Event early warning method, device, equipment and medium

Abstract

The invention provides an event early warning method, device and equipment and a medium, the method is applied to the technical field of communication, and the method comprises the following steps: acquiring scene data for a target object sent by data acquisition equipment, the scene data comprising image data and audio data; performing image recognition on the image data to obtain an image recognition result, and performing semantic recognition on the audio data to obtain a semantic recognition result; determining an event type to which an event indicated by the scene data belongs based on the image recognition result and the semantic recognition result; and based on the event type, event early warning information of the event is generated, the event early warning information is sent to the processing terminal, and the processing terminal intervenes the event based on the received event early warning information. According to the method, the event type judgment accuracy can be improved, and then the event intervention efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and more specifically, to an event warning method, device, equipment and medium in the field of communication technology. Background Art

[0002] In today's campuses, public places or online environments, bullying incidents occur from time to time, causing serious physical and mental harm to victims. Traditional methods of monitoring and intervening in bullying incidents often rely on manual reporting or post-investigation, which has problems such as slow response, limited coverage, and easy omissions. With the development of information technology, existing anti-bullying products mainly use voice recognition and alarms, and then enable security personnel to intervene in time to form a two-way conversation with the scene to prevent the situation from deteriorating. However, the collected audio data is a single data type, which is difficult to fully and accurately reflect the complexity and diversity of bullying incidents. Summary of the invention

[0003] The present application provides an event warning method, device, equipment and medium, which can improve the accuracy of event type judgment and thus improve the efficiency of event intervention.

[0004] In a first aspect, an event warning method is provided, the method comprising:

[0005] Acquire scene data for a target object sent by a data acquisition device, the scene data including image data and audio data; perform image recognition on the image data to obtain an image recognition result, and perform semantic recognition on the audio data to obtain a semantic recognition result; determine the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result; generate event warning information of the event based on the event type, and send the event warning information to a processing terminal, which intervenes in the event based on the received event warning information.

[0006] Through the above technical solution, by simultaneously collecting image data and audio data, and performing image recognition and semantic recognition respectively, the details of bullying incidents can be captured more comprehensively and accurately, and real-time monitoring of bullying behavior can be achieved. This multi-dimensional data fusion analysis method significantly improves the recognition rate and warning accuracy. Based on the image recognition results and semantic recognition results, a scientific bullying incident type confirmation system is constructed. Through the comprehensive analysis of scene data, the severity of bullying incidents can be accurately judged and corresponding incident warning information can be generated. This intelligent classification mechanism helps the processing terminal to take targeted intervention measures according to the event type and improve the intervention effect. The event warning information is sent to the processing terminal in a timely manner, realizing a rapid closed loop from data collection, identification and analysis to warning response. After receiving the warning information, the processing terminal can quickly start the emergency plan and organize relevant personnel to intervene and deal with it, effectively shortening the response time and reducing the harm of bullying incidents to victims.

[0007] In conjunction with the first aspect, in some possible implementations, performing image recognition on image data to obtain an image recognition result includes:

[0008] Perform image behavior recognition on the image data to obtain behavior feature information of the image data, perform face recognition on the image data to obtain face feature information; determine the behavior feature information and face feature information as the image recognition result.

[0009] Through the above technical solution, after performing more detailed image behavior recognition and face recognition on the image data, the obtained behavior feature information and face feature information can further improve the accuracy and comprehensiveness of event monitoring, providing more comprehensive and powerful support for subsequent intervention processing.

[0010] In combination with the first aspect and the above implementations, in some possible implementations, performing image behavior recognition on image data to obtain behavior feature information of the image data includes:

[0011] Target limb feature information matching preset limb feature information is extracted from the image data; facial expression features are extracted from the image data to obtain expression feature information of the image data; and the target limb feature information and expression feature information are determined as behavior feature information.

[0012] Through the above technical solution, by combining body features and facial features, the behavior of the target object in the image can be more comprehensively understood. This comprehensive judgment method is more accurate than relying on body or facial features alone.

[0013] In combination with the first aspect and the above implementation manner, in some possible implementation manners, performing semantic recognition on audio data to obtain a semantic recognition result includes:

[0014] The audio data is converted into text data, and sensitive words are matched on the text data based on a preset sensitive word library; if there are word segments in the text data that match sensitive words in the sensitive word library, language attack feature information of the text data is generated based on the word segmentation, and the language attack feature information is determined as a semantic recognition result.

[0015] Through the above technical solutions, audio data contains rich communication information. By converting it into text data and matching sensitive words, it is possible to capture verbal attacks that may not be directly reflected in image data. This broadens the scope of monitoring and enables the system to more comprehensively identify various manifestations of bullying incidents.

[0016] In combination with the first aspect and the above implementation manner, in some possible implementation manners, determining the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result includes:

[0017] If the target body feature information matches the preset attack behavior feature information, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the first type; if the target body feature information matches the attack behavior feature information, the facial expression feature information matches the preset emotional feature information, and the language attack feature information matches the attack feature information, then the event type of the event indicated by the scene data is determined to be the second type; if the target body feature information matches the attack behavior feature information, the facial expression feature information matches the preset emotional feature information, the face indicated by the facial feature information is recognized at least twice within a preset time period, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the third type.

[0018] Through the above technical solution, by combining information from multiple dimensions (body language, facial expressions, language and face), the system can more comprehensively assess the nature of the event and reduce misjudgments caused by insufficient or incorrect single information sources. Accurate classification: According to different combination conditions, the event is subdivided into different types (for example, the first event type, the second event type, and the third event type), which helps to handle and respond more accurately in the future.

