System

Abstract

To provide a system for protecting privacy and improving security by processing a landscape reflected in a pupil of a photographed image into a natural form.SOLUTION: A system comprising: means for acquiring an image; means for detecting a pupil in the acquired image; means for identifying a scenery portion reflected in the detected pupil; means for generating a natural pupil by deleting the identified scenery portion; means for merging the generated natural pupil with the original image; and means for providing the merged image to a user.SELECTED DRAWING: Figure 12

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] In recent years, advances in camera technology and image analysis technology have made it easier to analyze the scenery reflected in a photographed image, increasing the risk of violating personal privacy. This has resulted in various security issues, such as identifying a person's home, becoming a victim of stalking, and identifying people they are with. This invention aims to protect privacy and improve security by processing the scenery reflected in the eyes in a natural way. [Means for solving the problem]

[0005] The present invention solves the above-mentioned problems with a system that includes a means for acquiring an image, detecting pupils in the acquired image, and a means for identifying a landscape portion reflected in the detected pupils. The system also includes a means for removing the identified landscape portion and generating natural-looking pupils, a means for merging the generated natural-looking pupils with the original image, and a means for providing the merged image to a user. The system can process images sent from a user's device and is characterized by using natural-looking pupils generated using a generative adversarial network (GAN). This removes the landscape reflected in the pupils and generates a natural-looking pupil image based on the original pupil shape and light source information, thereby protecting privacy.

[0006] The "means for acquiring images" is a function for receiving image data from a user's terminal or other input device and importing it into the system.

[0007] The "means for detecting pupils in the captured image" is a function that uses an image analysis algorithm to accurately identify pupil areas of a face present in the captured image.

[0008] The "means for identifying the portion of the scenery reflected in the detected pupil" is a function for recognizing the external scenery or object reflected in the detected pupil and distinguishing it from other portions.

[0009] The "means for removing identified scene features and generating natural-looking pupils" refers to a function for removing identified scene features from an image and then reconstructing natural-looking pupils using other image generation techniques.

[0010] The "means for merging the generated natural pupil image with the original image" is a function for integrating the generated natural pupil image with the original image and processing it so as to maintain a natural appearance overall.

[0011] The "means for providing the merged image to the user" is a function for returning the processed image to the user so that the user can view or save it.

[0012] A "generative adversarial network (GAN)" is a machine learning model that generates realistic images and data by training a generative model against a decision model.

[0013] "User's device" refers to an electronic device such as a smartphone, computer, or tablet used to upload images. [Brief explanation of the drawings]

[0014] [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

[0015] 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.

[0016] First, the terms used in the following description will be explained.

[0017] 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).

[0018] 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.

[0019] 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.

[0020] 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.

[0021] 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."

[0022] [First embodiment]

[0023] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0024] 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.

[0025] 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).

[0026] 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.

[0027] 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.

[0028] 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.

[0029] 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.

[0030] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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."

[0035] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. The specific operation of each step of this system is shown below.

[0036] Program processing flow

[0037] 1. Upload an image

[0038] User: Uploads images that require privacy protection from their own device to the system. These images can be anything from selfies to group photos.

[0039] Server: Receives and temporarily stores the images uploaded above, preparing them for the subsequent analysis process.

[0040] 2. Image Analysis

[0041] Server: Inputs the received image into an image analysis algorithm, which performs facial recognition and identifies the location of facial features and eyes.

[0042] 3. Detection of the landscape part of the pupil

[0043] Server: Performs detailed analysis of the scenery reflected in the detected pupils and identifies the scenery at the pixel level. This information is used in subsequent processing steps.

[0044] 4. Eye Image Generation

[0045] Server: Removes the landscape and applies an AI algorithm (e.g., a generative adversarial network (GAN)) to generate natural-looking eye images. The AI ​​takes into account the original pupil shape and light source information to reconstruct natural-looking eyes.

[0046] 5. Image Merging

[0047] Server: Integrate the generated natural pupils into the original image, replacing the original eye regions with the new pupil image, making the whole image look natural.

[0048] 6. Providing edited images

[0049] Server: Returns the completed processed image to the user's device. In addition, it also provides comparison images before and after processing so that the user can check the processing results.

[0050] Specific examples

[0051] Example 1: Editing a selfie

[0052] User: Uploads a selfie to the application.

[0053] Server: Receives the image and starts eye detection.

[0054] Server: Analyzes and identifies the part of the scenery reflected in the eyes.

[0055] Server: Generates natural-looking pupils using image generation AI.

[0056] Server: Sends the processed image to the user.

[0057] User: Review the received images and ensure their privacy is protected.

[0058] Example 2: Editing a group photo

[0059] User: Uploads a group photo of family and friends to the system.

[0060] Server: Receives the image and starts eye detection.

[0061] Server: Identify the landscape portion of each pupil and generate natural-looking pupils using AI.

[0062] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[0063] User: Review the received images and ensure their privacy is protected.

[0064] This system allows users to share images safely while protecting their privacy, and is expected to be widely adopted by businesses as well.

[0065] The processing flow will be explained below.

[0066] Step 1:

[0067] User: Uploads images from their device to the system, specifically by selecting a selfie or group photo and sending it to the server through the application or web interface.

[0068] Step 2:

[0069] Server: Receives uploaded images and stores them in temporary storage. After storing them, the server prepares them for image analysis.

[0070] Step 3:

[0071] Server: Uses an image analysis algorithm to detect faces in the received image. The face detection algorithm identifies the position and contours of the face.

[0072] Step 4:

[0073] Server: Identify the position of the pupils based on the facial position information. The pupil detection algorithm extracts the eye feature points and identifies the pupil area at the pixel level.

[0074] Step 5:

[0075] Server: Analyzes and identifies the scenery reflected within the detected pupil area. Specifically, it analyzes the light reflected from the pupil and the surrounding scenery information for each pixel and extracts unnecessary information.

[0076] Step 6:

[0077] Server: Using image generation AI (e.g., generative adversarial networks (GANs)), the identified landscape features are removed and natural-looking pupils are generated, while maintaining a natural appearance by taking into account the original pupil shape and lighting information.

[0078] Step 7:

[0079] Server: Merge the generated natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[0080] Step 8:

[0081] Server: Returns the processed image to the user, along with a before and after comparison image so the user can see the changes.

[0082] Step 9:

[0083] User: Receives the processed image and checks the results. Once the user confirms that their privacy has been protected, they can save the image to their device or post it safely to social media.

[0084] Example 1

[0085] 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."

[0086] With conventional technology, the landscape portion of the eyes in an image could be a violation of privacy, making it difficult for users to share images with confidence. Furthermore, manually editing the eyes requires time and skill, making it inefficient. Furthermore, the image processing process can sometimes produce unnatural results, making the processed image appear unnatural. To solve these problems, a system is needed that can automatically detect and process the landscape portion of the eyes, maintaining a natural appearance.

[0087] 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.

[0088] In this invention, the server includes means for a user to upload an image from a terminal, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, and means for providing the merged image to the user. This enables users to easily edit and safely share images while ensuring privacy.

[0089] "Means for users to upload images from their devices" refers to a system that allows users to select image files using their own devices and send them to a server via a network.

[0090] The "means for detecting pupils in a captured image" is a system that has the function of identifying the positions of pupils in a captured image using image processing technology.

[0091] The "means for identifying the portion of the scenery reflected in the detected pupil" is a system that has the function of analyzing the portion of the scenery reflected in the pupil area and identifying that portion at the pixel level.

[0092] The "means for removing identified landscape parts and generating natural-looking pupils" refers to a system that uses AI technology and image generation algorithms to remove identified landscape parts and generate natural-looking pupils by taking into account the original pupil shape and light source information.

[0093] The "means for merging the generated natural pupils into the original image" is a system that has the function of integrating the newly generated natural pupils into the original image to make the image appear natural overall.

[0094] The "means for providing a merged image to a user" refers to a system that has a function for returning the processed image to a user so that the user can check the result.

[0095] A "generative adversarial network" is a type of deep learning model that can generate new, high-quality data by learning through a competitive process between a generative model and a discriminative model.

[0096] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. Specific embodiments for implementing this system are described below.

[0097] First, the user uploads an image that requires privacy protection to the system using the device. Specifically, the user opens the application's file selection dialog and selects an image file from local storage. This operation causes the device to send the image to the server. The image file data is sent to the server using an HTTP POST request.

[0098] The server temporarily stores the received image file in storage. The path of the stored file is retained for the next processing step. Specific hardware used is a server computer, and software used is an HTTP server and a storage management system.

[0099] Next, the server inputs the saved image into an image analysis algorithm. Specifically, it uses the OpenCV and dlib libraries to perform face recognition. The image is read using OpenCV's cv2.imread function and converted to grayscale using cv2.cvtColor. The dlib face detector is used to detect the position of the face, and the dlib landmark detector is used to identify the position of the eye. This series of processes results in the detection of the eye in the image.

[0100] To analyze the landscape part reflected in the detected pupil in detail, the server identifies the landscape part pixel by pixel. For this purpose, OpenCV's cv2.boundingRect function is used. The identified landscape part information is used in the subsequent processing steps.

[0101] The server then uses a generative adversarial network (GAN) to remove the landscape elements from the pupils and generate natural-looking pupils. To do this, the image is resized and normalized to an appropriate size. A pre-trained GAN model is then used. Specific software used is Tensorflow (registered trademark) and PyTorch.

[0102] To integrate the generated natural pupils into the original image, the server uses OpenCV's cv2.seamlessClone function, which blends the newly generated pupils seamlessly into the original image, maintaining an overall natural look.

[0103] Finally, the server returns the completed processed image to the user's device. The processed image is returned as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can then view and check the received image.

[0104] Specific examples

[0105] Selfie photo editing

[0106] The user uploads a selfie to the application. The device sends the image to the server, which uses OpenCV and dlib to initiate face and eye detection. The server uses a custom algorithm to identify the landscape portion of the eye and generates natural-looking eyes using a GAN. The server merges the new eye image with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[0107] Editing group photos

[0108] A user uploads a group photo of family and friends to the system. The device sends the image to the server, which uses OpenCV and dlib to initiate multiple face and eye detection. The server uses a custom algorithm to identify the landscape portion of each eye and uses GAN to generate natural-looking eyes for each eye. The server then merges the generated eyes with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[0109] This system allows users to edit and share images safely and easily while protecting their privacy.

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

[0111] Step 1:

[0112] The user uploads an image requiring privacy protection to the system using their own device. Specifically, they open the application's file selection dialog and select an image file from local storage. This operation causes the device to send the image data to the server via an HTTP POST request.

[0113] Input: An image file selected by the user.

[0114] Output: Image data sent to the server

[0115] Step 2:

[0116] The server temporarily stores the received image data. Specifically, it stores the received image file in storage and retains the file path for the next analysis step.

[0117] Input: Image data sent to the server

[0118] Output: Image file saved to storage

[0119] Step 3:

[0120] The server inputs the saved image into the image analysis algorithm. First, it reads the image using OpenCV's cv2.imread function, then converts the image to grayscale using the cv2.cvtColor function. Then it uses dlib's face detector (dlib.get_frontal_face_detector) to detect the face position. Finally, it uses dlib's landmark detector (dlib.shape_predictor) to identify the eye position.

[0121] Input: Image file saved in storage

[0122] Output: Face and eye position information

[0123] Step 4:

[0124] The server then performs a detailed analysis of the landscape reflected within the detected pupil. Specifically, it uses OpenCV's cv2.boundingRect function to extract the pupil area pixel by pixel. It then analyzes the pixel values ​​within this pupil area using a custom algorithm to identify the landscape.

[0125] Input: Face and eye position information

[0126] Output: Pixel information of the landscape

[0127] Step 5:

[0128] The server uses a generative adversarial network (GAN) to remove the identified landscape features and generate natural-looking pupils. First, the pupil regions are resized and normalized to an appropriate size as input for the GAN. Then, a pre-trained GAN model is used to generate natural-looking pupils. This process uses TensorFlow and PyTorch.

[0129] Input: Pixel information of the landscape

[0130] Output: Natural eye image

[0131] Step 6:

[0132] The server merges the generated natural pupils with the original image, using OpenCV's cv2.seamlessClone function to seamlessly integrate the newly generated pupil images into the original image, resulting in an overall image that maintains a natural appearance.

[0133] Input: natural eye image, original image

[0134] Output: Natural-looking processed image

[0135] Step 7:

[0136] The server returns the processed image to the user. The processed image is sent to the user's device as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can view the received image and confirm that their privacy has been protected.

[0137] Input: Natural-looking processed image

[0138] Output: The processed image sent to the user

[0139] (Application example 1)

[0140] 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."

[0141] In modern society, there is an increasing trend of capturing video footage using surveillance cameras and smart devices and using it for various purposes. However, these videos contain personal information, and the scene reflected in the eyes of people's eyes is particularly at high risk of being leaked. This raises concerns about privacy violations, and there is a demand for technology to safely and efficiently protect personal information.

[0142] 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.

[0143] In this invention, the server includes means for acquiring an image, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, means for providing the merged image to a user, means for providing comparison images before and after processing based on the generated privacy-protected image data, and means for performing the above processing on video data uploaded from a surveillance camera or a smart device. This makes it possible to achieve privacy protection while preventing the leakage of personal information in videos acquired from a surveillance camera or a smart device.

[0144] "Means for acquiring images" refers to a device or method for acquiring images or video using a surveillance camera or smart device and receiving them in a form that is input into the system.

[0145] The "means for detecting eyes" is an algorithm or technique for analyzing the facial area in the acquired image and specifically identifying the location of the eyes.

[0146] The "means for identifying the scenery portion" refers to a technology or algorithm that analyzes the scenery or background reflected in the detected pupil at the pixel level and identifies its range.

[0147] The "means for generating natural-looking pupils" refers to AI techniques such as generative adversarial networks (GANs) that remove detected landscape features and reconstruct a natural appearance based on the original pupil shape and light source information.

[0148] "Means for merging into original image" refers to the technique or process used to incorporate the generated natural pupils into the original image so that the overall image appears natural.

[0149] "Means for providing to the user" refers to a system or technology that returns the merged image to the user's terminal so that the user can view the results.

[0150] "Means for providing images before and after processing for comparison" refers to a display function for comparing the privacy-protected image with the original image, or technology that allows users to visually confirm the differences.

[0151] "Means for processing video data uploaded from surveillance cameras and smart devices" refers to a system or technology that receives video data sent from surveillance cameras, smartphones, tablets, etc., and automatically performs the series of analyses and processing described above.

[0152] A system embodying the invention includes a series of processing steps including image acquisition, pupil detection, scene location, natural pupil generation, merging with the original image, providing a processed image, and generating a comparison image.

[0153] First, the server has a means of acquiring images. These images may be uploaded from surveillance cameras or smart devices. Users upload images and video data from their own devices to the server. This data is temporarily stored on the cloud server.

[0154] Next, the server has a means to detect the eyes in the captured images. Specifically, it uses an image analysis algorithm to perform face recognition and identify the position of the eyes. For this, an image processing library such as OpenCV is used. A face detector and an eye detector (e.g., Haar Cascade Classifier) ​​are used together to identify the eyes.

[0155] The server then uses a method to identify the landscape portion reflected in the detected pupil, analyzing the landscape and background in the pupil at the pixel level and identifying its range. OpenCV is mainly used here as well.

[0156] The identified landscape elements are then removed using a server function. AI techniques such as generative adversarial networks (GANs) are then used to generate natural-looking pupils. Using GANs, natural-looking pupils are reconstructed based on the original pupil shape and lighting information.

[0157] The server then merges the generated natural pupils with the original image, which is a technique or process that incorporates the generated pupils into the original image so that the overall image looks natural. Again, OpenCV is used to combine the images.

[0158] The processed image is provided to the user's device from the server. Also, based on the generated privacy-protected image data, comparison images before and after processing are generated, allowing the user to visually compare the original and processed images.

[0159] As a concrete example, there is a process to upload security camera footage to a system, remove the scenery reflected in the eyes of each person in the footage, and generate natural-looking eyes. An example of a prompt for the program that performs this process is as follows:

[0160] Example prompt sentence:

[0161] Please convert the scenery reflected in the eyes in the image below into natural eyes.

[0162] Image: path / to / your / image.jpg

[0163] Algorithm: GANModel

[0164] Output: Privacy-preserving natural eye images

[0165] This makes it possible to protect privacy while preventing the leakage of personal information in images captured by surveillance cameras and smart devices.

[0166] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0167] Step 1:

[0168] Users upload images and video data captured from surveillance cameras or smart devices to a server. The input is image data sent from the user's device. The output is image data that is temporarily stored on the server. Specifically, the user uses a dedicated application to send the video data to the cloud server.

[0169] Step 2:

[0170] The server detects the eyes in the acquired images. The input is the image data saved in step 1. The output is face analysis data including eye position information. Specifically, the server uses the OpenCV library to perform face recognition and identify the eye positions. In this process, it applies face detection and eye detection algorithms such as the Haar Cascade Classifier.

[0171] Step 3:

[0172] The server identifies the part of the scenery reflected in the detected pupils. The input is face analysis data including pupil position information obtained in step 2. The output is data in which the scenery part is identified at the pixel level. Specifically, the server uses OpenCV to perform a detailed analysis of the scenery and background in each pupil and identify its range.

[0173] Step 4:

[0174] The server removes the identified scenery and generates natural-looking pupils. The input is image data including the scenery identified in step 3. The output is image data of natural-looking pupils generated using a generative adversarial network (GAN). Specifically, the server applies the GAN model to reconstruct natural-looking pupils based on the original pupil shape and light source information while removing the scenery.

[0175] Step 5:

[0176] The server merges the generated natural pupils with the original image. The input is the image data of the natural pupils generated in step 4. The output is the final merged image data. Specifically, the server uses OpenCV to incorporate the generated pupils into the original image and adjust it so that the whole image looks natural.

[0177] Step 6:

[0178] The server provides the merged image to the user and also generates before-and-after images for comparison. The input is the final image data obtained in step 5. The output is the final image sent to the user's device and the before-and-after images for comparison. Specifically, the server sends the image data to the user's device and prepares it so that the user can check the results.

