analysis results" refers to information or data derived from deep learning models or other analytical means.

[0452] A "terminal" is a device that communicates with a server via a network and displays or notifies the results of analysis.

[0453] The present invention provides a system that uses a retinal sensor to continuously acquire data and perform real-time image analysis. The system includes a server, a terminal, and a means for communicating between them.

[0454] Data Acquisition

[0455] The server receives the data stream sent from the retinal sensor in real time. Specifically, the data continuously output by the sensor is divided into multiple data packets and received via the network. The server stores these data packets in a buffer. For example, the server receives approximately 1,000 packets of data per second and stores each packet sequentially in the buffer.

[0456] Noise Reduction

[0457] The server removes noise from the data stored in the buffer. By applying a noise reduction filter, environmental noise and sensor noise components are reduced. The server's processing unit passes 50% of the data in the buffer through the noise reduction filter every 10 ms. The filtered data has 80% less noise compared to the original data.

[0458] Data normalization

[0459] The server scales the noise-removed data to a range of 0 to 1. This normalization process makes the data suitable for analytical models. The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of the data normalization.

[0460] Image analysis

[0461] The server uses the normalized data to perform image analysis using a deep learning model. This model is composed of a multi-layer neural network and performs highly accurate analysis using parameters based on the training data. The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step. During this process, the model maintains an accuracy of over 95%.

[0462] Send results

[0463] The server sends the analysis results to the device. The device receives the analysis results and notifies or displays them appropriately to the user. The notification content and display format change dynamically depending on the analysis results. For example, the device receives the analysis results from the server every second and displays a pop-up notification on the user's smartphone saying "A face has been detected." The user can immediately check the notification and take additional action if necessary.

[0464] Usage example

[0465] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The person in charge of monitoring (user) receives an alert from the device and can respond promptly.

[0466] Example of input prompt for generative AI model

[0467] Please explain the process of using a retinal sensor in a surveillance system to acquire data in real time and perform image analysis. Please describe in detail each step from data acquisition to sending the results, dividing it into the roles of the server, device, and user.

[0468] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0469] Step 1:

[0470] The server receives a continuous data stream from the retinal sensor in real time by dividing the data continuously output by the sensor into multiple data packets and receiving them over the network.

[0471] Input: Data packet from retinal sensor

[0472] Output: Buffered data packets

[0473] Specific operation: The server receives approximately 1,000 packets of data per second and stores each packet sequentially in a buffer.

[0474] Step 2:

[0475] The server applies a noise reduction filter to remove noise from the buffered data, which reduces environmental noise and sensor noise components.

[0476] Input: raw data stored in a buffer

[0477] Output: Clean data with noise removed

[0478] How it works: Every 10 ms, the server's processing unit passes 50% of the data in the buffer through a noise reduction filter. The filtered data has 80% less noise than the original data.

[0479] Step 3:

[0480] The server scales the denoised data to a range between 0 and 1. The normalization process involves scaling each data point based on the maximum and minimum values ​​of the data.

[0481] Input: Clean data with noise removed

[0482] Output: Normalized data

[0483] Specific operation: The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of data normalization.

[0484] Step 4:

[0485] The server uses the normalized data to perform image analysis using a deep learning model, which is composed of a multi-layer neural network and uses parameters based on the training data to perform highly accurate analysis.

[0486] Input: Normalized data

[0487] Output: Image analysis results

[0488] Specific operation: The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step, maintaining the model accuracy of over 95% during this process.

[0489] Step 5:

[0490] The server sends the results of the image analysis to the device. The device receives the results and notifies or displays the appropriate information to the user. The notification content and display format are dynamically changed depending on the analysis results.

[0491] Input: Image analysis results

[0492] Output: Information reported to the user

[0493] Specific operation: The device receives the analysis results from the server every second and displays a "Face detected" notification on the user's smartphone. The user can immediately check the notification and take additional action if necessary.

[0494] (Application example 1)

[0495] 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."

[0496] Conventional real-time image analysis systems often lack the efficiency and accuracy required for continuous data processing and noise removal, resulting in delays and false positives when detecting suspicious individuals, particularly in the security field. Furthermore, when notifying a device of analysis results, the system lacks the functionality to respond immediately, requiring a fast and accurate response.

[0497] 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.

[0498] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis using the normalized data, means for transmitting the analysis results to the terminal, means for generating a security notification based on the analysis results and displaying it on the terminal, and means for detecting suspicious activity based on the image analysis results in real time, thereby enabling highly efficient and highly accurate security monitoring.

[0499] A "retinal sensor" is a sensor that continuously acquires video data, unlike conventional frame-by-frame imaging methods, and generates a data stream that can be analyzed in real time.

[0500] A "continuous data stream" is a data format in which data is transmitted continuously, without interruption, rather than in specific frames.

[0501] "Noise removal" is the process of removing unnecessary environmental noise and sensor noise components from acquired data to obtain clean data.

[0502] "Data normalization" is the process of scaling noise-removed data to a range of 0 to 1 and converting it into a format suitable for analytical models.

[0503] "Image analysis" is the process of using data to detect and classify objects using algorithms such as deep learning models.

[0504] A "security notification" is a notification that notifies the user when suspicious activity or anomalies are detected based on the analysis results.

[0505] A "terminal" refers to an electronic device such as a smartphone or smart glasses that receives the analysis results and notifies the user.

[0506] "Real-time" refers to data acquisition and processing occurring immediately, with little or no delay.

[0507] "Detecting suspicious activity" is the process of identifying suspicious behavior or individuals based on the results of image analysis and providing information for responding.

[0508] The system for realizing this invention is composed of a retinal sensor, a server, and a terminal. The specific operation of each component and its program are described in detail below.

[0509] Hardware and Software Use Cases

[0510] This system uses the following hardware and software:

[0511] Hardware

[0512] Retinal sensor: A sensor that continuously acquires video data and generates a data stream that can be analyzed in real time.

[0513] Smartphone or smart glasses: A mobile device carried by the user.

[0514] Server: Receives data, processes it, and sends the analysis results.

[0515] software

[0516] Python: An entire programming language.

[0517] OpenCV: An image processing library, used for denoising and preprocessing data.

[0518] TensorFlow / Keras: Deep learning libraries used for image analysis models.

[0519] Flask: A web framework used for communication between the server and the terminal and for notifying results.

[0520] Data processing and calculation flow

[0521] The server starts by receiving a continuous data stream from the retinal sensor. The received data is first stored in a buffer. Next, noise is removed from the data using OpenCV to obtain clean data. The noise-removed data is normalized to a range of 0 to 1, and image analysis is performed using a deep learning model. If suspicious activity is detected as a result of the analysis, the results are sent to the device. This series of processes is performed in real time.

[0522] The device receives the analysis results from the server and generates and displays security notifications to the user as needed, such as a "Suspicious person detected" message on a smartphone or smart glasses.

[0523] Specific examples of processing

[0524] This system is used in a scenario where office entrances are monitored. When a retinal sensor installed in the smart glasses detects a suspicious person or movement, the data is sent to a server in real time. The server processes the data and notifies the user of the analysis results on their smartphone. This allows the user to take prompt action even when they are in a remote location.

[0525] Prompt Sentence Examples

[0526] "Implement a security application that uses a retinal sensor to monitor the surroundings in real time with high accuracy and detect suspicious individuals. The application should be installable on smartphones or smart glasses, using Python, OpenCV, TensorFlow, and Flask."

[0527] In this way, the present invention realizes continuous data stream processing using a retinal sensor, and can provide highly efficient and highly accurate security monitoring.

[0528] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0529] Step 1:

[0530] A continuous data stream is acquired from the retinal sensor.

[0531] Input: Raw data stream from the retinal sensor.

[0532] Data processing / calculation: The image data captured by the sensor is acquired continuously, rather than frame by frame.

[0533] Output: A continuous data stream.

[0534] Step 2:

[0535] The server receives the data stream and stores it in a buffer.

[0536] Input: A continuous data stream.

[0537] Data processing / calculation: The received data stream is temporarily stored in a buffer.

[0538] Output: The raw data stored in the buffer.

[0539] Step 3:

[0540] The server applies a noise reduction filter to the buffered data.

[0541] Input: Raw data stored in a buffer.

[0542] Data processing / calculation: Using OpenCV, unnecessary information such as environmental noise and sensor noise is filtered out.

[0543] Specific operation: Convert to grayscale and apply a filter such as GaussianBlur.

[0544] Output: Denoised data.

[0545] Step 4:

[0546] The server normalizes the denoised data.

[0547] Input: Denoised data.

[0548] Data processing / calculation: Scale the data to the range 0 to 1.

[0549] Specific operation: Normalization is performed using the maximum and minimum values ​​of the data.

[0550] Output: Normalized data.

[0551] Step 5:

[0552] The server inputs the normalized data into a deep learning model to perform image analysis.

[0553] Input: Normalized data.

[0554] Data processing / calculation: Image analysis is performed using a trained model using TensorFlow / Keras to detect suspicious individuals and abnormal behavior.

[0555] Specific operation: Analysis is performed using a neural network.

[0556] Output: Analysis results (such as suspicious person detection).

[0557] Step 6:

[0558] The server transmits the analysis results to the terminal.

[0559] Input: Image analysis results.

[0560] Data processing / calculation: Analysis results are sent to the terminal.

[0561] Specific operation: Send the result via HTTP request using Flask.

[0562] Output: The analysis results are sent to the terminal.

[0563] Step 7:

[0564] The device displays a notification to the user based on the received analysis results.

[0565] Input: Analysis results sent from the server.

[0566] Data processing / calculation: The received results are notified to the user in an easy-to-understand format.

[0567] Specific operation: Display an alert on your smartphone or smart glasses.

[0568] Output: A notification is displayed to the user.

[0569] In this way, the server, terminal, and user play their respective roles at each step, achieving highly accurate real-time image analysis.

[0570] 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.

[0571] The present invention is a system for seamlessly processing data acquired from a retinal sensor to perform image analysis and emotion recognition. Specifically, it includes a server, a terminal, and an emotion engine.

[0572] System configuration

[0573] 1. Data Acquisition

[0574] The server receives a continuous data stream from the retinal sensor, which stores the data in a buffer in real time.

[0575] Specific examples

[0576] The server continuously receives data from the retinal sensor every second and stores it in a buffer.

[0577] 2. Noise Reduction

[0578] The server retrieves the data stored in the buffer and applies a noise reduction filter to remove the noise components, resulting in clean data.

[0579] Specific examples

[0580] The server applies a noise reduction filter to the received data to remove environmental noise and sensor noise.

[0581] 3. Data normalization

[0582] The server normalizes the denoised data, scaling it to the range 0-1 and converting it into a form suitable for analytical models.

[0583] Specific examples

[0584] The server scales the denoised data based on the maximum and minimum values.

[0585] 4. Image Analysis

[0586] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model is composed of a multi-layered neural network, achieving highly accurate recognition.

[0587] Specific examples

[0588] The server inputs the normalized data into a deep learning model to detect faces and classify objects.

[0589] 5. Emotion recognition

[0590] The emotion engine uses the results of image analysis and additional sensor data to recognize the user's emotions. It uses deep learning models to analyze emotions from the user's facial expressions and voice.

[0591] Specific examples

[0592] The emotion engine uses deep learning models to analyze the user's facial expressions to determine joy, anger, sadness, and happiness.

[0593] 6. Send results

[0594] The server sends the results of image analysis and emotion recognition to the device. The analysis results are packaged in a predefined format and sent to the device in real time. The device then receives the analysis results and notifies the user.

