System

Abstract

A system is provided.SOLUTION: A system comprising: means for acquiring video data in real time from a video camera; means for preprocessing the acquired video data and performing noise removal and resolution adjustment; means for inputting the preprocessed video data to a generation model and detecting an abnormal behavior based on learning data; means for notifying and reporting to a related organization or a specific user when the abnormal behavior is detected; and means for recording and storing a detection result of the abnormal behavior and the video data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] Prevention of damage and labor shortages are important issues in the security and guarding industries. Current surveillance camera systems rely primarily on human eyes for monitoring, which takes time and effort and carries the risk of overlooking something. This often makes it difficult to respond quickly when an incident or accident occurs. The present invention solves these problems by providing a system that improves surveillance efficiency and enables quick and accurate responses. [Means for solving the problem]

[0005] The present invention provides a means for acquiring video data from a surveillance camera in real time and preprocessing the acquired video data to remove noise and adjust resolution. It also includes a means for inputting the preprocessed video data into a generative model and detecting anomalous behavior based on past training data. It also includes a means for immediately notifying and reporting relevant organizations or specific users when anomalous behavior is detected. It also includes a means for recording and saving the results of anomalous behavior detection and the video data, and a means for analyzing the saved data to retrain the generative model. This provides a system that improves surveillance efficiency, enables prevention of damage, and enables rapid response.

[0006] A "surveillance camera" is a camera device installed to monitor and record a specific area, and acquires video data in real time.

[0007] "Video data" refers to digital data of video images acquired from a surveillance camera.

[0008] "Preprocessing" is a processing step in which acquired video data is subjected to noise removal, resolution adjustment, etc. to make it suitable for analysis.

[0009] A "generative model" is a model that is trained using artificial intelligence or machine learning algorithms to identify specific patterns or anomalous behavior.

[0010] "Abnormal behavior" refers to behavior or situations that deviate from normal behavioral patterns (e.g., falling, running away, placing suspicious objects, etc.).

[0011] "Notification" is the act of communicating information to specific users or relevant organizations when abnormal behavior is detected.

[0012] "Reporting" is the act of quickly providing detailed information about abnormal behavior to relevant organizations or specific users when it occurs.

[0013] "Recording" refers to the act of storing detected anomalous behavior and associated video data in a digital format.

[0014] "Storage" refers to the act of retaining recorded data in a state that allows for later analysis and review.

[0015] "Relearning" is a retraining process that aims to improve the accuracy of a generative model by using previously detected abnormal behavior data.

[0016] "Related agencies" refers to external agencies such as security companies and police that respond to incidents and accidents. [Brief explanation of the drawings]

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

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

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

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

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

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

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

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

[0025] [First embodiment]

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

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

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

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

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

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

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

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

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

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

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

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

[0038] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[0039] System configuration

[0040] This system consists of a surveillance camera, a server, a terminal, and a user.

[0041] 1. Surveillance cameras

[0042] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0043] 2. Server

[0044] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0045] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0046] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0047] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0048] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0049] Data storage module: Stores all anomaly detection events and associated video data.

[0050] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0051] 3. Terminal

[0052] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0053] 4. Users

[0054] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0055] Explanation of program processing

[0056] Below, the operation of the system program is explained in natural language.

[0057] 1. Obtaining surveillance camera footage

[0058] The server acquires video data in real time from multiple surveillance cameras, which are connected to the server via a network and continuously transmit the video data to the server.

[0059] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[0060] 2. Preprocessing of video data

[0061] The server performs preprocessing on the acquired video data.

[0062] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[0063] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[0064] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[0065] 3. Anomaly Detection Using Generative Models

[0066] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[0067] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[0068] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[0069] 4. Notification and reporting when an abnormality is detected

[0070] When the server detects an abnormality, it immediately notifies and reports the relevant organizations and specific users.

[0071] The notification will include the time and location of the anomaly and any necessary action.

[0072] Reporting can be done via SMS, email, or a dedicated notification app.

[0073] 5. History storage and analysis

[0074] The server records and stores all anomaly detection events and associated video data.

[0075] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[0076] Specific examples

[0077] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[0078] 1. Video acquisition

[0079] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0080] 2. Preprocessing of video data

[0081] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0082] 3. Anomaly Detection

[0083] The server inputs the preprocessed video data into a generative model to detect falling behavior.

[0084] 4. Notifications and Reports

[0085] When a fall is detected, the server immediately notifies mall security staff, including the location and time of the fall, as well as a video snapshot of the frame in question.

[0086] 5. History storage and analysis

[0087] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generative model.

[0088] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[0089] The processing flow will be explained below.

[0090] Step 1:

[0091] The server acquires video data in real time from each surveillance camera. The surveillance cameras are connected to the server via an IP network and transmit the video data continuously in streaming format.

[0092] Step 2:

[0093] The server preprocesses the captured video data by performing noise reduction, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting the video data into a format suitable for analysis.

[0094] Step 3:

[0095] The server divides the pre-processed video data into frames, each of which is assigned a timestamp to make it uniquely identifiable.

[0096] Step 4:

[0097] The server inputs each preprocessed frame of data into a generative model, which then analyzes each frame based on pre-trained patterns of abnormal behavior.

[0098] Step 5:

[0099] The server receives the analysis results from the generative model and detects anomalous behavior. If anomalous behavior is detected, an anomalous flag is set for the corresponding frame. Abnormal behavior includes falling, fleeing, placing suspicious objects, etc.

[0100] Step 6:

[0101] The server will send a notification when an abnormality is detected. Specifically, it will send a notification containing detailed information about the abnormality to relevant organizations or specific users. Notification methods include SMS, email, or a dedicated notification app.

[0102] Step 7:

[0103] The server records and stores all anomaly detection events and associated video data for later analysis and review.

[0104] Step 8:

[0105] The server periodically analyzes the stored data and retrains the generative model, improving the model's accuracy and enabling more effective anomalous behavior detection.

[0106] Example 1

[0107] 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."

[0108] Conventional surveillance camera systems have the problem of being difficult to detect abnormal behavior in real time and quickly notify and report it. They also lack a mechanism for saving detection results and using them for re-learning to improve the accuracy of the system. This can result in insufficient accuracy in detecting abnormal behavior and rapid response.

[0109] 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.

[0110] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise, adjust resolution, and split frames, means for inputting the preprocessed video data into a generative AI model and detecting abnormal behavior based on the learning data, means for immediately notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results and video data, and for later analysis and re-training of the generative AI model. This enables highly accurate detection of abnormal behavior in real time and rapid response.

[0111] A "surveillance camera" is a device that captures real-time video data within a designated area and transmits the data over a network.

[0112] "Real-time video data" refers to video data that is continuously captured by a surveillance camera and instantly transmitted to a server.

[0113] "Preprocessing" refers to processing that removes noise, adjusts resolution, and divides frames from video data.

[0114] "Noise removal" is a process that removes unnecessary noise components from video data to improve the accuracy of analysis.

[0115] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[0116] "Frame division" is a process of breaking down video data into frames and assigning a timestamp to each frame.

[0117] A "generative AI model" is an artificial intelligence model for detecting abnormal behavior in video data based on past learning data.

[0118] "Abnormal behavior" refers to actions that would not normally occur, such as falling, running away, or suspicious behavior.

[0119] "Notification and reporting" refers to the act of sending information about abnormal behavior to relevant organizations or specific users when it is detected.

[0120] "Recording and storage" refers to storing abnormal behavior detection results and related video data in a database.

[0121] "Retraining" is the process of data analysis and retraining that utilizes stored data to improve the accuracy of generative AI models.

[0122] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[0123] The system consists of a surveillance camera, a server, a terminal, and a user.

[0124] Hardware and Software Details

[0125] surveillance cameras

[0126] A surveillance camera is a device that captures real-time video data within a specified area and transmits it to a server via a network. The video data is transmitted to the server in streaming format via an IP network. The camera used is, for example, a high-resolution IP camera (e.g., 1080p HD camera).

[0127] server

[0128] The server is a computer system that aggregates and stores video data acquired from surveillance cameras and performs various processing. The server has the following functions:

[0129] Data Acquisition Module: Acquires real-time video data from each surveillance camera using standard protocols (e.g., RTSP).

[0130] Pre-processing module: Performs noise reduction, resolution adjustment, and frame division on the acquired video data. Here, an image processing library (e.g., OpenCV) is used.

[0131] Noise Reduction: Apply a Gaussian filter to remove noise.

[0132] Resolution adjustment: The resize function converts video data to a uniform resolution (e.g., 720p).

[0133] Frame division: Obtains FPS information, divides the video data into frames, and adds timestamps.

[0134] Anomaly Detection Module: Preprocessed video data is fed into a generative AI model to detect anomalous behavior. Deep learning frameworks (e.g., TensorFlow, PyTorch) are used.

[0135] Send the following prompt to the generative model:

[0136] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[0137] Notification and reporting module: When an abnormality is detected, it immediately notifies and reports to relevant organizations or specific users via SMS, email, or a dedicated notification app.

[0138] For example, an email could be sent stating, "A fall occurred in the North Area on October 5, 2023 at 2:30 p.m. Action is required."

[0139] Data storage module: Records anomaly detection events and related video data and stores them in a database.

[0140] Retraining module: Retrains the generative AI model using the stored data. Periodically extracts historical data from the database to create a retraining dataset.

[0141] Terminal

[0142] The devices are used by surveillance camera system administrators and security staff to receive notifications and alerts when an abnormality is detected. A dedicated app is installed, allowing for real-time situation monitoring and response.

[0143] User

[0144] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users respond promptly based on the content of notifications and reports sent from the system.

[0145] Specific examples

[0146] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[0147] 1. Acquiring video data

[0148] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0149] 2. Pretreatment

[0150] The server performs noise reduction and resolution adjustment on the video data and adds a timestamp to the frame.

[0151] 3. Detecting Abnormal Behavior

[0152] The server inputs the preprocessed video data into a generative AI model to detect falling behavior.

[0153] 4. Notifications and Reports

[0154] When a fall is detected, the server immediately notifies the mall's security staff.

[0155] 5. History storage and analysis

[0156] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generated AI model.

[0157] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[0158] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0159] Step 1: Acquire video data

[0160] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera sends video data to the server in streaming format via an IP network. The input is real-time video data from the surveillance cameras, and the output is the acquired raw video data. The server receives the video data using a standard protocol (e.g., RTSP). Specifically, the server periodically checks the connection status of each camera and attempts to reconnect if there is an abnormality.

[0161] Step 2: Preprocessing

[0162] The server performs pre-processing on the captured video data, including noise reduction, resolution adjustment, and frame splitting. The input is the captured raw video data, and the output is the pre-processed video data.

[0163] Noise removal: The server uses an image processing library (e.g., OpenCV) to apply a Gaussian filter to remove noise from the video data.

[0164] Resolution Adjustment: The server uses a resize function to convert all video data to a uniform resolution (e.g., 720p).

[0165] Frame division: The server acquires FPS (frames per second) information and divides the video data into frames. At the same time, it assigns a timestamp to each frame. Specifically, the server analyzes the video data and extracts frames at the appropriate time.

[0166] Step 3: Detecting Abnormal Behavior

[0167] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The input is the preprocessed video data, and the output is the abnormal behavior detection results.

[0168] Generative AI models are built using deep learning frameworks (e.g., TensorFlow, PyTorch).

[0169] The server converts the video data into the model's input format, sets the following prompt text, and sends it to the model.

[0170] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[0171] The server receives the model output, and if an anomaly is detected, it sets an anomaly flag for the corresponding frame. Specifically, the server inputs the data converted into tensor format into the model, processes the results, and sets an anomaly flag.

[0172] Step 4: Notification

[0173] When the server detects abnormal behavior, it immediately notifies and reports to relevant organizations or specific users. The input is the abnormal behavior detection result, and the output is a notification / report message.

[0174] The notification will include the time and location of the anomaly and any necessary action.

[0175] Reporting methods include SMS, email, and a dedicated notification app.

[0176] For example, an email could be sent stating, "A fall occurred in the North Area at 2:30 PM on October 5, 2023. Action is required." Specifically, the server generates a notification message and sends it to the specified contacts. It also uses the API of a dedicated notification app to send notifications in real time.

[0177] Step 5: Store and analyze history

[0178] The server records and stores all anomaly detection events and associated video data. The input is the abnormal behavior detection results and associated video data, and the output is the stored historical data.

[0179] The stored data will be analyzed at a later date and used to improve the system and retrain the generative AI model. Specifically, the server stores the data in a database, periodically extracts historical data from the database, and creates a dataset for retraining.

[0180] This enables the entire system to detect abnormal behavior in real time with high accuracy and respond quickly.

[0181] (Application example 1)

[0182] 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."

[0183] Conventional surveillance camera systems have issues with delays in detecting and notifying abnormal behavior, making it difficult to respond quickly. Furthermore, because notifications of abnormal situations are limited to specific devices, information sharing is insufficient and there is a lack of means to quickly contact multiple parties involved. As a result, security staff and guards are often unable to respond in a timely manner.

[0184] 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.

[0185] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the learning data, means for notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, means for recording and saving the abnormal behavior detection results and the video data, means for sending the notification as a push notification to a mobile device of the relevant organization or specific user, and means for generating an audio alarm when abnormal behavior is detected, thereby enabling immediate notification to multiple relevant parties and appropriate and prompt response.

[0186] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[0187] "Real-time" refers to the immediate acquisition and processing of data without delay.

[0188] "Video data" is information in the form of video captured by a surveillance camera.

[0189] "Noise reduction" refers to the process of removing unnecessary signals and noise from video data.

[0190] "Resolution" is an index that indicates the level of detail and image quality of video data.

[0191] A "generative model" is an AI model used to detect anomalous behavior based on training data.

[0192] "Training data" is the historical dataset used to train the generative model.

[0193] "Abnormal behavior" refers to actions or events that are different from the norm, such as falling, running away, or suspicious behavior.

[0194] "Notification" refers to the act of notifying relevant organizations or specific users when abnormal behavior is detected.

[0195] "Reporting" is a method of alerting or reporting to relevant authorities when abnormal behavior is detected.

[0196] "Related agencies" are agencies or organizations involved in responding to abnormal behavior, such as security staff, guards, and police.

[0197] A "specific user" is an individual or organization that is designated in advance by the system to receive notifications.

[0198] "Recording" refers to the act of saving detected abnormal behavior and related data.

[0199] "Storage" refers to keeping recorded data for future reference.

[0200] A "mobile terminal" is a mobile communication device such as a smartphone or tablet.

[0201] "Push notification" is a function that delivers notifications from the system to mobile devices in real time.

[0202] An "audio alarm" is a means of alerting the user by sound when abnormal behavior is detected.

[0203] This invention provides a system that acquires video data from a surveillance camera in real time and detects, notifies, and reports abnormal behavior. This system can quickly and efficiently detect abnormal behavior and report it to relevant authorities and designated users in order to increase the safety of the monitored area. A specific embodiment of the invention will be described below.

[0204] System configuration

[0205] surveillance cameras

[0206] Surveillance cameras capture real-time video data within a specific area and are connected to a central server via a network, where the captured video data is sent in streaming format.

[0207] server

[0208] The server has the following functions:

[0209] Data acquisition module: Acquires video data from surveillance cameras in real time.

[0210] Pre-processing module: Performs noise reduction and resolution adjustment on the acquired video data.

[0211] Anomaly detection module: Preprocessed video data is input into the generative model, and anomalous behavior is detected based on the training data.

[0212] Notification and reporting module: When abnormal behavior is detected, the module notifies and reports relevant organizations or specific users. Notifications are sent as push notifications to the mobile devices of relevant organizations or users.

[0213] Audio alarm module: Generates an audio alarm when abnormal behavior is detected.

[0214] Data storage module: Records and stores abnormal behavior detection results and related video data.

[0215] Terminal

[0216] The devices are used by security staff and administrators, and have an app installed that receives push notifications when an abnormality is detected. If abnormal behavior is detected, a push notification is immediately sent to the device, prompting appropriate action to be taken.

[0217] User

[0218] Users include relevant organizations and security staff, guards, police officers, etc. who are required to receive specific notifications. Users are required to respond promptly to the situation based on the push notification.

[0219] Hardware and software used

[0220] Hardware: surveillance cameras, servers, smartphones (mobile devices)

[0221] Software: Flask (server-side framework), OpenCV (image processing library), TensorFlow (generative model), Firebase Cloud Messaging (push notifications)

[0222] How we process your data

[0223] The server receives streaming video data from the surveillance cameras, and the preprocessing module performs noise reduction and resolution adjustment. The preprocessed video data is then input into the generative model to detect anomalous behavior. Detected anomalous behavior is immediately recognized by the anomaly detection module.

[0224] When an abnormality is detected, the notification module is activated and sends a push notification to relevant organizations or specific users. At the same time, the audio alarm module is activated and generates an audio alarm to prompt a prompt response. In addition, the data storage module records and stores the abnormal behavior detection results and related video data.

[0225] For example, if someone falls in a shopping mall, the video from the surveillance camera is sent to the server in real time, and the fall is detected using a TensorFlow model. A push notification is then sent to the security guard's mobile device via Firebase Cloud Messaging, and an audio alarm sounds, enabling a prompt response.

[0226] Prompt Sentence Examples

[0227] "Model input is generated from preprocessed video data and fed into the generative model."

[0228] "Detects abnormal behavior based on the prediction results and notifies relevant authorities."