[0179] Through these steps, the system protects privacy while preventing the leakage of personal information.

[0180] 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.

[0181] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results. How this system is implemented will be specifically described below.

[0182] Program processing flow

[0183] 1. Upload an image

[0184] User: Uploads an image to be processed from their own device to the system. The image can be a selfie or a group photo.

[0185] Server: Receives uploaded images and stores them in temporary storage, then prepares them for the analysis process.

[0186] 2. Image Analysis

[0187] Server: Uses image analysis algorithms to detect faces in the received images and obtain their detailed location information.

[0188] 3. Detection of the landscape part of the pupil

[0189] Server: Based on the facial position information, the position of the eyes is identified, and the part of the scenery reflected in the eyes is further analyzed at the pixel level to identify it.

[0190] 4. Eye Image Generation

[0191] Server: Removes the identified landscape features and generates natural-looking pupils using image generation AI (e.g., generative adversarial networks (GANs)). During this process, the original pupil shape and lighting information are used as a reference to maintain a natural appearance.

[0192] 5. Applying the Emotion Engine

[0193] Server: In parallel with the eye detection, the emotion engine runs to analyze the user's facial expressions and recognize their emotions. This engine identifies emotions such as smile, sadness, surprise, etc.

[0194] Server: Adjust the color and shine of the automatically generated eyes based on the recognized emotion. For example, add a slight sparkle to the eyes when smiling.

[0195] 6. Image Merging

[0196] Server: The adjusted natural pupils are merged with the original image, and the whole image is processed to look natural.

[0197] 7. Providing edited images

[0198] Server: Returns the completed processed image to the user's device, along with comparison images of before and after processing, allowing the user to confirm the changes.

[0199] Specific examples

[0200] Example 1: Editing a selfie

[0201] User: Uploads a selfie to the application.

[0202] Server: Receives the image and starts face recognition and eye detection. At the same time, the emotion engine analyzes the user's facial expressions.

[0203] Server: Analyzes and identifies the part of the scenery reflected in the eyes. The emotion engine recognizes that the user is smiling.

[0204] Server: Uses image generation AI to generate natural-looking eyes while adding a slight glow effect.

[0205] Server: Sends the processed image to the user.

[0206] User: Review the received images and ensure privacy is maintained.

[0207] Example 2: Editing a group photo

[0208] User: Uploads a group photo of family and friends to the system.

[0209] Server: Receives the image and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[0210] Server: Identifies the part of the landscape reflected in each eye and uses AI to generate natural-looking eyes for each.

[0211] Server: Adds appropriate effects to the generated eyes based on the emotions recognized by the emotion engine.

[0212] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[0213] User: Review the received images and ensure their privacy is protected.

[0214] This system allows users to obtain images with natural-looking eyes that reflect emotions while preventing privacy violations due to the scenery reflected in the eyes. It can also be introduced to businesses, and is expected to be widely used.

[0215] The processing flow will be explained below.

[0216] Step 1:

[0217] User: Uploads images that require privacy protection and emotion recognition to the system from their own device, such as selfies or group photos, by selecting and submitting them through the application or web interface.

[0218] Step 2:

[0219] Server: Receives the uploaded images and stores the image data in temporary storage, which prepares it for the subsequent analysis process.

[0220] Step 3:

[0221] Server: Activates the image analysis algorithm to detect faces in the received image. The face detection algorithm identifies and records the position and contour information of the face.

[0222] Step 4:

[0223] Server: Based on the face position information, the eye position is identified. Here, the eye detection algorithm detects the eye feature points and identifies the pupil area at pixel level.

[0224] Step 5:

[0225] Server: Performs a detailed analysis of the external scenery reflected within the detected pupil area and identifies that area. During this process, the pupil reflection and surrounding scenery information are analyzed pixel by pixel to extract unnecessary information.

[0226] Step 6:

[0227] Server: Activates the emotion engine and analyzes facial expressions in the image. The emotion engine identifies emotions such as smile, sadness, surprise, etc. and records the data.

[0228] Step 7:

[0229] Server: Removes the identified landscape features and generates natural-looking eyes using image generation AI (e.g., generative adversarial networks (GANs)). This process takes into account not only the original eye shape and lighting information, but also emotion data recognized by the emotion engine, preserving a natural appearance.

[0230] Step 8:

[0231] Server: Adjusts the color tone and gloss of the generated eyes based on the emotion engine data. For example, if the person is smiling, the eyes will be given a slight sparkle, making adjustments according to the emotion.

[0232] Step 9:

[0233] Server: Merge the adjusted natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[0234] Step 10:

[0235] Server: Returns the processed image to the user's device, along with a comparison image of the original and the processed image, allowing the user to easily verify the changes.

[0236] Step 11:

[0237] User: Receives the processed image, checks the processing results and any additional changes, and once they have confirmed that privacy is protected and that natural-looking eyes have been generated according to the emotion, they can save the image to their device or post it safely to social media.

[0238] Example 2

[0239] 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."

[0240] In modern society, as opportunities for image sharing increase, protecting user privacy has become an important issue. In particular, the scenery reflected in the eyes of images can potentially infringe on a user's privacy. It is also necessary to generate natural-looking eyes that reflect the user's emotions and maintain the realism of the entire image. Therefore, there is a need for technology that can reliably remove the scenery reflected in the eyes and generate natural-looking eyes that take emotions into account.

[0241] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for recognizing emotions from the user's facial expression, means for adjusting the generated natural-looking pupils based on the recognized emotion, and means for providing the adjusted image to the user. This makes it possible to generate natural-looking pupils according to the emotion while protecting the user's privacy.

[0242] The "means for acquiring images" provides an interface for users to upload images taken from their own terminals to the server.

[0243] The "means for detecting pupils" uses an image analysis algorithm to identify the position of pupils in the acquired image.

[0244] The "means for identifying the scenery portion" is a means for analyzing and identifying the scenery portion reflected in the pupil at the pixel level based on the pupil position.

[0245] The "means for generating natural pupils" involves removing the identified scene features and generating new pupil images using a generative model.

[0246] The "means for merging with the original image" is a means for integrating the generated natural pupils with the original image, and processing the entire image so that it looks natural.

[0247] "Means for recognizing emotions from facial expressions" refers to algorithms or devices for analyzing a user's facial expressions and identifying emotions from those expressions.

[0248] The "means for adjusting the generated natural pupils" adjusts the color tone and gloss of the generated pupils based on the recognized emotion.

[0249] The "means for providing the adjusted image to the user" means transmitting the completed processed image to the user's terminal so that the user can check it.

[0250] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results.

[0251] Specific means for implementing this system include the following steps involving a server, a terminal, and a user.

[0252] First, the user uploads the image that needs to be processed from their own device (for example, a personal computer or smartphone) to the system. The uploaded image is sent to a server and saved in temporary storage such as cloud storage. A specific storage method here could be to use a cloud storage service such as Amazon S3.

[0253] Next, the server uses image analysis libraries such as OpenCV and dlib to detect faces in the received images. This process obtains the face's position information (bounding box) and identifies the position of the eyes based on that. dlib's landmark detection technology is effective for identifying the position of the eyes. At this stage, pixel-level analysis is used to identify the part of the landscape reflected in the eyes.

[0254] The server removes the scenery and generates new, natural-looking eyes using a generative adversarial network (GAN). The generated eyes maintain the original eye shape and lighting information, achieving a natural look. The server also simultaneously uses an emotion engine (such as Microsoft® Azure® Face API) to recognize emotions from the user's facial expressions. Based on the recognition results, the server adjusts the color tone and gloss of the generated natural-looking eyes. For example, when smiling, it can add a slight sparkle to the eyes.

[0255] Finally, the adjusted natural pupils are merged with the original image to create a natural-looking overall image. Alpha blending is used to combine the generated pupils with the original image. The completed processed image is sent from the server to the user's device, and comparison images before and after are also provided so the user can confirm the changes.

[0256] Specific examples

[0257] Example 1: Editing a selfie

[0258] 1. User: Uploads a selfie image from a smartphone application.

[0259] 2. Server: Receives the image and performs face recognition and eye detection, while analyzing the user's facial expressions using the emotion engine.

[0260] 3. Server: Identify the parts of the landscape that are reflected in the eyes and generate natural-looking eyes using GAN, adding sparkle to the eyes when smiling.

[0261] 4. Server: Merge the new pupils with the original image and send the processed image to the user.

[0262] 5. User: Check the edited image and ensure your privacy is protected.

[0263] Example 2: Editing a group photo

[0264] 1. User: Uploads a group photo of family and friends from a PC.

[0265] 2. Server: Receives the images and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[0266] 3. Server: Identify the part of the scene reflected in each eye and generate natural-looking eyes for each eye, for example, adding a glow to the emotion of joy.

[0267] 4. Server: Merge the generated pupils with the original image and send the processed image to the user.

[0268] 5. User: Check the edited image and ensure that your privacy is protected.

[0269] Example prompts to input to the generative AI model

[0270] Example prompt:

[0271] "Please provide a step-by-step outline of a system that removes the landscape reflected in the eyes from user-uploaded images and generates natural-looking eyes using a generative adversarial network (GAN). Also, please describe how to recognize the user's emotion and apply eye effects accordingly."

[0272] This system can obtain images with natural-looking eyes that correspond to emotions while preventing the invasion of privacy of the user's eyes. It can also be introduced to businesses, and is expected to be widely used.

[0273] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0274] Step 1:

[0275] Uploading an image

[0276] User: Upload an image to the system from their device.

[0277] Input: The image that needs to be enhanced (e.g. a selfie).

[0278] Specific behavior: Selects an image file using the device's file selection dialog.

[0279] Output: The image file is sent to the server.

[0280] Step 2:

[0281] Receiving and saving images

[0282] Server: Receives uploaded images and temporarily stores them in cloud storage.

[0283] Input: An image file submitted by the user.

[0284] Specific operation: The received image data is saved in cloud storage such as Amazon S3.

[0285] Output: The storage path of the saved image file.

[0286] Step 3:

[0287] Image analysis and facial recognition

[0288] Server: Detect faces using an image analysis library.

[0289] Input: Image files stored in cloud storage.

[0290] Specific operation: Detects faces using OpenCV's cv2.CascadeClassifier and obtains their position information as a bounding box.

[0291] Output: Face position information (bounding box).

[0292] Step 4:

[0293] Detecting the landscape part of the pupil

[0294] Server: Identifies the position of the eyes based on face position information and analyzes the part of the scenery reflected in the eyes.

[0295] Input: Face bounding box information and image files.

[0296] What it does: It uses dlib landmark detection to identify the position of the pupil, then analyzes pixel values ​​to extract the part of the landscape reflected in the pupil.

[0297] Output: Eye position and landscape pixel information.

[0298] Step 5:

[0299] Eye image generation

[0300] Server: Generates new natural-looking pupils using a generative model (GAN).

[0301] Input: Pixel information of the landscape part of the pupil.

[0302] How it works: First, mask out the landscape area, then input that area into the GAN, which then generates natural-looking pupils.

[0303] Output: Generated pupil image data.

[0304] Step 6:

[0305] Applying the Emotion Engine

[0306] Server: Analyzes facial expressions and recognizes emotions.

[0307] Input: Image file and face position information.

[0308] Specific operation: Using an emotion engine (for example, Microsoft Azure's Face API), emotions are recognized from facial expressions, and the color and gloss of the generated eyes are adjusted based on that information.

[0309] Output: Recognized emotion data and adjusted eye image data.

[0310] Step 7:

[0311] Image merging

[0312] Server: Integrate the adjusted natural pupils into the original image.

[0313] Input: Adjusted pupil image data and original image files.

[0314] What it does: Uses alpha blending to naturally merge the generated pupil with the original image.

[0315] Output: The finished processed image.

[0316] Step 8:

[0317] Providing processed images

[0318] Server: Sends the processed image to the user's device.

[0319] Input: The finished processed image.

[0320] Specific operation: Comparison images before and after processing are also generated and sent back to the user. WebSocket is used as a real-time notification method.

[0321] Output: The processed and comparison images sent to the user.

[0322] (Application example 2)

[0323] 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."

[0324] Because the scenery reflected in the eyes in an image can invade privacy, a means to prevent this is needed. Also, to make the image look more natural, it is necessary to adjust the color tone and brightness of the eyes according to the user's emotions. In brick-and-mortar stores, technology is needed to protect privacy from selfies and other people's gazes through mirrors while improving the user experience.

[0325] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for providing the merged image to the user, and means for recognizing the user's emotion and adjusting the color tone and gloss of the pupils based on the result. This makes it possible to adjust the pupils in real time according to the emotion while maintaining a natural appearance and protecting the user's privacy.

[0326] "Means for acquiring images" refers to a combination of hardware and software for capturing images from a user's information processing device or camera.

[0327] The "means for detecting pupils in the acquired image" is an algorithm for identifying pupils in the image and obtaining their position information.

[0328] The "means for identifying the portion of the scenery reflected in the detected pupil" is a process for identifying the external scenery or objects reflected in the pupil and analyzing it on a pixel-by-pixel basis.

[0329] The "means for removing identified landscape features and generating natural-looking eyes" is a system that uses AI techniques such as generative adversarial networks (GANs) to generate natural-looking eye images after removing the landscape features.

[0330] The "means for merging the generated natural pupils into the original image" is a technique for integrating the generated pupil images into the original photograph, making the whole look natural.

[0331] The "means for providing a merged image to a user" is a system for transmitting or displaying the processed image to a user.

[0332] "Means for recognizing the user's emotions and adjusting the color tone and gloss of the pupils based on the results" refers to a process that uses an emotion recognition engine to analyze the user's facial expression and appropriately corrects the characteristics of the pupils (color tone and gloss) generated based on that.

[0333] This invention is part of a smart mirror system that also protects privacy in brick-and-mortar stores. It removes the scenery reflected in the eyes in the image, generates natural-looking eyes based on the user's emotions, and provides real-time feedback.

[0334] Program processing flow

[0335] 1. Image Acquisition:

[0336] When a user stands in front of the smart mirror, the mirror's camera automatically captures an image, which is then seamlessly transmitted to the server.

[0337] 2. Eye detection:

[0338] The server uses a facial recognition algorithm such as OpenCV to detect faces and eyes in the image, which allows the precise location of the eyes.

[0339] 3. Identifying the landscape:

[0340] The server zooms in on the detected pupil area and performs pixel-by-pixel analysis, identifying specific parts of the scene or object and extracting that pixel information.

[0341] 4. Removing landscape elements and generating natural-looking eyes:

[0342] The server uses a generative adversarial network (GAN) to remove the landscape and generate natural-looking pupils, taking into account the original pupil shape and lighting information to maintain harmony across the entire image.

[0343] 5. Emotion Recognition:

[0344] The server uses an emotion recognition engine such as AWS® Rekognition to analyze the user's facial expressions and read their emotions. Based on this emotional information, the color and gloss of the eyes are adjusted appropriately.

[0345] 6. Image merging:

[0346] The server integrates the generated pupils into the original image and processes it so that the whole image looks natural.

[0347] 7. Provide processed images:

[0348] The server provides the completed image to the user in real time, allowing the user to instantly check the edited image.

[0349] Hardware and software used

[0350] Camera: A camera device attached to the smart mirror.

[0351] Face Recognition Algorithm: OpenCV

[0352] Generative AI Models: Generative Adversarial Networks (GANs)

[0353] Emotion recognition engine: AWS Rekognition

[0354] Server: Server for image analysis and data processing

[0355] Specific examples

[0356] A specific example in which a user actually stands in front of a mirror is shown below.

[0357] When a user stands in front of the smart mirror, the camera automatically recognizes their face and eyes and captures an image. The emotion recognition engine detects when the user is smiling and applies a slightly sparkling eye effect. This allows the user to view the processed image in real time, ensuring privacy while allowing them to see natural-looking eyes that reflect their emotions.

[0358] Prompt Sentence Examples

[0359] "Write a Python program to protect image privacy and generate natural-looking eye pupils based on emotions. The following requirements must be met:

[0360] 1. Upload an image

[0361] 2. Face and Eye Analysis

[0362] 3. Removing the scenery reflected in the eyes

[0363] 4. Emotion recognition and pupil generation

[0364] 5. Merging and delivering the final image

[0365] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0366] Step 1:

[0367] The server acquires images from the smart mirror's camera. When a user stands in front of the mirror, the camera automatically activates and captures an image containing the user's face. The input is the raw image from the camera, and the output is an image file sent to the server.

[0368] Step 2:

[0369] The server uses a face recognition algorithm (e.g., OpenCV) to detect the pupils in the captured image. The input is the image sent to the server, and the output is the coordinate information of the pupils. The server scans the entire face to identify the pupil positions as a data calculation, and identifies a rectangular area that includes the eye positions.

[0370] Step 3:

[0371] The server performs pixel-by-pixel analysis to identify the landscape portion of the detected pupil. The input is the coordinate information of the pupil and the image range of the detected face, and the output is pixel information of the landscape portion. The server enlarges the pupil area and analyzes the reflection and different color tones within it to obtain the landscape portion.

[0372] Step 4:

[0373] The server uses a generative adversarial network (GAN) to remove the identified scene and generate natural-looking pupils. The input is pixel information of the scene, and the output is an image of a natural-looking pupil. The server inputs the scene information into the GAN model and obtains pixel data of the generated natural-looking pupils in return.

[0374] Step 5:

[0375] The server uses an emotion recognition engine (e.g., AWS Rekognition) to analyze the user's facial expressions and recognize their emotions. The input is an image of the user's face, and the output is the user's emotion data. The server identifies emotions such as smile, sadness, or surprise from the user's facial expressions through natural language processing.

[0376] Step 6:

[0377] The server adjusts the color tone and gloss of the generated eyes based on the recognized emotion. The input is emotion data and a generated natural eye image, and the output is an eye image adjusted according to the emotion. The server analyzes the emotion data and applies effects such as brightening the eyes if the person is smiling.