[0595] Specific examples

[0596] The server sends the results of image analysis and emotion recognition to the device, which then displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0597] Usage example

[0598] Usage scenarios

[0599] The online education system analyzes students' facial expressions and voices in real time to understand their emotions. This allows instructors to understand the students' emotional state and adjust the educational content in real time. For example, if a student appears confused during a lecture, the data can be analyzed by another engine and a message will be sent to the instructor saying, "The student is confused." This allows the instructor to respond by providing additional explanations.

[0600] As described above, the present invention provides a system that uses a retinal sensor to realize real-time processing of continuous data streams, image analysis, and emotion recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[0601] The processing flow will be explained below.

[0602] Step 1:

[0603] The server receives a continuous data stream from the retinal sensor. The data generated by the sensor is sent in real time to the server, which stores it in a buffer where it is ready for the next processing step.

[0604] Step 2:

[0605] The server retrieves the buffered data and performs a denoising process, applying filtering algorithms to remove environmental and sensor noise. This clean data provides the basis for more accurate analysis.

[0606] Step 3:

[0607] The server then performs a normalization process on the denoised data, scaling it to a range of 0 to 1 and converting it into a form that is compatible with the analytical model. This normalization process allows the analytical model to function efficiently.

[0608] Step 4:

[0609] The server uses the normalized data to perform image analysis. It then uses deep learning models to analyze the data and perform object detection and classification. For example, the server uses deep learning models to detect human faces and classify objects.

[0610] Step 5:

[0611] The emotion engine uses image analysis results and additional sensor data to recognize the user's emotions. Deep learning models are applied to emotion recognition, analyzing the user's facial expressions and voice data. The analysis results are classified into emotion categories such as "happiness," "sadness," and "surprise."

[0612] Step 6:

[0613] The server sends the results of image analysis and emotion recognition to the device, which then packages the results in a predefined format and sends them to the device. The device processes the results and notifies the user.

[0614] Step 7:

[0615] The device displays the analysis results received from the server and notifies the user in a variety of notification formats, including visual alerts, audio notifications, and messages. The user can then take action based on the notification.

[0616] Step 8:

[0617] The user checks the display on the device and takes action as necessary. For example, in the case of a surveillance system, after receiving a notification, the user may take action such as zooming in on the area where suspicious activity was detected.

[0618] Through these steps, this system achieves real-time data processing, image analysis, and emotion recognition using a retinal sensor, enabling it to provide users with fast and accurate information. This overcomes the limitations of conventional frame-by-frame processing methods and enables more efficient operation.

[0619] Example 2

[0620] 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."

[0621] Conventional image analysis and emotion recognition systems face the challenge of processing the entire process, from data acquisition to analysis, recognition, and notification, in real time. This can lead to issues with reduced analysis accuracy, particularly when dealing with raw data containing large amounts of noise. Furthermore, there is a lack of effective means for quickly notifying users of analysis results. To address these issues, a system capable of high-precision data processing and analysis in real time is required.

[0622] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0623] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data in a buffer, means for applying a noise reduction filter to the data stored in the buffer, means for normalizing the noise-reduced data, means for performing image analysis using a deep learning model with the normalized data, means for performing emotion recognition using the image analysis results and additional sensor data, means for transmitting the analysis results and emotion recognition results to a terminal, and means for notifying the user of the results on the terminal. This enables noise reduction and data normalization to be performed in real time, achieving highly accurate image analysis and emotion recognition, and quickly notifying the user of the analysis results.

[0624] A "retinal sensor" is a sensor that detects light and outputs visual information as an electrical signal.

[0625] A "continuous data stream" is a continuous flow of data that is sent without interruption.

[0626] A "buffer" is a temporary storage area for data, and is an area used to adjust the difference between the data reception speed and the data processing speed.

[0627] A "noise removal filter" is an algorithm or hardware for removing unwanted noise components contained in data.

[0628] "Normalization" is the process of scaling data to a certain range (usually between 0 and 1).

[0629] A "deep learning model" is a machine learning model that uses a multi-layer neural network to learn the characteristics of data and perform analysis and prediction.

[0630] "Image analysis" is the process of detecting and classifying objects from image data, and is primarily performed using deep learning models.

[0631] "Emotion recognition" is the process of identifying a person's emotional state from image and audio data.

[0632] A "terminal" is an equipment or device that is directly operated by a user, and examples include smartphones and tablets.

[0633] MODE FOR CARRYING OUT THE INVENTION

[0634] The present invention provides a system for image analysis and emotion recognition that processes a continuous data stream acquired from a retinal sensor in real time. The system mainly includes a server, a terminal, and an emotion engine.

[0635] 1. Data Acquisition

[0636] The server receives the data sent from the retina sensor in real time and stores it in a buffer in memory, preventing data interruptions and ensuring a continuous data stream.

[0637] Specific examples

[0638] The server receives data from the retina sensor every second and stores it in a buffer.

[0639] 2. Noise Reduction

[0640] The server applies a noise reduction filter to the buffered data, removing sensor noise and environmental noise to obtain clean data.

[0641] Specific examples

[0642] The server applies a low-pass filter to the data in the buffer to remove unwanted high frequency components.

[0643] 3. Data normalization

[0644] The server normalizes the denoised data, specifically scaling the data values ​​to the range 0 to 1.

[0645] Specific examples

[0646] The server takes the maximum value of the data as 1 and the minimum value as 0 and scales each data point to fit into this range.

[0647] 4. Image Analysis

[0648] The server uses the normalized data to perform highly accurate image analysis using a deep learning model, which is built using a multi-layer neural network.

[0649] Specific examples

[0650] The server inputs the data into a neural network model built using a deep learning library (e.g., TensorFlow, PyTorch) to detect human faces and classify objects.

[0651] 5. Emotion recognition

[0652] The emotion engine uses a deep learning model to analyze the user's emotions based on the results of image analysis and additional sensor data, allowing it to identify emotions from the user's facial expressions and voice.

[0653] Specific examples

[0654] The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial expressions and voice data to identify emotions.

[0655] 6. Send results

[0656] The server sends the results of image analysis and emotion recognition to the device, which receives them and notifies the user of the results.

[0657] Specific examples

[0658] The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0659] Prompt Sentence Examples

[0660] Below are some examples of prompts to input to the generative AI model.

[0661] Example prompts

[0662] Please explain in detail how the system works, which continuously receives data from a retinal sensor and performs noise removal, data normalization, image analysis, and emotion recognition in real time.

[0663] The above is an embodiment of the system. This invention enables highly accurate data processing and analysis in real time, and allows analysis results to be quickly notified to the user.

[0664] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0665] Step 1:

[0666] The server receives a continuous data stream from the retinal sensor, which is buffered in real time.

[0667] Input: Raw data from the retinal sensor

[0668] Output: raw data stored in a buffer

[0669] Specific operation: The server receives data every second and immediately stores the data in a buffer to prevent data interruptions.

[0670] Step 2:

[0671] The server applies a noise reduction filter to the buffered data, which removes unwanted noise components.

[0672] Input: raw data stored in a buffer

[0673] Output: Clean data after noise removal

[0674] What it does: The server applies a low-pass filter to remove environmental noise and sensor noise, for example, a Butterworth filter to reduce noise in a specific frequency band.

[0675] Step 3:

[0676] The server normalizes the denoised data to a range of 0 to 1, converting it into a data format suitable for the analytical model.

[0677] Input: Clean data after noise removal

[0678] Output: Normalized data

[0679] What it does: The server scales each value of the data based on the maximum and minimum values. For example, if the original data ranges from 0 to 255, it divides each value by 255 to scale it to the range of 0 to 1.

[0680] Step 4:

[0681] The server runs a deep learning model on the normalized data to analyze the images, resulting in object detection and classification.

[0682] Input: Normalized data

[0683] Output: Image analysis results (e.g., object detection and classification results)

[0684] What it does: The server uses deep learning libraries (e.g., TensorFlow, PyTorch) to apply neural network models to perform image analysis, such as detecting human faces and classifying their attributes.

[0685] Step 5:

[0686] The emotion engine performs emotion recognition based on the results of image analysis and additional sensor data, making it possible to identify the user's emotions.

[0687] Input: Image analysis results, additional sensor data

[0688] Output: Emotion recognition results (e.g., joy, anger, sadness, and happiness)

[0689] Specific operation: The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial and voice data and identify emotions. For example, it determines that the user is "happy" based on their facial expression.

[0690] Step 6:

[0691] The server transmits the results of the image analysis and emotion recognition to the terminal, and the terminal notifies the user of the results.

[0692] Input: Image analysis results, emotion recognition results

[0693] Output: The result reported to the user

[0694] Specific operation: The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0695] (Application example 2)

[0696] 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."

[0697] Conventional factory robot systems have been unable to monitor worker emotions and fatigue levels in real time and provide appropriate feedback according to the situation. This has made it difficult to ensure worker safety and has left challenges in creating an efficient work environment. The present invention aims to solve this problem by providing a system that enables worker emotion recognition and fatigue level monitoring.

[0698] 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 receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis and emotion recognition using the normalized data, means for transmitting the analysis results and emotion recognition results to the terminal and providing feedback according to the user's situation based on the analysis results, means for continuously buffering preprocessed data, means for analyzing the data received from the retinal sensor in real time and monitoring the worker's emotion and fatigue level, and means for processing the received data stream in real time and notifying the worker to take a break based on the analysis results. This makes it possible to monitor the worker's emotion and fatigue level in real time and create a safe and efficient work environment.

[0699] A "retinal sensor" is a device that senses light, generates signals, and outputs a continuous data stream, mimicking the function of the human retina.

[0700] A "continuous data stream" refers to a data flow in which data is transmitted continuously rather than intermittently.

[0701] "Noise" refers to unnecessary components contained in sensor data that prevent accurate information from being obtained.

[0702] A "means for removing noise" is a method or device for removing unnecessary components from received data to obtain clean data.

[0703] A "means for normalizing data" is a method or device that scales data to a particular range and converts it into a form suitable for analysis.

[0704] "Image analysis" refers to the method or process of processing acquired image data to detect and classify objects.

[0705] "Emotion recognition" is a method of analyzing and identifying a user's emotional state using images and additional sensor data.

[0706] "Analysis results" refers to the information obtained as a result of data analysis, which is used to prompt decision-making and action.

[0707] "Feedback" refers to information or instructions provided to the user based on the analysis results, to help them maintain appropriate behavior or state.

[0708] "Terminal" refers to a computer or smart device that is directly operated by a user and is used to display analysis results and receive notifications.

[0709] A "buffer" is a storage device that temporarily stores data, allowing for smooth reception and processing of the data.

[0710] "Real-time" refers to data acquisition, processing, and feedback occurring immediately, without delay.

[0711] In this invention, a system is constructed that monitors the emotions and fatigue levels of factory workers in real time to provide a safe and efficient working environment. A detailed embodiment of this system will be described below.

[0712] The server first receives the data stream from the retinal sensor, which captures information such as the worker's facial expressions and gaze in real time. The captured data is temporarily stored in a buffer. A noise reduction filter is then applied to the stored data to remove unnecessary noise components, resulting in clean data.

[0713] The noise-removed data is then normalized to a range of 0 to 1. A deep learning model is then used to analyze the image and recognize emotions. The image analysis process detects the worker's face and facial expressions, and emotion recognition is performed based on this information.

[0714] The results of emotion recognition are sent from the server to the device in real time. Based on the analysis results, the device provides the worker with context-appropriate feedback. This feedback is displayed on the worker's smart glasses. For example, if the device determines that the worker is tired, a notification will be displayed encouraging them to take a break.