[0229] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0230] Step 1:

[0231] The server acquires video data from the surveillance cameras in real time. The surveillance cameras capture a specific area and continuously transmit the video data to the server via the network. The server receives the data and prepares it for the next step of processing. The input of this step is the raw data sent from the surveillance cameras, and the output is the video data for preprocessing.

[0232] Step 2:

[0233] The server performs preprocessing on the received video data. Specifically, it uses OpenCV to remove noise and adjust the resolution. This improves the quality of the video and increases the accuracy of abnormal behavior detection. The input of this step is raw data, and the output is video data with noise removed and resolution adjusted.

[0234] Step 3:

[0235] The server inputs the preprocessed video data into the generative model. Using TensorFlow, the data is analyzed frame by frame and anomalous behavior is detected based on the training data. The input to this step is the preprocessed video data, and the output is the abnormal behavior detection results. Specifically, the generative model identifies abnormal behavior.

[0236] Step 4:

[0237] The server issues notifications and reports when abnormal behavior is detected. It uses Firebase Cloud Messaging to send push notifications to the mobile devices of relevant organizations or specific users. This notification includes the time, location, and video snapshot of the abnormal behavior. The input to this step is the abnormal behavior detection result, and the output is the sent notification and report. Specifically, the notification and report module is activated and a push notification is sent.

[0238] Step 5:

[0239] The server generates an audio alarm when abnormal behavior is detected. The audio alarm module plays an alarm sound immediately when the abnormal behavior detection result is input. The input of this step is the abnormal behavior detection result, and the output is the generated alarm sound. Specifically, the audio alarm is activated to warn the user.

[0240] Step 6:

[0241] The server records and stores the abnormal behavior detection results and related video data. The data storage module stores the data in a database for future analysis and system improvement. The input of this step is the abnormal behavior detection results and related video data, and the output is the stored data. The specific operation is writing to the database.

[0242] Step 7:

[0243] The user receives a push notification sent to their mobile device and is prompted to take action. In this step, the user who received the push notification quickly heads to the scene and takes appropriate action. The input to this step is the push notification sent from the server, and the output is the user's on-site action. Specifically, the user checks the notification and takes the necessary action.

[0244] As described above, the present invention is a system that can detect abnormal behavior in real time and quickly notify and report the situation to relevant organizations and users, thereby significantly improving the safety of the monitored area.

[0245] 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.

[0246] The present invention provides a system that detects abnormal behavior in real time in a surveillance camera system and further enhances security and response to abnormalities by combining it with an emotion engine that recognizes user emotions. Specific embodiments of the present invention will be described below.

[0247] System configuration

[0248] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[0249] 1. Surveillance cameras

[0250] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0251] 2. Server

[0252] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0253] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0254] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0255] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0256] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0257] Data storage module: Stores all anomaly detection events and associated video data.

[0258] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0259] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[0260] 3. Terminal

[0261] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0262] 4. Users

[0263] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0264] 5. Emotion Engine

[0265] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[0266] Explanation of program processing

[0267] Below, the operation of the system program is explained in natural language.

[0268] 1. Obtaining surveillance camera footage

[0269] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera is connected to the server via a network, and the video data is continuously transmitted to the server.

[0270] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[0271] 2. Preprocessing of video data

[0272] The server performs preprocessing on the acquired video data.

[0273] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[0274] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[0275] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[0276] 3. Anomaly Detection Using Generative Models

[0277] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[0278] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[0279] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[0280] 4. Emotion analysis

[0281] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine.

[0282] The emotion engine uses facial recognition technology to identify emotions (e.g., surprise, anger, fear, etc.) from the user's facial expressions.

[0283] 5. Notification and reporting when an abnormality is detected

[0284] The server detects an abnormality and analyzes the user's emotions, then immediately notifies and reports the relevant authorities and specific users.

[0285] The notification will include detailed information about the time and location of the anomaly, as well as emotional data.

[0286] Reporting can be done via SMS, email, or a dedicated notification app.

[0287] 6. History storage and analysis

[0288] The server records and stores all anomaly detection events and associated emotion and video data.

[0289] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[0290] Specific examples

[0291] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[0292] 1. Video acquisition

[0293] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0294] 2. Preprocessing of video data

[0295] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0296] 3. Anomaly Detection and Sentiment Analysis

[0297] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[0298] 4. Notifications and Reports

[0299] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[0300] 5. History storage and analysis

[0301] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[0302] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[0303] The processing flow will be explained below.

[0304] Step 1:

[0305] The server acquires video data in real time from each surveillance camera. Each surveillance camera is connected to the server via a network, and the video data is continuously streamed to the server.

[0306] Step 2:

[0307] The server preprocesses the acquired video data, specifically by denoising the video data, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting it into a format that is easier to analyze.

[0308] Step 3:

[0309] The server divides the pre-processed video data into frames, and assigns a timestamp to each frame to make them individually identifiable.

[0310] Step 4:

[0311] The server inputs each preprocessed frame of data into a generative model, which analyzes each frame based on pre-trained abnormal behavior patterns.

[0312] Step 5:

[0313] The server receives the analysis results of the generative model and detects abnormal behavior (e.g., falls, escape, placing suspicious objects, etc.). For detected abnormal behavior, it sets an abnormal flag for the corresponding frame.

[0314] Step 6:

[0315] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to analyze the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.).

[0316] Step 7:

[0317] The server then notifies relevant organizations and specific users based on the results of the user's emotion analysis and abnormal behavior detection. The notification includes detailed information about the abnormal behavior and emotions. Notifications can be sent via SMS, email, or a dedicated notification app.

[0318] Step 8:

[0319] The server records and stores all anomaly detection events, emotion data, and associated video data for later analysis and review.

[0320] Step 9:

[0321] The server periodically analyzes the stored data and retrains the generative model and emotion engine, improving the accuracy of the model and enabling more advanced anomaly detection and emotion recognition.

[0322] Example 2

[0323] 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."

[0324] Conventional surveillance camera systems are required to only detect abnormal behavior and respond quickly to it. However, they do not take into account navigation information or the user's emotional state, resulting in insufficient response when an abnormality occurs. Furthermore, delays in response after abnormal behavior detection and unnecessary reports due to false positives are also problems. Another issue is the inefficiency of the re-learning process to improve the accuracy of abnormal behavior detection.

[0325] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data, means for analyzing user emotions using an emotion engine when detecting the abnormal behavior, means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results, emotion data, and video data. This enables more accurate anomaly detection and rapid response that takes user emotions into consideration. Furthermore, by analyzing the saved data and retraining the generative model, the accuracy of the system can be continuously improved.

[0326] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[0327] "Real-time" refers to immediate data acquisition and processing with little to no delay.

[0328] "Video data" refers to data that includes visual information acquired from a surveillance camera.

[0329] "Preprocessing" refers to initial processing of video data, such as noise removal, resolution adjustment, and frame division.

[0330] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[0331] "Resolution adjustment" is a process for uniformly adjusting the resolution of video data.

[0332] A "generative model" is a model that uses AI technology to detect abnormal behavior based on past learning data.

[0333] "Abnormal behavior" refers to behavior that differs from normal conditions, such as falling, running away, or placing suspicious objects.

[0334] The "Emotion Engine" is a system that uses facial recognition technology to analyze emotions from a user's facial expressions.

[0335] "Notification and reporting" refers to the act of informing relevant organizations or specific users of the occurrence of an abnormality.

[0336] "Relevant agencies" refer to agencies that need to respond to the occurrence of abnormal behavior (e.g., police, security companies, etc.).

[0337] "User" refers to the party who receives notifications and alerts from the system.

[0338] "Emotion data" is data that indicates the user's emotional state analyzed by the emotion engine.

[0339] "Recording and storage" refers to the act of retaining the results of abnormal behavior detection and related data for later use.

[0340] "Relearning" is the process of analyzing past data to improve the accuracy of the generative model.

[0341] The present invention is a system that further enhances security and response to abnormalities by detecting abnormal behavior in real time in a surveillance camera system and combining it with an emotion engine that recognizes user emotions.

[0342] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[0343] 1. Surveillance cameras

[0344] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0345] 2. Server

[0346] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0347] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0348] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0349] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0350] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0351] Data storage module: Stores all anomaly detection events and associated video data.

[0352] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0353] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[0354] 3. Terminal

[0355] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0356] 4. Users

[0357] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0358] 5. Emotion Engine

[0359] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[0360] Specific examples

[0361] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[0362] 1. Video acquisition

[0363] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0364] 2. Preprocessing of video data

[0365] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0366] 3. Anomaly Detection and Sentiment Analysis

[0367] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[0368] 4. Notifications and Reports

[0369] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[0370] 5. History storage and analysis

[0371] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[0372] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[0373] Prompt Sentence Examples

[0374] For example, below is a specific example of a prompt for a generative AI model to detect suspicious behavior in a shopping mall:

[0375] Example prompt sentence:

[0376] "Abnormal behavior must be detected in the following video frames. Abnormal behavior may include falling, fleeing, or placing a suspicious object. Please analyze the following frame: [Frame Data]"

[0377] This prompt allows the generative model to operate properly.

[0378] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0379] Step 1: Acquire video data

[0380] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network, and video data is continuously transmitted to the server. The server uses a data acquisition module to receive this video data and process it in streaming format. The input is the video data from the surveillance cameras, and the output is the video data stored in the server's internal database waiting to be processed.

[0381] Step 2: Preprocessing the video data

[0382] The server performs preprocessing on the captured video data. Specifically, it performs noise reduction, resolution adjustment, and frame splitting. Noise reduction filters out unnecessary noise from the captured video data to improve image quality. Resolution adjustment converts the video data to a uniform resolution (e.g., 720p). Frame splitting divides the video data into frames and assigns a timestamp to each frame. The input is the captured raw data, and the output is processed, clear video frames.

[0383] Step 3: Detecting Abnormal Behavior

[0384] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The generative model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on the training data. When abnormal behavior is detected, an abnormality flag is set for the corresponding frame. In addition, detailed information about the abnormality (such as the time and location where it occurred) is generated. The input is the preprocessed video frame, and the output is the abnormal behavior identification result and the frame with the abnormality flag set.

[0385] Step 4: Sentiment Analysis

[0386] When detecting anomalous behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to read the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.). The analysis results are combined with anomalous behavior data and compiled as detailed emotional information. The input is the frame in which anomalous behavior was detected and the user's facial image, and the output is the analyzed emotional data.

[0387] Step 5: Notification and reporting when an abnormality is detected

[0388] After detecting abnormal behavior and analyzing emotions, the server immediately notifies and alerts relevant organizations and specific users. The notification includes the location and time of the abnormal behavior, as well as the analyzed emotional data. The server sends this information via SMS, email, or a dedicated notification app. The input is the anomaly detection result and emotional data, and the output is a notification or alert message.

[0389] Step 6: Store and analyze history

[0390] The server records and stores all anomaly detection events and associated emotion and video data. The stored data is later used to improve the accuracy of the generative model using a re-learning module. The stored data is stored in local or cloud storage and analyzed as needed. The inputs are anomaly detection results, emotion data, and video data, and the output is the stored historical data.

[0391] In this way, by clarifying the specific processing performed at each step and the flow of its inputs and outputs, the system can quickly and accurately detect and respond to abnormal behavior and user emotions.

[0392] (Application example 2)

[0393] 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."

[0394] Conventional surveillance camera systems are not only unable to detect abnormal behavior, but also have difficulty analyzing the causes of abnormal behavior and the user's underlying emotions in real time, resulting in inadequate responses. Furthermore, delays in notifications and alerts make it difficult to respond quickly. This has led to a demand for improved security and appropriate, prompt responses to abnormal situations.

[0395] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on learning data, means for analyzing the abnormal behavior and user emotions, means including a video output device for displaying the detected abnormality and emotion data, and means for recording and saving the abnormal behavior detection results and the video data. This makes it possible to analyze abnormal behavior and the associated user emotions in real time and respond quickly and appropriately.

[0396] A "surveillance camera" is a device that captures real-time video data within a specific area and transmits it to a server via a network.

[0397] "Real-time" refers to acquiring data the moment an event occurs and processing and analyzing it almost simultaneously.

[0398] "Video data" refers to digital data including video and image information captured by surveillance cameras.

[0399] "Preprocessing" refers to processing of acquired video data, such as noise removal, resolution adjustment, and frame division.

[0400] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[0401] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[0402] A "generative model" is a predictive model that uses past training data to identify abnormal behavior from new data.

[0403] "Abnormal behavior" refers to movements or actions that deviate from normal behavior, such as falling, running away, or suspicious behavior.

[0404] "Emotion analysis" is a process of identifying and analyzing emotions (surprise, fear, anger, etc.) from the user's facial expressions.

[0405] A "communication device" is a means for notifying or reporting to relevant organizations or specific users when abnormal behavior is detected.

[0406] "User" refers to a person (e.g., security guard, police officer, parent, etc.) who is responsible for receiving notifications and reports sent from the system and responding promptly.

[0407] The "video output device for display" is a device for displaying abnormal behavior and emotion data to the user, and includes smart glasses and dedicated monitors.

[0408] "Recording and storage" refers to the process of storing detected abnormal behavior and associated video and emotional data for future analysis and relearning.

[0409] The present invention is a system that detects abnormal behavior in real time in a surveillance camera system and combines it with an emotion engine that recognizes user emotions to further enhance security and response to abnormalities. The specific configuration and implementation method of this system are described below.

[0410] System configuration

[0411] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine. Each element is as follows:

[0412] 1. Surveillance cameras

[0413] Surveillance cameras capture real-time video data within a specific area. Each surveillance camera is connected to a server via an IP network and transmits the video data in streaming format.

[0414] 2. Server

[0415] The server aggregates and stores video data acquired from surveillance cameras and implements the following functions:

[0416] Acquisition and preprocessing of video data: Acquire real-time video data from surveillance cameras and perform noise removal and resolution adjustment.

[0417] Anomalous behavior detection: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0418] Sentiment analysis: In addition to detecting abnormal behavior, we also use an emotion engine to analyze user emotions.

[0419] Notification and reporting: Based on detected abnormal behavior and emotion data, relevant authorities and specific users are notified and reported.

[0420] Data storage: Record and store anomaly detection results and video data.

[0421] 3. Terminal

[0422] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected, including the time, location, and emotional data of the abnormality.

[0423] 4. Users

[0424] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). They are required to respond promptly based on the content of notifications and reports sent from the system.

[0425] 5. Emotion Engine

[0426] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the user's emotion data is correlated with abnormal behavior data for analysis.

[0427] Specific examples

[0428] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[0429] 1. Video acquisition

[0430] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0431] 2. Preprocessing of video data

[0432] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0433] 3. Anomaly Detection and Sentiment Analysis

[0434] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[0435] 4. Notifications and Reports

[0436] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[0437] 5. History storage and analysis

[0438] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[0439] Prompt Sentence Examples

[0440] Here is an example of input to a generative AI model using the following prompts:

[0441] Develop an application that analyzes real-time video footage from a surveillance camera system, detects abnormal behavior, and recognizes and analyzes user emotions. The application will be for smart glasses, and will immediately notify the smart glasses when abnormal behavior is detected and display the user's emotional data. The server will preprocess the video data and incorporate functions to analyze abnormal behavior and emotions using a generative model.

[0442] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0443] Step 1:

[0444] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network and continuously transmits video data. The input is real-time video data transmitted from the surveillance cameras, and the output is raw data stored in the server.

[0445] Step 2:

[0446] The server performs preprocessing on the acquired video data. This preprocessing includes noise removal, resolution adjustment, and frame division. The input is real-time video data, and the output is preprocessed data after data processing such as noise removal, resolution adjustment, and adding a timestamp to each frame.

[0447] Step 3:

[0448] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. Based on past learning data, the generative model identifies abnormal behavior (e.g., falls, fleeing, suspicious behavior) in real time. The input is the preprocessed data, and abnormal behavior is identified as data calculation. The output is the frame data in which abnormal behavior is detected and an abnormality flag.

[0449] Step 4:

[0450] The server detects abnormal behavior and simultaneously analyzes the user's emotions using an emotion engine. It uses facial recognition technology to identify emotions (e.g., surprise, fear, anger, etc.) from facial expressions. The input is video data in which abnormal behavior was detected, and emotion identification is performed as part of the data analysis. The output is the user's emotional data.

[0451] Step 5:

[0452] The server immediately notifies and reports to relevant organizations and specific users based on abnormal behavior and emotion data. The notification content includes the time, location, and emotion data of the abnormality. The input is the abnormal behavior detection result and emotion data, and the notification message is generated and sent as the data output.

[0453] Step 6:

[0454] The server records and stores the detected abnormal behavior and emotion data. All anomaly detection events and associated video data are used for future analysis and retraining of the generative model. The inputs are the anomaly detection results and emotion data, which are recorded in the database as data storage.

[0455] 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.

[0456] 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.

[0457] 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.

[0458] [Second embodiment]

[0459] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0460] 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.

[0461] 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).

[0462] 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.

[0463] 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.

[0464] 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).

[0465] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0466] 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.

[0467] 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.

[0468] 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.

[0469] 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.

[0470] 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."

[0471] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[0472] System configuration

[0473] This system consists of a surveillance camera, a server, a terminal, and a user.

[0474] 1. Surveillance cameras

[0475] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0476] 2. Server

[0477] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0478] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0479] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0480] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0481] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0482] Data storage module: Stores all anomaly detection events and associated video data.