[0378] Step 7:

[0379] The server merges the adjusted natural pupils with the original image and processes it so that the whole image looks natural. The input is the adjusted natural pupil image and the original face image, and the output is the processed final image. The server processes the data by pasting the new pupils pixel by pixel to the original image.

[0380] Step 8:

[0381] The server provides the completed processed image to the user's information processing device. The input is the final processed image, and the output is image data sent to the user. The user can check the processed image on their own device, safely protecting their privacy while also being able to see natural-looking eyes based on their emotions.

[0382] 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.

[0383] 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.

[0384] 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.

[0385] [Second embodiment]

[0386] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0387] 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.

[0388] 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).

[0389] 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.

[0390] 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.

[0391] 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).

[0392] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0393] 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.

[0394] 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.

[0395] 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.

[0396] 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.

[0397] 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."

[0398] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. The specific operation of each step of this system is shown below.

[0399] Program processing flow

[0400] 1. Upload an image

[0401] User: Uploads images that require privacy protection from their own device to the system. These images can be anything from selfies to group photos.

[0402] Server: Receives and temporarily stores the images uploaded above, preparing them for the subsequent analysis process.

[0403] 2. Image Analysis

[0404] Server: Inputs the received image into an image analysis algorithm, which performs facial recognition and identifies the location of facial features and eyes.

[0405] 3. Detection of the landscape part of the pupil

[0406] Server: Performs detailed analysis of the scenery reflected in the detected pupils and identifies the scenery at the pixel level. This information is used in subsequent processing steps.

[0407] 4. Eye Image Generation

[0408] Server: Removes the landscape and applies an AI algorithm (e.g., a generative adversarial network (GAN)) to generate natural-looking eye images. The AI ​​takes into account the original pupil shape and light source information to reconstruct natural-looking eyes.

[0409] 5. Image Merging

[0410] Server: Integrate the generated natural pupils into the original image, replacing the original eye regions with the new pupil image, making the whole image look natural.

[0411] 6. Providing edited images

[0412] Server: Returns the completed processed image to the user's device. In addition, it also provides comparison images before and after processing so that the user can check the processing results.

[0413] Specific examples

[0414] Example 1: Editing a selfie

[0415] User: Uploads a selfie to the application.

[0416] Server: Receives the image and starts eye detection.

[0417] Server: Analyzes and identifies the part of the scenery reflected in the eyes.

[0418] Server: Generates natural-looking pupils using image generation AI.

[0419] Server: Sends the processed image to the user.

[0420] User: Review the received images and ensure their privacy is protected.

[0421] Example 2: Editing a group photo

[0422] User: Uploads a group photo of family and friends to the system.

[0423] Server: Receives the image and starts eye detection.

[0424] Server: Identify the landscape portion of each pupil and generate natural-looking pupils using AI.

[0425] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[0426] User: Review the received images and ensure their privacy is protected.

[0427] This system allows users to share images safely while protecting their privacy, and is expected to be widely adopted by businesses as well.

[0428] The processing flow will be explained below.

[0429] Step 1:

[0430] User: Uploads images from their device to the system, specifically by selecting a selfie or group photo and sending it to the server through the application or web interface.

[0431] Step 2:

[0432] Server: Receives uploaded images and stores them in temporary storage. After storing them, the server prepares them for image analysis.

[0433] Step 3:

[0434] Server: Uses an image analysis algorithm to detect faces in the received image. The face detection algorithm identifies the position and contours of the face.

[0435] Step 4:

[0436] Server: Identify the position of the pupils based on the facial position information. The pupil detection algorithm extracts the eye feature points and identifies the pupil area at the pixel level.

[0437] Step 5:

[0438] Server: Analyzes and identifies the scenery reflected within the detected pupil area. Specifically, it analyzes the light reflected from the pupil and the surrounding scenery information for each pixel and extracts unnecessary information.

[0439] Step 6:

[0440] Server: Using image generation AI (e.g., generative adversarial networks (GANs)), the identified landscape features are removed and natural-looking pupils are generated, while maintaining a natural appearance by taking into account the original pupil shape and lighting information.

[0441] Step 7:

[0442] Server: Merge the generated natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[0443] Step 8:

[0444] Server: Returns the processed image to the user, along with a before and after comparison image so the user can see the changes.

[0445] Step 9:

[0446] User: Receives the processed image and checks the results. Once the user confirms that their privacy has been protected, they can save the image to their device or post it safely to social media.

[0447] Example 1

[0448] 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."

[0449] With conventional technology, the landscape portion of the eyes in an image could be a violation of privacy, making it difficult for users to share images with confidence. Furthermore, manually editing the eyes requires time and skill, making it inefficient. Furthermore, the image processing process can sometimes produce unnatural results, making the processed image appear unnatural. To solve these problems, a system is needed that can automatically detect and process the landscape portion of the eyes, maintaining a natural appearance.

[0450] 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.

[0451] In this invention, the server includes means for a user to upload an image from a terminal, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, and means for providing the merged image to the user. This enables users to easily edit and safely share images while ensuring privacy.

[0452] "Means for users to upload images from their devices" refers to a system that allows users to select image files using their own devices and send them to a server via a network.

[0453] The "means for detecting pupils in a captured image" is a system that has the function of identifying the positions of pupils in a captured image using image processing technology.

[0454] The "means for identifying the portion of the scenery reflected in the detected pupil" is a system that has the function of analyzing the portion of the scenery reflected in the pupil area and identifying that portion at the pixel level.

[0455] The "means for removing identified landscape parts and generating natural-looking pupils" refers to a system that uses AI technology and image generation algorithms to remove identified landscape parts and generate natural-looking pupils by taking into account the original pupil shape and light source information.

[0456] The "means for merging the generated natural pupils into the original image" is a system that has the function of integrating the newly generated natural pupils into the original image to make the image appear natural overall.

[0457] The "means for providing a merged image to a user" refers to a system that has a function for returning the processed image to a user so that the user can check the result.

[0458] A "generative adversarial network" is a type of deep learning model that can generate new, high-quality data by learning through a competitive process between a generative model and a discriminative model.

[0459] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. Specific embodiments for implementing this system are described below.

[0460] First, the user uploads an image that requires privacy protection to the system using the device. Specifically, the user opens the application's file selection dialog and selects an image file from local storage. This operation causes the device to send the image to the server. The image file data is sent to the server using an HTTP POST request.

[0461] The server temporarily stores the received image file in storage. The path of the stored file is retained for the next processing step. Specific hardware used is a server computer, and software used is an HTTP server and a storage management system.

[0462] Next, the server inputs the saved image into an image analysis algorithm. Specifically, it uses the OpenCV and dlib libraries to perform face recognition. The image is read using OpenCV's cv2.imread function and converted to grayscale using cv2.cvtColor. The dlib face detector is used to detect the position of the face, and the dlib landmark detector is used to identify the position of the eye. This series of processes results in the detection of the eye in the image.

[0463] To analyze the landscape part reflected in the detected pupil in detail, the server identifies the landscape part pixel by pixel. For this purpose, OpenCV's cv2.boundingRect function is used. The identified landscape part information is used in the subsequent processing steps.

[0464] The server then uses a generative adversarial network (GAN) to remove the landscape elements from the pupils and generate natural-looking pupils. To do this, the image is resized and normalized to the appropriate size. A pre-trained GAN model is then used. Specific software used is TensorFlow and PyTorch.

[0465] To integrate the generated natural pupils into the original image, the server uses OpenCV's cv2.seamlessClone function, which blends the newly generated pupils seamlessly into the original image, maintaining an overall natural look.

[0466] Finally, the server returns the completed processed image to the user's device. The processed image is returned as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can then view and check the received image.

[0467] Specific examples

[0468] Selfie photo editing

[0469] The user uploads a selfie to the application. The device sends the image to the server, which uses OpenCV and dlib to initiate face and eye detection. The server uses a custom algorithm to identify the landscape portion of the eye and generates natural-looking eyes using a GAN. The server merges the new eye image with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[0470] Editing group photos

[0471] A user uploads a group photo of family and friends to the system. The device sends the image to the server, which uses OpenCV and dlib to initiate multiple face and eye detection. The server uses a custom algorithm to identify the landscape portion of each eye and uses GAN to generate natural-looking eyes for each eye. The server then merges the generated eyes with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[0472] This system allows users to edit and share images safely and easily while protecting their privacy.

[0473] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0474] Step 1:

[0475] The user uploads an image requiring privacy protection to the system using their own device. Specifically, they open the application's file selection dialog and select an image file from local storage. This operation causes the device to send the image data to the server via an HTTP POST request.

[0476] Input: An image file selected by the user.

[0477] Output: Image data sent to the server

[0478] Step 2:

[0479] The server temporarily stores the received image data. Specifically, it stores the received image file in storage and retains the file path for the next analysis step.

[0480] Input: Image data sent to the server

[0481] Output: Image file saved to storage

[0482] Step 3:

[0483] The server inputs the saved image into the image analysis algorithm. First, it reads the image using OpenCV's cv2.imread function, then converts the image to grayscale using the cv2.cvtColor function. Then it uses dlib's face detector (dlib.get_frontal_face_detector) to detect the face position. Finally, it uses dlib's landmark detector (dlib.shape_predictor) to identify the eye position.

[0484] Input: Image file saved in storage

[0485] Output: Face and eye position information

[0486] Step 4:

[0487] The server then performs a detailed analysis of the landscape reflected within the detected pupil. Specifically, it uses OpenCV's cv2.boundingRect function to extract the pupil area pixel by pixel. It then analyzes the pixel values ​​within this pupil area using a custom algorithm to identify the landscape.

[0488] Input: Face and eye position information

[0489] Output: Pixel information of the landscape

[0490] Step 5:

[0491] The server uses a generative adversarial network (GAN) to remove the identified landscape features and generate natural-looking pupils. First, the pupil regions are resized and normalized to an appropriate size as input for the GAN. Then, a pre-trained GAN model is used to generate natural-looking pupils. This process uses TensorFlow and PyTorch.

[0492] Input: Pixel information of the landscape

[0493] Output: Natural eye image

[0494] Step 6:

[0495] The server merges the generated natural pupils with the original image, using OpenCV's cv2.seamlessClone function to seamlessly integrate the newly generated pupil images into the original image, resulting in an overall image that maintains a natural appearance.

[0496] Input: natural eye image, original image

[0497] Output: Natural-looking processed image

[0498] Step 7:

[0499] The server returns the processed image to the user. The processed image is sent to the user's device as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can view the received image and confirm that their privacy has been protected.

[0500] Input: Natural-looking processed image

[0501] Output: The processed image sent to the user

[0502] (Application example 1)

[0503] 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."

[0504] In modern society, there is an increasing trend of capturing video footage using surveillance cameras and smart devices and using it for various purposes. However, these videos contain personal information, and the scene reflected in the eyes of people's eyes is particularly at high risk of being leaked. This raises concerns about privacy violations, and there is a demand for technology to safely and efficiently protect personal information.

[0505] 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.

[0506] In this invention, the server includes means for acquiring an image, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, means for providing the merged image to a user, means for providing comparison images before and after processing based on the generated privacy-protected image data, and means for performing the above processing on video data uploaded from a surveillance camera or a smart device. This makes it possible to achieve privacy protection while preventing the leakage of personal information in videos acquired from a surveillance camera or a smart device.

[0507] "Means for acquiring images" refers to a device or method for acquiring images or video using a surveillance camera or smart device and receiving them in a form that is input into the system.

[0508] The "means for detecting eyes" is an algorithm or technique for analyzing the facial area in the acquired image and specifically identifying the location of the eyes.

[0509] The "means for identifying the scenery portion" refers to a technology or algorithm that analyzes the scenery or background reflected in the detected pupil at the pixel level and identifies its range.

[0510] The "means for generating natural-looking pupils" refers to AI techniques such as generative adversarial networks (GANs) that remove detected landscape features and reconstruct a natural appearance based on the original pupil shape and light source information.

[0511] "Means for merging into original image" refers to the technique or process used to incorporate the generated natural pupils into the original image so that the overall image appears natural.

[0512] "Means for providing to the user" refers to a system or technology that returns the merged image to the user's terminal so that the user can view the results.

[0513] "Means for providing images before and after processing for comparison" refers to a display function for comparing the privacy-protected image with the original image, or technology that allows users to visually confirm the differences.

[0514] "Means for processing video data uploaded from surveillance cameras and smart devices" refers to a system or technology that receives video data sent from surveillance cameras, smartphones, tablets, etc., and automatically performs the series of analyses and processing described above.

[0515] A system embodying the invention includes a series of processing steps including image acquisition, pupil detection, scene location, natural pupil generation, merging with the original image, providing a processed image, and generating a comparison image.

[0516] First, the server has a means of acquiring images. These images may be uploaded from surveillance cameras or smart devices. Users upload images and video data from their own devices to the server. This data is temporarily stored on the cloud server.

[0517] Next, the server has a means to detect the eyes in the captured images. Specifically, it uses an image analysis algorithm to perform face recognition and identify the position of the eyes. For this, an image processing library such as OpenCV is used. A face detector and an eye detector (e.g., Haar Cascade Classifier) ​​are used together to identify the eyes.

[0518] The server then uses a method to identify the landscape portion reflected in the detected pupil, analyzing the landscape and background in the pupil at the pixel level and identifying its range. OpenCV is mainly used here as well.

[0519] The identified landscape elements are then removed using a server function. AI techniques such as generative adversarial networks (GANs) are then used to generate natural-looking pupils. Using GANs, natural-looking pupils are reconstructed based on the original pupil shape and lighting information.

[0520] The server then merges the generated natural pupils with the original image, which is a technique or process that incorporates the generated pupils into the original image so that the overall image looks natural. Again, OpenCV is used to combine the images.

[0521] The processed image is provided to the user's device from the server. Also, based on the generated privacy-protected image data, comparison images before and after processing are generated, allowing the user to visually compare the original and processed images.

[0522] As a concrete example, there is a process to upload security camera footage to a system, remove the scenery reflected in the eyes of each person in the footage, and generate natural-looking eyes. An example of a prompt for the program that performs this process is as follows:

[0523] Example prompt sentence:

[0524] Please convert the scenery reflected in the eyes in the image below into natural eyes.

[0525] Image: path / to / your / image.jpg

[0526] Algorithm: GANModel

[0527] Output: Privacy-preserving natural eye images

[0528] This makes it possible to protect privacy while preventing the leakage of personal information in images captured by surveillance cameras and smart devices.

[0529] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0530] Step 1:

[0531] Users upload images and video data captured from surveillance cameras or smart devices to a server. The input is image data sent from the user's device. The output is image data that is temporarily stored on the server. Specifically, the user uses a dedicated application to send the video data to the cloud server.

[0532] Step 2:

[0533] The server detects the eyes in the acquired images. The input is the image data saved in step 1. The output is face analysis data including eye position information. Specifically, the server uses the OpenCV library to perform face recognition and identify the eye positions. In this process, it applies face detection and eye detection algorithms such as the Haar Cascade Classifier.

[0534] Step 3:

[0535] The server identifies the part of the scenery reflected in the detected pupils. The input is face analysis data including pupil position information obtained in step 2. The output is data in which the scenery part is identified at the pixel level. Specifically, the server uses OpenCV to perform a detailed analysis of the scenery and background in each pupil and identify its range.

[0536] Step 4:

[0537] The server removes the identified scenery and generates natural-looking pupils. The input is image data including the scenery identified in step 3. The output is image data of natural-looking pupils generated using a generative adversarial network (GAN). Specifically, the server applies the GAN model to reconstruct natural-looking pupils based on the original pupil shape and light source information while removing the scenery.

[0538] Step 5:

[0539] The server merges the generated natural pupils with the original image. The input is the image data of the natural pupils generated in step 4. The output is the final merged image data. Specifically, the server uses OpenCV to incorporate the generated pupils into the original image and adjust it so that the whole image looks natural.

[0540] Step 6:

[0541] The server provides the merged image to the user and also generates before-and-after images for comparison. The input is the final image data obtained in step 5. The output is the final image sent to the user's device and the before-and-after images for comparison. Specifically, the server sends the image data to the user's device and prepares it so that the user can check the results.

[0542] Through these steps, the system protects privacy while preventing the leakage of personal information.

[0543] 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.

[0544] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results. How this system is implemented will be specifically described below.

[0545] Program processing flow

[0546] 1. Upload an image

[0547] User: Uploads an image to be processed from their own device to the system. The image can be a selfie or a group photo.

[0548] Server: Receives uploaded images and stores them in temporary storage, then prepares them for the analysis process.

[0549] 2. Image Analysis

[0550] Server: Uses image analysis algorithms to detect faces in the received images and obtain their detailed location information.

[0551] 3. Detection of the landscape part of the pupil

[0552] Server: Based on the facial position information, the position of the eyes is identified, and the part of the scenery reflected in the eyes is further analyzed at the pixel level to identify it.

[0553] 4. Eye Image Generation

[0554] Server: Removes the identified landscape features and generates natural-looking pupils using image generation AI (e.g., generative adversarial networks (GANs)). During this process, the original pupil shape and lighting information are used as a reference to maintain a natural appearance.

[0555] 5. Applying the Emotion Engine

[0556] Server: In parallel with the eye detection, the emotion engine runs to analyze the user's facial expressions and recognize their emotions. This engine identifies emotions such as smile, sadness, surprise, etc.

[0557] Server: Adjust the color and shine of the automatically generated eyes based on the recognized emotion. For example, add a slight sparkle to the eyes when smiling.

[0558] 6. Image Merging

[0559] Server: The adjusted natural pupils are merged with the original image, and the whole image is processed to look natural.

[0560] 7. Providing edited images

[0561] Server: Returns the completed processed image to the user's device, along with comparison images of before and after processing, allowing the user to confirm the changes.

[0562] Specific examples

[0563] Example 1: Editing a selfie

[0564] User: Uploads a selfie to the application.

[0565] Server: Receives the image and starts face recognition and eye detection. At the same time, the emotion engine analyzes the user's facial expressions.

[0566] Server: Analyzes and identifies the part of the scenery reflected in the eyes. The emotion engine recognizes that the user is smiling.