[0715] Specific examples of hardware and software used

[0716] Hardware: retinal sensor, smart glasses, server

[0717] Software: TensorFlow / Keras (for deep learning), OpenCV (for image processing), Flask (to build the API)

[0718] Specific examples

[0719] If a factory worker is operating a machine without expressing any emotion for a long period of time, the retinal sensor will capture this behavior and the server will analyze it in real time. If this analysis determines that the worker is fatigued, the system will display a notification on the smart glasses telling the worker to take a break.

[0720] Prompt Sentence Examples

[0721] Create a Python program that detects fatigue when a factory worker operates a machine without expressing any emotion for a long period of time and displays a notification encouraging them to take a break.

[0722] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0723] Step 1:

[0724] The server receives a continuous data stream from the retinal sensor. The input is the raw data from the retinal sensor, and the output is the received data stream. This data stream contains information such as the worker's facial expressions and gaze.

[0725] Step 2:

[0726] The server stores the received data stream in a buffer. The input is the data stream received in step 1, and the output is the data stored in the buffer. In this step, data is temporarily stored.

[0727] Step 3:

[0728] The server applies a noise reduction filter to the buffered data. The input is the raw data stored in the buffer, and the output is the clean, denoised data. This filtering process removes environmental noise and sensor noise.

[0729] Step 4:

[0730] The server normalizes the denoised data. The input is the denoised data and the output is the normalised data. The normalisation process scales the data to the range 0 to 1.

[0731] Step 5:

[0732] The server uses the normalized data to perform image analysis. The input is the normalized data, and the output is the result of image analysis. This image analysis uses a deep learning model to detect and classify objects.

[0733] Step 6:

[0734] The server performs emotion recognition based on the results of image analysis. The input is the result of image analysis, and the output is the result of emotion recognition. The type of emotion (joy, anger, sadness, happiness, etc.) and fatigue level are analyzed using a deep learning model.

[0735] Step 7:

[0736] The server sends the results of image analysis and emotion recognition to the terminal. The input is the emotion recognition result, and the output is the analysis result and emotion recognition result sent to the receiving terminal. This data is sent in real time.

[0737] Step 8:

[0738] The terminal provides feedback to the user based on the analysis results received from the server. The input is the analysis results and emotion recognition results sent from the server, and the output is a notification displayed to the user. As a specific example, a notification saying "Please take a break" is displayed on smart glasses. This notification allows the user (worker) to receive appropriate feedback.

[0739] 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.

[0740] 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.

[0741] 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.

[0742] [Third embodiment]

[0743] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0744] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0745] 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).

[0746] 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.

[0747] 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.

[0748] 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).

[0749] 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.

[0750] 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.

[0751] 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.

[0752] 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.

[0753] 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.

[0754] 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."

[0755] The present invention is a system that uses a retinal sensor to continuously acquire data and perform image analysis in real time. Specifically, the system includes a server, a terminal, and a means for communicating between them.

[0756] System configuration

[0757] 1. Data Acquisition

[0758] The server receives the data stream from the retinal sensor. The data output by the sensor is sent in real time, divided into multiple data packets rather than frame by frame. The server receives this data and stores it in a buffer for processing.

[0759] Specific examples

[0760] The server processes data received from the retinal sensor every second and stores it in a buffer.

[0761] 2. Noise Reduction

[0762] The server retrieves the buffered data and applies a noise reduction filter to it, eliminating environmental and sensor noise components to obtain clean data.

[0763] Specific examples

[0764] The server applies a noise reduction filter to the received data to smooth out highly variable pixel values.

[0765] 3. Data normalization

[0766] The server scales the denoised data to a range between 0 and 1. This normalization process puts the data in a form suitable for analytical models.

[0767] Specific examples

[0768] The server scales the denoised data based on the maximum and minimum values.

[0769] 4. Image Analysis

[0770] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model consists of a multi-layered neural network and uses pre-trained parameters to achieve high accuracy in recognition.

[0771] Specific examples

[0772] The server inputs the normalized data into a deep learning model to detect faces and recognize facial expressions.

[0773] 5. Send results

[0774] The server sends the analysis results to the terminal. The terminal receives them and notifies or displays them to the user. The notification content and display format are dynamically changed depending on the analysis results.

[0775] Specific examples

[0776] The server sends the analysis results to the device, which then displays a notification on the user's smartphone saying "A face has been detected."

[0777] Usage example

[0778] Usage scenarios

[0779] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The monitoring personnel (users) receive alerts from their devices and can take prompt action.

[0780] As described above, the present invention provides a system that realizes continuous data stream processing using a retinal sensor and supports highly efficient and highly accurate image recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[0781] The processing flow will be explained below.

[0782] Step 1:

[0783] The server receives a continuous data stream from the retinal sensor. The sensor sends data packets at regular intervals, which the server buffers in real time, ready for the next processing step.

[0784] Step 2:

[0785] The server retrieves the buffered data and performs a denoising process, applying a filtering algorithm to remove external environmental noise and sensor noise components. The denoised data is clean and suitable for subsequent normalization.

[0786] Step 3:

[0787] The server normalizes the denoised data, scaling it based on the minimum and maximum values ​​of the data to convert it into a range between 0 and 1. This normalization process optimizes the input data for the analytical model.

[0788] Step 4:

[0789] The server uses the normalized data to perform image analysis. The data is then fed into a deep learning model to perform object detection and classification. Specifically, a neural network analyzes the data using multiple layers of parameters to perform tasks such as face detection and facial expression recognition.

[0790] Step 5:

[0791] The server sends the analysis results to the terminal, which then packages the analysis results in a predefined format and sends them to the terminal through a communication channel, allowing the terminal to receive the analysis results in real time.

[0792] Step 6:

[0793] The device will notify the user of the analysis results it has received in various forms, including visual alerts, audio alerts, and messages, so that the user can take further action.

[0794] Step 7:

[0795] The user can check the display on the device and take action as needed. For example, in the case of a surveillance system, the user can take action such as zooming the camera to the area where the face was detected.

[0796] By going through this series of steps, this system achieves continuous, real-time data processing using a retinal sensor, enabling highly accurate image recognition.

[0797] Example 1

[0798] 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."

[0799] Conventional image analysis systems process data on a frame-by-frame basis, but this method makes real-time analysis difficult and can result in excessive delays. Furthermore, environmental noise and sensor noise components can reduce the accuracy of data analysis. The present invention aims to solve these problems and achieve high-precision image analysis in real time. Furthermore, the present invention aims to shorten response times in surveillance and security systems by quickly notifying users of analysis results.

[0800] 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.

[0801] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data stream in a buffer, means for removing noise from the data stored in the buffer, means for normalizing the data from which the noise has been removed, means for performing image analysis using a deep learning model using the normalized data, and means for transmitting the analysis results to the terminal, thereby enabling highly accurate data analysis in real time and prompt notification of the results.

[0802] A "retinal sensor" is a sensor device that receives light, converts it into an electrical signal, and outputs visual information as digital data.

[0803] A "data stream" is a continuously transmitted stream of signals or data.

[0804] A "buffer" is a memory area or storage that temporarily stores received data.

[0805] "Noise" and "noise removal filters" are algorithms or devices used to remove unwanted signals or noise components from data.

[0806] "Data normalization" is the process of converting data into a certain range, usually scaling it to the range 0 to 1.

[0807] A "deep learning model" is a machine learning model that uses a multi-layer neural network to analyze data.

[0808] "Image analysis" is the process of extracting information such as object detection and classification from image data.

[0809] "Analysis results" refers to information or data derived from deep learning models or other analytical means.

[0810] A "terminal" is a device that communicates with a server via a network and displays or notifies the results of analysis.

[0811] The present invention provides a system that uses a retinal sensor to continuously acquire data and perform real-time image analysis. The system includes a server, a terminal, and a means for communicating between them.

[0812] Data Acquisition

[0813] The server receives the data stream sent from the retinal sensor in real time. Specifically, the data continuously output by the sensor is divided into multiple data packets and received via the network. The server stores these data packets in a buffer. For example, the server receives approximately 1,000 packets of data per second and stores each packet sequentially in the buffer.

[0814] Noise Reduction

[0815] The server removes noise from the data stored in the buffer. By applying a noise reduction filter, environmental noise and sensor noise components are reduced. The server's processing unit passes 50% of the data in the buffer through the noise reduction filter every 10 ms. The filtered data has 80% less noise compared to the original data.

[0816] Data normalization

[0817] The server scales the noise-removed data to a range of 0 to 1. This normalization process makes the data suitable for analytical models. The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of the data normalization.

[0818] Image analysis

[0819] The server uses the normalized data to perform image analysis using a deep learning model. This model is composed of a multi-layer neural network and performs highly accurate analysis using parameters based on the training data. The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step. During this process, the model maintains an accuracy of over 95%.

[0820] Send results

[0821] The server sends the analysis results to the device. The device receives the analysis results and notifies or displays them appropriately to the user. The notification content and display format change dynamically depending on the analysis results. For example, the device receives the analysis results from the server every second and displays a pop-up notification on the user's smartphone saying "A face has been detected." The user can immediately check the notification and take additional action if necessary.

[0822] Usage example

[0823] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The person in charge of monitoring (user) receives an alert from the device and can respond promptly.

[0824] Example of input prompt for generative AI model

[0825] Please explain the process of using a retinal sensor in a surveillance system to acquire data in real time and perform image analysis. Please describe in detail each step from data acquisition to sending the results, dividing it into the roles of the server, device, and user.

[0826] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0827] Step 1:

[0828] The server receives a continuous data stream from the retinal sensor in real time by dividing the data continuously output by the sensor into multiple data packets and receiving them over the network.

[0829] Input: Data packet from retinal sensor

[0830] Output: Buffered data packets

[0831] Specific operation: The server receives approximately 1,000 packets of data per second and stores each packet sequentially in a buffer.

[0832] Step 2:

[0833] The server applies a noise reduction filter to remove noise from the buffered data, which reduces environmental noise and sensor noise components.

[0834] Input: raw data stored in a buffer

[0835] Output: Clean data with noise removed

[0836] How it works: Every 10 ms, the server's processing unit passes 50% of the data in the buffer through a noise reduction filter. The filtered data has 80% less noise than the original data.

[0837] Step 3:

[0838] The server scales the denoised data to a range between 0 and 1. The normalization process involves scaling each data point based on the maximum and minimum values ​​of the data.

[0839] Input: Clean data with noise removed

[0840] Output: Normalized data

[0841] Specific operation: The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of data normalization.

[0842] Step 4:

[0843] The server uses the normalized data to perform image analysis using a deep learning model, which is composed of a multi-layer neural network and uses parameters based on the training data to perform highly accurate analysis.

[0844] Input: Normalized data

[0845] Output: Image analysis results

[0846] Specific operation: The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step, maintaining the model accuracy of over 95% during this process.

[0847] Step 5:

[0848] The server sends the results of the image analysis to the device. The device receives the results and notifies or displays the appropriate information to the user. The notification content and display format are dynamically changed depending on the analysis results.

[0849] Input: Image analysis results

[0850] Output: Information reported to the user

[0851] Specific operation: The device receives the analysis results from the server every second and displays a "Face detected" notification on the user's smartphone. The user can immediately check the notification and take additional action if necessary.

[0852] (Application example 1)

[0853] 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."

[0854] Conventional real-time image analysis systems often lack the efficiency and accuracy required for continuous data processing and noise removal, resulting in delays and false positives when detecting suspicious individuals, particularly in the security field. Furthermore, when notifying a device of analysis results, the system lacks the functionality to respond immediately, requiring a fast and accurate response.

[0855] 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.