[0483] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0484] 3. Terminal

[0485] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0486] 4. Users

[0487] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0488] Explanation of program processing

[0489] Below, the operation of the system program is explained in natural language.

[0490] 1. Obtaining surveillance camera footage

[0491] The server acquires video data in real time from multiple surveillance cameras, which are connected to the server via a network and continuously transmit the video data to the server.

[0492] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[0493] 2. Preprocessing of video data

[0494] The server performs preprocessing on the acquired video data.

[0495] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[0496] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[0497] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[0498] 3. Anomaly Detection Using Generative Models

[0499] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[0500] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[0501] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[0502] 4. Notification and reporting when an abnormality is detected

[0503] When the server detects an abnormality, it immediately notifies and reports the relevant organizations and specific users.

[0504] The notification will include the time and location of the anomaly and any necessary action.

[0505] Reporting can be done via SMS, email, or a dedicated notification app.

[0506] 5. History storage and analysis

[0507] The server records and stores all anomaly detection events and associated video data.

[0508] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[0509] Specific examples

[0510] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[0511] 1. Video acquisition

[0512] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0513] 2. Preprocessing of video data

[0514] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0515] 3. Anomaly Detection

[0516] The server inputs the preprocessed video data into a generative model to detect falling behavior.

[0517] 4. Notifications and Reports

[0518] When a fall is detected, the server immediately notifies mall security staff, including the location and time of the fall, as well as a video snapshot of the frame in question.

[0519] 5. History storage and analysis

[0520] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generative model.

[0521] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[0522] The processing flow will be explained below.

[0523] Step 1:

[0524] The server acquires video data in real time from each surveillance camera. The surveillance cameras are connected to the server via an IP network and transmit the video data continuously in streaming format.

[0525] Step 2:

[0526] The server preprocesses the captured video data by performing noise reduction, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting the video data into a format suitable for analysis.

[0527] Step 3:

[0528] The server divides the pre-processed video data into frames, each of which is assigned a timestamp to make it uniquely identifiable.

[0529] Step 4:

[0530] The server inputs each preprocessed frame of data into a generative model, which then analyzes each frame based on pre-trained patterns of abnormal behavior.

[0531] Step 5:

[0532] The server receives the analysis results from the generative model and detects anomalous behavior. If anomalous behavior is detected, an anomalous flag is set for the corresponding frame. Abnormal behavior includes falling, fleeing, placing suspicious objects, etc.

[0533] Step 6:

[0534] The server will send a notification when an abnormality is detected. Specifically, it will send a notification containing detailed information about the abnormality to relevant organizations or specific users. Notification methods include SMS, email, or a dedicated notification app.

[0535] Step 7:

[0536] The server records and stores all anomaly detection events and associated video data for later analysis and review.

[0537] Step 8:

[0538] The server periodically analyzes the stored data and retrains the generative model, improving the model's accuracy and enabling more effective anomalous behavior detection.

[0539] Example 1

[0540] 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."

[0541] Conventional surveillance camera systems have the problem of being difficult to detect abnormal behavior in real time and quickly notify and report it. They also lack a mechanism for saving detection results and using them for re-learning to improve the accuracy of the system. This can result in insufficient accuracy in detecting abnormal behavior and rapid response.

[0542] 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.

[0543] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise, adjust resolution, and split frames, means for inputting the preprocessed video data into a generative AI model and detecting abnormal behavior based on the learning data, means for immediately notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results and video data, and for later analysis and re-training of the generative AI model. This enables highly accurate detection of abnormal behavior in real time and rapid response.

[0544] A "surveillance camera" is a device that captures real-time video data within a designated area and transmits the data over a network.

[0545] "Real-time video data" refers to video data that is continuously captured by a surveillance camera and instantly transmitted to a server.

[0546] "Preprocessing" refers to processing that removes noise, adjusts resolution, and divides frames from video data.

[0547] "Noise removal" is a process that removes unnecessary noise components from video data to improve the accuracy of analysis.

[0548] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[0549] "Frame division" is a process of breaking down video data into frames and assigning a timestamp to each frame.

[0550] A "generative AI model" is an artificial intelligence model for detecting abnormal behavior in video data based on past learning data.

[0551] "Abnormal behavior" refers to actions that would not normally occur, such as falling, running away, or suspicious behavior.

[0552] "Notification and reporting" refers to the act of sending information about abnormal behavior to relevant organizations or specific users when it is detected.

[0553] "Recording and storage" refers to storing abnormal behavior detection results and related video data in a database.

[0554] "Retraining" is the process of data analysis and retraining that utilizes stored data to improve the accuracy of generative AI models.

[0555] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[0556] The system consists of a surveillance camera, a server, a terminal, and a user.

[0557] Hardware and Software Details

[0558] surveillance cameras

[0559] A surveillance camera is a device that captures real-time video data within a specified area and transmits it to a server via a network. The video data is transmitted to the server in streaming format via an IP network. The camera used is, for example, a high-resolution IP camera (e.g., 1080p HD camera).

[0560] server

[0561] The server is a computer system that aggregates and stores video data acquired from surveillance cameras and performs various processing. The server has the following functions:

[0562] Data Acquisition Module: Acquires real-time video data from each surveillance camera using standard protocols (e.g., RTSP).

[0563] Pre-processing module: Performs noise reduction, resolution adjustment, and frame division on the acquired video data. Here, an image processing library (e.g., OpenCV) is used.

[0564] Noise Reduction: Apply a Gaussian filter to remove noise.

[0565] Resolution adjustment: The resize function converts video data to a uniform resolution (e.g., 720p).

[0566] Frame division: Obtains FPS information, divides the video data into frames, and adds timestamps.

[0567] Anomaly Detection Module: Preprocessed video data is fed into a generative AI model to detect anomalous behavior. Deep learning frameworks (e.g., TensorFlow, PyTorch) are used.

[0568] Send the following prompt to the generative model:

[0569] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[0570] Notification and reporting module: When an abnormality is detected, it immediately notifies and reports to relevant organizations or specific users via SMS, email, or a dedicated notification app.

[0571] For example, an email could be sent stating, "A fall occurred in the North Area on October 5, 2023 at 2:30 p.m. Action is required."

[0572] Data storage module: Records anomaly detection events and related video data and stores them in a database.

[0573] Retraining module: Retrains the generative AI model using the stored data. Periodically extracts historical data from the database to create a retraining dataset.

[0574] Terminal

[0575] The devices are used by surveillance camera system administrators and security staff to receive notifications and alerts when an abnormality is detected. A dedicated app is installed, allowing for real-time situation monitoring and response.

[0576] User

[0577] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users respond promptly based on the content of notifications and reports sent from the system.

[0578] Specific examples

[0579] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[0580] 1. Acquiring video data

[0581] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0582] 2. Pretreatment

[0583] The server performs noise reduction and resolution adjustment on the video data and adds a timestamp to the frame.

[0584] 3. Detecting Abnormal Behavior

[0585] The server inputs the preprocessed video data into a generative AI model to detect falling behavior.

[0586] 4. Notifications and Reports

[0587] When a fall is detected, the server immediately notifies the mall's security staff.

[0588] 5. History storage and analysis

[0589] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generated AI model.

[0590] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[0591] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0592] Step 1: Acquire video data

[0593] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera sends video data to the server in streaming format via an IP network. The input is real-time video data from the surveillance cameras, and the output is the acquired raw video data. The server receives the video data using a standard protocol (e.g., RTSP). Specifically, the server periodically checks the connection status of each camera and attempts to reconnect if there is an abnormality.

[0594] Step 2: Preprocessing

[0595] The server performs pre-processing on the captured video data, including noise reduction, resolution adjustment, and frame splitting. The input is the captured raw video data, and the output is the pre-processed video data.

[0596] Noise removal: The server uses an image processing library (e.g., OpenCV) to apply a Gaussian filter to remove noise from the video data.

[0597] Resolution Adjustment: The server uses a resize function to convert all video data to a uniform resolution (e.g., 720p).

[0598] Frame division: The server acquires FPS (frames per second) information and divides the video data into frames. At the same time, it assigns a timestamp to each frame. Specifically, the server analyzes the video data and extracts frames at the appropriate time.

[0599] Step 3: Detecting Abnormal Behavior

[0600] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The input is the preprocessed video data, and the output is the abnormal behavior detection results.

[0601] Generative AI models are built using deep learning frameworks (e.g., TensorFlow, PyTorch).

[0602] The server converts the video data into the model's input format, sets the following prompt text, and sends it to the model.

[0603] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[0604] The server receives the model output, and if an anomaly is detected, it sets an anomaly flag for the corresponding frame. Specifically, the server inputs the data converted into tensor format into the model, processes the results, and sets an anomaly flag.

[0605] Step 4: Notification

[0606] When the server detects abnormal behavior, it immediately notifies and reports to relevant organizations or specific users. The input is the abnormal behavior detection result, and the output is a notification / report message.

[0607] The notification will include the time and location of the anomaly and any necessary action.

[0608] Reporting methods include SMS, email, and a dedicated notification app.

[0609] For example, an email could be sent stating, "A fall occurred in the North Area at 2:30 PM on October 5, 2023. Action is required." Specifically, the server generates a notification message and sends it to the specified contacts. It also uses the API of a dedicated notification app to send notifications in real time.

[0610] Step 5: Store and analyze history

[0611] The server records and stores all anomaly detection events and associated video data. The input is the abnormal behavior detection results and associated video data, and the output is the stored historical data.

[0612] The stored data will be analyzed at a later date and used to improve the system and retrain the generative AI model. Specifically, the server stores the data in a database, periodically extracts historical data from the database, and creates a dataset for retraining.

[0613] This enables the entire system to detect abnormal behavior in real time with high accuracy and respond quickly.

[0614] (Application example 1)

[0615] 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."

[0616] Conventional surveillance camera systems have issues with delays in detecting and notifying abnormal behavior, making it difficult to respond quickly. Furthermore, because notifications of abnormal situations are limited to specific devices, information sharing is insufficient and there is a lack of means to quickly contact multiple parties involved. As a result, security staff and guards are often unable to respond in a timely manner.

[0617] 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.

[0618] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the learning data, means for notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, means for recording and saving the abnormal behavior detection results and the video data, means for sending the notification as a push notification to a mobile device of the relevant organization or specific user, and means for generating an audio alarm when abnormal behavior is detected, thereby enabling immediate notification to multiple relevant parties and appropriate and prompt response.

[0619] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[0620] "Real-time" refers to the immediate acquisition and processing of data without delay.

[0621] "Video data" is information in the form of video captured by a surveillance camera.

[0622] "Noise reduction" refers to the process of removing unnecessary signals and noise from video data.

[0623] "Resolution" is an index that indicates the level of detail and image quality of video data.

[0624] A "generative model" is an AI model used to detect anomalous behavior based on training data.

[0625] "Training data" is the historical dataset used to train the generative model.

[0626] "Abnormal behavior" refers to actions or events that are different from the norm, such as falling, running away, or suspicious behavior.

[0627] "Notification" refers to the act of notifying relevant organizations or specific users when abnormal behavior is detected.

[0628] "Reporting" is a method of alerting or reporting to relevant authorities when abnormal behavior is detected.

[0629] "Related agencies" are agencies or organizations involved in responding to abnormal behavior, such as security staff, guards, and police.

[0630] A "specific user" is an individual or organization that is designated in advance by the system to receive notifications.

[0631] "Recording" refers to the act of saving detected abnormal behavior and related data.

[0632] "Storage" refers to keeping recorded data for future reference.

[0633] A "mobile terminal" is a mobile communication device such as a smartphone or tablet.

[0634] "Push notification" is a function that delivers notifications from the system to mobile devices in real time.

[0635] An "audio alarm" is a means of alerting the user by sound when abnormal behavior is detected.

[0636] This invention provides a system that acquires video data from a surveillance camera in real time and detects, notifies, and reports abnormal behavior. This system can quickly and efficiently detect abnormal behavior and report it to relevant authorities and designated users in order to increase the safety of the monitored area. A specific embodiment of the invention will be described below.

[0637] System configuration

[0638] surveillance cameras

[0639] Surveillance cameras capture real-time video data within a specific area and are connected to a central server via a network, where the captured video data is sent in streaming format.

[0640] server

[0641] The server has the following functions:

[0642] Data acquisition module: Acquires video data from surveillance cameras in real time.

[0643] Pre-processing module: Performs noise reduction and resolution adjustment on the acquired video data.

[0644] Anomaly detection module: Preprocessed video data is input into the generative model, and anomalous behavior is detected based on the training data.

[0645] Notification and reporting module: When abnormal behavior is detected, the module notifies and reports relevant organizations or specific users. Notifications are sent as push notifications to the mobile devices of relevant organizations or users.

[0646] Audio alarm module: Generates an audio alarm when abnormal behavior is detected.

[0647] Data storage module: Records and stores abnormal behavior detection results and related video data.

[0648] Terminal

[0649] The devices are used by security staff and administrators, and have an app installed that receives push notifications when an abnormality is detected. If abnormal behavior is detected, a push notification is immediately sent to the device, prompting appropriate action to be taken.

[0650] User

[0651] Users include relevant organizations and security staff, guards, police officers, etc. who are required to receive specific notifications. Users are required to respond promptly to the situation based on the push notification.

[0652] Hardware and software used

[0653] Hardware: surveillance cameras, servers, smartphones (mobile devices)

[0654] Software: Flask (server-side framework), OpenCV (image processing library), TensorFlow (generative model), Firebase Cloud Messaging (push notifications)

[0655] How we process your data

[0656] The server receives streaming video data from the surveillance cameras, and the preprocessing module performs noise reduction and resolution adjustment. The preprocessed video data is then input into the generative model to detect anomalous behavior. Detected anomalous behavior is immediately recognized by the anomaly detection module.

[0657] When an abnormality is detected, the notification module is activated and sends a push notification to relevant organizations or specific users. At the same time, the audio alarm module is activated and generates an audio alarm to prompt a prompt response. In addition, the data storage module records and stores the abnormal behavior detection results and related video data.

[0658] For example, if someone falls in a shopping mall, the video from the surveillance camera is sent to the server in real time, and the fall is detected using a TensorFlow model. A push notification is then sent to the security guard's mobile device via Firebase Cloud Messaging, and an audio alarm sounds, enabling a prompt response.

[0659] Prompt Sentence Examples

[0660] "Model input is generated from preprocessed video data and fed into the generative model."

[0661] "Detects abnormal behavior based on the prediction results and notifies relevant authorities."

[0662] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0663] Step 1:

[0664] The server acquires video data from the surveillance cameras in real time. The surveillance cameras capture a specific area and continuously transmit the video data to the server via the network. The server receives the data and prepares it for the next step of processing. The input of this step is the raw data sent from the surveillance cameras, and the output is the video data for preprocessing.

[0665] Step 2:

[0666] The server performs preprocessing on the received video data. Specifically, it uses OpenCV to remove noise and adjust the resolution. This improves the quality of the video and increases the accuracy of abnormal behavior detection. The input of this step is raw data, and the output is video data with noise removed and resolution adjusted.

[0667] Step 3:

[0668] The server inputs the preprocessed video data into the generative model. Using TensorFlow, the data is analyzed frame by frame and anomalous behavior is detected based on the training data. The input to this step is the preprocessed video data, and the output is the abnormal behavior detection results. Specifically, the generative model identifies abnormal behavior.

[0669] Step 4:

[0670] The server issues notifications and reports when abnormal behavior is detected. It uses Firebase Cloud Messaging to send push notifications to the mobile devices of relevant organizations or specific users. This notification includes the time, location, and video snapshot of the abnormal behavior. The input to this step is the abnormal behavior detection result, and the output is the sent notification and report. Specifically, the notification and report module is activated and a push notification is sent.

[0671] Step 5:

[0672] The server generates an audio alarm when abnormal behavior is detected. The audio alarm module plays an alarm sound immediately when the abnormal behavior detection result is input. The input of this step is the abnormal behavior detection result, and the output is the generated alarm sound. Specifically, the audio alarm is activated to warn the user.

[0673] Step 6:

[0674] The server records and stores the abnormal behavior detection results and related video data. The data storage module stores the data in a database for future analysis and system improvement. The input of this step is the abnormal behavior detection results and related video data, and the output is the stored data. The specific operation is writing to the database.

[0675] Step 7:

[0676] The user receives a push notification sent to their mobile device and is prompted to take action. In this step, the user who received the push notification quickly heads to the scene and takes appropriate action. The input to this step is the push notification sent from the server, and the output is the user's on-site action. Specifically, the user checks the notification and takes the necessary action.

[0677] As described above, the present invention is a system that can detect abnormal behavior in real time and quickly notify and report the situation to relevant organizations and users, thereby significantly improving the safety of the monitored area.

[0678] 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.

[0679] The present invention provides a system that detects abnormal behavior in real time in a surveillance camera system and further enhances security and response to abnormalities by combining it with an emotion engine that recognizes user emotions. Specific embodiments of the present invention will be described below.

[0680] System configuration

[0681] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[0682] 1. Surveillance cameras

[0683] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0684] 2. Server

[0685] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0686] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0687] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0688] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0689] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0690] Data storage module: Stores all anomaly detection events and associated video data.

[0691] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0692] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[0693] 3. Terminal

[0694] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0695] 4. Users

[0696] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0697] 5. Emotion Engine

[0698] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[0699] Explanation of program processing

[0700] Below, the operation of the system program is explained in natural language.