[0567] Server: Uses image generation AI to generate natural-looking eyes while adding a slight glow effect.

[0568] Server: Sends the processed image to the user.

[0569] User: Review the received images and ensure privacy is maintained.

[0570] Example 2: Editing a group photo

[0571] User: Uploads a group photo of family and friends to the system.

[0572] Server: Receives the image and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[0573] Server: Identifies the part of the landscape reflected in each eye and uses AI to generate natural-looking eyes for each.

[0574] Server: Adds appropriate effects to the generated eyes based on the emotions recognized by the emotion engine.

[0575] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[0576] User: Review the received images and ensure their privacy is protected.

[0577] This system allows users to obtain images with natural-looking eyes that reflect emotions while preventing privacy violations due to the scenery reflected in the eyes. It can also be introduced to businesses, and is expected to be widely used.

[0578] The processing flow will be explained below.

[0579] Step 1:

[0580] User: Uploads images that require privacy protection and emotion recognition to the system from their own device, such as selfies or group photos, by selecting and submitting them through the application or web interface.

[0581] Step 2:

[0582] Server: Receives the uploaded images and stores the image data in temporary storage, which prepares it for the subsequent analysis process.

[0583] Step 3:

[0584] Server: Activates the image analysis algorithm to detect faces in the received image. The face detection algorithm identifies and records the position and contour information of the face.

[0585] Step 4:

[0586] Server: Based on the face position information, the eye position is identified. Here, the eye detection algorithm detects the eye feature points and identifies the pupil area at pixel level.

[0587] Step 5:

[0588] Server: Performs a detailed analysis of the external scenery reflected within the detected pupil area and identifies that area. During this process, the pupil reflection and surrounding scenery information are analyzed pixel by pixel to extract unnecessary information.

[0589] Step 6:

[0590] Server: Activates the emotion engine and analyzes facial expressions in the image. The emotion engine identifies emotions such as smile, sadness, surprise, etc. and records the data.

[0591] Step 7:

[0592] Server: Removes the identified landscape features and generates natural-looking eyes using image generation AI (e.g., generative adversarial networks (GANs)). This process takes into account not only the original eye shape and lighting information, but also emotion data recognized by the emotion engine, preserving a natural appearance.

[0593] Step 8:

[0594] Server: Adjusts the color tone and gloss of the generated eyes based on the emotion engine data. For example, if the person is smiling, the eyes will be given a slight sparkle, making adjustments according to the emotion.

[0595] Step 9:

[0596] Server: Merge the adjusted natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[0597] Step 10:

[0598] Server: Returns the processed image to the user's device, along with a comparison image of the original and the processed image, allowing the user to easily verify the changes.

[0599] Step 11:

[0600] User: Receives the processed image, checks the processing results and any additional changes, and once they have confirmed that privacy is protected and that natural-looking eyes have been generated according to the emotion, they can save the image to their device or post it safely to social media.

[0601] Example 2

[0602] 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."

[0603] In modern society, as opportunities for image sharing increase, protecting user privacy has become an important issue. In particular, the scenery reflected in the eyes of images can potentially infringe on a user's privacy. It is also necessary to generate natural-looking eyes that reflect the user's emotions and maintain the realism of the entire image. Therefore, there is a need for technology that can reliably remove the scenery reflected in the eyes and generate natural-looking eyes that take emotions into account.

[0604] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for recognizing emotions from the user's facial expression, means for adjusting the generated natural-looking pupils based on the recognized emotion, and means for providing the adjusted image to the user. This makes it possible to generate natural-looking pupils according to the emotion while protecting the user's privacy.

[0605] The "means for acquiring images" provides an interface for users to upload images taken from their own terminals to the server.

[0606] The "means for detecting pupils" uses an image analysis algorithm to identify the position of pupils in the acquired image.

[0607] The "means for identifying the scenery portion" is a means for analyzing and identifying the scenery portion reflected in the pupil at the pixel level based on the pupil position.

[0608] The "means for generating natural pupils" involves removing the identified scene features and generating new pupil images using a generative model.

[0609] The "means for merging with the original image" is a means for integrating the generated natural pupils with the original image, and processing the entire image so that it looks natural.

[0610] "Means for recognizing emotions from facial expressions" refers to algorithms or devices for analyzing a user's facial expressions and identifying emotions from those expressions.

[0611] The "means for adjusting the generated natural pupils" adjusts the color tone and gloss of the generated pupils based on the recognized emotion.

[0612] The "means for providing the adjusted image to the user" means transmitting the completed processed image to the user's terminal so that the user can check it.

[0613] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results.

[0614] Specific means for implementing this system include the following steps involving a server, a terminal, and a user.

[0615] First, the user uploads the image that needs to be processed from their own device (for example, a personal computer or smartphone) to the system. The uploaded image is sent to a server and saved in temporary storage such as cloud storage. A specific storage method here could be to use a cloud storage service such as Amazon S3.

[0616] Next, the server uses image analysis libraries such as OpenCV and dlib to detect faces in the received images. This process obtains the face's position information (bounding box) and identifies the position of the eyes based on that. dlib's landmark detection technology is effective for identifying the position of the eyes. At this stage, pixel-level analysis is used to identify the part of the landscape reflected in the eyes.

[0617] The server removes the scenery and generates new, natural-looking eyes using a generative adversarial network (GAN). The generated eyes maintain the original eye shape and lighting information, achieving a natural look. The server also simultaneously uses an emotion engine (such as Microsoft Azure's Face API) to recognize emotions from the user's facial expressions. Based on the recognition results, the server adjusts the color tone and gloss of the generated natural-looking eyes. For example, when smiling, it can add a slight sparkle to the eyes.

[0618] Finally, the adjusted natural pupils are merged with the original image to create a natural-looking overall image. Alpha blending is used to combine the generated pupils with the original image. The completed processed image is sent from the server to the user's device, and comparison images before and after are also provided so the user can confirm the changes.

[0619] Specific examples

[0620] Example 1: Editing a selfie

[0621] 1. User: Uploads a selfie image from a smartphone application.

[0622] 2. Server: Receives the image and performs face recognition and eye detection, while analyzing the user's facial expressions using the emotion engine.

[0623] 3. Server: Identify the parts of the landscape that are reflected in the eyes and generate natural-looking eyes using GAN, adding sparkle to the eyes when smiling.

[0624] 4. Server: Merge the new pupils with the original image and send the processed image to the user.

[0625] 5. User: Check the edited image and ensure your privacy is protected.

[0626] Example 2: Editing a group photo

[0627] 1. User: Uploads a group photo of family and friends from a PC.

[0628] 2. Server: Receives the images and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[0629] 3. Server: Identify the part of the scene reflected in each eye and generate natural-looking eyes for each eye, for example, adding a glow to the emotion of joy.

[0630] 4. Server: Merge the generated pupils with the original image and send the processed image to the user.

[0631] 5. User: Check the edited image and ensure that your privacy is protected.

[0632] Example prompts to input to the generative AI model

[0633] Example prompt:

[0634] "Please provide a step-by-step outline of a system that removes the landscape reflected in the eyes from user-uploaded images and generates natural-looking eyes using a generative adversarial network (GAN). Also, please describe how to recognize the user's emotion and apply eye effects accordingly."

[0635] This system can obtain images with natural-looking eyes that correspond to emotions while preventing the invasion of privacy of the user's eyes. It can also be introduced to businesses, and is expected to be widely used.

[0636] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0637] Step 1:

[0638] Uploading an image

[0639] User: Upload an image to the system from their device.

[0640] Input: The image that needs to be enhanced (e.g. a selfie).

[0641] Specific behavior: Selects an image file using the device's file selection dialog.

[0642] Output: The image file is sent to the server.

[0643] Step 2:

[0644] Receiving and saving images

[0645] Server: Receives uploaded images and temporarily stores them in cloud storage.

[0646] Input: An image file submitted by the user.

[0647] Specific operation: The received image data is saved in cloud storage such as Amazon S3.

[0648] Output: The storage path of the saved image file.

[0649] Step 3:

[0650] Image analysis and facial recognition

[0651] Server: Detect faces using an image analysis library.

[0652] Input: Image files stored in cloud storage.

[0653] Specific operation: Detects faces using OpenCV's cv2.CascadeClassifier and obtains their position information as a bounding box.

[0654] Output: Face position information (bounding box).

[0655] Step 4:

[0656] Detecting the landscape part of the pupil

[0657] Server: Identifies the position of the eyes based on face position information and analyzes the part of the scenery reflected in the eyes.

[0658] Input: Face bounding box information and image files.

[0659] What it does: It uses dlib landmark detection to identify the position of the pupil, then analyzes pixel values ​​to extract the part of the landscape reflected in the pupil.

[0660] Output: Eye position and landscape pixel information.

[0661] Step 5:

[0662] Eye image generation

[0663] Server: Generates new natural-looking pupils using a generative model (GAN).

[0664] Input: Pixel information of the landscape part of the pupil.

[0665] How it works: First, mask out the landscape area, then input that area into the GAN, which then generates natural-looking pupils.

[0666] Output: Generated pupil image data.

[0667] Step 6:

[0668] Applying the Emotion Engine

[0669] Server: Analyzes facial expressions and recognizes emotions.

[0670] Input: Image file and face position information.

[0671] Specific operation: Using an emotion engine (for example, Microsoft Azure's Face API), emotions are recognized from facial expressions, and the color and gloss of the generated eyes are adjusted based on that information.

[0672] Output: Recognized emotion data and adjusted eye image data.

[0673] Step 7:

[0674] Image merging

[0675] Server: Integrate the adjusted natural pupils into the original image.

[0676] Input: Adjusted pupil image data and original image files.

[0677] What it does: Uses alpha blending to naturally merge the generated pupil with the original image.

[0678] Output: The finished processed image.

[0679] Step 8:

[0680] Providing processed images

[0681] Server: Sends the processed image to the user's device.

[0682] Input: The finished processed image.

[0683] Specific operation: Comparison images before and after processing are also generated and sent back to the user. WebSocket is used as a real-time notification method.

[0684] Output: The processed and comparison images sent to the user.

[0685] (Application example 2)

[0686] 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."

[0687] Because the scenery reflected in the eyes in an image can invade privacy, a means to prevent this is needed. Also, to make the image look more natural, it is necessary to adjust the color tone and brightness of the eyes according to the user's emotions. In brick-and-mortar stores, technology is needed to protect privacy from selfies and other people's gazes through mirrors while improving the user experience.

[0688] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for providing the merged image to the user, and means for recognizing the user's emotion and adjusting the color tone and gloss of the pupils based on the result. This makes it possible to adjust the pupils in real time according to the emotion while maintaining a natural appearance and protecting the user's privacy.

[0689] "Means for acquiring images" refers to a combination of hardware and software for capturing images from a user's information processing device or camera.

[0690] The "means for detecting pupils in the acquired image" is an algorithm for identifying pupils in the image and obtaining their position information.

[0691] The "means for identifying the portion of the scenery reflected in the detected pupil" is a process for identifying the external scenery or objects reflected in the pupil and analyzing it on a pixel-by-pixel basis.

[0692] The "means for removing identified landscape features and generating natural-looking eyes" is a system that uses AI techniques such as generative adversarial networks (GANs) to generate natural-looking eye images after removing the landscape features.

[0693] The "means for merging the generated natural pupils into the original image" is a technique for integrating the generated pupil images into the original photograph, making the whole look natural.

[0694] The "means for providing a merged image to a user" is a system for transmitting or displaying the processed image to a user.

[0695] "Means for recognizing the user's emotions and adjusting the color tone and gloss of the pupils based on the results" refers to a process that uses an emotion recognition engine to analyze the user's facial expression and appropriately corrects the characteristics of the pupils (color tone and gloss) generated based on that.

[0696] This invention is part of a smart mirror system that also protects privacy in brick-and-mortar stores. It removes the scenery reflected in the eyes in the image, generates natural-looking eyes based on the user's emotions, and provides real-time feedback.

[0697] Program processing flow

[0698] 1. Image Acquisition:

[0699] When a user stands in front of the smart mirror, the mirror's camera automatically captures an image, which is then seamlessly transmitted to the server.

[0700] 2. Eye detection:

[0701] The server uses a facial recognition algorithm such as OpenCV to detect faces and eyes in the image, which allows the precise location of the eyes.

[0702] 3. Identifying the landscape:

[0703] The server zooms in on the detected pupil area and performs pixel-by-pixel analysis, identifying specific parts of the scene or object and extracting that pixel information.

[0704] 4. Removing landscape elements and generating natural-looking eyes:

[0705] The server uses a generative adversarial network (GAN) to remove the landscape and generate natural-looking pupils, taking into account the original pupil shape and lighting information to maintain harmony across the entire image.

[0706] 5. Emotion Recognition:

[0707] The server uses an emotion recognition engine such as AWS Rekognition to analyze the user's facial expressions and read their emotions, then adjusts the color and gloss of the eyes appropriately based on the emotional information.

[0708] 6. Image merging:

[0709] The server integrates the generated pupils into the original image and processes it so that the whole image looks natural.

[0710] 7. Provide processed images:

[0711] The server provides the completed image to the user in real time, allowing the user to instantly check the edited image.

[0712] Hardware and software used

[0713] Camera: A camera device attached to the smart mirror.

[0714] Face Recognition Algorithm: OpenCV

[0715] Generative AI Models: Generative Adversarial Networks (GANs)

[0716] Emotion recognition engine: AWS Rekognition

[0717] Server: Server for image analysis and data processing

[0718] Specific examples

[0719] A specific example in which a user actually stands in front of a mirror is shown below.

[0720] When a user stands in front of the smart mirror, the camera automatically recognizes their face and eyes and captures an image. The emotion recognition engine detects when the user is smiling and applies a slightly sparkling eye effect. This allows the user to view the processed image in real time, ensuring privacy while allowing them to see natural-looking eyes that reflect their emotions.

[0721] Prompt Sentence Examples

[0722] "Write a Python program to protect image privacy and generate natural-looking eye pupils based on emotions. The following requirements must be met:

[0723] 1. Upload an image

[0724] 2. Face and Eye Analysis

[0725] 3. Removing the scenery reflected in the eyes

[0726] 4. Emotion recognition and pupil generation

[0727] 5. Merging and delivering the final image

[0728] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0729] Step 1:

[0730] The server acquires images from the smart mirror's camera. When a user stands in front of the mirror, the camera automatically activates and captures an image containing the user's face. The input is the raw image from the camera, and the output is an image file sent to the server.

[0731] Step 2:

[0732] The server uses a face recognition algorithm (e.g., OpenCV) to detect the pupils in the captured image. The input is the image sent to the server, and the output is the coordinate information of the pupils. The server scans the entire face to identify the pupil positions as a data calculation, and identifies a rectangular area that includes the eye positions.

[0733] Step 3:

[0734] The server performs pixel-by-pixel analysis to identify the landscape portion of the detected pupil. The input is the coordinate information of the pupil and the image range of the detected face, and the output is pixel information of the landscape portion. The server enlarges the pupil area and analyzes the reflection and different color tones within it to obtain the landscape portion.

[0735] Step 4:

[0736] The server uses a generative adversarial network (GAN) to remove the identified scene and generate natural-looking pupils. The input is pixel information of the scene, and the output is an image of a natural-looking pupil. The server inputs the scene information into the GAN model and obtains pixel data of the generated natural-looking pupils in return.

[0737] Step 5:

[0738] The server uses an emotion recognition engine (e.g., AWS Rekognition) to analyze the user's facial expressions and recognize their emotions. The input is an image of the user's face, and the output is the user's emotion data. The server identifies emotions such as smile, sadness, or surprise from the user's facial expressions through natural language processing.

[0739] Step 6:

[0740] The server adjusts the color tone and gloss of the generated eyes based on the recognized emotion. The input is emotion data and a generated natural eye image, and the output is an eye image adjusted according to the emotion. The server analyzes the emotion data and applies effects such as brightening the eyes if the person is smiling.

[0741] Step 7:

[0742] The server merges the adjusted natural pupils with the original image and processes it so that the whole image looks natural. The input is the adjusted natural pupil image and the original face image, and the output is the processed final image. The server processes the data by pasting the new pupils pixel by pixel to the original image.

[0743] Step 8:

[0744] The server provides the completed processed image to the user's information processing device. The input is the final processed image, and the output is image data sent to the user. The user can check the processed image on their own device, safely protecting their privacy while also being able to see natural-looking eyes based on their emotions.

[0745] 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.

[0746] 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.

[0747] 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.

[0748] [Third embodiment]

[0749] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0750] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0751] 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).

[0752] 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.

[0753] 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.

[0754] 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).

[0755] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0756] 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.

[0757] 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.

[0758] 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.

[0759] 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.

[0760] 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."

[0761] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. The specific operation of each step of this system is shown below.

[0762] Program processing flow

[0763] 1. Upload an image

[0764] User: Uploads images that require privacy protection from their own device to the system. These images can be anything from selfies to group photos.

[0765] Server: Receives and temporarily stores the images uploaded above, preparing them for the subsequent analysis process.

[0766] 2. Image Analysis

[0767] Server: Inputs the received image into an image analysis algorithm, which performs facial recognition and identifies the location of facial features and eyes.

[0768] 3. Detection of the landscape part of the pupil

[0769] Server: Performs detailed analysis of the scenery reflected in the detected pupils and identifies the scenery at the pixel level. This information is used in subsequent processing steps.

[0770] 4. Eye Image Generation

[0771] Server: Removes the landscape and applies an AI algorithm (e.g., a generative adversarial network (GAN)) to generate natural-looking eye images. The AI ​​takes into account the original pupil shape and light source information to reconstruct natural-looking eyes.

[0772] 5. Image Merging

[0773] Server: Integrate the generated natural pupils into the original image, replacing the original eye regions with the new pupil image, making the whole image look natural.

[0774] 6. Providing edited images

[0775] Server: Returns the completed processed image to the user's device. In addition, it also provides comparison images before and after processing so that the user can check the processing results.

[0776] Specific examples

[0777] Example 1: Editing a selfie

[0778] User: Uploads a selfie to the application.

[0779] Server: Receives the image and starts eye detection.