[0856] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis using the normalized data, means for transmitting the analysis results to the terminal, means for generating a security notification based on the analysis results and displaying it on the terminal, and means for detecting suspicious activity based on the image analysis results in real time, thereby enabling highly efficient and highly accurate security monitoring.

[0857] A "retinal sensor" is a sensor that continuously acquires video data, unlike conventional frame-by-frame imaging methods, and generates a data stream that can be analyzed in real time.

[0858] A "continuous data stream" is a data format in which data is transmitted continuously, without interruption, rather than in specific frames.

[0859] "Noise removal" is the process of removing unnecessary environmental noise and sensor noise components from acquired data to obtain clean data.

[0860] "Data normalization" is the process of scaling noise-removed data to a range of 0 to 1 and converting it into a format suitable for analytical models.

[0861] "Image analysis" is the process of using data to detect and classify objects using algorithms such as deep learning models.

[0862] A "security notification" is a notification that notifies the user when suspicious activity or anomalies are detected based on the analysis results.

[0863] A "terminal" refers to an electronic device such as a smartphone or smart glasses that receives the analysis results and notifies the user.

[0864] "Real-time" refers to data acquisition and processing occurring immediately, with little or no delay.

[0865] "Detecting suspicious activity" is the process of identifying suspicious behavior or individuals based on the results of image analysis and providing information for responding.

[0866] The system for realizing this invention is composed of a retinal sensor, a server, and a terminal. The specific operation of each component and its program are described in detail below.

[0867] Hardware and Software Use Cases

[0868] This system uses the following hardware and software:

[0869] Hardware

[0870] Retinal sensor: A sensor that continuously acquires video data and generates a data stream that can be analyzed in real time.

[0871] Smartphone or smart glasses: A mobile device carried by the user.

[0872] Server: Receives data, processes it, and sends the analysis results.

[0873] software

[0874] Python: An entire programming language.

[0875] OpenCV: An image processing library, used for denoising and preprocessing data.

[0876] TensorFlow / Keras: Deep learning libraries used for image analysis models.

[0877] Flask: A web framework used for communication between the server and the terminal and for notifying results.

[0878] Data processing and calculation flow

[0879] The server starts by receiving a continuous data stream from the retinal sensor. The received data is first stored in a buffer. Next, noise is removed from the data using OpenCV to obtain clean data. The noise-removed data is normalized to a range of 0 to 1, and image analysis is performed using a deep learning model. If suspicious activity is detected as a result of the analysis, the results are sent to the device. This series of processes is performed in real time.

[0880] The device receives the analysis results from the server and generates and displays security notifications to the user as needed, such as a "Suspicious person detected" message on a smartphone or smart glasses.

[0881] Specific examples of processing

[0882] This system is used in a scenario where office entrances are monitored. When a retinal sensor installed in the smart glasses detects a suspicious person or movement, the data is sent to a server in real time. The server processes the data and notifies the user of the analysis results on their smartphone. This allows the user to take prompt action even when they are in a remote location.

[0883] Prompt Sentence Examples

[0884] "Implement a security application that uses a retinal sensor to monitor the surroundings in real time with high accuracy and detect suspicious individuals. The application should be installable on smartphones or smart glasses, using Python, OpenCV, TensorFlow, and Flask."

[0885] In this way, the present invention realizes continuous data stream processing using a retinal sensor, and can provide highly efficient and highly accurate security monitoring.

[0886] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0887] Step 1:

[0888] A continuous data stream is acquired from the retinal sensor.

[0889] Input: Raw data stream from the retinal sensor.

[0890] Data processing / calculation: The image data captured by the sensor is acquired continuously, rather than frame by frame.

[0891] Output: A continuous data stream.

[0892] Step 2:

[0893] The server receives the data stream and stores it in a buffer.

[0894] Input: A continuous data stream.

[0895] Data processing / calculation: The received data stream is temporarily stored in a buffer.

[0896] Output: The raw data stored in the buffer.

[0897] Step 3:

[0898] The server applies a noise reduction filter to the buffered data.

[0899] Input: Raw data stored in a buffer.

[0900] Data processing / calculation: Using OpenCV, unnecessary information such as environmental noise and sensor noise is filtered out.

[0901] Specific operation: Convert to grayscale and apply a filter such as GaussianBlur.

[0902] Output: Denoised data.

[0903] Step 4:

[0904] The server normalizes the denoised data.

[0905] Input: Denoised data.

[0906] Data processing / calculation: Scale the data to the range 0 to 1.

[0907] Specific operation: Normalization is performed using the maximum and minimum values ​​of the data.

[0908] Output: Normalized data.

[0909] Step 5:

[0910] The server inputs the normalized data into a deep learning model to perform image analysis.

[0911] Input: Normalized data.

[0912] Data processing / calculation: Image analysis is performed using a trained model using TensorFlow / Keras to detect suspicious individuals and abnormal behavior.

[0913] Specific operation: Analysis is performed using a neural network.

[0914] Output: Analysis results (such as suspicious person detection).

[0915] Step 6:

[0916] The server transmits the analysis results to the terminal.

[0917] Input: Image analysis results.

[0918] Data processing / calculation: Analysis results are sent to the terminal.

[0919] Specific operation: Send the result via HTTP request using Flask.

[0920] Output: The analysis results are sent to the terminal.

[0921] Step 7:

[0922] The device displays a notification to the user based on the received analysis results.

[0923] Input: Analysis results sent from the server.

[0924] Data processing / calculation: The received results are notified to the user in an easy-to-understand format.

[0925] Specific operation: Display an alert on your smartphone or smart glasses.

[0926] Output: A notification is displayed to the user.

[0927] In this way, the server, terminal, and user play their respective roles at each step, achieving highly accurate real-time image analysis.

[0928] 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.

[0929] The present invention is a system for seamlessly processing data acquired from a retinal sensor to perform image analysis and emotion recognition. Specifically, it includes a server, a terminal, and an emotion engine.

[0930] System configuration

[0931] 1. Data Acquisition

[0932] The server receives a continuous data stream from the retinal sensor, which stores the data in a buffer in real time.

[0933] Specific examples

[0934] The server continuously receives data from the retinal sensor every second and stores it in a buffer.

[0935] 2. Noise Reduction

[0936] The server retrieves the data stored in the buffer and applies a noise reduction filter to remove the noise components, resulting in clean data.

[0937] Specific examples

[0938] The server applies a noise reduction filter to the received data to remove environmental noise and sensor noise.

[0939] 3. Data normalization

[0940] The server normalizes the denoised data, scaling it to the range 0-1 and converting it into a form suitable for analytical models.

[0941] Specific examples

[0942] The server scales the denoised data based on the maximum and minimum values.

[0943] 4. Image Analysis

[0944] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model is composed of a multi-layered neural network, achieving highly accurate recognition.

[0945] Specific examples

[0946] The server inputs the normalized data into a deep learning model to detect faces and classify objects.

[0947] 5. Emotion recognition

[0948] The emotion engine uses the results of image analysis and additional sensor data to recognize the user's emotions. It uses deep learning models to analyze emotions from the user's facial expressions and voice.

[0949] Specific examples

[0950] The emotion engine uses deep learning models to analyze the user's facial expressions to determine joy, anger, sadness, and happiness.

[0951] 6. Send results

[0952] The server sends the results of image analysis and emotion recognition to the device. The analysis results are packaged in a predefined format and sent to the device in real time. The device then receives the analysis results and notifies the user.

[0953] Specific examples

[0954] The server sends the results of image analysis and emotion recognition to the device, which then displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[0955] Usage example

[0956] Usage scenarios

[0957] The online education system analyzes students' facial expressions and voices in real time to understand their emotions. This allows instructors to understand the students' emotional state and adjust the educational content in real time. For example, if a student appears confused during a lecture, the data can be analyzed by another engine and a message will be sent to the instructor saying, "The student is confused." This allows the instructor to respond by providing additional explanations.

[0958] As described above, the present invention provides a system that uses a retinal sensor to realize real-time processing of continuous data streams, image analysis, and emotion recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[0959] The processing flow will be explained below.

[0960] Step 1:

[0961] The server receives a continuous data stream from the retinal sensor. The data generated by the sensor is sent in real time to the server, which stores it in a buffer where it is ready for the next processing step.

[0962] Step 2:

[0963] The server retrieves the buffered data and performs a denoising process, applying filtering algorithms to remove environmental and sensor noise. This clean data provides the basis for more accurate analysis.

[0964] Step 3:

[0965] The server then performs a normalization process on the denoised data, scaling it to a range of 0 to 1 and converting it into a form that is compatible with the analytical model. This normalization process allows the analytical model to function efficiently.

[0966] Step 4:

[0967] The server uses the normalized data to perform image analysis. It then uses deep learning models to analyze the data and perform object detection and classification. For example, the server uses deep learning models to detect human faces and classify objects.

[0968] Step 5:

[0969] The emotion engine uses image analysis results and additional sensor data to recognize the user's emotions. Deep learning models are applied to emotion recognition, analyzing the user's facial expressions and voice data. The analysis results are classified into emotion categories such as "happiness," "sadness," and "surprise."

[0970] Step 6:

[0971] The server sends the results of image analysis and emotion recognition to the device, which then packages the results in a predefined format and sends them to the device. The device processes the results and notifies the user.

[0972] Step 7:

[0973] The device displays the analysis results received from the server and notifies the user in a variety of notification formats, including visual alerts, audio notifications, and messages. The user can then take action based on the notification.

[0974] Step 8:

[0975] The user checks the display on the device and takes action as necessary. For example, in the case of a surveillance system, after receiving a notification, the user may take action such as zooming in on the area where suspicious activity was detected.

[0976] Through these steps, this system achieves real-time data processing, image analysis, and emotion recognition using a retinal sensor, enabling it to provide users with fast and accurate information. This overcomes the limitations of conventional frame-by-frame processing methods and enables more efficient operation.

[0977] Example 2

[0978] 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."

[0979] Conventional image analysis and emotion recognition systems face the challenge of processing the entire process, from data acquisition to analysis, recognition, and notification, in real time. This can lead to issues with reduced analysis accuracy, particularly when dealing with raw data containing large amounts of noise. Furthermore, there is a lack of effective means for quickly notifying users of analysis results. To address these issues, a system capable of high-precision data processing and analysis in real time is required.

[0980] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0981] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data in a buffer, means for applying a noise reduction filter to the data stored in the buffer, means for normalizing the noise-reduced data, means for performing image analysis using a deep learning model with the normalized data, means for performing emotion recognition using the image analysis results and additional sensor data, means for transmitting the analysis results and emotion recognition results to a terminal, and means for notifying the user of the results on the terminal. This enables noise reduction and data normalization to be performed in real time, achieving highly accurate image analysis and emotion recognition, and quickly notifying the user of the analysis results.

[0982] A "retinal sensor" is a sensor that detects light and outputs visual information as an electrical signal.

[0983] A "continuous data stream" is a continuous flow of data that is sent without interruption.

[0984] A "buffer" is a temporary storage area for data, and is an area used to adjust the difference between the data reception speed and the data processing speed.

[0985] A "noise removal filter" is an algorithm or hardware for removing unwanted noise components contained in data.

[0986] "Normalization" is the process of scaling data to a certain range (usually between 0 and 1).

[0987] A "deep learning model" is a machine learning model that uses a multi-layer neural network to learn the characteristics of data and perform analysis and prediction.

[0988] "Image analysis" is the process of detecting and classifying objects from image data, and is primarily performed using deep learning models.

[0989] "Emotion recognition" is the process of identifying a person's emotional state from image and audio data.

[0990] A "terminal" is an equipment or device that is directly operated by a user, and examples include smartphones and tablets.