[0701] 1. Obtaining surveillance camera footage

[0702] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera is connected to the server via a network, and the video data is continuously transmitted to the server.

[0703] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[0704] 2. Preprocessing of video data

[0705] The server performs preprocessing on the acquired video data.

[0706] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[0707] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[0708] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[0709] 3. Anomaly Detection Using Generative Models

[0710] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[0711] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[0712] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[0713] 4. Emotion analysis

[0714] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine.

[0715] The emotion engine uses facial recognition technology to identify emotions (e.g., surprise, anger, fear, etc.) from the user's facial expressions.

[0716] 5. Notification and reporting when an abnormality is detected

[0717] The server detects an abnormality and analyzes the user's emotions, then immediately notifies and reports the relevant authorities and specific users.

[0718] The notification will include detailed information about the time and location of the anomaly, as well as emotional data.

[0719] Reporting can be done via SMS, email, or a dedicated notification app.

[0720] 6. History storage and analysis

[0721] The server records and stores all anomaly detection events and associated emotion and video data.

[0722] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[0723] Specific examples

[0724] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[0725] 1. Video acquisition

[0726] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0727] 2. Preprocessing of video data

[0728] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0729] 3. Anomaly Detection and Sentiment Analysis

[0730] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[0731] 4. Notifications and Reports

[0732] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[0733] 5. History storage and analysis

[0734] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[0735] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[0736] The processing flow will be explained below.

[0737] Step 1:

[0738] The server acquires video data in real time from each surveillance camera. Each surveillance camera is connected to the server via a network, and the video data is continuously streamed to the server.

[0739] Step 2:

[0740] The server preprocesses the acquired video data, specifically by denoising the video data, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting it into a format that is easier to analyze.

[0741] Step 3:

[0742] The server divides the pre-processed video data into frames, and assigns a timestamp to each frame to make them individually identifiable.

[0743] Step 4:

[0744] The server inputs each preprocessed frame of data into a generative model, which analyzes each frame based on pre-trained abnormal behavior patterns.

[0745] Step 5:

[0746] The server receives the analysis results of the generative model and detects abnormal behavior (e.g., falls, escape, placing suspicious objects, etc.). For detected abnormal behavior, it sets an abnormal flag for the corresponding frame.

[0747] Step 6:

[0748] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to analyze the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.).

[0749] Step 7:

[0750] The server then notifies relevant organizations and specific users based on the results of the user's emotion analysis and abnormal behavior detection. The notification includes detailed information about the abnormal behavior and emotions. Notifications can be sent via SMS, email, or a dedicated notification app.

[0751] Step 8:

[0752] The server records and stores all anomaly detection events, emotion data, and associated video data for later analysis and review.

[0753] Step 9:

[0754] The server periodically analyzes the stored data and retrains the generative model and emotion engine, improving the accuracy of the model and enabling more advanced anomaly detection and emotion recognition.

[0755] Example 2

[0756] 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."

[0757] Conventional surveillance camera systems are required to only detect abnormal behavior and respond quickly to it. However, they do not take into account navigation information or the user's emotional state, resulting in insufficient response when an abnormality occurs. Furthermore, delays in response after abnormal behavior detection and unnecessary reports due to false positives are also problems. Another issue is the inefficiency of the re-learning process to improve the accuracy of abnormal behavior detection.

[0758] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data, means for analyzing user emotions using an emotion engine when detecting the abnormal behavior, means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results, emotion data, and video data. This enables more accurate anomaly detection and rapid response that takes user emotions into consideration. Furthermore, by analyzing the saved data and retraining the generative model, the accuracy of the system can be continuously improved.

[0759] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[0760] "Real-time" refers to immediate data acquisition and processing with little to no delay.

[0761] "Video data" refers to data that includes visual information acquired from a surveillance camera.

[0762] "Preprocessing" refers to initial processing of video data, such as noise removal, resolution adjustment, and frame division.

[0763] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[0764] "Resolution adjustment" is a process for uniformly adjusting the resolution of video data.

[0765] A "generative model" is a model that uses AI technology to detect abnormal behavior based on past learning data.

[0766] "Abnormal behavior" refers to behavior that differs from normal conditions, such as falling, running away, or placing suspicious objects.

[0767] The "Emotion Engine" is a system that uses facial recognition technology to analyze emotions from a user's facial expressions.

[0768] "Notification and reporting" refers to the act of informing relevant organizations or specific users of the occurrence of an abnormality.

[0769] "Relevant agencies" refer to agencies that need to respond to the occurrence of abnormal behavior (e.g., police, security companies, etc.).

[0770] "User" refers to the party who receives notifications and alerts from the system.

[0771] "Emotion data" is data that indicates the user's emotional state analyzed by the emotion engine.

[0772] "Recording and storage" refers to the act of retaining the results of abnormal behavior detection and related data for later use.

[0773] "Relearning" is the process of analyzing past data to improve the accuracy of the generative model.

[0774] The present invention is a system that further enhances security and response to abnormalities by detecting abnormal behavior in real time in a surveillance camera system and combining it with an emotion engine that recognizes user emotions.

[0775] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[0776] 1. Surveillance cameras

[0777] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0778] 2. Server

[0779] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0780] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0781] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0782] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0783] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0784] Data storage module: Stores all anomaly detection events and associated video data.

[0785] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0786] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[0787] 3. Terminal

[0788] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0789] 4. Users

[0790] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0791] 5. Emotion Engine

[0792] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[0793] Specific examples

[0794] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[0795] 1. Video acquisition

[0796] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0797] 2. Preprocessing of video data

[0798] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0799] 3. Anomaly Detection and Sentiment Analysis

[0800] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[0801] 4. Notifications and Reports

[0802] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[0803] 5. History storage and analysis

[0804] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[0805] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[0806] Prompt Sentence Examples

[0807] For example, below is a specific example of a prompt for a generative AI model to detect suspicious behavior in a shopping mall:

[0808] Example prompt sentence:

[0809] "Abnormal behavior must be detected in the following video frames. Abnormal behavior may include falling, fleeing, or placing a suspicious object. Please analyze the following frame: [Frame Data]"

[0810] This prompt allows the generative model to operate properly.

[0811] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0812] Step 1: Acquire video data

[0813] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network, and video data is continuously transmitted to the server. The server uses a data acquisition module to receive this video data and process it in streaming format. The input is the video data from the surveillance cameras, and the output is the video data stored in the server's internal database waiting to be processed.

[0814] Step 2: Preprocessing the video data

[0815] The server performs preprocessing on the captured video data. Specifically, it performs noise reduction, resolution adjustment, and frame splitting. Noise reduction filters out unnecessary noise from the captured video data to improve image quality. Resolution adjustment converts the video data to a uniform resolution (e.g., 720p). Frame splitting divides the video data into frames and assigns a timestamp to each frame. The input is the captured raw data, and the output is processed, clear video frames.

[0816] Step 3: Detecting Abnormal Behavior

[0817] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The generative model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on the training data. When abnormal behavior is detected, an abnormality flag is set for the corresponding frame. In addition, detailed information about the abnormality (such as the time and location where it occurred) is generated. The input is the preprocessed video frame, and the output is the abnormal behavior identification result and the frame with the abnormality flag set.

[0818] Step 4: Sentiment Analysis

[0819] When detecting anomalous behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to read the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.). The analysis results are combined with anomalous behavior data and compiled as detailed emotional information. The input is the frame in which anomalous behavior was detected and the user's facial image, and the output is the analyzed emotional data.

[0820] Step 5: Notification and reporting when an abnormality is detected

[0821] After detecting abnormal behavior and analyzing emotions, the server immediately notifies and alerts relevant organizations and specific users. The notification includes the location and time of the abnormal behavior, as well as the analyzed emotional data. The server sends this information via SMS, email, or a dedicated notification app. The input is the anomaly detection result and emotional data, and the output is a notification or alert message.

[0822] Step 6: Store and analyze history

[0823] The server records and stores all anomaly detection events and associated emotion and video data. The stored data is later used to improve the accuracy of the generative model using a re-learning module. The stored data is stored in local or cloud storage and analyzed as needed. The inputs are anomaly detection results, emotion data, and video data, and the output is the stored historical data.

[0824] In this way, by clarifying the specific processing performed at each step and the flow of its inputs and outputs, the system can quickly and accurately detect and respond to abnormal behavior and user emotions.

[0825] (Application example 2)

[0826] 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."

[0827] Conventional surveillance camera systems are not only unable to detect abnormal behavior, but also have difficulty analyzing the causes of abnormal behavior and the user's underlying emotions in real time, resulting in inadequate responses. Furthermore, delays in notifications and alerts make it difficult to respond quickly. This has led to a demand for improved security and appropriate, prompt responses to abnormal situations.

[0828] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on learning data, means for analyzing the abnormal behavior and user emotions, means including a video output device for displaying the detected abnormality and emotion data, and means for recording and saving the abnormal behavior detection results and the video data. This makes it possible to analyze abnormal behavior and the associated user emotions in real time and respond quickly and appropriately.

[0829] A "surveillance camera" is a device that captures real-time video data within a specific area and transmits it to a server via a network.

[0830] "Real-time" refers to acquiring data the moment an event occurs and processing and analyzing it almost simultaneously.

[0831] "Video data" refers to digital data including video and image information captured by surveillance cameras.

[0832] "Preprocessing" refers to processing of acquired video data, such as noise removal, resolution adjustment, and frame division.

[0833] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[0834] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[0835] A "generative model" is a predictive model that uses past training data to identify abnormal behavior from new data.

[0836] "Abnormal behavior" refers to movements or actions that deviate from normal behavior, such as falling, running away, or suspicious behavior.

[0837] "Emotion analysis" is a process of identifying and analyzing emotions (surprise, fear, anger, etc.) from the user's facial expressions.

[0838] A "communication device" is a means for notifying or reporting to relevant organizations or specific users when abnormal behavior is detected.

[0839] "User" refers to a person (e.g., security guard, police officer, parent, etc.) who is responsible for receiving notifications and reports sent from the system and responding promptly.

[0840] The "video output device for display" is a device for displaying abnormal behavior and emotion data to the user, and includes smart glasses and dedicated monitors.

[0841] "Recording and storage" refers to the process of storing detected abnormal behavior and associated video and emotional data for future analysis and relearning.

[0842] The present invention is a system that detects abnormal behavior in real time in a surveillance camera system and combines it with an emotion engine that recognizes user emotions to further enhance security and response to abnormalities. The specific configuration and implementation method of this system are described below.

[0843] System configuration

[0844] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine. Each element is as follows:

[0845] 1. Surveillance cameras

[0846] Surveillance cameras capture real-time video data within a specific area. Each surveillance camera is connected to a server via an IP network and transmits the video data in streaming format.

[0847] 2. Server

[0848] The server aggregates and stores video data acquired from surveillance cameras and implements the following functions:

[0849] Acquisition and preprocessing of video data: Acquire real-time video data from surveillance cameras and perform noise removal and resolution adjustment.

[0850] Anomalous behavior detection: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0851] Sentiment analysis: In addition to detecting abnormal behavior, we also use an emotion engine to analyze user emotions.

[0852] Notification and reporting: Based on detected abnormal behavior and emotion data, relevant authorities and specific users are notified and reported.

[0853] Data storage: Record and store anomaly detection results and video data.

[0854] 3. Terminal

[0855] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected, including the time, location, and emotional data of the abnormality.

[0856] 4. Users

[0857] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). They are required to respond promptly based on the content of notifications and reports sent from the system.

[0858] 5. Emotion Engine

[0859] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the user's emotion data is correlated with abnormal behavior data for analysis.

[0860] Specific examples

[0861] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[0862] 1. Video acquisition

[0863] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0864] 2. Preprocessing of video data

[0865] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0866] 3. Anomaly Detection and Sentiment Analysis

[0867] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[0868] 4. Notifications and Reports

[0869] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[0870] 5. History storage and analysis

[0871] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[0872] Prompt Sentence Examples

[0873] Here is an example of input to a generative AI model using the following prompts:

[0874] Develop an application that analyzes real-time video footage from a surveillance camera system, detects abnormal behavior, and recognizes and analyzes user emotions. The application will be for smart glasses, and will immediately notify the smart glasses when abnormal behavior is detected and display the user's emotional data. The server will preprocess the video data and incorporate functions to analyze abnormal behavior and emotions using a generative model.

[0875] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0876] Step 1:

[0877] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network and continuously transmits video data. The input is real-time video data transmitted from the surveillance cameras, and the output is raw data stored in the server.

[0878] Step 2:

[0879] The server performs preprocessing on the acquired video data. This preprocessing includes noise removal, resolution adjustment, and frame division. The input is real-time video data, and the output is preprocessed data after data processing such as noise removal, resolution adjustment, and adding a timestamp to each frame.

[0880] Step 3:

[0881] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. Based on past learning data, the generative model identifies abnormal behavior (e.g., falls, fleeing, suspicious behavior) in real time. The input is the preprocessed data, and abnormal behavior is identified as data calculation. The output is the frame data in which abnormal behavior is detected and an abnormality flag.

[0882] Step 4:

[0883] The server detects abnormal behavior and simultaneously analyzes the user's emotions using an emotion engine. It uses facial recognition technology to identify emotions (e.g., surprise, fear, anger, etc.) from facial expressions. The input is video data in which abnormal behavior was detected, and emotion identification is performed as part of the data analysis. The output is the user's emotional data.

[0884] Step 5:

[0885] The server immediately notifies and reports to relevant organizations and specific users based on abnormal behavior and emotion data. The notification content includes the time, location, and emotion data of the abnormality. The input is the abnormal behavior detection result and emotion data, and the notification message is generated and sent as the data output.

[0886] Step 6:

[0887] The server records and stores the detected abnormal behavior and emotion data. All anomaly detection events and associated video data are used for future analysis and retraining of the generative model. The inputs are the anomaly detection results and emotion data, which are recorded in the database as data storage.

[0888] 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.

[0889] 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.

[0890] 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.

[0891] [Third embodiment]

[0892] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0893] 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.

[0894] 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).

[0895] 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.

[0896] 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.

[0897] 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).

[0898] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0899] 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.

[0900] 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.

[0901] 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.

[0902] 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.

[0903] 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."

[0904] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[0905] System configuration

[0906] This system consists of a surveillance camera, a server, a terminal, and a user.

[0907] 1. Surveillance cameras

[0908] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[0909] 2. Server

[0910] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[0911] Data acquisition module: Acquires real-time video data from each surveillance camera.

[0912] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[0913] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[0914] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[0915] Data storage module: Stores all anomaly detection events and associated video data.

[0916] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[0917] 3. Terminal

[0918] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[0919] 4. Users

[0920] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[0921] Explanation of program processing

[0922] Below, the operation of the system program is explained in natural language.

[0923] 1. Obtaining surveillance camera footage

[0924] The server acquires video data in real time from multiple surveillance cameras, which are connected to the server via a network and continuously transmit the video data to the server.

[0925] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[0926] 2. Preprocessing of video data

[0927] The server performs preprocessing on the acquired video data.

[0928] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[0929] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[0930] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[0931] 3. Anomaly Detection Using Generative Models

[0932] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[0933] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[0934] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[0935] 4. Notification and reporting when an abnormality is detected

[0936] When the server detects an abnormality, it immediately notifies and reports the relevant organizations and specific users.

[0937] The notification will include the time and location of the anomaly and any necessary action.

[0938] Reporting can be done via SMS, email, or a dedicated notification app.

[0939] 5. History storage and analysis

[0940] The server records and stores all anomaly detection events and associated video data.

[0941] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[0942] Specific examples

[0943] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[0944] 1. Video acquisition

[0945] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[0946] 2. Preprocessing of video data

[0947] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[0948] 3. Anomaly Detection

[0949] The server inputs the preprocessed video data into a generative model to detect falling behavior.

[0950] 4. Notifications and Reports

[0951] When a fall is detected, the server immediately notifies mall security staff, including the location and time of the fall, as well as a video snapshot of the frame in question.

[0952] 5. History storage and analysis

[0953] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generative model.

[0954] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[0955] The processing flow will be explained below.

[0956] Step 1:

[0957] The server acquires video data in real time from each surveillance camera. The surveillance cameras are connected to the server via an IP network and transmit the video data continuously in streaming format.

[0958] Step 2:

[0959] The server preprocesses the captured video data by performing noise reduction, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting the video data into a format suitable for analysis.

[0960] Step 3:

[0961] The server divides the pre-processed video data into frames, each of which is assigned a timestamp to make it uniquely identifiable.

[0962] Step 4:

[0963] The server inputs each preprocessed frame of data into a generative model, which then analyzes each frame based on pre-trained patterns of abnormal behavior.

[0964] Step 5:

[0965] The server receives the analysis results from the generative model and detects anomalous behavior. If anomalous behavior is detected, an anomalous flag is set for the corresponding frame. Abnormal behavior includes falling, fleeing, placing suspicious objects, etc.

[0966] Step 6:

[0967] The server will send a notification when an abnormality is detected. Specifically, it will send a notification containing detailed information about the abnormality to relevant organizations or specific users. Notification methods include SMS, email, or a dedicated notification app.

[0968] Step 7:

[0969] The server records and stores all anomaly detection events and associated video data for later analysis and review.

[0970] Step 8:

[0971] The server periodically analyzes the stored data and retrains the generative model, improving the model's accuracy and enabling more effective anomalous behavior detection.

[0972] Example 1

[0973] 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."