[0780] Server: Analyzes and identifies the part of the scenery reflected in the eyes.

[0781] Server: Generates natural-looking pupils using image generation AI.

[0782] Server: Sends the processed image to the user.

[0783] User: Review the received images and ensure their privacy is protected.

[0784] Example 2: Editing a group photo

[0785] User: Uploads a group photo of family and friends to the system.

[0786] Server: Receives the image and starts eye detection.

[0787] Server: Identify the landscape portion of each pupil and generate natural-looking pupils using AI.

[0788] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[0789] User: Review the received images and ensure their privacy is protected.

[0790] This system allows users to share images safely while protecting their privacy, and is expected to be widely adopted by businesses as well.

[0791] The processing flow will be explained below.

[0792] Step 1:

[0793] User: Uploads images from their device to the system, specifically by selecting a selfie or group photo and sending it to the server through the application or web interface.

[0794] Step 2:

[0795] Server: Receives uploaded images and stores them in temporary storage. After storing them, the server prepares them for image analysis.

[0796] Step 3:

[0797] Server: Uses an image analysis algorithm to detect faces in the received image. The face detection algorithm identifies the position and contours of the face.

[0798] Step 4:

[0799] Server: Identify the position of the pupils based on the facial position information. The pupil detection algorithm extracts the eye feature points and identifies the pupil area at the pixel level.

[0800] Step 5:

[0801] Server: Analyzes and identifies the scenery reflected within the detected pupil area. Specifically, it analyzes the light reflected from the pupil and the surrounding scenery information for each pixel and extracts unnecessary information.

[0802] Step 6:

[0803] Server: Using image generation AI (e.g., generative adversarial networks (GANs)), the identified landscape features are removed and natural-looking pupils are generated, while maintaining a natural appearance by taking into account the original pupil shape and lighting information.

[0804] Step 7:

[0805] Server: Merge the generated natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[0806] Step 8:

[0807] Server: Returns the processed image to the user, along with a before and after comparison image so the user can see the changes.

[0808] Step 9:

[0809] User: Receives the processed image and checks the results. Once the user confirms that their privacy has been protected, they can save the image to their device or post it safely to social media.

[0810] Example 1

[0811] 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."

[0812] With conventional technology, the landscape portion of the eyes in an image could be a violation of privacy, making it difficult for users to share images with confidence. Furthermore, manually editing the eyes requires time and skill, making it inefficient. Furthermore, the image processing process can sometimes produce unnatural results, making the processed image appear unnatural. To solve these problems, a system is needed that can automatically detect and process the landscape portion of the eyes, maintaining a natural appearance.

[0813] 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.

[0814] In this invention, the server includes means for a user to upload an image from a terminal, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, and means for providing the merged image to the user. This enables users to easily edit and safely share images while ensuring privacy.

[0815] "Means for users to upload images from their devices" refers to a system that allows users to select image files using their own devices and send them to a server via a network.

[0816] The "means for detecting pupils in a captured image" is a system that has the function of identifying the positions of pupils in a captured image using image processing technology.

[0817] The "means for identifying the portion of the scenery reflected in the detected pupil" is a system that has the function of analyzing the portion of the scenery reflected in the pupil area and identifying that portion at the pixel level.

[0818] The "means for removing identified landscape parts and generating natural-looking pupils" refers to a system that uses AI technology and image generation algorithms to remove identified landscape parts and generate natural-looking pupils by taking into account the original pupil shape and light source information.

[0819] The "means for merging the generated natural pupils into the original image" is a system that has the function of integrating the newly generated natural pupils into the original image to make the image appear natural overall.

[0820] The "means for providing a merged image to a user" refers to a system that has a function for returning the processed image to a user so that the user can check the result.

[0821] A "generative adversarial network" is a type of deep learning model that can generate new, high-quality data by learning through a competitive process between a generative model and a discriminative model.

[0822] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. Specific embodiments for implementing this system are described below.

[0823] First, the user uploads an image that requires privacy protection to the system using the device. Specifically, the user opens the application's file selection dialog and selects an image file from local storage. This operation causes the device to send the image to the server. The image file data is sent to the server using an HTTP POST request.

[0824] The server temporarily stores the received image file in storage. The path of the stored file is retained for the next processing step. Specific hardware used is a server computer, and software used is an HTTP server and a storage management system.

[0825] Next, the server inputs the saved image into an image analysis algorithm. Specifically, it uses the OpenCV and dlib libraries to perform face recognition. The image is read using OpenCV's cv2.imread function and converted to grayscale using cv2.cvtColor. The dlib face detector is used to detect the position of the face, and the dlib landmark detector is used to identify the position of the eye. This series of processes results in the detection of the eye in the image.

[0826] To analyze the landscape part reflected in the detected pupil in detail, the server identifies the landscape part pixel by pixel. For this purpose, OpenCV's cv2.boundingRect function is used. The identified landscape part information is used in the subsequent processing steps.

[0827] The server then uses a generative adversarial network (GAN) to remove the landscape elements from the pupils and generate natural-looking pupils. To do this, the image is resized and normalized to the appropriate size. A pre-trained GAN model is then used. Specific software used is TensorFlow and PyTorch.

[0828] To integrate the generated natural pupils into the original image, the server uses OpenCV's cv2.seamlessClone function, which blends the newly generated pupils seamlessly into the original image, maintaining an overall natural look.

[0829] Finally, the server returns the completed processed image to the user's device. The processed image is returned as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can then view and check the received image.

[0830] Specific examples

[0831] Selfie photo editing

[0832] The user uploads a selfie to the application. The device sends the image to the server, which uses OpenCV and dlib to initiate face and eye detection. The server uses a custom algorithm to identify the landscape portion of the eye and generates natural-looking eyes using a GAN. The server merges the new eye image with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[0833] Editing group photos

[0834] A user uploads a group photo of family and friends to the system. The device sends the image to the server, which uses OpenCV and dlib to initiate multiple face and eye detection. The server uses a custom algorithm to identify the landscape portion of each eye and uses GAN to generate natural-looking eyes for each eye. The server then merges the generated eyes with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[0835] This system allows users to edit and share images safely and easily while protecting their privacy.

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

[0837] Step 1:

[0838] The user uploads an image requiring privacy protection to the system using their own device. Specifically, they open the application's file selection dialog and select an image file from local storage. This operation causes the device to send the image data to the server via an HTTP POST request.

[0839] Input: An image file selected by the user.

[0840] Output: Image data sent to the server

[0841] Step 2:

[0842] The server temporarily stores the received image data. Specifically, it stores the received image file in storage and retains the file path for the next analysis step.

[0843] Input: Image data sent to the server

[0844] Output: Image file saved to storage

[0845] Step 3:

[0846] The server inputs the saved image into the image analysis algorithm. First, it reads the image using OpenCV's cv2.imread function, then converts the image to grayscale using the cv2.cvtColor function. Then it uses dlib's face detector (dlib.get_frontal_face_detector) to detect the face position. Finally, it uses dlib's landmark detector (dlib.shape_predictor) to identify the eye position.

[0847] Input: Image file saved in storage

[0848] Output: Face and eye position information

[0849] Step 4:

[0850] The server then performs a detailed analysis of the landscape reflected within the detected pupil. Specifically, it uses OpenCV's cv2.boundingRect function to extract the pupil area pixel by pixel. It then analyzes the pixel values ​​within this pupil area using a custom algorithm to identify the landscape.

[0851] Input: Face and eye position information

[0852] Output: Pixel information of the landscape

[0853] Step 5:

[0854] The server uses a generative adversarial network (GAN) to remove the identified landscape features and generate natural-looking pupils. First, the pupil regions are resized and normalized to an appropriate size as input for the GAN. Then, a pre-trained GAN model is used to generate natural-looking pupils. This process uses TensorFlow and PyTorch.

[0855] Input: Pixel information of the landscape

[0856] Output: Natural eye image

[0857] Step 6:

[0858] The server merges the generated natural pupils with the original image, using OpenCV's cv2.seamlessClone function to seamlessly integrate the newly generated pupil images into the original image, resulting in an overall image that maintains a natural appearance.

[0859] Input: natural eye image, original image

[0860] Output: Natural-looking processed image

[0861] Step 7:

[0862] The server returns the processed image to the user. The processed image is sent to the user's device as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can view the received image and confirm that their privacy has been protected.

[0863] Input: Natural-looking processed image

[0864] Output: The processed image sent to the user

[0865] (Application example 1)

[0866] 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."

[0867] In modern society, there is an increasing trend of capturing video footage using surveillance cameras and smart devices and using it for various purposes. However, these videos contain personal information, and the scene reflected in the eyes of people's eyes is particularly at high risk of being leaked. This raises concerns about privacy violations, and there is a demand for technology to safely and efficiently protect personal information.

[0868] 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.

[0869] In this invention, the server includes means for acquiring an image, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, means for providing the merged image to a user, means for providing comparison images before and after processing based on the generated privacy-protected image data, and means for performing the above processing on video data uploaded from a surveillance camera or a smart device. This makes it possible to achieve privacy protection while preventing the leakage of personal information in videos acquired from a surveillance camera or a smart device.

[0870] "Means for acquiring images" refers to a device or method for acquiring images or video using a surveillance camera or smart device and receiving them in a form that is input into the system.

[0871] The "means for detecting eyes" is an algorithm or technique for analyzing the facial area in the acquired image and specifically identifying the location of the eyes.

[0872] The "means for identifying the scenery portion" refers to a technology or algorithm that analyzes the scenery or background reflected in the detected pupil at the pixel level and identifies its range.

[0873] The "means for generating natural-looking pupils" refers to AI techniques such as generative adversarial networks (GANs) that remove detected landscape features and reconstruct a natural appearance based on the original pupil shape and light source information.

[0874] "Means for merging into original image" refers to the technique or process used to incorporate the generated natural pupils into the original image so that the overall image appears natural.

[0875] "Means for providing to the user" refers to a system or technology that returns the merged image to the user's terminal so that the user can view the results.

[0876] "Means for providing images before and after processing for comparison" refers to a display function for comparing the privacy-protected image with the original image, or technology that allows users to visually confirm the differences.

[0877] "Means for processing video data uploaded from surveillance cameras and smart devices" refers to a system or technology that receives video data sent from surveillance cameras, smartphones, tablets, etc., and automatically performs the series of analyses and processing described above.

[0878] A system embodying the invention includes a series of processing steps including image acquisition, pupil detection, scene location, natural pupil generation, merging with the original image, providing a processed image, and generating a comparison image.

[0879] First, the server has a means of acquiring images. These images may be uploaded from surveillance cameras or smart devices. Users upload images and video data from their own devices to the server. This data is temporarily stored on the cloud server.

[0880] Next, the server has a means to detect the eyes in the captured images. Specifically, it uses an image analysis algorithm to perform face recognition and identify the position of the eyes. For this, an image processing library such as OpenCV is used. A face detector and an eye detector (e.g., Haar Cascade Classifier) ​​are used together to identify the eyes.

[0881] The server then uses a method to identify the landscape portion reflected in the detected pupil, analyzing the landscape and background in the pupil at the pixel level and identifying its range. OpenCV is mainly used here as well.

[0882] The identified landscape elements are then removed using a server function. AI techniques such as generative adversarial networks (GANs) are then used to generate natural-looking pupils. Using GANs, natural-looking pupils are reconstructed based on the original pupil shape and lighting information.

[0883] The server then merges the generated natural pupils with the original image, which is a technique or process that incorporates the generated pupils into the original image so that the overall image looks natural. Again, OpenCV is used to combine the images.

[0884] The processed image is provided to the user's device from the server. Also, based on the generated privacy-protected image data, comparison images before and after processing are generated, allowing the user to visually compare the original and processed images.

[0885] As a concrete example, there is a process to upload security camera footage to a system, remove the scenery reflected in the eyes of each person in the footage, and generate natural-looking eyes. An example of a prompt for the program that performs this process is as follows:

[0886] Example prompt sentence:

[0887] Please convert the scenery reflected in the eyes in the image below into natural eyes.

[0888] Image: path / to / your / image.jpg

[0889] Algorithm: GANModel

[0890] Output: Privacy-preserving natural eye images

[0891] This makes it possible to protect privacy while preventing the leakage of personal information in images captured by surveillance cameras and smart devices.

[0892] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0893] Step 1:

[0894] Users upload images and video data captured from surveillance cameras or smart devices to a server. The input is image data sent from the user's device. The output is image data that is temporarily stored on the server. Specifically, the user uses a dedicated application to send the video data to the cloud server.

[0895] Step 2:

[0896] The server detects the eyes in the acquired images. The input is the image data saved in step 1. The output is face analysis data including eye position information. Specifically, the server uses the OpenCV library to perform face recognition and identify the eye positions. In this process, it applies face detection and eye detection algorithms such as the Haar Cascade Classifier.

[0897] Step 3:

[0898] The server identifies the part of the scenery reflected in the detected pupils. The input is face analysis data including pupil position information obtained in step 2. The output is data in which the scenery part is identified at the pixel level. Specifically, the server uses OpenCV to perform a detailed analysis of the scenery and background in each pupil and identify its range.

[0899] Step 4:

[0900] The server removes the identified scenery and generates natural-looking pupils. The input is image data including the scenery identified in step 3. The output is image data of natural-looking pupils generated using a generative adversarial network (GAN). Specifically, the server applies the GAN model to reconstruct natural-looking pupils based on the original pupil shape and light source information while removing the scenery.

[0901] Step 5:

[0902] The server merges the generated natural pupils with the original image. The input is the image data of the natural pupils generated in step 4. The output is the final merged image data. Specifically, the server uses OpenCV to incorporate the generated pupils into the original image and adjust it so that the whole image looks natural.

[0903] Step 6:

[0904] The server provides the merged image to the user and also generates before-and-after images for comparison. The input is the final image data obtained in step 5. The output is the final image sent to the user's device and the before-and-after images for comparison. Specifically, the server sends the image data to the user's device and prepares it so that the user can check the results.

[0905] Through these steps, the system protects privacy while preventing the leakage of personal information.

[0906] 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.

[0907] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results. How this system is implemented will be specifically described below.

[0908] Program processing flow

[0909] 1. Upload an image

[0910] User: Uploads an image to be processed from their own device to the system. The image can be a selfie or a group photo.

[0911] Server: Receives uploaded images and stores them in temporary storage, then prepares them for the analysis process.

[0912] 2. Image Analysis

[0913] Server: Uses image analysis algorithms to detect faces in the received images and obtain their detailed location information.

[0914] 3. Detection of the landscape part of the pupil

[0915] Server: Based on the facial position information, the position of the eyes is identified, and the part of the scenery reflected in the eyes is further analyzed at the pixel level to identify it.

[0916] 4. Eye Image Generation

[0917] Server: Removes the identified landscape features and generates natural-looking pupils using image generation AI (e.g., generative adversarial networks (GANs)). During this process, the original pupil shape and lighting information are used as a reference to maintain a natural appearance.

[0918] 5. Applying the Emotion Engine

[0919] Server: In parallel with the eye detection, the emotion engine runs to analyze the user's facial expressions and recognize their emotions. This engine identifies emotions such as smile, sadness, surprise, etc.

[0920] Server: Adjust the color and shine of the automatically generated eyes based on the recognized emotion. For example, add a slight sparkle to the eyes when smiling.

[0921] 6. Image Merging

[0922] Server: The adjusted natural pupils are merged with the original image, and the whole image is processed to look natural.

[0923] 7. Providing edited images

[0924] Server: Returns the completed processed image to the user's device, along with comparison images of before and after processing, allowing the user to confirm the changes.

[0925] Specific examples

[0926] Example 1: Editing a selfie

[0927] User: Uploads a selfie to the application.

[0928] Server: Receives the image and starts face recognition and eye detection. At the same time, the emotion engine analyzes the user's facial expressions.

[0929] Server: Analyzes and identifies the part of the scenery reflected in the eyes. The emotion engine recognizes that the user is smiling.

[0930] Server: Uses image generation AI to generate natural-looking eyes while adding a slight glow effect.

[0931] Server: Sends the processed image to the user.

[0932] User: Review the received images and ensure privacy is maintained.

[0933] Example 2: Editing a group photo

[0934] User: Uploads a group photo of family and friends to the system.

[0935] Server: Receives the image and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[0936] Server: Identifies the part of the landscape reflected in each eye and uses AI to generate natural-looking eyes for each.

[0937] Server: Adds appropriate effects to the generated eyes based on the emotions recognized by the emotion engine.

[0938] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[0939] User: Review the received images and ensure their privacy is protected.

[0940] This system allows users to obtain images with natural-looking eyes that reflect emotions while preventing privacy violations due to the scenery reflected in the eyes. It can also be introduced to businesses, and is expected to be widely used.

[0941] The processing flow will be explained below.

[0942] Step 1:

[0943] User: Uploads images that require privacy protection and emotion recognition to the system from their own device, such as selfies or group photos, by selecting and submitting them through the application or web interface.

[0944] Step 2:

[0945] Server: Receives the uploaded images and stores the image data in temporary storage, which prepares it for the subsequent analysis process.

[0946] Step 3:

[0947] Server: Activates the image analysis algorithm to detect faces in the received image. The face detection algorithm identifies and records the position and contour information of the face.

[0948] Step 4:

[0949] Server: Based on the face position information, the eye position is identified. Here, the eye detection algorithm detects the eye feature points and identifies the pupil area at pixel level.

[0950] Step 5:

[0951] Server: Performs a detailed analysis of the external scenery reflected within the detected pupil area and identifies that area. During this process, the pupil reflection and surrounding scenery information are analyzed pixel by pixel to extract unnecessary information.

[0952] Step 6:

[0953] Server: Activates the emotion engine and analyzes facial expressions in the image. The emotion engine identifies emotions such as smile, sadness, surprise, etc. and records the data.

[0954] Step 7:

[0955] Server: Removes the identified landscape features and generates natural-looking eyes using image generation AI (e.g., generative adversarial networks (GANs)). This process takes into account not only the original eye shape and lighting information, but also emotion data recognized by the emotion engine, preserving a natural appearance.