[0991] MODE FOR CARRYING OUT THE INVENTION

[0992] The present invention provides a system for image analysis and emotion recognition that processes a continuous data stream acquired from a retinal sensor in real time. The system mainly includes a server, a terminal, and an emotion engine.

[0993] 1. Data Acquisition

[0994] The server receives the data sent from the retina sensor in real time and stores it in a buffer in memory, preventing data interruptions and ensuring a continuous data stream.

[0995] Specific examples

[0996] The server receives data from the retina sensor every second and stores it in a buffer.

[0997] 2. Noise Reduction

[0998] The server applies a noise reduction filter to the buffered data, removing sensor noise and environmental noise to obtain clean data.

[0999] Specific examples

[1000] The server applies a low-pass filter to the data in the buffer to remove unwanted high frequency components.

[1001] 3. Data normalization

[1002] The server normalizes the denoised data, specifically scaling the data values ​​to the range 0 to 1.

[1003] Specific examples

[1004] The server takes the maximum value of the data as 1 and the minimum value as 0 and scales each data point to fit into this range.

[1005] 4. Image Analysis

[1006] The server uses the normalized data to perform highly accurate image analysis using a deep learning model, which is built using a multi-layer neural network.

[1007] Specific examples

[1008] The server inputs the data into a neural network model built using a deep learning library (e.g., TensorFlow, PyTorch) to detect human faces and classify objects.

[1009] 5. Emotion recognition

[1010] The emotion engine uses a deep learning model to analyze the user's emotions based on the results of image analysis and additional sensor data, allowing it to identify emotions from the user's facial expressions and voice.

[1011] Specific examples

[1012] The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial expressions and voice data to identify emotions.

[1013] 6. Send results

[1014] The server sends the results of image analysis and emotion recognition to the device, which receives them and notifies the user of the results.

[1015] Specific examples

[1016] The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[1017] Prompt Sentence Examples

[1018] Below are some examples of prompts to input to the generative AI model.

[1019] Example prompts

[1020] Please explain in detail how the system works, which continuously receives data from a retinal sensor and performs noise removal, data normalization, image analysis, and emotion recognition in real time.

[1021] The above is an embodiment of the system. This invention enables highly accurate data processing and analysis in real time, and allows analysis results to be quickly notified to the user.

[1022] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1023] Step 1:

[1024] The server receives a continuous data stream from the retinal sensor, which is buffered in real time.

[1025] Input: Raw data from the retinal sensor

[1026] Output: raw data stored in a buffer

[1027] Specific operation: The server receives data every second and immediately stores the data in a buffer to prevent data interruptions.

[1028] Step 2:

[1029] The server applies a noise reduction filter to the buffered data, which removes unwanted noise components.

[1030] Input: raw data stored in a buffer

[1031] Output: Clean data after noise removal

[1032] What it does: The server applies a low-pass filter to remove environmental noise and sensor noise, for example, a Butterworth filter to reduce noise in a specific frequency band.

[1033] Step 3:

[1034] The server normalizes the denoised data to a range of 0 to 1, converting it into a data format suitable for the analytical model.

[1035] Input: Clean data after noise removal

[1036] Output: Normalized data

[1037] What it does: The server scales each value of the data based on the maximum and minimum values. For example, if the original data ranges from 0 to 255, it divides each value by 255 to scale it to the range of 0 to 1.

[1038] Step 4:

[1039] The server runs a deep learning model on the normalized data to analyze the images, resulting in object detection and classification.

[1040] Input: Normalized data

[1041] Output: Image analysis results (e.g., object detection and classification results)

[1042] What it does: The server uses deep learning libraries (e.g., TensorFlow, PyTorch) to apply neural network models to perform image analysis, such as detecting human faces and classifying their attributes.

[1043] Step 5:

[1044] The emotion engine performs emotion recognition based on the results of image analysis and additional sensor data, making it possible to identify the user's emotions.

[1045] Input: Image analysis results, additional sensor data

[1046] Output: Emotion recognition results (e.g., joy, anger, sadness, and happiness)

[1047] Specific operation: The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial and voice data and identify emotions. For example, it determines that the user is "happy" based on their facial expression.

[1048] Step 6:

[1049] The server transmits the results of the image analysis and emotion recognition to the terminal, and the terminal notifies the user of the results.

[1050] Input: Image analysis results, emotion recognition results

[1051] Output: The result reported to the user

[1052] Specific operation: The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[1053] (Application example 2)

[1054] 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."

[1055] Conventional factory robot systems have been unable to monitor worker emotions and fatigue levels in real time and provide appropriate feedback according to the situation. This has made it difficult to ensure worker safety and has left challenges in creating an efficient work environment. The present invention aims to solve this problem by providing a system that enables worker emotion recognition and fatigue level monitoring.

[1056] 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 receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis and emotion recognition using the normalized data, means for transmitting the analysis results and emotion recognition results to the terminal and providing feedback according to the user's situation based on the analysis results, means for continuously buffering preprocessed data, means for analyzing the data received from the retinal sensor in real time and monitoring the worker's emotion and fatigue level, and means for processing the received data stream in real time and notifying the worker to take a break based on the analysis results. This makes it possible to monitor the worker's emotion and fatigue level in real time and create a safe and efficient work environment.

[1057] A "retinal sensor" is a device that senses light, generates signals, and outputs a continuous data stream, mimicking the function of the human retina.

[1058] A "continuous data stream" refers to a data flow in which data is transmitted continuously rather than intermittently.

[1059] "Noise" refers to unnecessary components contained in sensor data that prevent accurate information from being obtained.

[1060] A "means for removing noise" is a method or device for removing unnecessary components from received data to obtain clean data.

[1061] A "means for normalizing data" is a method or device that scales data to a particular range and converts it into a form suitable for analysis.

[1062] "Image analysis" refers to the method or process of processing acquired image data to detect and classify objects.

[1063] "Emotion recognition" is a method of analyzing and identifying a user's emotional state using images and additional sensor data.

[1064] "Analysis results" refers to the information obtained as a result of data analysis, which is used to prompt decision-making and action.

[1065] "Feedback" refers to information or instructions provided to the user based on the analysis results, to help them maintain appropriate behavior or state.

[1066] "Terminal" refers to a computer or smart device that is directly operated by a user and is used to display analysis results and receive notifications.

[1067] A "buffer" is a storage device that temporarily stores data, allowing for smooth reception and processing of the data.

[1068] "Real-time" refers to data acquisition, processing, and feedback occurring immediately, without delay.

[1069] In this invention, a system is constructed that monitors the emotions and fatigue levels of factory workers in real time to provide a safe and efficient working environment. A detailed embodiment of this system will be described below.

[1070] The server first receives the data stream from the retinal sensor, which captures information such as the worker's facial expressions and gaze in real time. The captured data is temporarily stored in a buffer. A noise reduction filter is then applied to the stored data to remove unnecessary noise components, resulting in clean data.

[1071] The noise-removed data is then normalized to a range of 0 to 1. A deep learning model is then used to analyze the image and recognize emotions. The image analysis process detects the worker's face and facial expressions, and emotion recognition is performed based on this information.

[1072] The results of emotion recognition are sent from the server to the device in real time. Based on the analysis results, the device provides the worker with context-appropriate feedback. This feedback is displayed on the worker's smart glasses. For example, if the device determines that the worker is tired, a notification will be displayed encouraging them to take a break.

[1073] Specific examples of hardware and software used

[1074] Hardware: retinal sensor, smart glasses, server

[1075] Software: TensorFlow / Keras (for deep learning), OpenCV (for image processing), Flask (to build the API)

[1076] Specific examples

[1077] If a factory worker is operating a machine without expressing any emotion for a long period of time, the retinal sensor will capture this behavior and the server will analyze it in real time. If this analysis determines that the worker is fatigued, the system will display a notification on the smart glasses telling the worker to take a break.

[1078] Prompt Sentence Examples

[1079] Create a Python program that detects fatigue when a factory worker operates a machine without expressing any emotion for a long period of time and displays a notification encouraging them to take a break.

[1080] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1081] Step 1:

[1082] The server receives a continuous data stream from the retinal sensor. The input is the raw data from the retinal sensor, and the output is the received data stream. This data stream contains information such as the worker's facial expressions and gaze.

[1083] Step 2:

[1084] The server stores the received data stream in a buffer. The input is the data stream received in step 1, and the output is the data stored in the buffer. In this step, data is temporarily stored.

[1085] Step 3:

[1086] The server applies a noise reduction filter to the buffered data. The input is the raw data stored in the buffer, and the output is the clean, denoised data. This filtering process removes environmental noise and sensor noise.

[1087] Step 4:

[1088] The server normalizes the denoised data. The input is the denoised data and the output is the normalised data. The normalisation process scales the data to the range 0 to 1.

[1089] Step 5:

[1090] The server uses the normalized data to perform image analysis. The input is the normalized data, and the output is the result of image analysis. This image analysis uses a deep learning model to detect and classify objects.

[1091] Step 6:

[1092] The server performs emotion recognition based on the results of image analysis. The input is the result of image analysis, and the output is the result of emotion recognition. The type of emotion (joy, anger, sadness, happiness, etc.) and fatigue level are analyzed using a deep learning model.

[1093] Step 7:

[1094] The server sends the results of image analysis and emotion recognition to the terminal. The input is the emotion recognition result, and the output is the analysis result and emotion recognition result sent to the receiving terminal. This data is sent in real time.

[1095] Step 8:

[1096] The terminal provides feedback to the user based on the analysis results received from the server. The input is the analysis results and emotion recognition results sent from the server, and the output is a notification displayed to the user. As a specific example, a notification saying "Please take a break" is displayed on smart glasses. This notification allows the user (worker) to receive appropriate feedback.

[1097] 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.

[1098] 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.

[1099] 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.

[1100] [Fourth embodiment]

[1101] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1102] 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.

[1103] 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).

[1104] 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.

[1105] 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.

[1106] 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).

[1107] 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.

[1108] 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.

[1109] 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.

[1110] 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.

[1111] 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.

[1112] 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.

[1113] 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."

[1114] The present invention is a system that uses a retinal sensor to continuously acquire data and perform image analysis in real time. Specifically, the system includes a server, a terminal, and a means for communicating between them.

[1115] System configuration

[1116] 1. Data Acquisition

[1117] The server receives the data stream from the retinal sensor. The data output by the sensor is sent in real time, divided into multiple data packets rather than frame by frame. The server receives this data and stores it in a buffer for processing.

[1118] Specific examples

[1119] The server processes data received from the retinal sensor every second and stores it in a buffer.

[1120] 2. Noise Reduction

[1121] The server retrieves the buffered data and applies a noise reduction filter to it, eliminating environmental and sensor noise components to obtain clean data.

[1122] Specific examples

[1123] The server applies a noise reduction filter to the received data to smooth out highly variable pixel values.

[1124] 3. Data normalization

[1125] The server scales the denoised data to a range between 0 and 1. This normalization process puts the data in a form suitable for analytical models.

[1126] Specific examples

[1127] The server scales the denoised data based on the maximum and minimum values.

[1128] 4. Image Analysis

[1129] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model consists of a multi-layered neural network and uses pre-trained parameters to achieve high accuracy in recognition.

[1130] Specific examples

[1131] The server inputs the normalized data into a deep learning model to detect faces and recognize facial expressions.

[1132] 5. Send results

[1133] The server sends the analysis results to the terminal. The terminal receives them and notifies or displays them to the user. The notification content and display format are dynamically changed depending on the analysis results.