[0974] Conventional surveillance camera systems have the problem of being difficult to detect abnormal behavior in real time and quickly notify and report it. They also lack a mechanism for saving detection results and using them for re-learning to improve the accuracy of the system. This can result in insufficient accuracy in detecting abnormal behavior and rapid response.

[0975] 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.

[0976] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise, adjust resolution, and split frames, means for inputting the preprocessed video data into a generative AI model and detecting abnormal behavior based on the learning data, means for immediately notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results and video data, and for later analysis and re-training of the generative AI model. This enables highly accurate detection of abnormal behavior in real time and rapid response.

[0977] A "surveillance camera" is a device that captures real-time video data within a designated area and transmits the data over a network.

[0978] "Real-time video data" refers to video data that is continuously captured by a surveillance camera and instantly transmitted to a server.

[0979] "Preprocessing" refers to processing that removes noise, adjusts resolution, and divides frames from video data.

[0980] "Noise removal" is a process that removes unnecessary noise components from video data to improve the accuracy of analysis.

[0981] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[0982] "Frame division" is a process of breaking down video data into frames and assigning a timestamp to each frame.

[0983] A "generative AI model" is an artificial intelligence model for detecting abnormal behavior in video data based on past learning data.

[0984] "Abnormal behavior" refers to actions that would not normally occur, such as falling, running away, or suspicious behavior.

[0985] "Notification and reporting" refers to the act of sending information about abnormal behavior to relevant organizations or specific users when it is detected.

[0986] "Recording and storage" refers to storing abnormal behavior detection results and related video data in a database.

[0987] "Retraining" is the process of data analysis and retraining that utilizes stored data to improve the accuracy of generative AI models.

[0988] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[0989] The system consists of a surveillance camera, a server, a terminal, and a user.

[0990] Hardware and Software Details

[0991] surveillance cameras

[0992] A surveillance camera is a device that captures real-time video data within a specified area and transmits it to a server via a network. The video data is transmitted to the server in streaming format via an IP network. The camera used is, for example, a high-resolution IP camera (e.g., 1080p HD camera).

[0993] server

[0994] The server is a computer system that aggregates and stores video data acquired from surveillance cameras and performs various processing. The server has the following functions:

[0995] Data Acquisition Module: Acquires real-time video data from each surveillance camera using standard protocols (e.g., RTSP).

[0996] Pre-processing module: Performs noise reduction, resolution adjustment, and frame division on the acquired video data. Here, an image processing library (e.g., OpenCV) is used.

[0997] Noise Reduction: Apply a Gaussian filter to remove noise.

[0998] Resolution adjustment: The resize function converts video data to a uniform resolution (e.g., 720p).

[0999] Frame division: Obtains FPS information, divides the video data into frames, and adds timestamps.

[1000] Anomaly Detection Module: Preprocessed video data is fed into a generative AI model to detect anomalous behavior. Deep learning frameworks (e.g., TensorFlow, PyTorch) are used.

[1001] Send the following prompt to the generative model:

[1002] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[1003] Notification and reporting module: When an abnormality is detected, it immediately notifies and reports to relevant organizations or specific users via SMS, email, or a dedicated notification app.

[1004] For example, an email could be sent stating, "A fall occurred in the North Area on October 5, 2023 at 2:30 p.m. Action is required."

[1005] Data storage module: Records anomaly detection events and related video data and stores them in a database.

[1006] Retraining module: Retrains the generative AI model using the stored data. Periodically extracts historical data from the database to create a retraining dataset.

[1007] Terminal

[1008] The devices are used by surveillance camera system administrators and security staff to receive notifications and alerts when an abnormality is detected. A dedicated app is installed, allowing for real-time situation monitoring and response.

[1009] User

[1010] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users respond promptly based on the content of notifications and reports sent from the system.

[1011] Specific examples

[1012] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[1013] 1. Acquiring video data

[1014] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1015] 2. Pretreatment

[1016] The server performs noise reduction and resolution adjustment on the video data and adds a timestamp to the frame.

[1017] 3. Detecting Abnormal Behavior

[1018] The server inputs the preprocessed video data into a generative AI model to detect falling behavior.

[1019] 4. Notifications and Reports

[1020] When a fall is detected, the server immediately notifies the mall's security staff.

[1021] 5. History storage and analysis

[1022] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generated AI model.

[1023] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[1024] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1025] Step 1: Acquire video data

[1026] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera sends video data to the server in streaming format via an IP network. The input is real-time video data from the surveillance cameras, and the output is the acquired raw video data. The server receives the video data using a standard protocol (e.g., RTSP). Specifically, the server periodically checks the connection status of each camera and attempts to reconnect if there is an abnormality.

[1027] Step 2: Preprocessing

[1028] The server performs pre-processing on the captured video data, including noise reduction, resolution adjustment, and frame splitting. The input is the captured raw video data, and the output is the pre-processed video data.

[1029] Noise removal: The server uses an image processing library (e.g., OpenCV) to apply a Gaussian filter to remove noise from the video data.

[1030] Resolution Adjustment: The server uses a resize function to convert all video data to a uniform resolution (e.g., 720p).

[1031] Frame division: The server acquires FPS (frames per second) information and divides the video data into frames. At the same time, it assigns a timestamp to each frame. Specifically, the server analyzes the video data and extracts frames at the appropriate time.

[1032] Step 3: Detecting Abnormal Behavior

[1033] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The input is the preprocessed video data, and the output is the abnormal behavior detection results.

[1034] Generative AI models are built using deep learning frameworks (e.g., TensorFlow, PyTorch).

[1035] The server converts the video data into the model's input format, sets the following prompt text, and sends it to the model.

[1036] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[1037] The server receives the model output, and if an anomaly is detected, it sets an anomaly flag for the corresponding frame. Specifically, the server inputs the data converted into tensor format into the model, processes the results, and sets an anomaly flag.

[1038] Step 4: Notification

[1039] When the server detects abnormal behavior, it immediately notifies and reports to relevant organizations or specific users. The input is the abnormal behavior detection result, and the output is a notification / report message.

[1040] The notification will include the time and location of the anomaly and any necessary action.

[1041] Reporting methods include SMS, email, and a dedicated notification app.

[1042] For example, an email could be sent stating, "A fall occurred in the North Area at 2:30 PM on October 5, 2023. Action is required." Specifically, the server generates a notification message and sends it to the specified contacts. It also uses the API of a dedicated notification app to send notifications in real time.

[1043] Step 5: Store and analyze history

[1044] The server records and stores all anomaly detection events and associated video data. The input is the abnormal behavior detection results and associated video data, and the output is the stored historical data.

[1045] The stored data will be analyzed at a later date and used to improve the system and retrain the generative AI model. Specifically, the server stores the data in a database, periodically extracts historical data from the database, and creates a dataset for retraining.

[1046] This enables the entire system to detect abnormal behavior in real time with high accuracy and respond quickly.

[1047] (Application example 1)

[1048] 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."

[1049] Conventional surveillance camera systems have issues with delays in detecting and notifying abnormal behavior, making it difficult to respond quickly. Furthermore, because notifications of abnormal situations are limited to specific devices, information sharing is insufficient and there is a lack of means to quickly contact multiple parties involved. As a result, security staff and guards are often unable to respond in a timely manner.

[1050] 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.

[1051] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the learning data, means for notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, means for recording and saving the abnormal behavior detection results and the video data, means for sending the notification as a push notification to a mobile device of the relevant organization or specific user, and means for generating an audio alarm when abnormal behavior is detected, thereby enabling immediate notification to multiple relevant parties and appropriate and prompt response.

[1052] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[1053] "Real-time" refers to the immediate acquisition and processing of data without delay.

[1054] "Video data" is information in the form of video captured by a surveillance camera.

[1055] "Noise reduction" refers to the process of removing unnecessary signals and noise from video data.

[1056] "Resolution" is an index that indicates the level of detail and image quality of video data.

[1057] A "generative model" is an AI model used to detect anomalous behavior based on training data.

[1058] "Training data" is the historical dataset used to train the generative model.

[1059] "Abnormal behavior" refers to actions or events that are different from the norm, such as falling, running away, or suspicious behavior.

[1060] "Notification" refers to the act of notifying relevant organizations or specific users when abnormal behavior is detected.

[1061] "Reporting" is a method of alerting or reporting to relevant authorities when abnormal behavior is detected.

[1062] "Related agencies" are agencies or organizations involved in responding to abnormal behavior, such as security staff, guards, and police.

[1063] A "specific user" is an individual or organization that is designated in advance by the system to receive notifications.

[1064] "Recording" refers to the act of saving detected abnormal behavior and related data.

[1065] "Storage" refers to keeping recorded data for future reference.

[1066] A "mobile terminal" is a mobile communication device such as a smartphone or tablet.

[1067] "Push notification" is a function that delivers notifications from the system to mobile devices in real time.

[1068] An "audio alarm" is a means of alerting the user by sound when abnormal behavior is detected.

[1069] This invention provides a system that acquires video data from a surveillance camera in real time and detects, notifies, and reports abnormal behavior. This system can quickly and efficiently detect abnormal behavior and report it to relevant authorities and designated users in order to increase the safety of the monitored area. A specific embodiment of the invention will be described below.

[1070] System configuration

[1071] surveillance cameras

[1072] Surveillance cameras capture real-time video data within a specific area and are connected to a central server via a network, where the captured video data is sent in streaming format.

[1073] server

[1074] The server has the following functions:

[1075] Data acquisition module: Acquires video data from surveillance cameras in real time.

[1076] Pre-processing module: Performs noise reduction and resolution adjustment on the acquired video data.

[1077] Anomaly detection module: Preprocessed video data is input into the generative model, and anomalous behavior is detected based on the training data.

[1078] Notification and reporting module: When abnormal behavior is detected, the module notifies and reports relevant organizations or specific users. Notifications are sent as push notifications to the mobile devices of relevant organizations or users.

[1079] Audio alarm module: Generates an audio alarm when abnormal behavior is detected.

[1080] Data storage module: Records and stores abnormal behavior detection results and related video data.

[1081] Terminal

[1082] The devices are used by security staff and administrators, and have an app installed that receives push notifications when an abnormality is detected. If abnormal behavior is detected, a push notification is immediately sent to the device, prompting appropriate action to be taken.

[1083] User

[1084] Users include relevant organizations and security staff, guards, police officers, etc. who are required to receive specific notifications. Users are required to respond promptly to the situation based on the push notification.

[1085] Hardware and software used

[1086] Hardware: surveillance cameras, servers, smartphones (mobile devices)

[1087] Software: Flask (server-side framework), OpenCV (image processing library), TensorFlow (generative model), Firebase Cloud Messaging (push notifications)

[1088] How we process your data

[1089] The server receives streaming video data from the surveillance cameras, and the preprocessing module performs noise reduction and resolution adjustment. The preprocessed video data is then input into the generative model to detect anomalous behavior. Detected anomalous behavior is immediately recognized by the anomaly detection module.

[1090] When an abnormality is detected, the notification module is activated and sends a push notification to relevant organizations or specific users. At the same time, the audio alarm module is activated and generates an audio alarm to prompt a prompt response. In addition, the data storage module records and stores the abnormal behavior detection results and related video data.

[1091] For example, if someone falls in a shopping mall, the video from the surveillance camera is sent to the server in real time, and the fall is detected using a TensorFlow model. A push notification is then sent to the security guard's mobile device via Firebase Cloud Messaging, and an audio alarm sounds, enabling a prompt response.

[1092] Prompt Sentence Examples

[1093] "Model input is generated from preprocessed video data and fed into the generative model."

[1094] "Detects abnormal behavior based on the prediction results and notifies relevant authorities."

[1095] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1096] Step 1:

[1097] The server acquires video data from the surveillance cameras in real time. The surveillance cameras capture a specific area and continuously transmit the video data to the server via the network. The server receives the data and prepares it for the next step of processing. The input of this step is the raw data sent from the surveillance cameras, and the output is the video data for preprocessing.

[1098] Step 2:

[1099] The server performs preprocessing on the received video data. Specifically, it uses OpenCV to remove noise and adjust the resolution. This improves the quality of the video and increases the accuracy of abnormal behavior detection. The input of this step is raw data, and the output is video data with noise removed and resolution adjusted.

[1100] Step 3:

[1101] The server inputs the preprocessed video data into the generative model. Using TensorFlow, the data is analyzed frame by frame and anomalous behavior is detected based on the training data. The input to this step is the preprocessed video data, and the output is the abnormal behavior detection results. Specifically, the generative model identifies abnormal behavior.

[1102] Step 4:

[1103] The server issues notifications and reports when abnormal behavior is detected. It uses Firebase Cloud Messaging to send push notifications to the mobile devices of relevant organizations or specific users. This notification includes the time, location, and video snapshot of the abnormal behavior. The input to this step is the abnormal behavior detection result, and the output is the sent notification and report. Specifically, the notification and report module is activated and a push notification is sent.

[1104] Step 5:

[1105] The server generates an audio alarm when abnormal behavior is detected. The audio alarm module plays an alarm sound immediately when the abnormal behavior detection result is input. The input of this step is the abnormal behavior detection result, and the output is the generated alarm sound. Specifically, the audio alarm is activated to warn the user.

[1106] Step 6:

[1107] The server records and stores the abnormal behavior detection results and related video data. The data storage module stores the data in a database for future analysis and system improvement. The input of this step is the abnormal behavior detection results and related video data, and the output is the stored data. The specific operation is writing to the database.

[1108] Step 7:

[1109] The user receives a push notification sent to their mobile device and is prompted to take action. In this step, the user who received the push notification quickly heads to the scene and takes appropriate action. The input to this step is the push notification sent from the server, and the output is the user's on-site action. Specifically, the user checks the notification and takes the necessary action.

[1110] As described above, the present invention is a system that can detect abnormal behavior in real time and quickly notify and report the situation to relevant organizations and users, thereby significantly improving the safety of the monitored area.

[1111] 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.

[1112] The present invention provides a system that detects abnormal behavior in real time in a surveillance camera system and further enhances security and response to abnormalities by combining it with an emotion engine that recognizes user emotions. Specific embodiments of the present invention will be described below.

[1113] System configuration

[1114] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[1115] 1. Surveillance cameras

[1116] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[1117] 2. Server

[1118] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[1119] Data acquisition module: Acquires real-time video data from each surveillance camera.

[1120] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[1121] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1122] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[1123] Data storage module: Stores all anomaly detection events and associated video data.

[1124] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[1125] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[1126] 3. Terminal

[1127] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[1128] 4. Users

[1129] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[1130] 5. Emotion Engine

[1131] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[1132] Explanation of program processing

[1133] Below, the operation of the system program is explained in natural language.

[1134] 1. Obtaining surveillance camera footage

[1135] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera is connected to the server via a network, and the video data is continuously transmitted to the server.

[1136] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[1137] 2. Preprocessing of video data

[1138] The server performs preprocessing on the acquired video data.

[1139] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[1140] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[1141] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[1142] 3. Anomaly Detection Using Generative Models

[1143] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[1144] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[1145] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[1146] 4. Emotion analysis

[1147] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine.

[1148] The emotion engine uses facial recognition technology to identify emotions (e.g., surprise, anger, fear, etc.) from the user's facial expressions.

[1149] 5. Notification and reporting when an abnormality is detected

[1150] The server detects an abnormality and analyzes the user's emotions, then immediately notifies and reports the relevant authorities and specific users.

[1151] The notification will include detailed information about the time and location of the anomaly, as well as emotional data.

[1152] Reporting can be done via SMS, email, or a dedicated notification app.

[1153] 6. History storage and analysis

[1154] The server records and stores all anomaly detection events and associated emotion and video data.

[1155] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[1156] Specific examples

[1157] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[1158] 1. Video acquisition

[1159] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1160] 2. Preprocessing of video data

[1161] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1162] 3. Anomaly Detection and Sentiment Analysis

[1163] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[1164] 4. Notifications and Reports

[1165] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[1166] 5. History storage and analysis

[1167] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[1168] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[1169] The processing flow will be explained below.

[1170] Step 1:

[1171] The server acquires video data in real time from each surveillance camera. Each surveillance camera is connected to the server via a network, and the video data is continuously streamed to the server.

[1172] Step 2:

[1173] The server preprocesses the acquired video data, specifically by denoising the video data, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting it into a format that is easier to analyze.

[1174] Step 3:

[1175] The server divides the pre-processed video data into frames, and assigns a timestamp to each frame to make them individually identifiable.

[1176] Step 4:

[1177] The server inputs each preprocessed frame of data into a generative model, which analyzes each frame based on pre-trained abnormal behavior patterns.

[1178] Step 5:

[1179] The server receives the analysis results of the generative model and detects abnormal behavior (e.g., falls, escape, placing suspicious objects, etc.). For detected abnormal behavior, it sets an abnormal flag for the corresponding frame.

[1180] Step 6:

[1181] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to analyze the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.).

[1182] Step 7:

[1183] The server then notifies relevant organizations and specific users based on the results of the user's emotion analysis and abnormal behavior detection. The notification includes detailed information about the abnormal behavior and emotions. Notifications can be sent via SMS, email, or a dedicated notification app.

[1184] Step 8:

[1185] The server records and stores all anomaly detection events, emotion data, and associated video data for later analysis and review.

[1186] Step 9:

[1187] The server periodically analyzes the stored data and retrains the generative model and emotion engine, improving the accuracy of the model and enabling more advanced anomaly detection and emotion recognition.

[1188] Example 2

[1189] 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."