[0956] Step 8:

[0957] Server: Adjusts the color tone and gloss of the generated eyes based on the emotion engine data. For example, if the person is smiling, the eyes will be given a slight sparkle, making adjustments according to the emotion.

[0958] Step 9:

[0959] Server: Merge the adjusted natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[0960] Step 10:

[0961] Server: Returns the processed image to the user's device, along with a comparison image of the original and the processed image, allowing the user to easily verify the changes.

[0962] Step 11:

[0963] User: Receives the processed image, checks the processing results and any additional changes, and once they have confirmed that privacy is protected and that natural-looking eyes have been generated according to the emotion, they can save the image to their device or post it safely to social media.

[0964] Example 2

[0965] 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."

[0966] In modern society, as opportunities for image sharing increase, protecting user privacy has become an important issue. In particular, the scenery reflected in the eyes of images can potentially infringe on a user's privacy. It is also necessary to generate natural-looking eyes that reflect the user's emotions and maintain the realism of the entire image. Therefore, there is a need for technology that can reliably remove the scenery reflected in the eyes and generate natural-looking eyes that take emotions into account.

[0967] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for recognizing emotions from the user's facial expression, means for adjusting the generated natural-looking pupils based on the recognized emotion, and means for providing the adjusted image to the user. This makes it possible to generate natural-looking pupils according to the emotion while protecting the user's privacy.

[0968] The "means for acquiring images" provides an interface for users to upload images taken from their own terminals to the server.

[0969] The "means for detecting pupils" uses an image analysis algorithm to identify the position of pupils in the acquired image.

[0970] The "means for identifying the scenery portion" is a means for analyzing and identifying the scenery portion reflected in the pupil at the pixel level based on the pupil position.

[0971] The "means for generating natural pupils" involves removing the identified scene features and generating new pupil images using a generative model.

[0972] The "means for merging with the original image" is a means for integrating the generated natural pupils with the original image, and processing the entire image so that it looks natural.

[0973] "Means for recognizing emotions from facial expressions" refers to algorithms or devices for analyzing a user's facial expressions and identifying emotions from those expressions.

[0974] The "means for adjusting the generated natural pupils" adjusts the color tone and gloss of the generated pupils based on the recognized emotion.

[0975] The "means for providing the adjusted image to the user" means transmitting the completed processed image to the user's terminal so that the user can check it.

[0976] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results.

[0977] Specific means for implementing this system include the following steps involving a server, a terminal, and a user.

[0978] First, the user uploads the image that needs to be processed from their own device (for example, a personal computer or smartphone) to the system. The uploaded image is sent to a server and saved in temporary storage such as cloud storage. A specific storage method here could be to use a cloud storage service such as Amazon S3.

[0979] Next, the server uses image analysis libraries such as OpenCV and dlib to detect faces in the received images. This process obtains the face's position information (bounding box) and identifies the position of the eyes based on that. dlib's landmark detection technology is effective for identifying the position of the eyes. At this stage, pixel-level analysis is used to identify the part of the landscape reflected in the eyes.

[0980] The server removes the scenery and generates new, natural-looking eyes using a generative adversarial network (GAN). The generated eyes maintain the original eye shape and lighting information, achieving a natural look. The server also simultaneously uses an emotion engine (such as Microsoft Azure's Face API) to recognize emotions from the user's facial expressions. Based on the recognition results, the server adjusts the color tone and gloss of the generated natural-looking eyes. For example, when smiling, it can add a slight sparkle to the eyes.

[0981] Finally, the adjusted natural pupils are merged with the original image to create a natural-looking overall image. Alpha blending is used to combine the generated pupils with the original image. The completed processed image is sent from the server to the user's device, and comparison images before and after are also provided so the user can confirm the changes.

[0982] Specific examples

[0983] Example 1: Editing a selfie

[0984] 1. User: Uploads a selfie image from a smartphone application.

[0985] 2. Server: Receives the image and performs face recognition and eye detection, while analyzing the user's facial expressions using the emotion engine.

[0986] 3. Server: Identify the parts of the landscape that are reflected in the eyes and generate natural-looking eyes using GAN, adding sparkle to the eyes when smiling.

[0987] 4. Server: Merge the new pupils with the original image and send the processed image to the user.

[0988] 5. User: Check the edited image and ensure your privacy is protected.

[0989] Example 2: Editing a group photo

[0990] 1. User: Uploads a group photo of family and friends from a PC.

[0991] 2. Server: Receives the images and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[0992] 3. Server: Identify the part of the scene reflected in each eye and generate natural-looking eyes for each eye, for example, adding a glow to the emotion of joy.

[0993] 4. Server: Merge the generated pupils with the original image and send the processed image to the user.

[0994] 5. User: Check the edited image and ensure that your privacy is protected.

[0995] Example prompts to input to the generative AI model

[0996] Example prompt:

[0997] "Please provide a step-by-step outline of a system that removes the landscape reflected in the eyes from user-uploaded images and generates natural-looking eyes using a generative adversarial network (GAN). Also, please describe how to recognize the user's emotion and apply eye effects accordingly."

[0998] This system can obtain images with natural-looking eyes that correspond to emotions while preventing the invasion of privacy of the user's eyes. It can also be introduced to businesses, and is expected to be widely used.

[0999] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1000] Step 1:

[1001] Uploading an image

[1002] User: Upload an image to the system from their device.

[1003] Input: The image that needs to be enhanced (e.g. a selfie).

[1004] Specific behavior: Selects an image file using the device's file selection dialog.

[1005] Output: The image file is sent to the server.

[1006] Step 2:

[1007] Receiving and saving images

[1008] Server: Receives uploaded images and temporarily stores them in cloud storage.

[1009] Input: An image file submitted by the user.

[1010] Specific operation: The received image data is saved in cloud storage such as Amazon S3.

[1011] Output: The storage path of the saved image file.

[1012] Step 3:

[1013] Image analysis and facial recognition

[1014] Server: Detect faces using an image analysis library.

[1015] Input: Image files stored in cloud storage.

[1016] Specific operation: Detects faces using OpenCV's cv2.CascadeClassifier and obtains their position information as a bounding box.

[1017] Output: Face position information (bounding box).

[1018] Step 4:

[1019] Detecting the landscape part of the pupil

[1020] Server: Identifies the position of the eyes based on face position information and analyzes the part of the scenery reflected in the eyes.

[1021] Input: Face bounding box information and image files.

[1022] What it does: It uses dlib landmark detection to identify the position of the pupil, then analyzes pixel values ​​to extract the part of the landscape reflected in the pupil.

[1023] Output: Eye position and landscape pixel information.

[1024] Step 5:

[1025] Eye image generation

[1026] Server: Generates new natural-looking pupils using a generative model (GAN).

[1027] Input: Pixel information of the landscape part of the pupil.

[1028] How it works: First, mask out the landscape area, then input that area into the GAN, which then generates natural-looking pupils.

[1029] Output: Generated pupil image data.

[1030] Step 6:

[1031] Applying the Emotion Engine

[1032] Server: Analyzes facial expressions and recognizes emotions.

[1033] Input: Image file and face position information.

[1034] Specific operation: Using an emotion engine (for example, Microsoft Azure's Face API), emotions are recognized from facial expressions, and the color and gloss of the generated eyes are adjusted based on that information.

[1035] Output: Recognized emotion data and adjusted eye image data.

[1036] Step 7:

[1037] Image merging

[1038] Server: Integrate the adjusted natural pupils into the original image.

[1039] Input: Adjusted pupil image data and original image files.

[1040] What it does: Uses alpha blending to naturally merge the generated pupil with the original image.

[1041] Output: The finished processed image.

[1042] Step 8:

[1043] Providing processed images

[1044] Server: Sends the processed image to the user's device.

[1045] Input: The finished processed image.

[1046] Specific operation: Comparison images before and after processing are also generated and sent back to the user. WebSocket is used as a real-time notification method.

[1047] Output: The processed and comparison images sent to the user.

[1048] (Application example 2)

[1049] 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."

[1050] Because the scenery reflected in the eyes in an image can invade privacy, a means to prevent this is needed. Also, to make the image look more natural, it is necessary to adjust the color tone and brightness of the eyes according to the user's emotions. In brick-and-mortar stores, technology is needed to protect privacy from selfies and other people's gazes through mirrors while improving the user experience.

[1051] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for providing the merged image to the user, and means for recognizing the user's emotion and adjusting the color tone and gloss of the pupils based on the result. This makes it possible to adjust the pupils in real time according to the emotion while maintaining a natural appearance and protecting the user's privacy.

[1052] "Means for acquiring images" refers to a combination of hardware and software for capturing images from a user's information processing device or camera.

[1053] The "means for detecting pupils in the acquired image" is an algorithm for identifying pupils in the image and obtaining their position information.

[1054] The "means for identifying the portion of the scenery reflected in the detected pupil" is a process for identifying the external scenery or objects reflected in the pupil and analyzing it on a pixel-by-pixel basis.

[1055] The "means for removing identified landscape features and generating natural-looking eyes" is a system that uses AI techniques such as generative adversarial networks (GANs) to generate natural-looking eye images after removing the landscape features.

[1056] The "means for merging the generated natural pupils into the original image" is a technique for integrating the generated pupil images into the original photograph, making the whole look natural.

[1057] The "means for providing a merged image to a user" is a system for transmitting or displaying the processed image to a user.

[1058] "Means for recognizing the user's emotions and adjusting the color tone and gloss of the pupils based on the results" refers to a process that uses an emotion recognition engine to analyze the user's facial expression and appropriately corrects the characteristics of the pupils (color tone and gloss) generated based on that.

[1059] This invention is part of a smart mirror system that also protects privacy in brick-and-mortar stores. It removes the scenery reflected in the eyes in the image, generates natural-looking eyes based on the user's emotions, and provides real-time feedback.

[1060] Program processing flow

[1061] 1. Image Acquisition:

[1062] When a user stands in front of the smart mirror, the mirror's camera automatically captures an image, which is then seamlessly transmitted to the server.

[1063] 2. Eye detection:

[1064] The server uses a facial recognition algorithm such as OpenCV to detect faces and eyes in the image, which allows the precise location of the eyes.

[1065] 3. Identifying the landscape:

[1066] The server zooms in on the detected pupil area and performs pixel-by-pixel analysis, identifying specific parts of the scene or object and extracting that pixel information.

[1067] 4. Removing landscape elements and generating natural-looking eyes:

[1068] The server uses a generative adversarial network (GAN) to remove the landscape and generate natural-looking pupils, taking into account the original pupil shape and lighting information to maintain harmony across the entire image.

[1069] 5. Emotion Recognition:

[1070] The server uses an emotion recognition engine such as AWS Rekognition to analyze the user's facial expressions and read their emotions, then adjusts the color and gloss of the eyes appropriately based on the emotional information.

[1071] 6. Image merging:

[1072] The server integrates the generated pupils into the original image and processes it so that the whole image looks natural.

[1073] 7. Provide processed images:

[1074] The server provides the completed image to the user in real time, allowing the user to instantly check the edited image.

[1075] Hardware and software used

[1076] Camera: A camera device attached to the smart mirror.

[1077] Face Recognition Algorithm: OpenCV

[1078] Generative AI Models: Generative Adversarial Networks (GANs)

[1079] Emotion recognition engine: AWS Rekognition

[1080] Server: Server for image analysis and data processing

[1081] Specific examples

[1082] A specific example in which a user actually stands in front of a mirror is shown below.

[1083] When a user stands in front of the smart mirror, the camera automatically recognizes their face and eyes and captures an image. The emotion recognition engine detects when the user is smiling and applies a slightly sparkling eye effect. This allows the user to view the processed image in real time, ensuring privacy while allowing them to see natural-looking eyes that reflect their emotions.

[1084] Prompt Sentence Examples

[1085] "Write a Python program to protect image privacy and generate natural-looking eye pupils based on emotions. The following requirements must be met:

[1086] 1. Upload an image

[1087] 2. Face and Eye Analysis

[1088] 3. Removing the scenery reflected in the eyes

[1089] 4. Emotion recognition and pupil generation

[1090] 5. Merging and delivering the final image

[1091] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1092] Step 1:

[1093] The server acquires images from the smart mirror's camera. When a user stands in front of the mirror, the camera automatically activates and captures an image containing the user's face. The input is the raw image from the camera, and the output is an image file sent to the server.

[1094] Step 2:

[1095] The server uses a face recognition algorithm (e.g., OpenCV) to detect the pupils in the captured image. The input is the image sent to the server, and the output is the coordinate information of the pupils. The server scans the entire face to identify the pupil positions as a data calculation, and identifies a rectangular area that includes the eye positions.

[1096] Step 3:

[1097] The server performs pixel-by-pixel analysis to identify the landscape portion of the detected pupil. The input is the coordinate information of the pupil and the image range of the detected face, and the output is pixel information of the landscape portion. The server enlarges the pupil area and analyzes the reflection and different color tones within it to obtain the landscape portion.

[1098] Step 4:

[1099] The server uses a generative adversarial network (GAN) to remove the identified scene and generate natural-looking pupils. The input is pixel information of the scene, and the output is an image of a natural-looking pupil. The server inputs the scene information into the GAN model and obtains pixel data of the generated natural-looking pupils in return.

[1100] Step 5:

[1101] The server uses an emotion recognition engine (e.g., AWS Rekognition) to analyze the user's facial expressions and recognize their emotions. The input is an image of the user's face, and the output is the user's emotion data. The server identifies emotions such as smile, sadness, or surprise from the user's facial expressions through natural language processing.

[1102] Step 6:

[1103] The server adjusts the color tone and gloss of the generated eyes based on the recognized emotion. The input is emotion data and a generated natural eye image, and the output is an eye image adjusted according to the emotion. The server analyzes the emotion data and applies effects such as brightening the eyes if the person is smiling.

[1104] Step 7:

[1105] The server merges the adjusted natural pupils with the original image and processes it so that the whole image looks natural. The input is the adjusted natural pupil image and the original face image, and the output is the processed final image. The server processes the data by pasting the new pupils pixel by pixel to the original image.

[1106] Step 8:

[1107] The server provides the completed processed image to the user's information processing device. The input is the final processed image, and the output is image data sent to the user. The user can check the processed image on their own device, safely protecting their privacy while also being able to see natural-looking eyes based on their emotions.

[1108] 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.

[1109] 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.

[1110] 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.

[1111] [Fourth embodiment]

[1112] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1113] 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.

[1114] 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).

[1115] 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.

[1116] 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.

[1117] 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).

[1118] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1119] 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.

[1120] 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.

[1121] 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.

[1122] 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.

[1123] 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.

[1124] 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."

[1125] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. The specific operation of each step of this system is shown below.

[1126] Program processing flow

[1127] 1. Upload an image

[1128] User: Uploads images that require privacy protection from their own device to the system. These images can be anything from selfies to group photos.

[1129] Server: Receives and temporarily stores the images uploaded above, preparing them for the subsequent analysis process.

[1130] 2. Image Analysis

[1131] Server: Inputs the received image into an image analysis algorithm, which performs facial recognition and identifies the location of facial features and eyes.

[1132] 3. Detection of the landscape part of the pupil

[1133] Server: Performs detailed analysis of the scenery reflected in the detected pupils and identifies the scenery at the pixel level. This information is used in subsequent processing steps.

[1134] 4. Eye Image Generation

[1135] Server: Removes the landscape and applies an AI algorithm (e.g., a generative adversarial network (GAN)) to generate natural-looking eye images. The AI ​​takes into account the original pupil shape and light source information to reconstruct natural-looking eyes.

[1136] 5. Image Merging

[1137] Server: Integrate the generated natural pupils into the original image, replacing the original eye regions with the new pupil image, making the whole image look natural.

[1138] 6. Providing edited images

[1139] Server: Returns the completed processed image to the user's device. In addition, it also provides comparison images before and after processing so that the user can check the processing results.

[1140] Specific examples

[1141] Example 1: Editing a selfie

[1142] User: Uploads a selfie to the application.

[1143] Server: Receives the image and starts eye detection.

[1144] Server: Analyzes and identifies the part of the scenery reflected in the eyes.

[1145] Server: Generates natural-looking pupils using image generation AI.

[1146] Server: Sends the processed image to the user.

[1147] User: Review the received images and ensure their privacy is protected.

[1148] Example 2: Editing a group photo

[1149] User: Uploads a group photo of family and friends to the system.

[1150] Server: Receives the image and starts eye detection.

[1151] Server: Identify the landscape portion of each pupil and generate natural-looking pupils using AI.

[1152] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[1153] User: Review the received images and ensure their privacy is protected.

[1154] This system allows users to share images safely while protecting their privacy, and is expected to be widely adopted by businesses as well.

[1155] The processing flow will be explained below.

[1156] Step 1:

[1157] User: Uploads images from their device to the system, specifically by selecting a selfie or group photo and sending it to the server through the application or web interface.

[1158] Step 2:

[1159] Server: Receives uploaded images and stores them in temporary storage. After storing them, the server prepares them for image analysis.

[1160] Step 3:

[1161] Server: Uses an image analysis algorithm to detect faces in the received image. The face detection algorithm identifies the position and contours of the face.

[1162] Step 4:

[1163] Server: Identify the position of the pupils based on the facial position information. The pupil detection algorithm extracts the eye feature points and identifies the pupil area at the pixel level.

[1164] Step 5:

[1165] Server: Analyzes and identifies the scenery reflected within the detected pupil area. Specifically, it analyzes the light reflected from the pupil and the surrounding scenery information for each pixel and extracts unnecessary information.

[1166] Step 6:

[1167] Server: Using image generation AI (e.g., generative adversarial networks (GANs)), the identified landscape features are removed and natural-looking pupils are generated, while maintaining a natural appearance by taking into account the original pupil shape and lighting information.

[1168] Step 7:

[1169] Server: Merge the generated natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[1170] Step 8:

[1171] Server: Returns the processed image to the user, along with a before and after comparison image so the user can see the changes.

[1172] Step 9:

[1173] User: Receives the processed image and checks the results. Once the user confirms that their privacy has been protected, they can save the image to their device or post it safely to social media.