[1134] Specific examples

[1135] The server sends the analysis results to the device, which then displays a notification on the user's smartphone saying "A face has been detected."

[1136] Usage example

[1137] Usage scenarios

[1138] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The monitoring personnel (users) receive alerts from their devices and can take prompt action.

[1139] As described above, the present invention provides a system that realizes continuous data stream processing using a retinal sensor and supports highly efficient and highly accurate image recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[1140] The processing flow will be explained below.

[1141] Step 1:

[1142] The server receives a continuous data stream from the retinal sensor. The sensor sends data packets at regular intervals, which the server buffers in real time, ready for the next processing step.

[1143] Step 2:

[1144] The server retrieves the buffered data and performs a denoising process, applying a filtering algorithm to remove external environmental noise and sensor noise components. The denoised data is clean and suitable for subsequent normalization.

[1145] Step 3:

[1146] The server normalizes the denoised data, scaling it based on the minimum and maximum values ​​of the data to convert it into a range between 0 and 1. This normalization process optimizes the input data for the analytical model.

[1147] Step 4:

[1148] The server uses the normalized data to perform image analysis. The data is then fed into a deep learning model to perform object detection and classification. Specifically, a neural network analyzes the data using multiple layers of parameters to perform tasks such as face detection and facial expression recognition.

[1149] Step 5:

[1150] The server sends the analysis results to the terminal, which then packages the analysis results in a predefined format and sends them to the terminal through a communication channel, allowing the terminal to receive the analysis results in real time.

[1151] Step 6:

[1152] The device will notify the user of the analysis results it has received in various forms, including visual alerts, audio alerts, and messages, so that the user can take further action.

[1153] Step 7:

[1154] The user can check the display on the device and take action as needed. For example, in the case of a surveillance system, the user can take action such as zooming the camera to the area where the face was detected.

[1155] By going through this series of steps, this system achieves continuous, real-time data processing using a retinal sensor, enabling highly accurate image recognition.

[1156] Example 1

[1157] 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."

[1158] Conventional image analysis systems process data on a frame-by-frame basis, but this method makes real-time analysis difficult and can result in excessive delays. Furthermore, environmental noise and sensor noise components can reduce the accuracy of data analysis. The present invention aims to solve these problems and achieve high-precision image analysis in real time. Furthermore, the present invention aims to shorten response times in surveillance and security systems by quickly notifying users of analysis results.

[1159] 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.

[1160] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data stream in a buffer, means for removing noise from the data stored in the buffer, means for normalizing the data from which the noise has been removed, means for performing image analysis using a deep learning model using the normalized data, and means for transmitting the analysis results to the terminal, thereby enabling highly accurate data analysis in real time and prompt notification of the results.

[1161] A "retinal sensor" is a sensor device that receives light, converts it into an electrical signal, and outputs visual information as digital data.

[1162] A "data stream" is a continuously transmitted stream of signals or data.

[1163] A "buffer" is a memory area or storage that temporarily stores received data.

[1164] "Noise" and "noise removal filters" are algorithms or devices used to remove unwanted signals or noise components from data.

[1165] "Data normalization" is the process of converting data into a certain range, usually scaling it to the range 0 to 1.

[1166] A "deep learning model" is a machine learning model that uses a multi-layer neural network to analyze data.

[1167] "Image analysis" is the process of extracting information such as object detection and classification from image data.

[1168] "Analysis results" refers to information or data derived from deep learning models or other analytical means.

[1169] A "terminal" is a device that communicates with a server via a network and displays or notifies the results of analysis.

[1170] The present invention provides a system that uses a retinal sensor to continuously acquire data and perform real-time image analysis. The system includes a server, a terminal, and a means for communicating between them.

[1171] Data Acquisition

[1172] The server receives the data stream sent from the retinal sensor in real time. Specifically, the data continuously output by the sensor is divided into multiple data packets and received via the network. The server stores these data packets in a buffer. For example, the server receives approximately 1,000 packets of data per second and stores each packet sequentially in the buffer.

[1173] Noise Reduction

[1174] The server removes noise from the data stored in the buffer. By applying a noise reduction filter, environmental noise and sensor noise components are reduced. The server's processing unit passes 50% of the data in the buffer through the noise reduction filter every 10 ms. The filtered data has 80% less noise compared to the original data.

[1175] Data normalization

[1176] The server scales the noise-removed data to a range of 0 to 1. This normalization process makes the data suitable for analytical models. The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of the data normalization.

[1177] Image analysis

[1178] The server uses the normalized data to perform image analysis using a deep learning model. This model is composed of a multi-layer neural network and performs highly accurate analysis using parameters based on the training data. The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step. During this process, the model maintains an accuracy of over 95%.

[1179] Send results

[1180] The server sends the analysis results to the device. The device receives the analysis results and notifies or displays them appropriately to the user. The notification content and display format change dynamically depending on the analysis results. For example, the device receives the analysis results from the server every second and displays a pop-up notification on the user's smartphone saying "A face has been detected." The user can immediately check the notification and take additional action if necessary.

[1181] Usage example

[1182] One surveillance system uses a retinal sensor to monitor the surroundings in real time. The images captured by the sensor are sent to a server, which processes them in the following order: noise removal, data normalization, and image analysis. If any suspicious activity is detected, the results are sent to the relevant device. The person in charge of monitoring (user) receives an alert from the device and can respond promptly.

[1183] Example of input prompt for generative AI model

[1184] Please explain the process of using a retinal sensor in a surveillance system to acquire data in real time and perform image analysis. Please describe in detail each step from data acquisition to sending the results, dividing it into the roles of the server, device, and user.

[1185] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1186] Step 1:

[1187] The server receives a continuous data stream from the retinal sensor in real time by dividing the data continuously output by the sensor into multiple data packets and receiving them over the network.

[1188] Input: Data packet from retinal sensor

[1189] Output: Buffered data packets

[1190] Specific operation: The server receives approximately 1,000 packets of data per second and stores each packet sequentially in a buffer.

[1191] Step 2:

[1192] The server applies a noise reduction filter to remove noise from the buffered data, which reduces environmental noise and sensor noise components.

[1193] Input: raw data stored in a buffer

[1194] Output: Clean data with noise removed

[1195] How it works: Every 10 ms, the server's processing unit passes 50% of the data in the buffer through a noise reduction filter. The filtered data has 80% less noise than the original data.

[1196] Step 3:

[1197] The server scales the denoised data to a range between 0 and 1. The normalization process involves scaling each data point based on the maximum and minimum values ​​of the data.

[1198] Input: Clean data with noise removed

[1199] Output: Normalized data

[1200] Specific operation: The server updates the maximum and minimum values ​​for scaling every second to maintain the accuracy of data normalization.

[1201] Step 4:

[1202] The server uses the normalized data to perform image analysis using a deep learning model, which is composed of a multi-layer neural network and uses parameters based on the training data to perform highly accurate analysis.

[1203] Input: Normalized data

[1204] Output: Image analysis results

[1205] Specific operation: The server analyzes one frame of image data in an average of 0.02 seconds and passes the analysis results to the next step, maintaining the model accuracy of over 95% during this process.

[1206] Step 5:

[1207] The server sends the results of the image analysis to the device. The device receives the results and notifies or displays the appropriate information to the user. The notification content and display format are dynamically changed depending on the analysis results.

[1208] Input: Image analysis results

[1209] Output: Information reported to the user

[1210] Specific operation: The device receives the analysis results from the server every second and displays a "Face detected" notification on the user's smartphone. The user can immediately check the notification and take additional action if necessary.

[1211] (Application example 1)

[1212] 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."

[1213] Conventional real-time image analysis systems often lack the efficiency and accuracy required for continuous data processing and noise removal, resulting in delays and false positives when detecting suspicious individuals, particularly in the security field. Furthermore, when notifying a device of analysis results, the system lacks the functionality to respond immediately, requiring a fast and accurate response.

[1214] 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.

[1215] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis using the normalized data, means for transmitting the analysis results to the terminal, means for generating a security notification based on the analysis results and displaying it on the terminal, and means for detecting suspicious activity based on the image analysis results in real time, thereby enabling highly efficient and highly accurate security monitoring.

[1216] A "retinal sensor" is a sensor that continuously acquires video data, unlike conventional frame-by-frame imaging methods, and generates a data stream that can be analyzed in real time.

[1217] A "continuous data stream" is a data format in which data is transmitted continuously, without interruption, rather than in specific frames.

[1218] "Noise removal" is the process of removing unnecessary environmental noise and sensor noise components from acquired data to obtain clean data.

[1219] "Data normalization" is the process of scaling noise-removed data to a range of 0 to 1 and converting it into a format suitable for analytical models.

[1220] "Image analysis" is the process of using data to detect and classify objects using algorithms such as deep learning models.

[1221] A "security notification" is a notification that notifies the user when suspicious activity or anomalies are detected based on the analysis results.

[1222] A "terminal" refers to an electronic device such as a smartphone or smart glasses that receives the analysis results and notifies the user.

[1223] "Real-time" refers to data acquisition and processing occurring immediately, with little or no delay.

[1224] "Detecting suspicious activity" is the process of identifying suspicious behavior or individuals based on the results of image analysis and providing information for responding.

[1225] The system for realizing this invention is composed of a retinal sensor, a server, and a terminal. The specific operation of each component and its program are described in detail below.

[1226] Hardware and Software Use Cases

[1227] This system uses the following hardware and software:

[1228] Hardware

[1229] Retinal sensor: A sensor that continuously acquires video data and generates a data stream that can be analyzed in real time.

[1230] Smartphone or smart glasses: A mobile device carried by the user.

[1231] Server: Receives data, processes it, and sends the analysis results.

[1232] software

[1233] Python: An entire programming language.

[1234] OpenCV: An image processing library, used for denoising and preprocessing data.

[1235] TensorFlow / Keras: Deep learning libraries used for image analysis models.

[1236] Flask: A web framework used for communication between the server and the terminal and for notifying results.

[1237] Data processing and calculation flow

[1238] The server starts by receiving a continuous data stream from the retinal sensor. The received data is first stored in a buffer. Next, noise is removed from the data using OpenCV to obtain clean data. The noise-removed data is normalized to a range of 0 to 1, and image analysis is performed using a deep learning model. If suspicious activity is detected as a result of the analysis, the results are sent to the device. This series of processes is performed in real time.

[1239] The device receives the analysis results from the server and generates and displays security notifications to the user as needed, such as a "Suspicious person detected" message on a smartphone or smart glasses.

[1240] Specific examples of processing

[1241] This system is used in a scenario where office entrances are monitored. When a retinal sensor installed in the smart glasses detects a suspicious person or movement, the data is sent to a server in real time. The server processes the data and notifies the user of the analysis results on their smartphone. This allows the user to take prompt action even when they are in a remote location.

[1242] Prompt Sentence Examples

[1243] "Implement a security application that uses a retinal sensor to monitor the surroundings in real time with high accuracy and detect suspicious individuals. The application should be installable on smartphones or smart glasses, using Python, OpenCV, TensorFlow, and Flask."

[1244] In this way, the present invention realizes continuous data stream processing using a retinal sensor, and can provide highly efficient and highly accurate security monitoring.

[1245] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1246] Step 1:

[1247] A continuous data stream is acquired from the retinal sensor.

[1248] Input: Raw data stream from the retinal sensor.

[1249] Data processing / calculation: The image data captured by the sensor is acquired continuously, rather than frame by frame.

[1250] Output: A continuous data stream.

[1251] Step 2:

[1252] The server receives the data stream and stores it in a buffer.

[1253] Input: A continuous data stream.

[1254] Data processing / calculation: The received data stream is temporarily stored in a buffer.