[1190] Conventional surveillance camera systems are required to only detect abnormal behavior and respond quickly to it. However, they do not take into account navigation information or the user's emotional state, resulting in insufficient response when an abnormality occurs. Furthermore, delays in response after abnormal behavior detection and unnecessary reports due to false positives are also problems. Another issue is the inefficiency of the re-learning process to improve the accuracy of abnormal behavior detection.

[1191] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data, means for analyzing user emotions using an emotion engine when detecting the abnormal behavior, means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results, emotion data, and video data. This enables more accurate anomaly detection and rapid response that takes user emotions into consideration. Furthermore, by analyzing the saved data and retraining the generative model, the accuracy of the system can be continuously improved.

[1192] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[1193] "Real-time" refers to immediate data acquisition and processing with little to no delay.

[1194] "Video data" refers to data that includes visual information acquired from a surveillance camera.

[1195] "Preprocessing" refers to initial processing of video data, such as noise removal, resolution adjustment, and frame division.

[1196] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[1197] "Resolution adjustment" is a process for uniformly adjusting the resolution of video data.

[1198] A "generative model" is a model that uses AI technology to detect abnormal behavior based on past learning data.

[1199] "Abnormal behavior" refers to behavior that differs from normal conditions, such as falling, running away, or placing suspicious objects.

[1200] The "Emotion Engine" is a system that uses facial recognition technology to analyze emotions from a user's facial expressions.

[1201] "Notification and reporting" refers to the act of informing relevant organizations or specific users of the occurrence of an abnormality.

[1202] "Relevant agencies" refer to agencies that need to respond to the occurrence of abnormal behavior (e.g., police, security companies, etc.).

[1203] "User" refers to the party who receives notifications and alerts from the system.

[1204] "Emotion data" is data that indicates the user's emotional state analyzed by the emotion engine.

[1205] "Recording and storage" refers to the act of retaining the results of abnormal behavior detection and related data for later use.

[1206] "Relearning" is the process of analyzing past data to improve the accuracy of the generative model.

[1207] The present invention is a system that further enhances security and response to abnormalities by detecting abnormal behavior in real time in a surveillance camera system and combining it with an emotion engine that recognizes user emotions.

[1208] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[1209] 1. Surveillance cameras

[1210] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[1211] 2. Server

[1212] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[1213] Data acquisition module: Acquires real-time video data from each surveillance camera.

[1214] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[1215] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1216] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[1217] Data storage module: Stores all anomaly detection events and associated video data.

[1218] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[1219] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[1220] 3. Terminal

[1221] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[1222] 4. Users

[1223] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[1224] 5. Emotion Engine

[1225] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[1226] Specific examples

[1227] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[1228] 1. Video acquisition

[1229] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1230] 2. Preprocessing of video data

[1231] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1232] 3. Anomaly Detection and Sentiment Analysis

[1233] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[1234] 4. Notifications and Reports

[1235] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[1236] 5. History storage and analysis

[1237] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[1238] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[1239] Prompt Sentence Examples

[1240] For example, below is a specific example of a prompt for a generative AI model to detect suspicious behavior in a shopping mall:

[1241] Example prompt sentence:

[1242] "Abnormal behavior must be detected in the following video frames. Abnormal behavior may include falling, fleeing, or placing a suspicious object. Please analyze the following frame: [Frame Data]"

[1243] This prompt allows the generative model to operate properly.

[1244] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1245] Step 1: Acquire video data

[1246] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network, and video data is continuously transmitted to the server. The server uses a data acquisition module to receive this video data and process it in streaming format. The input is the video data from the surveillance cameras, and the output is the video data stored in the server's internal database waiting to be processed.

[1247] Step 2: Preprocessing the video data

[1248] The server performs preprocessing on the captured video data. Specifically, it performs noise reduction, resolution adjustment, and frame splitting. Noise reduction filters out unnecessary noise from the captured video data to improve image quality. Resolution adjustment converts the video data to a uniform resolution (e.g., 720p). Frame splitting divides the video data into frames and assigns a timestamp to each frame. The input is the captured raw data, and the output is processed, clear video frames.

[1249] Step 3: Detecting Abnormal Behavior

[1250] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The generative model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on the training data. When abnormal behavior is detected, an abnormality flag is set for the corresponding frame. In addition, detailed information about the abnormality (such as the time and location where it occurred) is generated. The input is the preprocessed video frame, and the output is the abnormal behavior identification result and the frame with the abnormality flag set.

[1251] Step 4: Sentiment Analysis

[1252] When detecting anomalous behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to read the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.). The analysis results are combined with anomalous behavior data and compiled as detailed emotional information. The input is the frame in which anomalous behavior was detected and the user's facial image, and the output is the analyzed emotional data.

[1253] Step 5: Notification and reporting when an abnormality is detected

[1254] After detecting abnormal behavior and analyzing emotions, the server immediately notifies and alerts relevant organizations and specific users. The notification includes the location and time of the abnormal behavior, as well as the analyzed emotional data. The server sends this information via SMS, email, or a dedicated notification app. The input is the anomaly detection result and emotional data, and the output is a notification or alert message.

[1255] Step 6: Store and analyze history

[1256] The server records and stores all anomaly detection events and associated emotion and video data. The stored data is later used to improve the accuracy of the generative model using a re-learning module. The stored data is stored in local or cloud storage and analyzed as needed. The inputs are anomaly detection results, emotion data, and video data, and the output is the stored historical data.

[1257] In this way, by clarifying the specific processing performed at each step and the flow of its inputs and outputs, the system can quickly and accurately detect and respond to abnormal behavior and user emotions.

[1258] (Application example 2)

[1259] 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."

[1260] Conventional surveillance camera systems are not only unable to detect abnormal behavior, but also have difficulty analyzing the causes of abnormal behavior and the user's underlying emotions in real time, resulting in inadequate responses. Furthermore, delays in notifications and alerts make it difficult to respond quickly. This has led to a demand for improved security and appropriate, prompt responses to abnormal situations.

[1261] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on learning data, means for analyzing the abnormal behavior and user emotions, means including a video output device for displaying the detected abnormality and emotion data, and means for recording and saving the abnormal behavior detection results and the video data. This makes it possible to analyze abnormal behavior and the associated user emotions in real time and respond quickly and appropriately.

[1262] A "surveillance camera" is a device that captures real-time video data within a specific area and transmits it to a server via a network.

[1263] "Real-time" refers to acquiring data the moment an event occurs and processing and analyzing it almost simultaneously.

[1264] "Video data" refers to digital data including video and image information captured by surveillance cameras.

[1265] "Preprocessing" refers to processing of acquired video data, such as noise removal, resolution adjustment, and frame division.

[1266] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[1267] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[1268] A "generative model" is a predictive model that uses past training data to identify abnormal behavior from new data.

[1269] "Abnormal behavior" refers to movements or actions that deviate from normal behavior, such as falling, running away, or suspicious behavior.

[1270] "Emotion analysis" is a process of identifying and analyzing emotions (surprise, fear, anger, etc.) from the user's facial expressions.

[1271] A "communication device" is a means for notifying or reporting to relevant organizations or specific users when abnormal behavior is detected.

[1272] "User" refers to a person (e.g., security guard, police officer, parent, etc.) who is responsible for receiving notifications and reports sent from the system and responding promptly.

[1273] The "video output device for display" is a device for displaying abnormal behavior and emotion data to the user, and includes smart glasses and dedicated monitors.

[1274] "Recording and storage" refers to the process of storing detected abnormal behavior and associated video and emotional data for future analysis and relearning.

[1275] The present invention is a system that detects abnormal behavior in real time in a surveillance camera system and combines it with an emotion engine that recognizes user emotions to further enhance security and response to abnormalities. The specific configuration and implementation method of this system are described below.

[1276] System configuration

[1277] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine. Each element is as follows:

[1278] 1. Surveillance cameras

[1279] Surveillance cameras capture real-time video data within a specific area. Each surveillance camera is connected to a server via an IP network and transmits the video data in streaming format.

[1280] 2. Server

[1281] The server aggregates and stores video data acquired from surveillance cameras and implements the following functions:

[1282] Acquisition and preprocessing of video data: Acquire real-time video data from surveillance cameras and perform noise removal and resolution adjustment.

[1283] Anomalous behavior detection: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1284] Sentiment analysis: In addition to detecting abnormal behavior, we also use an emotion engine to analyze user emotions.

[1285] Notification and reporting: Based on detected abnormal behavior and emotion data, relevant authorities and specific users are notified and reported.

[1286] Data storage: Record and store anomaly detection results and video data.

[1287] 3. Terminal

[1288] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected, including the time, location, and emotional data of the abnormality.

[1289] 4. Users

[1290] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). They are required to respond promptly based on the content of notifications and reports sent from the system.

[1291] 5. Emotion Engine

[1292] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the user's emotion data is correlated with abnormal behavior data for analysis.

[1293] Specific examples

[1294] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[1295] 1. Video acquisition

[1296] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1297] 2. Preprocessing of video data

[1298] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1299] 3. Anomaly Detection and Sentiment Analysis

[1300] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[1301] 4. Notifications and Reports

[1302] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[1303] 5. History storage and analysis

[1304] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[1305] Prompt Sentence Examples

[1306] Here is an example of input to a generative AI model using the following prompts:

[1307] Develop an application that analyzes real-time video footage from a surveillance camera system, detects abnormal behavior, and recognizes and analyzes user emotions. The application will be for smart glasses, and will immediately notify the smart glasses when abnormal behavior is detected and display the user's emotional data. The server will preprocess the video data and incorporate functions to analyze abnormal behavior and emotions using a generative model.

[1308] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1309] Step 1:

[1310] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network and continuously transmits video data. The input is real-time video data transmitted from the surveillance cameras, and the output is raw data stored in the server.

[1311] Step 2:

[1312] The server performs preprocessing on the acquired video data. This preprocessing includes noise removal, resolution adjustment, and frame division. The input is real-time video data, and the output is preprocessed data after data processing such as noise removal, resolution adjustment, and adding a timestamp to each frame.

[1313] Step 3:

[1314] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. Based on past learning data, the generative model identifies abnormal behavior (e.g., falls, fleeing, suspicious behavior) in real time. The input is the preprocessed data, and abnormal behavior is identified as data calculation. The output is the frame data in which abnormal behavior is detected and an abnormality flag.

[1315] Step 4:

[1316] The server detects abnormal behavior and simultaneously analyzes the user's emotions using an emotion engine. It uses facial recognition technology to identify emotions (e.g., surprise, fear, anger, etc.) from facial expressions. The input is video data in which abnormal behavior was detected, and emotion identification is performed as part of the data analysis. The output is the user's emotional data.

[1317] Step 5:

[1318] The server immediately notifies and reports to relevant organizations and specific users based on abnormal behavior and emotion data. The notification content includes the time, location, and emotion data of the abnormality. The input is the abnormal behavior detection result and emotion data, and the notification message is generated and sent as the data output.

[1319] Step 6:

[1320] The server records and stores the detected abnormal behavior and emotion data. All anomaly detection events and associated video data are used for future analysis and retraining of the generative model. The inputs are the anomaly detection results and emotion data, which are recorded in the database as data storage.

[1321] 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.

[1322] 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.

[1323] 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.

[1324] [Fourth embodiment]

[1325] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1326] 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.

[1327] 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).

[1328] 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.

[1329] 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.

[1330] 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).

[1331] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1332] 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.

[1333] 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.

[1334] 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.

[1335] 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.

[1336] 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.

[1337] 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."

[1338] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[1339] System configuration

[1340] This system consists of a surveillance camera, a server, a terminal, and a user.

[1341] 1. Surveillance cameras

[1342] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[1343] 2. Server

[1344] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[1345] Data acquisition module: Acquires real-time video data from each surveillance camera.

[1346] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[1347] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1348] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[1349] Data storage module: Stores all anomaly detection events and associated video data.

[1350] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[1351] 3. Terminal

[1352] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[1353] 4. Users

[1354] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[1355] Explanation of program processing

[1356] Below, the operation of the system program is explained in natural language.

[1357] 1. Obtaining surveillance camera footage

[1358] The server acquires video data in real time from multiple surveillance cameras, which are connected to the server via a network and continuously transmit the video data to the server.

[1359] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[1360] 2. Preprocessing of video data

[1361] The server performs preprocessing on the acquired video data.

[1362] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[1363] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[1364] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[1365] 3. Anomaly Detection Using Generative Models

[1366] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[1367] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[1368] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[1369] 4. Notification and reporting when an abnormality is detected

[1370] When the server detects an abnormality, it immediately notifies and reports the relevant organizations and specific users.

[1371] The notification will include the time and location of the anomaly and any necessary action.

[1372] Reporting can be done via SMS, email, or a dedicated notification app.

[1373] 5. History storage and analysis

[1374] The server records and stores all anomaly detection events and associated video data.

[1375] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[1376] Specific examples

[1377] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[1378] 1. Video acquisition

[1379] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1380] 2. Preprocessing of video data

[1381] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1382] 3. Anomaly Detection

[1383] The server inputs the preprocessed video data into a generative model to detect falling behavior.

[1384] 4. Notifications and Reports

[1385] When a fall is detected, the server immediately notifies mall security staff, including the location and time of the fall, as well as a video snapshot of the frame in question.

[1386] 5. History storage and analysis

[1387] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generative model.

[1388] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[1389] The processing flow will be explained below.

[1390] Step 1:

[1391] The server acquires video data in real time from each surveillance camera. The surveillance cameras are connected to the server via an IP network and transmit the video data continuously in streaming format.

[1392] Step 2:

[1393] The server preprocesses the captured video data by performing noise reduction, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting the video data into a format suitable for analysis.

[1394] Step 3:

[1395] The server divides the pre-processed video data into frames, each of which is assigned a timestamp to make it uniquely identifiable.

[1396] Step 4:

[1397] The server inputs each preprocessed frame of data into a generative model, which then analyzes each frame based on pre-trained patterns of abnormal behavior.

[1398] Step 5:

[1399] The server receives the analysis results from the generative model and detects anomalous behavior. If anomalous behavior is detected, an anomalous flag is set for the corresponding frame. Abnormal behavior includes falling, fleeing, placing suspicious objects, etc.

[1400] Step 6:

[1401] The server will send a notification when an abnormality is detected. Specifically, it will send a notification containing detailed information about the abnormality to relevant organizations or specific users. Notification methods include SMS, email, or a dedicated notification app.

[1402] Step 7:

[1403] The server records and stores all anomaly detection events and associated video data for later analysis and review.

[1404] Step 8:

[1405] The server periodically analyzes the stored data and retrains the generative model, improving the model's accuracy and enabling more effective anomalous behavior detection.

[1406] Example 1

[1407] 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."

[1408] Conventional surveillance camera systems have the problem of being difficult to detect abnormal behavior in real time and quickly notify and report it. They also lack a mechanism for saving detection results and using them for re-learning to improve the accuracy of the system. This can result in insufficient accuracy in detecting abnormal behavior and rapid response.

[1409] 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.

[1410] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise, adjust resolution, and split frames, means for inputting the preprocessed video data into a generative AI model and detecting abnormal behavior based on the learning data, means for immediately notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results and video data, and for later analysis and re-training of the generative AI model. This enables highly accurate detection of abnormal behavior in real time and rapid response.

[1411] A "surveillance camera" is a device that captures real-time video data within a designated area and transmits the data over a network.

[1412] "Real-time video data" refers to video data that is continuously captured by a surveillance camera and instantly transmitted to a server.

[1413] "Preprocessing" refers to processing that removes noise, adjusts resolution, and divides frames from video data.

[1414] "Noise removal" is a process that removes unnecessary noise components from video data to improve the accuracy of analysis.

[1415] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[1416] "Frame division" is a process of breaking down video data into frames and assigning a timestamp to each frame.

[1417] A "generative AI model" is an artificial intelligence model for detecting abnormal behavior in video data based on past learning data.

[1418] "Abnormal behavior" refers to actions that would not normally occur, such as falling, running away, or suspicious behavior.

[1419] "Notification and reporting" refers to the act of sending information about abnormal behavior to relevant organizations or specific users when it is detected.

[1420] "Recording and storage" refers to storing abnormal behavior detection results and related video data in a database.

[1421] "Retraining" is the process of data analysis and retraining that utilizes stored data to improve the accuracy of generative AI models.

[1422] The present invention provides a system for acquiring video data from a surveillance camera in real time, detecting abnormal behavior based on the data, and notifying and reporting the abnormal behavior. Specific embodiments of the present invention will be described below.

[1423] The system consists of a surveillance camera, a server, a terminal, and a user.

[1424] Hardware and Software Details

[1425] surveillance cameras

[1426] A surveillance camera is a device that captures real-time video data within a specified area and transmits it to a server via a network. The video data is transmitted to the server in streaming format via an IP network. The camera used is, for example, a high-resolution IP camera (e.g., 1080p HD camera).

[1427] server

[1428] The server is a computer system that aggregates and stores video data acquired from surveillance cameras and performs various processing. The server has the following functions:

[1429] Data Acquisition Module: Acquires real-time video data from each surveillance camera using standard protocols (e.g., RTSP).

[1430] Pre-processing module: Performs noise reduction, resolution adjustment, and frame division on the acquired video data. Here, an image processing library (e.g., OpenCV) is used.

[1431] Denoise: Apply a Gaussian filter to remove noise.

[1432] Resolution adjustment: The resize function converts video data to a uniform resolution (e.g., 720p).