[1174] Example 1

[1175] 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."

[1176] With conventional technology, the landscape portion of the eyes in an image could be a violation of privacy, making it difficult for users to share images with confidence. Furthermore, manually editing the eyes requires time and skill, making it inefficient. Furthermore, the image processing process can sometimes produce unnatural results, making the processed image appear unnatural. To solve these problems, a system is needed that can automatically detect and process the landscape portion of the eyes, maintaining a natural appearance.

[1177] 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.

[1178] In this invention, the server includes means for a user to upload an image from a terminal, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, and means for providing the merged image to the user. This enables users to easily edit and safely share images while ensuring privacy.

[1179] "Means for users to upload images from their devices" refers to a system that allows users to select image files using their own devices and send them to a server via a network.

[1180] The "means for detecting pupils in a captured image" is a system that has the function of identifying the positions of pupils in a captured image using image processing technology.

[1181] The "means for identifying the portion of the scenery reflected in the detected pupil" is a system that has the function of analyzing the portion of the scenery reflected in the pupil area and identifying that portion at the pixel level.

[1182] The "means for removing identified landscape parts and generating natural-looking pupils" refers to a system that uses AI technology and image generation algorithms to remove identified landscape parts and generate natural-looking pupils by taking into account the original pupil shape and light source information.

[1183] The "means for merging the generated natural pupils into the original image" is a system that has the function of integrating the newly generated natural pupils into the original image to make the image appear natural overall.

[1184] The "means for providing a merged image to a user" refers to a system that has a function for returning the processed image to a user so that the user can check the result.

[1185] A "generative adversarial network" is a type of deep learning model that can generate new, high-quality data by learning through a competitive process between a generative model and a discriminative model.

[1186] This invention is a system that processes the scenery reflected in the eyes of an image to protect the user's privacy. This system automatically performs a series of steps from image acquisition to processing and provision to the user. Specific embodiments for implementing this system are described below.

[1187] First, the user uploads an image that requires privacy protection to the system using the device. Specifically, the user opens the application's file selection dialog and selects an image file from local storage. This operation causes the device to send the image to the server. The image file data is sent to the server using an HTTP POST request.

[1188] The server temporarily stores the received image file in storage. The path of the stored file is retained for the next processing step. Specific hardware used is a server computer, and software used is an HTTP server and a storage management system.

[1189] Next, the server inputs the saved image into an image analysis algorithm. Specifically, it uses the OpenCV and dlib libraries to perform face recognition. The image is read using OpenCV's cv2.imread function and converted to grayscale using cv2.cvtColor. The dlib face detector is used to detect the position of the face, and the dlib landmark detector is used to identify the position of the eye. This series of processes results in the detection of the eye in the image.

[1190] To analyze the landscape part reflected in the detected pupil in detail, the server identifies the landscape part pixel by pixel. For this purpose, OpenCV's cv2.boundingRect function is used. The identified landscape part information is used in the subsequent processing steps.

[1191] The server then uses a generative adversarial network (GAN) to remove the landscape elements from the pupils and generate natural-looking pupils. To do this, the image is resized and normalized to the appropriate size. A pre-trained GAN model is then used. Specific software used is TensorFlow and PyTorch.

[1192] To integrate the generated natural pupils into the original image, the server uses OpenCV's cv2.seamlessClone function, which blends the newly generated pupils seamlessly into the original image, maintaining an overall natural look.

[1193] Finally, the server returns the completed processed image to the user's device. The processed image is returned as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can then view and check the received image.

[1194] Specific examples

[1195] Selfie photo editing

[1196] The user uploads a selfie to the application. The device sends the image to the server, which uses OpenCV and dlib to initiate face and eye detection. The server uses a custom algorithm to identify the landscape portion of the eye and generates natural-looking eyes using a GAN. The server merges the new eye image with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[1197] Editing group photos

[1198] A user uploads a group photo of family and friends to the system. The device sends the image to the server, which uses OpenCV and dlib to initiate multiple face and eye detection. The server uses a custom algorithm to identify the landscape portion of each eye and uses GAN to generate natural-looking eyes for each eye. The server then merges the generated eyes with the original image using OpenCV's cv2.seamlessClone function. The server then sends the processed image to the user, who can review the received image to ensure privacy is protected.

[1199] This system allows users to edit and share images safely and easily while protecting their privacy.

[1200] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1201] Step 1:

[1202] The user uploads an image requiring privacy protection to the system using their own device. Specifically, they open the application's file selection dialog and select an image file from local storage. This operation causes the device to send the image data to the server via an HTTP POST request.

[1203] Input: An image file selected by the user.

[1204] Output: Image data sent to the server

[1205] Step 2:

[1206] The server temporarily stores the received image data. Specifically, it stores the received image file in storage and retains the file path for the next analysis step.

[1207] Input: Image data sent to the server

[1208] Output: Image file saved to storage

[1209] Step 3:

[1210] The server inputs the saved image into the image analysis algorithm. First, it reads the image using OpenCV's cv2.imread function, then converts the image to grayscale using the cv2.cvtColor function. Then it uses dlib's face detector (dlib.get_frontal_face_detector) to detect the face position. Finally, it uses dlib's landmark detector (dlib.shape_predictor) to identify the eye position.

[1211] Input: Image file saved in storage

[1212] Output: Face and eye position information

[1213] Step 4:

[1214] The server then performs a detailed analysis of the landscape reflected within the detected pupil. Specifically, it uses OpenCV's cv2.boundingRect function to extract the pupil area pixel by pixel. It then analyzes the pixel values ​​within this pupil area using a custom algorithm to identify the landscape.

[1215] Input: Face and eye position information

[1216] Output: Pixel information of the landscape

[1217] Step 5:

[1218] The server uses a generative adversarial network (GAN) to remove the identified landscape features and generate natural-looking pupils. First, the pupil regions are resized and normalized to an appropriate size as input for the GAN. Then, a pre-trained GAN model is used to generate natural-looking pupils. This process uses TensorFlow and PyTorch.

[1219] Input: Pixel information of the landscape

[1220] Output: Natural eye image

[1221] Step 6:

[1222] The server merges the generated natural pupils with the original image, using OpenCV's cv2.seamlessClone function to seamlessly integrate the newly generated pupil images into the original image, resulting in an overall image that maintains a natural appearance.

[1223] Input: natural eye image, original image

[1224] Output: Natural-looking processed image

[1225] Step 7:

[1226] The server returns the processed image to the user. The processed image is sent to the user's device as an HTTP response and, if necessary, saved in cloud storage, with a link provided to the user. The user can view the received image and confirm that their privacy has been protected.

[1227] Input: Natural-looking processed image

[1228] Output: The processed image sent to the user

[1229] (Application example 1)

[1230] 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."

[1231] In modern society, there is an increasing trend of capturing video footage using surveillance cameras and smart devices and using it for various purposes. However, these videos contain personal information, and the scene reflected in the eyes of people's eyes is particularly at high risk of being leaked. This raises concerns about privacy violations, and there is a demand for technology to safely and efficiently protect personal information.

[1232] 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.

[1233] In this invention, the server includes means for acquiring an image, means for detecting an eye in the acquired image, means for identifying a landscape portion reflected in the detected eye, means for deleting the identified landscape portion and generating a natural-looking eye, means for merging the generated natural-looking eye with the original image, means for providing the merged image to a user, means for providing comparison images before and after processing based on the generated privacy-protected image data, and means for performing the above processing on video data uploaded from a surveillance camera or a smart device. This makes it possible to achieve privacy protection while preventing the leakage of personal information in videos acquired from a surveillance camera or a smart device.

[1234] "Means for acquiring images" refers to a device or method for acquiring images or video using a surveillance camera or smart device and receiving them in a form that is input into the system.

[1235] The "means for detecting eyes" is an algorithm or technique for analyzing the facial area in the acquired image and specifically identifying the location of the eyes.

[1236] The "means for identifying the scenery portion" refers to a technology or algorithm that analyzes the scenery or background reflected in the detected pupil at the pixel level and identifies its range.

[1237] The "means for generating natural-looking pupils" refers to AI techniques such as generative adversarial networks (GANs) that remove detected landscape features and reconstruct a natural appearance based on the original pupil shape and light source information.

[1238] "Means for merging into original image" refers to the technique or process used to incorporate the generated natural pupils into the original image so that the overall image appears natural.

[1239] "Means for providing to the user" refers to a system or technology that returns the merged image to the user's terminal so that the user can view the results.

[1240] "Means for providing images before and after processing for comparison" refers to a display function for comparing the privacy-protected image with the original image, or technology that allows users to visually confirm the differences.

[1241] "Means for processing video data uploaded from surveillance cameras and smart devices" refers to a system or technology that receives video data sent from surveillance cameras, smartphones, tablets, etc., and automatically performs the series of analyses and processing described above.

[1242] A system embodying the invention includes a series of processing steps including image acquisition, pupil detection, scene location, natural pupil generation, merging with the original image, providing a processed image, and generating a comparison image.

[1243] First, the server has a means of acquiring images. These images may be uploaded from surveillance cameras or smart devices. Users upload images and video data from their own devices to the server. This data is temporarily stored on the cloud server.

[1244] Next, the server has a means to detect the eyes in the captured images. Specifically, it uses an image analysis algorithm to perform face recognition and identify the position of the eyes. For this, an image processing library such as OpenCV is used. A face detector and an eye detector (e.g., Haar Cascade Classifier) ​​are used together to identify the eyes.

[1245] The server then uses a method to identify the landscape portion reflected in the detected pupil, analyzing the landscape and background in the pupil at the pixel level and identifying its range. OpenCV is mainly used here as well.

[1246] The identified landscape elements are then removed using a server function. AI techniques such as generative adversarial networks (GANs) are then used to generate natural-looking pupils. Using GANs, natural-looking pupils are reconstructed based on the original pupil shape and lighting information.

[1247] The server then merges the generated natural pupils with the original image, which is a technique or process that incorporates the generated pupils into the original image so that the overall image looks natural. Again, OpenCV is used to combine the images.

[1248] The processed image is provided to the user's device from the server. Also, based on the generated privacy-protected image data, comparison images before and after processing are generated, allowing the user to visually compare the original and processed images.

[1249] As a concrete example, there is a process to upload security camera footage to a system, remove the scenery reflected in the eyes of each person in the footage, and generate natural-looking eyes. An example of a prompt for the program that performs this process is as follows:

[1250] Example prompt sentence:

[1251] Please convert the scenery reflected in the eyes in the image below into natural eyes.

[1252] Image: path / to / your / image.jpg

[1253] Algorithm: GANModel

[1254] Output: Privacy-preserving natural eye images

[1255] This makes it possible to protect privacy while preventing the leakage of personal information in images captured by surveillance cameras and smart devices.

[1256] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1257] Step 1:

[1258] Users upload images and video data captured from surveillance cameras or smart devices to a server. The input is image data sent from the user's device. The output is image data that is temporarily stored on the server. Specifically, the user uses a dedicated application to send the video data to the cloud server.

[1259] Step 2:

[1260] The server detects the eyes in the acquired images. The input is the image data saved in step 1. The output is face analysis data including eye position information. Specifically, the server uses the OpenCV library to perform face recognition and identify the eye positions. In this process, it applies face detection and eye detection algorithms such as the Haar Cascade Classifier.

[1261] Step 3:

[1262] The server identifies the part of the scenery reflected in the detected pupils. The input is face analysis data including pupil position information obtained in step 2. The output is data in which the scenery part is identified at the pixel level. Specifically, the server uses OpenCV to perform a detailed analysis of the scenery and background in each pupil and identify its range.

[1263] Step 4:

[1264] The server removes the identified scenery and generates natural-looking pupils. The input is image data including the scenery identified in step 3. The output is image data of natural-looking pupils generated using a generative adversarial network (GAN). Specifically, the server applies the GAN model to reconstruct natural-looking pupils based on the original pupil shape and light source information while removing the scenery.

[1265] Step 5:

[1266] The server merges the generated natural pupils with the original image. The input is the image data of the natural pupils generated in step 4. The output is the final merged image data. Specifically, the server uses OpenCV to incorporate the generated pupils into the original image and adjust it so that the whole image looks natural.

[1267] Step 6:

[1268] The server provides the merged image to the user and also generates before-and-after images for comparison. The input is the final image data obtained in step 5. The output is the final image sent to the user's device and the before-and-after images for comparison. Specifically, the server sends the image data to the user's device and prepares it so that the user can check the results.

[1269] Through these steps, the system protects privacy while preventing the leakage of personal information.

[1270] 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.

[1271] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results. How this system is implemented will be specifically described below.

[1272] Program processing flow

[1273] 1. Upload an image

[1274] User: Uploads an image to be processed from their own device to the system. The image can be a selfie or a group photo.

[1275] Server: Receives uploaded images and stores them in temporary storage, then prepares them for the analysis process.

[1276] 2. Image Analysis

[1277] Server: Uses image analysis algorithms to detect faces in the received images and obtain their detailed location information.

[1278] 3. Detection of the landscape part of the pupil

[1279] Server: Based on the facial position information, the position of the eyes is identified, and the part of the scenery reflected in the eyes is further analyzed at the pixel level to identify it.

[1280] 4. Eye Image Generation

[1281] Server: Removes the identified landscape features and generates natural-looking pupils using image generation AI (e.g., generative adversarial networks (GANs)). During this process, the original pupil shape and lighting information are used as a reference to maintain a natural appearance.

[1282] 5. Applying the Emotion Engine

[1283] Server: In parallel with the eye detection, the emotion engine runs to analyze the user's facial expressions and recognize their emotions. This engine identifies emotions such as smile, sadness, surprise, etc.

[1284] Server: Adjust the color and shine of the automatically generated eyes based on the recognized emotion. For example, add a slight sparkle to the eyes when smiling.

[1285] 6. Image Merging

[1286] Server: The adjusted natural pupils are merged with the original image, and the whole image is processed to look natural.

[1287] 7. Providing edited images

[1288] Server: Returns the completed processed image to the user's device, along with comparison images of before and after processing, allowing the user to confirm the changes.

[1289] Specific examples

[1290] Example 1: Editing a selfie

[1291] User: Uploads a selfie to the application.

[1292] Server: Receives the image and starts face recognition and eye detection. At the same time, the emotion engine analyzes the user's facial expressions.

[1293] Server: Analyzes and identifies the part of the scenery reflected in the eyes. The emotion engine recognizes that the user is smiling.

[1294] Server: Uses image generation AI to generate natural-looking eyes while adding a slight glow effect.

[1295] Server: Sends the processed image to the user.

[1296] User: Review the received images and ensure privacy is maintained.

[1297] Example 2: Editing a group photo

[1298] User: Uploads a group photo of family and friends to the system.

[1299] Server: Receives the image and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[1300] Server: Identifies the part of the landscape reflected in each eye and uses AI to generate natural-looking eyes for each.

[1301] Server: Adds appropriate effects to the generated eyes based on the emotions recognized by the emotion engine.

[1302] Server: Merges the generated pupils with the original image and sends the processed image to the user.

[1303] User: Review the received images and ensure their privacy is protected.

[1304] This system allows users to obtain images with natural-looking eyes that reflect emotions while preventing privacy violations due to the scenery reflected in the eyes. It can also be introduced to businesses, and is expected to be widely used.

[1305] The processing flow will be explained below.

[1306] Step 1:

[1307] User: Uploads images that require privacy protection and emotion recognition to the system from their own device, such as selfies or group photos, by selecting and submitting them through the application or web interface.

[1308] Step 2:

[1309] Server: Receives the uploaded images and stores the image data in temporary storage, which prepares it for the subsequent analysis process.

[1310] Step 3:

[1311] Server: Activates the image analysis algorithm to detect faces in the received image. The face detection algorithm identifies and records the position and contour information of the face.

[1312] Step 4:

[1313] Server: Based on the face position information, the eye position is identified. Here, the eye detection algorithm detects the eye feature points and identifies the pupil area at pixel level.

[1314] Step 5:

[1315] Server: Performs a detailed analysis of the external scenery reflected within the detected pupil area and identifies that area. During this process, the pupil reflection and surrounding scenery information are analyzed pixel by pixel to extract unnecessary information.

[1316] Step 6:

[1317] Server: Activates the emotion engine and analyzes facial expressions in the image. The emotion engine identifies emotions such as smile, sadness, surprise, etc. and records the data.

[1318] Step 7:

[1319] Server: Removes the identified landscape features and generates natural-looking eyes using image generation AI (e.g., generative adversarial networks (GANs)). This process takes into account not only the original eye shape and lighting information, but also emotion data recognized by the emotion engine, preserving a natural appearance.

[1320] Step 8:

[1321] Server: Adjusts the color tone and gloss of the generated eyes based on the emotion engine data. For example, if the person is smiling, the eyes will be given a slight sparkle, making adjustments according to the emotion.

[1322] Step 9:

[1323] Server: Merge the adjusted natural pupils into the original image, replacing the original pupil regions with the new pupil image, so that the whole image looks natural.

[1324] Step 10:

[1325] Server: Returns the processed image to the user's device, along with a comparison image of the original and the processed image, allowing the user to easily verify the changes.

[1326] Step 11:

[1327] User: Receives the processed image, checks the processing results and any additional changes, and once they have confirmed that privacy is protected and that natural-looking eyes have been generated according to the emotion, they can save the image to their device or post it safely to social media.

[1328] Example 2

[1329] 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."

[1330] In modern society, as opportunities for image sharing increase, protecting user privacy has become an important issue. In particular, the scenery reflected in the eyes of images can potentially infringe on a user's privacy. It is also necessary to generate natural-looking eyes that reflect the user's emotions and maintain the realism of the entire image. Therefore, there is a need for technology that can reliably remove the scenery reflected in the eyes and generate natural-looking eyes that take emotions into account.

[1331] The identification process by the identification processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for recognizing emotions from the user's facial expression, means for adjusting the generated natural-looking pupils based on the recognized emotion, and means for providing the adjusted image to the user. This makes it possible to generate natural-looking pupils according to the emotion while protecting the user's privacy.

[1332] The "means for acquiring images" provides an interface for users to upload images taken from their own terminals to the server.