[1255] Output: The raw data stored in the buffer.

[1256] Step 3:

[1257] The server applies a noise reduction filter to the buffered data.

[1258] Input: Raw data stored in a buffer.

[1259] Data processing / calculation: Using OpenCV, unnecessary information such as environmental noise and sensor noise is filtered out.

[1260] Specific operation: Convert to grayscale and apply a filter such as GaussianBlur.

[1261] Output: Denoised data.

[1262] Step 4:

[1263] The server normalizes the denoised data.

[1264] Input: Denoised data.

[1265] Data processing / calculation: Scale the data to the range 0 to 1.

[1266] Specific operation: Normalization is performed using the maximum and minimum values ​​of the data.

[1267] Output: Normalized data.

[1268] Step 5:

[1269] The server inputs the normalized data into a deep learning model to perform image analysis.

[1270] Input: Normalized data.

[1271] Data processing / calculation: Image analysis is performed using a trained model using TensorFlow / Keras to detect suspicious individuals and abnormal behavior.

[1272] Specific operation: Analysis is performed using a neural network.

[1273] Output: Analysis results (such as suspicious person detection).

[1274] Step 6:

[1275] The server transmits the analysis results to the terminal.

[1276] Input: Image analysis results.

[1277] Data processing / calculation: Analysis results are sent to the terminal.

[1278] Specific operation: Send the result via HTTP request using Flask.

[1279] Output: The analysis results are sent to the terminal.

[1280] Step 7:

[1281] The device displays a notification to the user based on the received analysis results.

[1282] Input: Analysis results sent from the server.

[1283] Data processing / calculation: The received results are notified to the user in an easy-to-understand format.

[1284] Specific operation: Display an alert on your smartphone or smart glasses.

[1285] Output: A notification is displayed to the user.

[1286] In this way, the server, terminal, and user play their respective roles at each step, achieving highly accurate real-time image analysis.

[1287] 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.

[1288] The present invention is a system for seamlessly processing data acquired from a retinal sensor to perform image analysis and emotion recognition. Specifically, it includes a server, a terminal, and an emotion engine.

[1289] System configuration

[1290] 1. Data Acquisition

[1291] The server receives a continuous data stream from the retinal sensor, which stores the data in a buffer in real time.

[1292] Specific examples

[1293] The server continuously receives data from the retinal sensor every second and stores it in a buffer.

[1294] 2. Noise Reduction

[1295] The server retrieves the data stored in the buffer and applies a noise reduction filter to remove the noise components, resulting in clean data.

[1296] Specific examples

[1297] The server applies a noise reduction filter to the received data to remove environmental noise and sensor noise.

[1298] 3. Data normalization

[1299] The server normalizes the denoised data, scaling it to the range 0-1 and converting it into a form suitable for analytical models.

[1300] Specific examples

[1301] The server scales the denoised data based on the maximum and minimum values.

[1302] 4. Image Analysis

[1303] The server uses the normalized data to perform image analysis, and uses a deep learning model to detect and classify objects. This model is composed of a multi-layered neural network, achieving highly accurate recognition.

[1304] Specific examples

[1305] The server inputs the normalized data into a deep learning model to detect faces and classify objects.

[1306] 5. Emotion recognition

[1307] The emotion engine uses the results of image analysis and additional sensor data to recognize the user's emotions. It uses deep learning models to analyze emotions from the user's facial expressions and voice.

[1308] Specific examples

[1309] The emotion engine uses deep learning models to analyze the user's facial expressions to determine joy, anger, sadness, and happiness.

[1310] 6. Send results

[1311] The server sends the results of image analysis and emotion recognition to the device. The analysis results are packaged in a predefined format and sent to the device in real time. The device then receives the analysis results and notifies the user.

[1312] Specific examples

[1313] The server sends the results of image analysis and emotion recognition to the device, which then displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[1314] Usage example

[1315] Usage scenarios

[1316] The online education system analyzes students' facial expressions and voices in real time to understand their emotions. This allows instructors to understand the students' emotional state and adjust the educational content in real time. For example, if a student appears confused during a lecture, the data can be analyzed by another engine and a message will be sent to the instructor saying, "The student is confused." This allows the instructor to respond by providing additional explanations.

[1317] As described above, the present invention provides a system that uses a retinal sensor to realize real-time processing of continuous data streams, image analysis, and emotion recognition. This system overcomes the limitations of conventional frame-by-frame processing methods and enables a wide variety of applications.

[1318] The processing flow will be explained below.

[1319] Step 1:

[1320] The server receives a continuous data stream from the retinal sensor. The data generated by the sensor is sent in real time to the server, which stores it in a buffer where it is ready for the next processing step.

[1321] Step 2:

[1322] The server retrieves the buffered data and performs a denoising process, applying filtering algorithms to remove environmental and sensor noise. This clean data provides the basis for more accurate analysis.

[1323] Step 3:

[1324] The server then performs a normalization process on the denoised data, scaling it to a range of 0 to 1 and converting it into a form that is compatible with the analytical model. This normalization process allows the analytical model to function efficiently.

[1325] Step 4:

[1326] The server uses the normalized data to perform image analysis. It then uses deep learning models to analyze the data and perform object detection and classification. For example, the server uses deep learning models to detect human faces and classify objects.

[1327] Step 5:

[1328] The emotion engine uses image analysis results and additional sensor data to recognize the user's emotions. Deep learning models are applied to emotion recognition, analyzing the user's facial expressions and voice data. The analysis results are classified into emotion categories such as "happiness," "sadness," and "surprise."

[1329] Step 6:

[1330] The server sends the results of image analysis and emotion recognition to the device, which then packages the results in a predefined format and sends them to the device. The device processes the results and notifies the user.

[1331] Step 7:

[1332] The device displays the analysis results received from the server and notifies the user in a variety of notification formats, including visual alerts, audio notifications, and messages. The user can then take action based on the notification.

[1333] Step 8:

[1334] The user checks the display on the device and takes action as necessary. For example, in the case of a surveillance system, after receiving a notification, the user may take action such as zooming in on the area where suspicious activity was detected.

[1335] Through these steps, this system achieves real-time data processing, image analysis, and emotion recognition using a retinal sensor, enabling it to provide users with fast and accurate information. This overcomes the limitations of conventional frame-by-frame processing methods and enables more efficient operation.

[1336] Example 2

[1337] 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."

[1338] Conventional image analysis and emotion recognition systems face the challenge of processing the entire process, from data acquisition to analysis, recognition, and notification, in real time. This can lead to issues with reduced analysis accuracy, particularly when dealing with raw data containing large amounts of noise. Furthermore, there is a lack of effective means for quickly notifying users of analysis results. To address these issues, a system capable of high-precision data processing and analysis in real time is required.

[1339] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1340] In this invention, the server includes means for receiving a continuous data stream from the retinal sensor, means for storing the received data in a buffer, means for applying a noise reduction filter to the data stored in the buffer, means for normalizing the noise-reduced data, means for performing image analysis using a deep learning model with the normalized data, means for performing emotion recognition using the image analysis results and additional sensor data, means for transmitting the analysis results and emotion recognition results to a terminal, and means for notifying the user of the results on the terminal. This enables noise reduction and data normalization to be performed in real time, achieving highly accurate image analysis and emotion recognition, and quickly notifying the user of the analysis results.

[1341] A "retinal sensor" is a sensor that detects light and outputs visual information as an electrical signal.

[1342] A "continuous data stream" is a continuous flow of data that is sent without interruption.

[1343] A "buffer" is a temporary storage area for data, and is an area used to adjust the difference between the data reception speed and the data processing speed.

[1344] A "noise removal filter" is an algorithm or hardware for removing unwanted noise components contained in data.

[1345] "Normalization" is the process of scaling data to a certain range (usually between 0 and 1).

[1346] A "deep learning model" is a machine learning model that uses a multi-layer neural network to learn the characteristics of data and perform analysis and prediction.

[1347] "Image analysis" is the process of detecting and classifying objects from image data, and is primarily performed using deep learning models.

[1348] "Emotion recognition" is the process of identifying a person's emotional state from image and audio data.

[1349] A "terminal" is an equipment or device that is directly operated by a user, and examples include smartphones and tablets.

[1350] MODE FOR CARRYING OUT THE INVENTION

[1351] The present invention provides a system for image analysis and emotion recognition that processes a continuous data stream acquired from a retinal sensor in real time. The system mainly includes a server, a terminal, and an emotion engine.

[1352] 1. Data Acquisition

[1353] The server receives the data sent from the retina sensor in real time and stores it in a buffer in memory, preventing data interruptions and ensuring a continuous data stream.

[1354] Specific examples

[1355] The server receives data from the retina sensor every second and stores it in a buffer.

[1356] 2. Noise Reduction

[1357] The server applies a noise reduction filter to the buffered data, removing sensor noise and environmental noise to obtain clean data.

[1358] Specific examples

[1359] The server applies a low-pass filter to the data in the buffer to remove unwanted high frequency components.

[1360] 3. Data normalization

[1361] The server normalizes the denoised data, specifically scaling the data values ​​to the range 0 to 1.

[1362] Specific examples

[1363] The server takes the maximum value of the data as 1 and the minimum value as 0 and scales each data point to fit into this range.

[1364] 4. Image Analysis

[1365] The server uses the normalized data to perform highly accurate image analysis using a deep learning model, which is built using a multi-layer neural network.

[1366] Specific examples

[1367] The server inputs the data into a neural network model built using a deep learning library (e.g., TensorFlow, PyTorch) to detect human faces and classify objects.

[1368] 5. Emotion recognition

[1369] The emotion engine uses a deep learning model to analyze the user's emotions based on the results of image analysis and additional sensor data, allowing it to identify emotions from the user's facial expressions and voice.

[1370] Specific examples

[1371] The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial expressions and voice data to identify emotions.

[1372] 6. Send results

[1373] The server sends the results of image analysis and emotion recognition to the device, which receives them and notifies the user of the results.

[1374] Specific examples

[1375] The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[1376] Prompt Sentence Examples

[1377] Below are some examples of prompts to input to the generative AI model.

[1378] Example prompts

[1379] Please explain in detail how the system works, which continuously receives data from a retinal sensor and performs noise removal, data normalization, image analysis, and emotion recognition in real time.

[1380] The above is an embodiment of the system. This invention enables highly accurate data processing and analysis in real time, and allows analysis results to be quickly notified to the user.

[1381] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1382] Step 1:

[1383] The server receives a continuous data stream from the retinal sensor, which is buffered in real time.

[1384] Input: Raw data from the retinal sensor

[1385] Output: raw data stored in a buffer

[1386] Specific operation: The server receives data every second and immediately stores the data in a buffer to prevent data interruptions.

[1387] Step 2:

[1388] The server applies a noise reduction filter to the buffered data, which removes unwanted noise components.

[1389] Input: raw data stored in a buffer

[1390] Output: Clean data after noise removal

[1391] What it does: The server applies a low-pass filter to remove environmental noise and sensor noise, for example, a Butterworth filter to reduce noise in a specific frequency band.

[1392] Step 3:

[1393] The server normalizes the denoised data to a range of 0 to 1, converting it into a data format suitable for the analytical model.

[1394] Input: Clean data after noise removal

[1395] Output: Normalized data

[1396] What it does: The server scales each value of the data based on the maximum and minimum values. For example, if the original data ranges from 0 to 255, it divides each value by 255 to scale it to the range of 0 to 1.

[1397] Step 4:

[1398] The server runs a deep learning model on the normalized data to analyze the images, resulting in object detection and classification.