[1433] Frame division: Obtains FPS information, divides the video data into frames, and adds timestamps.

[1434] Anomaly Detection Module: Preprocessed video data is fed into a generative AI model to detect anomalous behavior. Deep learning frameworks (e.g., TensorFlow, PyTorch) are used.

[1435] Send the following prompt to the generative model:

[1436] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[1437] Notification and reporting module: When an abnormality is detected, it immediately notifies and reports to relevant organizations or specific users via SMS, email, or a dedicated notification app.

[1438] For example, an email could be sent stating, "A fall occurred in the North Area on October 5, 2023 at 2:30 p.m. Action is required."

[1439] Data storage module: Records anomaly detection events and related video data and stores them in a database.

[1440] Retraining module: Retrains the generative AI model using the stored data. Periodically extracts historical data from the database to create a retraining dataset.

[1441] Terminal

[1442] The devices are used by surveillance camera system administrators and security staff to receive notifications and alerts when an abnormality is detected. A dedicated app is installed, allowing for real-time situation monitoring and response.

[1443] User

[1444] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users respond promptly based on the content of notifications and reports sent from the system.

[1445] Specific examples

[1446] For example, when detecting a fall accident in a shopping mall, the process proceeds as follows:

[1447] 1. Acquiring video data

[1448] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1449] 2. Pretreatment

[1450] The server performs noise reduction and resolution adjustment on the video data and adds a timestamp to the frame.

[1451] 3. Detecting Abnormal Behavior

[1452] The server inputs the preprocessed video data into a generative AI model to detect falling behavior.

[1453] 4. Notifications and Reports

[1454] When a fall is detected, the server immediately notifies the mall's security staff.

[1455] 5. History storage and analysis

[1456] The server stores the fall detection results and video data, and analyzes them at a later date to retrain the generated AI model.

[1457] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior.

[1458] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1459] Step 1: Acquire video data

[1460] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera sends video data to the server in streaming format via an IP network. The input is real-time video data from the surveillance cameras, and the output is the acquired raw video data. The server receives the video data using a standard protocol (e.g., RTSP). Specifically, the server periodically checks the connection status of each camera and attempts to reconnect if there is an abnormality.

[1461] Step 2: Preprocessing

[1462] The server performs pre-processing on the captured video data, including noise reduction, resolution adjustment, and frame splitting. The input is the captured raw video data, and the output is the pre-processed video data.

[1463] Noise removal: The server uses an image processing library (e.g., OpenCV) to apply a Gaussian filter to remove noise from the video data.

[1464] Resolution Adjustment: The server uses a resize function to convert all video data to a uniform resolution (e.g., 720p).

[1465] Frame division: The server acquires FPS (frames per second) information and divides the video data into frames. At the same time, it assigns a timestamp to each frame. Specifically, the server analyzes the video data and extracts frames at the appropriate time.

[1466] Step 3: Detecting Abnormal Behavior

[1467] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The input is the preprocessed video data, and the output is the abnormal behavior detection results.

[1468] Generative AI models are built using deep learning frameworks (e.g., TensorFlow, PyTorch).

[1469] The server converts the video data into the model's input format, sets the following prompt text, and sends it to the model.

[1470] "Detect any abnormal behavior (falls, escapes, placing suspicious objects, etc.) within this frame."

[1471] The server receives the model output, and if an anomaly is detected, it sets an anomaly flag for the corresponding frame. Specifically, the server inputs the data converted into tensor format into the model, processes the results, and sets an anomaly flag.

[1472] Step 4: Notification

[1473] When the server detects abnormal behavior, it immediately notifies and reports to relevant organizations or specific users. The input is the abnormal behavior detection result, and the output is a notification / report message.

[1474] The notification will include the time and location of the anomaly and any necessary action.

[1475] Reporting methods include SMS, email, and a dedicated notification app.

[1476] For example, an email could be sent stating, "A fall occurred in the North Area at 2:30 PM on October 5, 2023. Action is required." Specifically, the server generates a notification message and sends it to the specified contacts. It also uses the API of a dedicated notification app to send notifications in real time.

[1477] Step 5: Store and analyze history

[1478] The server records and stores all anomaly detection events and associated video data. The input is the abnormal behavior detection results and associated video data, and the output is the stored historical data.

[1479] The stored data will be analyzed at a later date and used to improve the system and retrain the generative AI model. Specifically, the server stores the data in a database, periodically extracts historical data from the database, and creates a dataset for retraining.

[1480] This enables the entire system to detect abnormal behavior in real time with high accuracy and respond quickly.

[1481] (Application example 1)

[1482] 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."

[1483] Conventional surveillance camera systems have issues with delays in detecting and notifying abnormal behavior, making it difficult to respond quickly. Furthermore, because notifications of abnormal situations are limited to specific devices, information sharing is insufficient and there is a lack of means to quickly contact multiple parties involved. As a result, security staff and guards are often unable to respond in a timely manner.

[1484] 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.

[1485] In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the learning data, means for notifying and reporting to relevant organizations or specific users when abnormal behavior is detected, means for recording and saving the abnormal behavior detection results and the video data, means for sending the notification as a push notification to a mobile device of the relevant organization or specific user, and means for generating an audio alarm when abnormal behavior is detected, thereby enabling immediate notification to multiple relevant parties and appropriate and prompt response.

[1486] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[1487] "Real-time" refers to the immediate acquisition and processing of data without delay.

[1488] "Video data" is information in the form of video captured by a surveillance camera.

[1489] "Noise reduction" refers to the process of removing unnecessary signals and noise from video data.

[1490] "Resolution" is an index that indicates the level of detail and image quality of video data.

[1491] A "generative model" is an AI model used to detect anomalous behavior based on training data.

[1492] "Training data" is the historical dataset used to train the generative model.

[1493] "Abnormal behavior" refers to actions or events that are different from the norm, such as falling, running away, or suspicious behavior.

[1494] "Notification" refers to the act of notifying relevant organizations or specific users when abnormal behavior is detected.

[1495] "Reporting" is a method of alerting or reporting to relevant authorities when abnormal behavior is detected.

[1496] "Related agencies" are agencies or organizations involved in responding to abnormal behavior, such as security staff, guards, and police.

[1497] A "specific user" is an individual or organization that is designated in advance by the system to receive notifications.

[1498] "Recording" refers to the act of saving detected abnormal behavior and related data.

[1499] "Storage" refers to keeping recorded data for future reference.

[1500] A "mobile terminal" is a mobile communication device such as a smartphone or tablet.

[1501] "Push notification" is a function that delivers notifications from the system to mobile devices in real time.

[1502] An "audio alarm" is a means of alerting the user by sound when abnormal behavior is detected.

[1503] This invention provides a system that acquires video data from a surveillance camera in real time and detects, notifies, and reports abnormal behavior. This system can quickly and efficiently detect abnormal behavior and report it to relevant authorities and designated users in order to increase the safety of the monitored area. A specific embodiment of the invention will be described below.

[1504] System configuration

[1505] surveillance cameras

[1506] Surveillance cameras capture real-time video data within a specific area and are connected to a central server via a network, where the captured video data is sent in streaming format.

[1507] server

[1508] The server has the following functions:

[1509] Data acquisition module: Acquires video data from surveillance cameras in real time.

[1510] Pre-processing module: Performs noise reduction and resolution adjustment on the acquired video data.

[1511] Anomaly detection module: Preprocessed video data is input into the generative model, and anomalous behavior is detected based on the training data.

[1512] Notification and reporting module: When abnormal behavior is detected, the module notifies and reports relevant organizations or specific users. Notifications are sent as push notifications to the mobile devices of relevant organizations or users.

[1513] Audio alarm module: Generates an audio alarm when abnormal behavior is detected.

[1514] Data storage module: Records and stores abnormal behavior detection results and related video data.

[1515] Terminal

[1516] The devices are used by security staff and administrators, and have an app installed that receives push notifications when an abnormality is detected. If abnormal behavior is detected, a push notification is immediately sent to the device, prompting appropriate action to be taken.

[1517] User

[1518] Users include relevant organizations and security staff, guards, police officers, etc. who are required to receive specific notifications. Users are required to respond promptly to the situation based on the push notification.

[1519] Hardware and software used

[1520] Hardware: surveillance cameras, servers, smartphones (mobile devices)

[1521] Software: Flask (server-side framework), OpenCV (image processing library), TensorFlow (generative model), Firebase Cloud Messaging (push notifications)

[1522] How we process your data

[1523] The server receives streaming video data from the surveillance cameras, and the preprocessing module performs noise reduction and resolution adjustment. The preprocessed video data is then input into the generative model to detect anomalous behavior. Detected anomalous behavior is immediately recognized by the anomaly detection module.

[1524] When an abnormality is detected, the notification module is activated and sends a push notification to relevant organizations or specific users. At the same time, the audio alarm module is activated and generates an audio alarm to prompt a prompt response. In addition, the data storage module records and stores the abnormal behavior detection results and related video data.

[1525] For example, if someone falls in a shopping mall, the video from the surveillance camera is sent to the server in real time, and the fall is detected using a TensorFlow model. A push notification is then sent to the security guard's mobile device via Firebase Cloud Messaging, and an audio alarm sounds, enabling a prompt response.

[1526] Prompt Sentence Examples

[1527] "Model input is generated from preprocessed video data and fed into the generative model."

[1528] "Detects abnormal behavior based on the prediction results and notifies relevant authorities."

[1529] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1530] Step 1:

[1531] The server acquires video data from the surveillance cameras in real time. The surveillance cameras capture a specific area and continuously transmit the video data to the server via the network. The server receives the data and prepares it for the next step of processing. The input of this step is the raw data sent from the surveillance cameras, and the output is the video data for preprocessing.

[1532] Step 2:

[1533] The server performs preprocessing on the received video data. Specifically, it uses OpenCV to remove noise and adjust the resolution. This improves the quality of the video and increases the accuracy of abnormal behavior detection. The input of this step is raw data, and the output is video data with noise removed and resolution adjusted.

[1534] Step 3:

[1535] The server inputs the preprocessed video data into the generative model. Using TensorFlow, the data is analyzed frame by frame and anomalous behavior is detected based on the training data. The input to this step is the preprocessed video data, and the output is the abnormal behavior detection results. Specifically, the generative model identifies abnormal behavior.

[1536] Step 4:

[1537] The server issues notifications and reports when abnormal behavior is detected. It uses Firebase Cloud Messaging to send push notifications to the mobile devices of relevant organizations or specific users. This notification includes the time, location, and video snapshot of the abnormal behavior. The input to this step is the abnormal behavior detection result, and the output is the sent notification and report. Specifically, the notification and report module is activated and a push notification is sent.

[1538] Step 5:

[1539] The server generates an audio alarm when abnormal behavior is detected. The audio alarm module plays an alarm sound immediately when the abnormal behavior detection result is input. The input of this step is the abnormal behavior detection result, and the output is the generated alarm sound. Specifically, the audio alarm is activated to warn the user.

[1540] Step 6:

[1541] The server records and stores the abnormal behavior detection results and related video data. The data storage module stores the data in a database for future analysis and system improvement. The input of this step is the abnormal behavior detection results and related video data, and the output is the stored data. The specific operation is writing to the database.

[1542] Step 7:

[1543] The user receives a push notification sent to their mobile device and is prompted to take action. In this step, the user who received the push notification quickly heads to the scene and takes appropriate action. The input to this step is the push notification sent from the server, and the output is the user's on-site action. Specifically, the user checks the notification and takes the necessary action.

[1544] As described above, the present invention is a system that can detect abnormal behavior in real time and quickly notify and report the situation to relevant organizations and users, thereby significantly improving the safety of the monitored area.

[1545] 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.

[1546] The present invention provides a system that detects abnormal behavior in real time in a surveillance camera system and further enhances security and response to abnormalities by combining it with an emotion engine that recognizes user emotions. Specific embodiments of the present invention will be described below.

[1547] System configuration

[1548] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[1549] 1. Surveillance cameras

[1550] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[1551] 2. Server

[1552] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[1553] Data acquisition module: Acquires real-time video data from each surveillance camera.

[1554] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[1555] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1556] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[1557] Data storage module: Stores all anomaly detection events and associated video data.

[1558] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[1559] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[1560] 3. Terminal

[1561] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[1562] 4. Users

[1563] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[1564] 5. Emotion Engine

[1565] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[1566] Explanation of program processing

[1567] Below, the operation of the system program is explained in natural language.

[1568] 1. Obtaining surveillance camera footage

[1569] The server acquires video data in real time from multiple surveillance cameras. Each surveillance camera is connected to the server via a network, and the video data is continuously transmitted to the server.

[1570] The terminal monitors the status of the surveillance camera and sends an alert to the server if there is an abnormality in the quality of the video.

[1571] 2. Preprocessing of video data

[1572] The server performs preprocessing on the acquired video data.

[1573] Noise removal: Removes unnecessary noise from video data to improve analysis accuracy.

[1574] Resolution adjustment: All video data is set to a uniform resolution (e.g., 720p) to facilitate analysis.

[1575] Frame division: Video data is divided into frames and a timestamp is assigned to each frame.

[1576] 3. Anomaly Detection Using Generative Models

[1577] The server inputs the preprocessed video data into a generative model to detect anomalous behavior.

[1578] The AI ​​model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on past learning data.

[1579] If an anomaly is detected, the server sets an anomaly flag in the corresponding frame.

[1580] 4. Emotion analysis

[1581] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine.

[1582] The emotion engine uses facial recognition technology to identify emotions (e.g., surprise, anger, fear, etc.) from the user's facial expressions.

[1583] 5. Notification and reporting when an abnormality is detected

[1584] The server detects an abnormality and analyzes the user's emotions, then immediately notifies and reports the relevant authorities and specific users.

[1585] The notification will include detailed information about the time and location of the anomaly, as well as emotional data.

[1586] Reporting can be done via SMS, email, or a dedicated notification app.

[1587] 6. History storage and analysis

[1588] The server records and stores all anomaly detection events and associated emotion and video data.

[1589] The stored data will be analyzed at a later date and used to improve the system and retrain the generative model.

[1590] Specific examples

[1591] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[1592] 1. Video acquisition

[1593] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1594] 2. Preprocessing of video data

[1595] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1596] 3. Anomaly Detection and Sentiment Analysis

[1597] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[1598] 4. Notifications and Reports

[1599] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[1600] 5. History storage and analysis

[1601] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[1602] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[1603] The processing flow will be explained below.

[1604] Step 1:

[1605] The server acquires video data in real time from each surveillance camera. Each surveillance camera is connected to the server via a network, and the video data is continuously streamed to the server.

[1606] Step 2:

[1607] The server preprocesses the acquired video data, specifically by denoising the video data, filtering out low-frequency noise and unwanted background sounds, standardizing the resolution (e.g., 720p), and converting it into a format that is easier to analyze.

[1608] Step 3:

[1609] The server divides the pre-processed video data into frames, and assigns a timestamp to each frame to make them individually identifiable.

[1610] Step 4:

[1611] The server inputs each preprocessed frame of data into a generative model, which analyzes each frame based on pre-trained abnormal behavior patterns.

[1612] Step 5:

[1613] The server receives the analysis results of the generative model and detects abnormal behavior (e.g., falls, escape, placing suspicious objects, etc.). For detected abnormal behavior, it sets an abnormal flag for the corresponding frame.

[1614] Step 6:

[1615] When detecting abnormal behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to analyze the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.).

[1616] Step 7:

[1617] The server then notifies relevant organizations and specific users based on the results of the user's emotion analysis and abnormal behavior detection. The notification includes detailed information about the abnormal behavior and emotions. Notifications can be sent via SMS, email, or a dedicated notification app.

[1618] Step 8:

[1619] The server records and stores all anomaly detection events, emotion data, and associated video data for later analysis and review.

[1620] Step 9:

[1621] The server periodically analyzes the stored data and retrains the generative model and emotion engine, improving the accuracy of the model and enabling more advanced anomaly detection and emotion recognition.

[1622] Example 2

[1623] 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."

[1624] Conventional surveillance camera systems are required to only detect abnormal behavior and respond quickly to it. However, they do not take into account navigation information or the user's emotional state, resulting in insufficient response when an abnormality occurs. Furthermore, delays in response after abnormal behavior detection and unnecessary reports due to false positives are also problems. Another issue is the inefficiency of the re-learning process to improve the accuracy of abnormal behavior detection.

[1625] The identification process by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data, means for analyzing user emotions using an emotion engine when detecting the abnormal behavior, means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected, and means for recording and saving the abnormal behavior detection results, emotion data, and video data. This enables more accurate anomaly detection and rapid response that takes user emotions into consideration. Furthermore, by analyzing the saved data and retraining the generative model, the accuracy of the system can be continuously improved.

[1626] A "surveillance camera" is a device for acquiring real-time video data within a specific area.

[1627] "Real-time" refers to immediate data acquisition and processing with little to no delay.

[1628] "Video data" refers to data that includes visual information acquired from a surveillance camera.

[1629] "Preprocessing" refers to initial processing of video data, such as noise removal, resolution adjustment, and frame division.

[1630] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[1631] "Resolution adjustment" is a process for uniformly adjusting the resolution of video data.

[1632] A "generative model" is a model that uses AI technology to detect abnormal behavior based on past learning data.

[1633] "Abnormal behavior" refers to behavior that differs from normal conditions, such as falling, running away, or placing suspicious objects.

[1634] The "Emotion Engine" is a system that uses facial recognition technology to analyze emotions from a user's facial expressions.