[1333] The "means for detecting pupils" uses an image analysis algorithm to identify the position of pupils in the acquired image.

[1334] The "means for identifying the scenery portion" is a means for analyzing and identifying the scenery portion reflected in the pupil at the pixel level based on the pupil position.

[1335] The "means for generating natural pupils" involves removing the identified scene features and generating new pupil images using a generative model.

[1336] The "means for merging with the original image" is a means for integrating the generated natural pupils with the original image, and processing the entire image so that it looks natural.

[1337] "Means for recognizing emotions from facial expressions" refers to algorithms or devices for analyzing a user's facial expressions and identifying emotions from those expressions.

[1338] The "means for adjusting the generated natural pupils" adjusts the color tone and gloss of the generated pupils based on the recognized emotion.

[1339] The "means for providing the adjusted image to the user" means transmitting the completed processed image to the user's terminal so that the user can check it.

[1340] This invention is a system that processes the scenery reflected in the eyes in an image to protect the user's privacy, and further recognizes the user's emotions and generates natural-looking eyes based on the results.

[1341] Specific means for implementing this system include the following steps involving a server, a terminal, and a user.

[1342] First, the user uploads the image that needs to be processed from their own device (for example, a personal computer or smartphone) to the system. The uploaded image is sent to a server and saved in temporary storage such as cloud storage. A specific storage method here could be to use a cloud storage service such as Amazon S3.

[1343] Next, the server uses image analysis libraries such as OpenCV and dlib to detect faces in the received images. This process obtains the face's position information (bounding box) and identifies the position of the eyes based on that. dlib's landmark detection technology is effective for identifying the position of the eyes. At this stage, pixel-level analysis is used to identify the part of the landscape reflected in the eyes.

[1344] The server removes the scenery and generates new, natural-looking eyes using a generative adversarial network (GAN). The generated eyes maintain the original eye shape and lighting information, achieving a natural look. The server also simultaneously uses an emotion engine (such as Microsoft Azure's Face API) to recognize emotions from the user's facial expressions. Based on the recognition results, the server adjusts the color tone and gloss of the generated natural-looking eyes. For example, when smiling, it can add a slight sparkle to the eyes.

[1345] Finally, the adjusted natural pupils are merged with the original image to create a natural-looking overall image. Alpha blending is used to combine the generated pupils with the original image. The completed processed image is sent from the server to the user's device, and comparison images before and after are also provided so the user can confirm the changes.

[1346] Specific examples

[1347] Example 1: Editing a selfie

[1348] 1. User: Uploads a selfie image from a smartphone application.

[1349] 2. Server: Receives the image and performs face recognition and eye detection, while analyzing the user's facial expressions using the emotion engine.

[1350] 3. Server: Identify the parts of the landscape that are reflected in the eyes and generate natural-looking eyes using GAN, adding sparkle to the eyes when smiling.

[1351] 4. Server: Merge the new pupils with the original image and send the processed image to the user.

[1352] 5. User: Check the edited image and ensure your privacy is protected.

[1353] Example 2: Editing a group photo

[1354] 1. User: Uploads a group photo of family and friends from a PC.

[1355] 2. Server: Receives the images and starts face recognition and eye detection. Each person's facial expression is recognized by the emotion engine.

[1356] 3. Server: Identify the part of the scene reflected in each eye and generate natural-looking eyes for each eye, for example, adding a glow to the emotion of joy.

[1357] 4. Server: Merge the generated pupils with the original image and send the processed image to the user.

[1358] 5. User: Check the edited image and ensure that your privacy is protected.

[1359] Example prompts to input to the generative AI model

[1360] Example prompt:

[1361] "Please provide a step-by-step outline of a system that removes the landscape reflected in the eyes from user-uploaded images and generates natural-looking eyes using a generative adversarial network (GAN). Also, please describe how to recognize the user's emotion and apply eye effects accordingly."

[1362] This system can obtain images with natural-looking eyes that correspond to emotions while preventing the invasion of privacy of the user's eyes. It can also be introduced to businesses, and is expected to be widely used.

[1363] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1364] Step 1:

[1365] Uploading an image

[1366] User: Upload an image to the system from their device.

[1367] Input: The image that needs to be enhanced (e.g. a selfie).

[1368] Specific behavior: Selects an image file using the device's file selection dialog.

[1369] Output: The image file is sent to the server.

[1370] Step 2:

[1371] Receiving and saving images

[1372] Server: Receives uploaded images and temporarily stores them in cloud storage.

[1373] Input: An image file submitted by the user.

[1374] Specific operation: The received image data is saved in cloud storage such as Amazon S3.

[1375] Output: The storage path of the saved image file.

[1376] Step 3:

[1377] Image analysis and facial recognition

[1378] Server: Detect faces using an image analysis library.

[1379] Input: Image files stored in cloud storage.

[1380] Specific operation: Detects faces using OpenCV's cv2.CascadeClassifier and obtains their position information as a bounding box.

[1381] Output: Face position information (bounding box).

[1382] Step 4:

[1383] Detecting the landscape part of the pupil

[1384] Server: Identifies the position of the eyes based on face position information and analyzes the part of the scenery reflected in the eyes.

[1385] Input: Face bounding box information and image files.

[1386] What it does: It uses dlib landmark detection to identify the position of the pupil, then analyzes pixel values ​​to extract the part of the landscape reflected in the pupil.

[1387] Output: Eye position and landscape pixel information.

[1388] Step 5:

[1389] Eye image generation

[1390] Server: Generates new natural-looking pupils using a generative model (GAN).

[1391] Input: Pixel information of the landscape part of the pupil.

[1392] How it works: First, mask out the landscape area, then input that area into the GAN, which then generates natural-looking pupils.

[1393] Output: Generated pupil image data.

[1394] Step 6:

[1395] Applying the Emotion Engine

[1396] Server: Analyzes facial expressions and recognizes emotions.

[1397] Input: Image file and face position information.

[1398] Specific operation: Using an emotion engine (for example, Microsoft Azure's Face API), emotions are recognized from facial expressions, and the color and gloss of the generated eyes are adjusted based on that information.

[1399] Output: Recognized emotion data and adjusted eye image data.

[1400] Step 7:

[1401] Image merging

[1402] Server: Integrate the adjusted natural pupils into the original image.

[1403] Input: Adjusted pupil image data and original image files.

[1404] What it does: Uses alpha blending to naturally merge the generated pupil with the original image.

[1405] Output: The finished processed image.

[1406] Step 8:

[1407] Providing processed images

[1408] Server: Sends the processed image to the user's device.

[1409] Input: The finished processed image.

[1410] Specific operation: Comparison images before and after processing are also generated and sent back to the user. WebSocket is used as a real-time notification method.

[1411] Output: The processed and comparison images sent to the user.

[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] Because the scenery reflected in the eyes in an image can invade privacy, a means to prevent this is needed. Also, to make the image look more natural, it is necessary to adjust the color tone and brightness of the eyes according to the user's emotions. In brick-and-mortar stores, technology is needed to protect privacy from selfies and other people's gazes through mirrors while improving the user experience.

[1415] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring an image, means for detecting pupils in the acquired image, means for identifying a landscape portion reflected in the detected pupils, means for deleting the identified landscape portion and generating natural-looking pupils, means for merging the generated natural-looking pupils with the original image, means for providing the merged image to the user, and means for recognizing the user's emotion and adjusting the color tone and gloss of the pupils based on the result. This makes it possible to adjust the pupils in real time according to the emotion while maintaining a natural appearance and protecting the user's privacy.

[1416] "Means for acquiring images" refers to a combination of hardware and software for capturing images from a user's information processing device or camera.

[1417] The "means for detecting pupils in the acquired image" is an algorithm for identifying pupils in the image and obtaining their position information.

[1418] The "means for identifying the portion of the scenery reflected in the detected pupil" is a process for identifying the external scenery or objects reflected in the pupil and analyzing it on a pixel-by-pixel basis.

[1419] The "means for removing identified landscape features and generating natural-looking eyes" is a system that uses AI techniques such as generative adversarial networks (GANs) to generate natural-looking eye images after removing the landscape features.

[1420] The "means for merging the generated natural pupils into the original image" is a technique for integrating the generated pupil images into the original photograph, making the whole look natural.

[1421] The "means for providing a merged image to a user" is a system for transmitting or displaying the processed image to a user.

[1422] "Means for recognizing the user's emotions and adjusting the color tone and gloss of the pupils based on the results" refers to a process that uses an emotion recognition engine to analyze the user's facial expression and appropriately corrects the characteristics of the pupils (color tone and gloss) generated based on that.

[1423] This invention is part of a smart mirror system that also protects privacy in brick-and-mortar stores. It removes the scenery reflected in the eyes in the image, generates natural-looking eyes based on the user's emotions, and provides real-time feedback.

[1424] Program processing flow

[1425] 1. Image Acquisition:

[1426] When a user stands in front of the smart mirror, the mirror's camera automatically captures an image, which is then seamlessly transmitted to the server.

[1427] 2. Eye detection:

[1428] The server uses a facial recognition algorithm such as OpenCV to detect faces and eyes in the image, which allows the precise location of the eyes.

[1429] 3. Identifying the landscape:

[1430] The server zooms in on the detected pupil area and performs pixel-by-pixel analysis, identifying specific parts of the scene or object and extracting that pixel information.

[1431] 4. Removing landscape elements and generating natural-looking eyes:

[1432] The server uses a generative adversarial network (GAN) to remove the landscape and generate natural-looking pupils, taking into account the original pupil shape and lighting information to maintain harmony across the entire image.

[1433] 5. Emotion Recognition:

[1434] The server uses an emotion recognition engine such as AWS Rekognition to analyze the user's facial expressions and read their emotions, then adjusts the color and gloss of the eyes appropriately based on the emotional information.

[1435] 6. Image merging:

[1436] The server integrates the generated pupils into the original image and processes it so that the whole image looks natural.

[1437] 7. Provide processed images:

[1438] The server provides the completed image to the user in real time, allowing the user to instantly check the edited image.

[1439] Hardware and software used

[1440] Camera: A camera device attached to the smart mirror.

[1441] Face Recognition Algorithm: OpenCV

[1442] Generative AI Models: Generative Adversarial Networks (GANs)

[1443] Emotion recognition engine: AWS Rekognition

[1444] Server: Server for image analysis and data processing

[1445] Specific examples

[1446] A specific example in which a user actually stands in front of a mirror is shown below.

[1447] When a user stands in front of the smart mirror, the camera automatically recognizes their face and eyes and captures an image. The emotion recognition engine detects when the user is smiling and applies a slightly sparkling eye effect. This allows the user to view the processed image in real time, ensuring privacy while allowing them to see natural-looking eyes that reflect their emotions.

[1448] Prompt Sentence Examples

[1449] "Write a Python program to protect image privacy and generate natural-looking eye pupils based on emotions. The following requirements must be met:

[1450] 1. Upload an image

[1451] 2. Face and Eye Analysis

[1452] 3. Removing the scenery reflected in the eyes

[1453] 4. Emotion recognition and pupil generation

[1454] 5. Merging and delivering the final image

[1455] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1456] Step 1:

[1457] The server acquires images from the smart mirror's camera. When a user stands in front of the mirror, the camera automatically activates and captures an image containing the user's face. The input is the raw image from the camera, and the output is an image file sent to the server.

[1458] Step 2:

[1459] The server uses a face recognition algorithm (e.g., OpenCV) to detect the pupils in the captured image. The input is the image sent to the server, and the output is the coordinate information of the pupils. The server scans the entire face to identify the pupil positions as a data calculation, and identifies a rectangular area that includes the eye positions.

[1460] Step 3:

[1461] The server performs pixel-by-pixel analysis to identify the landscape portion of the detected pupil. The input is the coordinate information of the pupil and the image range of the detected face, and the output is pixel information of the landscape portion. The server enlarges the pupil area and analyzes the reflection and different color tones within it to obtain the landscape portion.

[1462] Step 4:

[1463] The server uses a generative adversarial network (GAN) to remove the identified scene and generate natural-looking pupils. The input is pixel information of the scene, and the output is an image of a natural-looking pupil. The server inputs the scene information into the GAN model and obtains pixel data of the generated natural-looking pupils in return.

[1464] Step 5:

[1465] The server uses an emotion recognition engine (e.g., AWS Rekognition) to analyze the user's facial expressions and recognize their emotions. The input is an image of the user's face, and the output is the user's emotion data. The server identifies emotions such as smile, sadness, or surprise from the user's facial expressions through natural language processing.

[1466] Step 6:

[1467] The server adjusts the color tone and gloss of the generated eyes based on the recognized emotion. The input is emotion data and a generated natural eye image, and the output is an eye image adjusted according to the emotion. The server analyzes the emotion data and applies effects such as brightening the eyes if the person is smiling.

[1468] Step 7:

[1469] The server merges the adjusted natural pupils with the original image and processes it so that the whole image looks natural. The input is the adjusted natural pupil image and the original face image, and the output is the processed final image. The server processes the data by pasting the new pupils pixel by pixel to the original image.

[1470] Step 8:

[1471] The server provides the completed processed image to the user's information processing device. The input is the final processed image, and the output is image data sent to the user. The user can check the processed image on their own device, safely protecting their privacy while also being able to see natural-looking eyes based on their emotions.

[1472] 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.

[1473] 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.

[1474] 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 robot 414.

[1475] 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.

[1476] FIG. 9 illustrates 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 behaviors 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.

[1477] 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.

[1478] 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).

[1479] 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.

[1480] 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."

[1481] 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.

[1482] 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).

[1483] 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.

[1484] 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.

[1485] 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.

[1486] 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.

[1487] 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.

[1488] 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.

[1489] 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.

[1490] 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.

[1491] 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.

[1492] 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.

[1493] The following is further disclosed regarding the above embodiment.

[1494] (Claim 1)

[1495] a means for acquiring an image;

[1496] means for detecting pupils in the acquired image;

[1497] A means for identifying a portion of a landscape reflected in the detected pupil;

[1498] A means for removing the identified scenery portion and generating natural-looking pupils;

[1499] a means for merging the generated natural pupils with the original image;

[1500] means for providing the merged image to a user;

[1501] A system including:

[1502] (Claim 2)

[1503] 2. The system according to claim 1, wherein the acquired image is an image transmitted from a user terminal.

[1504] (Claim 3)

[1505] 10. The system of claim 1, wherein the generated natural-looking pupils are generated using a generative adversarial network (GAN).

[1506] "Example 1"

[1507] (Claim 1)

[1508] a means for a user to upload an image from a terminal;

[1509] means for detecting pupils in the acquired image;

[1510] A means for identifying a portion of a landscape reflected in the detected pupil;

[1511] A means for removing the identified scenery portion and generating natural-looking pupils;

[1512] a means for merging the generated natural pupils with the original image;

[1513] means for providing the merged image to a user;

[1514] A system including:

[1515] (Claim 2)

[1516] 2. The system of claim 1, wherein the captured image is transmitted from a user terminal.

[1517] (Claim 3)

[1518] 10. The system of claim 1, wherein the generated natural-looking pupils are generated using a generative adversarial network.

[1519] "Application Example 1"

[1520] (Claim 1)

[1521] a means for acquiring an image;

[1522] means for detecting pupils in the acquired image;

[1523] A means for identifying a portion of a landscape reflected in the detected pupil;

[1524] A means for removing the identified scenery portion and generating natural-looking pupils;

[1525] a means for merging the generated natural pupils with the original image;

[1526] means for providing the merged image to a user;

[1527] A means for providing a comparison before and after image based on the generated privacy-protected image data;

[1528] A means for performing the above processing on video data uploaded from a surveillance camera or a smart device;

[1529] A system including:

[1530] (Claim 2)

[1531] 2. The system according to claim 1, wherein the acquired images are images transmitted from a surveillance camera or a smart device.

[1532] (Claim 3)

[1533] 10. The system of claim 1, wherein the generated natural-looking pupils are generated using a generative adversarial network (GAN).

[1534] "Example 2: Combining Emotion Engines"

[1535] (Claim 1)

[1536] a means for acquiring an image;

[1537] means for detecting pupils in the acquired image;

[1538] A means for identifying a portion of a landscape reflected in the detected pupil;

[1539] A means for removing the identified scenery portion and generating natural-looking pupils;

[1540] a means for merging the generated natural pupils with the original image;

[1541] means for recognizing emotions from the facial expressions of a user;

[1542] means for adjusting the generated natural pupils based on the recognized emotion;

[1543] means for providing the adjusted image to a user;

[1544] A system including:

[1545] (Claim 2)

[1546] 2. The system according to claim 1, wherein the acquired image is an image transmitted from an information terminal.

[1547] (Claim 3)

[1548] 10. The system of claim 1, wherein the generated natural pupil is generated using a generative model.

[1549] "Application example 2 when combining emotion engines"

[1550] (Claim 1)

[1551] a means for acquiring an image;

[1552] means for detecting pupils in the acquired image;

[1553] A means for identifying a portion of a landscape reflected in the detected pupil;

[1554] A means for removing the identified scenery portion and generating natural-looking pupils;

[1555] a means for merging the generated natural pupils with the original image;

[1556] means for providing the merged image to a user;

[1557] a means for recognizing a user's emotion and adjusting the color tone and gloss of the pupil based on the result of the recognition;

[1558] A system including:

[1559] (Claim 2)

[1560] 2. The system according to claim 1, wherein the acquired image is an image transmitted from a user's information processing device.

[1561] (Claim 3)

[1562] 10. The system of claim 1, wherein the generated natural-looking pupils are generated using a generative adversarial network. [Explanation of symbols]

[1563] 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. a means for acquiring an image; means for detecting pupils in the acquired image; A means for identifying a portion of a landscape reflected in the detected pupil; A means for removing the identified scenery portion and generating natural-looking pupils; a means for merging the generated natural pupils with the original image; means for providing the merged image to a user; A system including:

2. 2. The system of claim 1, wherein the captured image is an image transmitted from a user terminal.

3. 10. The system of claim 1, wherein the generated natural-looking pupils are generated using a generative adversarial network.

Images