[1399] Input: Normalized data

[1400] Output: Image analysis results (e.g., object detection and classification results)

[1401] What it does: The server uses deep learning libraries (e.g., TensorFlow, PyTorch) to apply neural network models to perform image analysis, such as detecting human faces and classifying their attributes.

[1402] Step 5:

[1403] The emotion engine performs emotion recognition based on the results of image analysis and additional sensor data, making it possible to identify the user's emotions.

[1404] Input: Image analysis results, additional sensor data

[1405] Output: Emotion recognition results (e.g., joy, anger, sadness, and happiness)

[1406] Specific operation: The emotion engine uses deep learning models (e.g., RNN, LSTM) to analyze the user's facial and voice data and identify emotions. For example, it determines that the user is "happy" based on their facial expression.

[1407] Step 6:

[1408] The server transmits the results of the image analysis and emotion recognition to the terminal, and the terminal notifies the user of the results.

[1409] Input: Image analysis results, emotion recognition results

[1410] Output: The result reported to the user

[1411] Specific operation: The server sends the analysis results to the device in a defined format such as JSON, and the device displays a notification on the user's smartphone saying, "A face has been detected and the emotion is joy."

[1412] (Application example 2)

[1413] 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."

[1414] Conventional factory robot systems have been unable to monitor worker emotions and fatigue levels in real time and provide appropriate feedback according to the situation. This has made it difficult to ensure worker safety and has left challenges in creating an efficient work environment. The present invention aims to solve this problem by providing a system that enables worker emotion recognition and fatigue level monitoring.

[1415] 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 receiving a continuous data stream from the retinal sensor, means for removing noise from the received data stream, means for normalizing the noise-removed data, means for performing image analysis and emotion recognition using the normalized data, means for transmitting the analysis results and emotion recognition results to the terminal and providing feedback according to the user's situation based on the analysis results, means for continuously buffering preprocessed data, means for analyzing the data received from the retinal sensor in real time and monitoring the worker's emotion and fatigue level, and means for processing the received data stream in real time and notifying the worker to take a break based on the analysis results. This makes it possible to monitor the worker's emotion and fatigue level in real time and create a safe and efficient work environment.

[1416] A "retinal sensor" is a device that senses light, generates signals, and outputs a continuous data stream, mimicking the function of the human retina.

[1417] A "continuous data stream" refers to a data flow in which data is transmitted continuously rather than intermittently.

[1418] "Noise" refers to unnecessary components contained in sensor data that prevent accurate information from being obtained.

[1419] A "means for removing noise" is a method or device for removing unnecessary components from received data to obtain clean data.

[1420] A "means for normalizing data" is a method or device that scales data to a particular range and converts it into a form suitable for analysis.

[1421] "Image analysis" refers to the method or process of processing acquired image data to detect and classify objects.

[1422] "Emotion recognition" is a method of analyzing and identifying a user's emotional state using images and additional sensor data.

[1423] "Analysis results" refers to the information obtained as a result of data analysis, which is used to prompt decision-making and action.

[1424] "Feedback" refers to information or instructions provided to the user based on the analysis results, to help them maintain appropriate behavior or state.

[1425] "Terminal" refers to a computer or smart device that is directly operated by a user and is used to display analysis results and receive notifications.

[1426] A "buffer" is a storage device that temporarily stores data, allowing for smooth reception and processing of the data.

[1427] "Real-time" refers to data acquisition, processing, and feedback occurring immediately, without delay.

[1428] In this invention, a system is constructed that monitors the emotions and fatigue levels of factory workers in real time to provide a safe and efficient working environment. A detailed embodiment of this system will be described below.

[1429] The server first receives the data stream from the retinal sensor, which captures information such as the worker's facial expressions and gaze in real time. The captured data is temporarily stored in a buffer. A noise reduction filter is then applied to the stored data to remove unnecessary noise components, resulting in clean data.

[1430] The noise-removed data is then normalized to a range of 0 to 1. A deep learning model is then used to analyze the image and recognize emotions. The image analysis process detects the worker's face and facial expressions, and emotion recognition is performed based on this information.

[1431] The results of emotion recognition are sent from the server to the device in real time. Based on the analysis results, the device provides the worker with context-appropriate feedback. This feedback is displayed on the worker's smart glasses. For example, if the device determines that the worker is tired, a notification will be displayed encouraging them to take a break.

[1432] Specific examples of hardware and software used

[1433] Hardware: retinal sensor, smart glasses, server

[1434] Software: TensorFlow / Keras (for deep learning), OpenCV (for image processing), Flask (to build the API)

[1435] Specific examples

[1436] If a factory worker is operating a machine without expressing any emotion for a long period of time, the retinal sensor will capture this behavior and the server will analyze it in real time. If this analysis determines that the worker is fatigued, the system will display a notification on the smart glasses telling the worker to take a break.

[1437] Prompt Sentence Examples

[1438] Create a Python program that detects fatigue when a factory worker operates a machine without expressing any emotion for a long period of time and displays a notification encouraging them to take a break.

[1439] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1440] Step 1:

[1441] The server receives a continuous data stream from the retinal sensor. The input is the raw data from the retinal sensor, and the output is the received data stream. This data stream contains information such as the worker's facial expressions and gaze.

[1442] Step 2:

[1443] The server stores the received data stream in a buffer. The input is the data stream received in step 1, and the output is the data stored in the buffer. In this step, data is temporarily stored.

[1444] Step 3:

[1445] The server applies a noise reduction filter to the buffered data. The input is the raw data stored in the buffer, and the output is the clean, denoised data. This filtering process removes environmental noise and sensor noise.

[1446] Step 4:

[1447] The server normalizes the denoised data. The input is the denoised data and the output is the normalised data. The normalisation process scales the data to the range 0 to 1.

[1448] Step 5:

[1449] The server uses the normalized data to perform image analysis. The input is the normalized data, and the output is the result of image analysis. This image analysis uses a deep learning model to detect and classify objects.

[1450] Step 6:

[1451] The server performs emotion recognition based on the results of image analysis. The input is the result of image analysis, and the output is the result of emotion recognition. The type of emotion (joy, anger, sadness, happiness, etc.) and fatigue level are analyzed using a deep learning model.

[1452] Step 7:

[1453] The server sends the results of image analysis and emotion recognition to the terminal. The input is the emotion recognition result, and the output is the analysis result and emotion recognition result sent to the receiving terminal. This data is sent in real time.

[1454] Step 8:

[1455] The terminal provides feedback to the user based on the analysis results received from the server. The input is the analysis results and emotion recognition results sent from the server, and the output is a notification displayed to the user. As a specific example, a notification saying "Please take a break" is displayed on smart glasses. This notification allows the user (worker) to receive appropriate feedback.

[1456] 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.

[1457] 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.

[1458] 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.

[1459] 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.

[1460] 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.

[1461] 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.

[1462] 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).

[1463] 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.

[1464] 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."

[1465] 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.

[1466] 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).

[1467] 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.

[1468] 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.

[1469] 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.

[1470] 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.

[1471] 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.

[1472] 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.

[1473] 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.

[1474] 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.

[1475] 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.

[1476] 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.

[1477] The following is further disclosed regarding the above embodiment.

[1478] (Claim 1)

[1479] means for receiving a continuous data stream from the retinal sensor;

[1480] means for removing noise from the received data stream;

[1481] a means for normalizing the denoised data;

[1482] means for performing image analysis using the normalized data;

[1483] The system includes a means for transmitting the analysis results to a terminal.

[1484] (Claim 2)

[1485] 10. The system of claim 1, further comprising means for continuously buffering the preprocessed data.

[1486] (Claim 3)

[1487] 10. The system of claim 1, further comprising means for processing the received data stream in real time.

[1488] (Claim 4)

[1489] 10. The system of claim 1, further comprising means for performing image analysis using a deep learning model.

[1490] (Claim 5)

[1491] 10. The system of claim 1, further comprising means for notifying a user of the analysis results.

[1492] (Claim 6)

[1493] 10. The system of claim 1, further comprising means for wireless communication between the retinal sensor and the server.

[1494] "Example 1"

[1495] (Claim 1)

[1496] means for receiving a continuous data stream from the retinal sensor;

[1497] means for buffering the received data stream;

[1498] means for removing noise from the buffered data;

[1499] a means for normalizing the denoised data;

[1500] A means for performing image analysis using a deep learning model using the normalized data;

[1501] means for transmitting the analysis results to a terminal;

[1502] A system including a means for the terminal to notify or display the analysis results to the user.

[1503] (Claim 2)

[1504] 10. The system of claim 1, further comprising means for buffering the preprocessed data in real time.

[1505] (Claim 3)

[1506] 10. The system of claim 1, further comprising means for processing the received data stream in real time.

[1507] "Application Example 1"

[1508] (Claim 1)

[1509] means for receiving a continuous data stream from the retinal sensor;

[1510] means for removing noise from the received data stream;

[1511] a means for normalizing the denoised data;

[1512] means for performing image analysis using the normalized data;

[1513] means for transmitting the analysis results to a terminal;

[1514] a means for generating and displaying a security notification on the terminal based on the analysis result;

[1515] A means for detecting suspicious activity based on image analysis results in real time;

[1516] A system including:

[1517] (Claim 2)

[1518] 10. The system of claim 1, further comprising means for continuously buffering the preprocessed data.

[1519] (Claim 3)

[1520] 10. The system of claim 1, further comprising means for processing the received data stream in real time.

[1521] "Example 2: Combining Emotion Engines"

[1522] (Claim 1)

[1523] means for receiving a continuous data stream from the retinal sensor;

[1524] means for storing received data in a buffer;

[1525] means for applying a noise reduction filter to the buffered data;

[1526] a means for normalizing the denoised data;

[1527] a means for performing image analysis using a deep learning model with the normalized data;

[1528] a means for performing emotion recognition using the results of the image analysis and additional sensor data;

[1529] means for transmitting the analysis result and the emotion recognition result to a terminal;

[1530] The system includes a means for notifying the user of the results at the terminal.

[1531] (Claim 2)

[1532] 10. The system of claim 1, further comprising means for continuously buffering the preprocessed data.

[1533] (Claim 3)

[1534] 10. The system of claim 1, further comprising means for processing the received data stream in real time.

[1535] "Application example 2 when combining emotion engines"

[1536] (Claim 1)

[1537] means for receiving a continuous data stream from the retinal sensor;

[1538] means for removing noise from the received data stream;

[1539] a means for normalizing the denoised data;

[1540] means for performing image analysis and emotion recognition using the normalized data;

[1541] The system includes a means for transmitting the analysis results and emotion recognition results to a terminal and providing feedback according to the user's situation based on the analysis results.

[1542] (Claim 2)

[1543] 10. The system of claim 1, further comprising: means for continuously buffering the preprocessed data; and means for analyzing data received from the retinal sensor in real time to monitor worker emotions and fatigue levels.

[1544] (Claim 3)

[1545] 10. The system of claim 1, further comprising means for processing the received data stream in real time and notifying workers to take a break based on the analysis results. [Explanation of symbols]

[1546] 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 receiving a continuous data stream from the retinal sensor; means for removing noise from the received data stream; a means for normalizing the denoised data; means for performing image analysis using the normalized data; The system includes a means for transmitting the analysis results to a terminal.

2. 10. The system of claim 1, further comprising means for continuously storing the preprocessed data in a buffer.

3. 10. The system of claim 1, further comprising means for processing the received data stream in real time.

4. The system of claim 1 , further comprising means for performing image analysis using a deep learning model.

5. The system of claim 1 , further comprising means for notifying a user of the analysis results.

6. The system of claim 1 , further comprising means for wirelessly communicating between the retinal sensor and the server.

Images