[1635] "Notification and reporting" refers to the act of informing relevant organizations or specific users of the occurrence of an abnormality.

[1636] "Relevant agencies" refer to agencies that need to respond to the occurrence of abnormal behavior (e.g., police, security companies, etc.).

[1637] "User" refers to the party who receives notifications and alerts from the system.

[1638] "Emotion data" is data that indicates the user's emotional state analyzed by the emotion engine.

[1639] "Recording and storage" refers to the act of retaining the results of abnormal behavior detection and related data for later use.

[1640] "Relearning" is the process of analyzing past data to improve the accuracy of the generative model.

[1641] The present invention is a system that further enhances security and response to abnormalities by detecting abnormal behavior in real time in a surveillance camera system and combining it with an emotion engine that recognizes user emotions.

[1642] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine.

[1643] 1. Surveillance cameras

[1644] Surveillance cameras capture real-time video data within a specific area, and each camera is connected to a central server via an IP network to transmit the video data in streaming format.

[1645] 2. Server

[1646] The server aggregates and stores the video data acquired from the surveillance cameras. The server has the following functions implemented:

[1647] Data acquisition module: Acquires real-time video data from each surveillance camera.

[1648] Pre-processing module: Performs noise removal, resolution adjustment, and frame division on the acquired video data.

[1649] Anomaly detection module: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1650] Notification and reporting module: Notifies and reports relevant organizations and specific users of detected anomalies.

[1651] Data storage module: Stores all anomaly detection events and associated video data.

[1652] Retraining module: Analyzes stored data and improves the accuracy of the generative model.

[1653] Sentiment Analysis Module: Implements the emotion engine that recognizes and analyzes user emotions.

[1654] 3. Terminal

[1655] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected.

[1656] 4. Users

[1657] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). Users are required to respond promptly based on the content of notifications and reports sent from the system.

[1658] 5. Emotion Engine

[1659] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the emotion data of the user is correlated with abnormal behavior data for analysis.

[1660] Specific examples

[1661] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[1662] 1. Video acquisition

[1663] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1664] 2. Preprocessing of video data

[1665] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1666] 3. Anomaly Detection and Sentiment Analysis

[1667] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[1668] 4. Notifications and Reports

[1669] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[1670] 5. History storage and analysis

[1671] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[1672] In this way, this system aims to improve security efficiency and safety by analyzing surveillance camera footage in real time and quickly detecting, notifying, and reporting abnormal behavior and user emotions.

[1673] Prompt Sentence Examples

[1674] For example, below is a specific example of a prompt for a generative AI model to detect suspicious behavior in a shopping mall:

[1675] Example prompt sentence:

[1676] "Abnormal behavior must be detected in the following video frames. Abnormal behavior may include falling, fleeing, or placing a suspicious object. Please analyze the following frame: [Frame Data]"

[1677] This prompt allows the generative model to operate properly.

[1678] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1679] Step 1: Acquire video data

[1680] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network, and video data is continuously transmitted to the server. The server uses a data acquisition module to receive this video data and process it in streaming format. The input is the video data from the surveillance cameras, and the output is the video data stored in the server's internal database waiting to be processed.

[1681] Step 2: Preprocessing the video data

[1682] The server performs preprocessing on the captured video data. Specifically, it performs noise reduction, resolution adjustment, and frame splitting. Noise reduction filters out unnecessary noise from the captured video data to improve image quality. Resolution adjustment converts the video data to a uniform resolution (e.g., 720p). Frame splitting divides the video data into frames and assigns a timestamp to each frame. The input is the captured raw data, and the output is processed, clear video frames.

[1683] Step 3: Detecting Abnormal Behavior

[1684] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. The generative model identifies abnormal behavior (e.g., falls, fleeing, placing suspicious objects, etc.) in real time based on the training data. When abnormal behavior is detected, an abnormality flag is set for the corresponding frame. In addition, detailed information about the abnormality (such as the time and location where it occurred) is generated. The input is the preprocessed video frame, and the output is the abnormal behavior identification result and the frame with the abnormality flag set.

[1685] Step 4: Sentiment Analysis

[1686] When detecting anomalous behavior, the server simultaneously analyzes the user's emotions using an emotion engine. The emotion engine uses facial recognition technology to read the user's facial expressions and identify emotions (e.g., surprise, anger, fear, etc.). The analysis results are combined with anomalous behavior data and compiled as detailed emotional information. The input is the frame in which anomalous behavior was detected and the user's facial image, and the output is the analyzed emotional data.

[1687] Step 5: Notification and reporting when an abnormality is detected

[1688] After detecting abnormal behavior and analyzing emotions, the server immediately notifies and alerts relevant organizations and specific users. The notification includes the location and time of the abnormal behavior, as well as the analyzed emotional data. The server sends this information via SMS, email, or a dedicated notification app. The input is the anomaly detection result and emotional data, and the output is a notification or alert message.

[1689] Step 6: Store and analyze history

[1690] The server records and stores all anomaly detection events and associated emotion and video data. The stored data is later used to improve the accuracy of the generative model using a re-learning module. The stored data is stored in local or cloud storage and analyzed as needed. The inputs are anomaly detection results, emotion data, and video data, and the output is the stored historical data.

[1691] In this way, by clarifying the specific processing performed at each step and the flow of its inputs and outputs, the system can realize a method for quickly and accurately detecting and responding to abnormal behavior and user emotions.

[1692] (Application example 2)

[1693] 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."

[1694] Conventional surveillance camera systems are not only unable to detect abnormal behavior, but also have difficulty analyzing the causes of abnormal behavior and the user's underlying emotions in real time, resulting in inadequate responses. Furthermore, delays in notifications and alerts make it difficult to respond quickly. This has led to a demand for improved security and appropriate, prompt responses to abnormal situations.

[1695] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring video data from a surveillance camera in real time, means for preprocessing the acquired video data to remove noise and adjust resolution, means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on learning data, means for analyzing the abnormal behavior and user emotions, means including a video output device for displaying the detected abnormality and emotion data, and means for recording and saving the abnormal behavior detection results and the video data. This makes it possible to analyze abnormal behavior and the associated user emotions in real time and respond quickly and appropriately.

[1696] A "surveillance camera" is a device that captures real-time video data within a specific area and transmits it to a server via a network.

[1697] "Real-time" refers to acquiring data the moment an event occurs and processing and analyzing it almost simultaneously.

[1698] "Video data" refers to digital data including video and image information captured by surveillance cameras.

[1699] "Preprocessing" refers to processing of acquired video data, such as noise removal, resolution adjustment, and frame division.

[1700] "Noise removal" is a process that removes unnecessary noise from video data to improve the accuracy of analysis.

[1701] "Resolution adjustment" is a process of changing the image quality of video data to a uniform resolution.

[1702] A "generative model" is a predictive model that uses past training data to identify abnormal behavior from new data.

[1703] "Abnormal behavior" refers to movements or actions that deviate from normal behavior, such as falling, running away, or suspicious behavior.

[1704] "Emotion analysis" is a process of identifying and analyzing emotions (surprise, fear, anger, etc.) from the user's facial expressions.

[1705] A "communication device" is a means for notifying or reporting to relevant organizations or specific users when abnormal behavior is detected.

[1706] "User" refers to a person (e.g., security guard, police officer, parent, etc.) who is responsible for receiving notifications and reports sent from the system and responding promptly.

[1707] The "video output device for display" is a device for displaying abnormal behavior and emotion data to the user, and includes smart glasses and dedicated monitors.

[1708] "Recording and storage" refers to the process of storing detected abnormal behavior and associated video and emotional data for future analysis and relearning.

[1709] The present invention is a system that detects abnormal behavior in real time in a surveillance camera system and combines it with an emotion engine that recognizes user emotions to further enhance security and response to abnormalities. The specific configuration and implementation method of this system are described below.

[1710] System configuration

[1711] This system consists of a surveillance camera, a server, a terminal, a user, and an emotion engine. Each element is as follows:

[1712] 1. Surveillance cameras

[1713] Surveillance cameras capture real-time video data within a specific area. Each surveillance camera is connected to a server via an IP network and transmits the video data in streaming format.

[1714] 2. Server

[1715] The server aggregates and stores video data acquired from surveillance cameras and implements the following functions:

[1716] Acquisition and preprocessing of video data: Acquire real-time video data from surveillance cameras and perform noise removal and resolution adjustment.

[1717] Anomalous behavior detection: Preprocessed video data is input into a generative model to detect anomalous behavior.

[1718] Sentiment analysis: In addition to detecting abnormal behavior, we also use an emotion engine to analyze user emotions.

[1719] Notification and reporting: Based on detected abnormal behavior and emotion data, relevant authorities and specific users are notified and reported.

[1720] Data storage: Record and store anomaly detection results and video data.

[1721] 3. Terminal

[1722] The devices are used by surveillance camera system administrators and security staff, and come with a dedicated app installed to receive notifications and alerts when an abnormality is detected, including the time, location, and emotional data of the abnormality.

[1723] 4. Users

[1724] Users are the parties who receive notifications and reports (e.g., security guards, police officers, guardians, etc.). They are required to respond promptly based on the content of notifications and reports sent from the system.

[1725] 5. Emotion Engine

[1726] The emotion engine uses facial recognition technology to analyze emotions from facial expressions. When an anomaly is detected, the user's emotion data is correlated with abnormal behavior data for analysis.

[1727] Specific examples

[1728] For example, the following measures are taken in response to suspicious behavior in a shopping mall:

[1729] 1. Video acquisition

[1730] The server acquires video data in real time from the surveillance cameras in the shopping mall.

[1731] 2. Preprocessing of video data

[1732] The server performs noise removal and resolution adjustment on the acquired video data and assigns a timestamp to each frame.

[1733] 3. Anomaly Detection and Sentiment Analysis

[1734] The server inputs the preprocessed video data into a generative model to detect suspicious behavior, and also uses an emotion engine to analyze the user's emotions such as surprise and fear.

[1735] 4. Notifications and Reports

[1736] When suspicious behavior is detected, the server immediately notifies the mall's security staff, including the location and time of the suspicious behavior, as well as the user's emotional state.

[1737] 5. History storage and analysis

[1738] The server stores the detection results and emotion data, analyzes them later, and retrains the generative model.

[1739] Prompt Sentence Examples

[1740] Here is an example of input to a generative AI model using the following prompts:

[1741] Develop an application that analyzes real-time video footage from a surveillance camera system, detects abnormal behavior, and recognizes and analyzes user emotions. The application will be for smart glasses, and will immediately notify the smart glasses when abnormal behavior is detected and display the user's emotional data. The server will preprocess the video data and incorporate functions to analyze abnormal behavior and emotions using a generative model.

[1742] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1743] Step 1:

[1744] The server acquires video data from the surveillance cameras in real time. Each surveillance camera is connected to the server via an IP network and continuously transmits video data. The input is the real-time video data transmitted from the surveillance camera, and the output is the raw data stored in the server.

[1745] Step 2:

[1746] The server performs preprocessing on the acquired video data. This preprocessing includes noise removal, resolution adjustment, and frame division. The input is real-time video data, and the output is preprocessed data after data processing such as noise removal, resolution adjustment, and adding a timestamp to each frame.

[1747] Step 3:

[1748] The server inputs the preprocessed video data into a generative AI model to detect abnormal behavior. Based on past learning data, the generative model identifies abnormal behavior (e.g., falls, fleeing, suspicious behavior) in real time. The input is the preprocessed data, and abnormal behavior is identified as data calculation. The output is the frame data in which abnormal behavior is detected and an abnormality flag.

[1749] Step 4:

[1750] The server detects abnormal behavior and simultaneously analyzes the user's emotions using an emotion engine. It uses facial recognition technology to identify emotions (e.g., surprise, fear, anger, etc.) from facial expressions. The input is video data in which abnormal behavior was detected, and emotion identification is performed as part of the data analysis. The output is the user's emotional data.

[1751] Step 5:

[1752] The server immediately notifies and reports to relevant organizations and specific users based on abnormal behavior and emotion data. The notification content includes the time, location, and emotion data of the abnormality. The input is the abnormal behavior detection result and emotion data, and the data output is a notification message that is generated and sent.

[1753] Step 6:

[1754] The server records and stores the detected abnormal behavior and emotion data. All anomaly detection events and associated video data are used for future analysis and retraining of the generative model. The inputs are the anomaly detection results and emotion data, which are recorded in the database as data storage.

[1755] 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.

[1756] 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.

[1757] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1758] 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.

[1759] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1760] 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.

[1761] 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).

[1762] 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.

[1763] 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."

[1764] 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.

[1765] 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).

[1766] 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.

[1767] 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.

[1768] 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.

[1769] 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.

[1770] 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.

[1771] 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.

[1772] 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.

[1773] 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.

[1774] 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.

[1775] 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.

[1776] The following is further disclosed regarding the above embodiment.

[1777] (Claim 1)

[1778] A means for acquiring video data from a surveillance camera in real time;

[1779] means for pre-processing the acquired image data to remove noise and adjust resolution;

[1780] a means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data;

[1781] a means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected;

[1782] A system including a means for recording and storing the abnormal behavior detection results and video data.

[1783] (Claim 2)

[1784] 2. The system of claim 1, wherein the abnormal behavior detection means identifies falls, escapes, and suspicious behavior from video data from a surveillance camera.

[1785] (Claim 3)

[1786] The system of claim 1, further comprising means for analyzing the stored data and retraining the generative model.

[1787] "Example 1"

[1788] (Claim 1)

[1789] A means for acquiring video data from a surveillance camera in real time;

[1790] means for preprocessing the acquired video data to remove noise, adjust resolution, and divide the video data into frames;

[1791] a means for inputting the preprocessed video data into a generative AI model and detecting abnormal behavior based on the learning data;

[1792] A means for immediately notifying and reporting relevant organizations or specific users when abnormal behavior is detected;

[1793] A system that includes a means for recording and storing the abnormal behavior detection results and video data, and for later analysis and retraining of the generating AI model.

[1794] (Claim 2)

[1795] 2. The system of claim 1, wherein the abnormal behavior detection means identifies falls, escapes, and suspicious behavior from video data from a surveillance camera.

[1796] (Claim 3)

[1797] 10. The system of claim 1, further comprising means for analyzing the stored data and retraining the generative AI model.

[1798] "Application Example 1"

[1799] (Claim 1)

[1800] A means for acquiring video data from a surveillance camera in real time;

[1801] means for pre-processing the acquired image data to remove noise and adjust resolution;

[1802] a means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data;

[1803] a means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected;

[1804] a means for recording and storing the abnormal behavior detection result and video data;

[1805] means for sending the notification as a push notification to a mobile device of a relevant organization or a specific user;

[1806] means for generating an audio alarm when the abnormal behavior is detected;

[1807] A system including:

[1808] (Claim 2)

[1809] 2. The system according to claim 1, wherein the abnormal behavior detection means identifies falls, escapes, and suspicious behavior from video data from a surveillance camera, and displays the detected abnormal behavior on a mobile terminal.

[1810] (Claim 3)

[1811] The system according to claim 1, further comprising means for analyzing the stored data and re-training the generative model, and displaying the results of the re-training on a mobile terminal.

[1812] "Example 2: Combining Emotion Engines"

[1813] (Claim 1)

[1814] A means for acquiring video data from a surveillance camera in real time;

[1815] means for pre-processing the acquired image data to remove noise and adjust resolution;

[1816] a means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data;

[1817] means for analyzing the user's emotions using an emotion engine when detecting the abnormal behavior;

[1818] a means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected;

[1819] The system includes a means for recording and storing the abnormal behavior detection results, emotion data, and video data.

[1820] (Claim 2)

[1821] 2. The system of claim 1, wherein the abnormal behavior detection means identifies falls, escapes, and suspicious behavior from video data from a surveillance camera.

[1822] (Claim 3)

[1823] The system of claim 1, further comprising means for analyzing the stored data and retraining the generative model.

[1824] "Application example 2 when combining emotion engines"

[1825] (Claim 1)

[1826] A means for acquiring video data from a surveillance camera in real time;

[1827] means for pre-processing the acquired image data to remove noise and adjust resolution;

[1828] a means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data;

[1829] a means for notifying and reporting the communication device or a specific user when the abnormal behavior is detected;

[1830] means for analyzing the abnormal behavior and the user's emotions;

[1831] means including a video output device for displaying anomaly detection and emotion data;

[1832] A system including a means for recording and storing the abnormal behavior detection results and video data.

[1833] (Claim 2)

[1834] 2. The system according to claim 1, wherein the abnormal behavior detection means identifies falls, escapes, and suspicious behavior from video data from a surveillance camera, and further analyzes the user's emotions of surprise, fear, and anger.

[1835] (Claim 3)

[1836] The system of claim 1, further comprising means for analyzing the stored data and retraining the generative model. [Explanation of symbols]

[1837] 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 video data from a surveillance camera in real time; means for pre-processing the acquired image data to remove noise and adjust resolution; a means for inputting the preprocessed video data into a generative model and detecting abnormal behavior based on the training data; a means for notifying and reporting the abnormal behavior to relevant organizations or specific users when the abnormal behavior is detected; A system including a means for recording and storing the abnormal behavior detection results and video data.

2. The system according to claim 1 , wherein the abnormal behavior detection means identifies falls, escapes, and suspicious behavior from video data from a surveillance camera.

3. The system of claim 1 , further comprising means for analyzing the stored data and retraining a generative model.

Images