[0019] In combination with the first aspect and the above implementations, in some possible implementations, generating event warning information of an event based on the event type includes:

[0020] Obtain warning suggestion information that matches the event type from the preset warning template library; generate event warning information corresponding to the event based on the image recognition results, semantic recognition results and warning suggestion information.

[0021] Through the above technical solution, through the preset warning template library, the system can quickly match the warning suggestion information corresponding to the event type, reducing the time of manual judgment and improving the response speed.

[0022] In combination with the first aspect and the above implementation manner, in some possible implementation manners, the method further includes:

[0023] Acquire facial data and a terminal identifier of a processing terminal, wherein the face indicated by the facial data includes the face in the image data; and establish an association relationship between the facial data and the corresponding processing terminal based on the terminal identifier;

[0024] Send event warning information to the processing terminal, including:

[0025] Based on the association relationship and facial feature information, the event warning information is sent to the corresponding processing terminal.

[0026] Through the above technical solution, by establishing the association between face data and processing terminals, it can be ensured that when an event related to a specific face is detected, the warning information can be accurately sent to the processing terminal associated with the face. This avoids the mis-sending or missing of the warning information and improves the accuracy and reliability of the warning system.

[0027] In a second aspect, an event warning device is provided, the device comprising:

[0028] A data acquisition unit, used to acquire scene data for a target object sent by a data acquisition device, wherein the scene data includes image data and audio data;

[0029] The recognition unit is used to perform image recognition on the image data to obtain an image recognition result, and perform semantic recognition on the audio data to obtain a semantic recognition result;

[0030] An event type determination unit, used to determine the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result;

[0031] The information generating unit is used to generate event warning information of the event based on the event type, and send the event warning information to the processing terminal, and the processing terminal intervenes in the event based on the received event warning information.

[0032] According to a third aspect, an electronic device is provided, including:

[0033] A memory for storing executable program codes;

[0034] A processor is used to call and run executable program code from a memory, so that the electronic device executes the method in the above-mentioned first aspect or any possible implementation of the first aspect, or executes the method in the second aspect or any possible implementation of the second aspect.

[0035] In a fourth aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation of the first aspect.

[0036] In a fifth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the above-mentioned first aspect or any possible implementation manner of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a scenario diagram of an event warning method provided in an embodiment of the present application;

[0038] Figure 2 It is a flow chart of an event warning method provided in an embodiment of the present application;

[0039] Figure 3 It is a flow chart of an event warning method provided in an embodiment of the present application;

[0040] Figure 4 It is a flow chart of an event warning method provided in an embodiment of the present application;

[0041] Figure 5 It is a schematic diagram of the platform architecture decomposition of an event warning method provided in an embodiment of the present application;

[0042] Figure 6 It is a structural schematic diagram of an event warning device provided in an embodiment of the present application;

[0043] Figure 7 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application;

[0044] Figure 8 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The technical solution in the present application will be described clearly and in detail below in conjunction with the accompanying drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.

[0046] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0047] Figure 1 It is a scenario diagram of an event warning method provided in an embodiment of the present application.

[0048] For example, existing anti-bullying products mainly use voice recognition and alarm, and then enable security personnel to intervene in time to establish a two-way conversation with the scene to prevent the situation from deteriorating. However, the collected audio data is a single data type, which is difficult to fully and accurately reflect the complexity and diversity of bullying incidents, resulting in low accuracy in event type determination and inability to achieve precise intervention in events.

[0049] Based on the above situation, the present application embodiment proposes an event warning method, which can be Figure 1 The electronic device 100 in the apparatus obtains scene data for a target object sent by a data acquisition device 200, wherein the scene data includes image data and audio data; performs image recognition on the image data to obtain an image recognition result, and performs semantic recognition on the audio data to obtain a semantic recognition result; determines the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result; generates event warning information of the event based on the event type, and sends the event warning information to a processing terminal, and the processing terminal intervenes in the event based on the received event warning information.

[0050] By simultaneously collecting image data and audio data, and performing image recognition and semantic recognition respectively, this method can capture the details of bullying incidents more comprehensively and accurately, and realize real-time monitoring of bullying behavior. This multi-dimensional data fusion analysis method significantly improves the recognition rate and warning accuracy. Based on the image recognition results and semantic recognition results, a scientific bullying incident type confirmation system is constructed. Through the comprehensive analysis of scene data, the severity of bullying incidents can be accurately judged and corresponding incident warning information can be generated. This intelligent classification mechanism helps the processing terminal to take targeted intervention measures according to the event type and improve the intervention effect. The event warning information is sent to the processing terminal in a timely manner, realizing a rapid closed loop from data collection, identification and analysis to warning response. After receiving the warning information, the processing terminal can quickly activate the emergency plan and organize relevant personnel to intervene and deal with it, effectively shortening the response time and reducing the harm of bullying incidents to victims.

[0051] based on Figure 1 The scene shown is shown below. Figure 2-Figure 4 , the event warning method provided in the embodiment of the present application is introduced in detail.

[0052] See also Figure 2 , which is a flow chart of an event warning method provided in an embodiment of the present application. Figure 2 As shown, the method of the embodiment of the present application may include the following steps S101-S105.

[0053] S101, acquiring scene data for a target object sent by a data acquisition device, where the scene data includes image data and audio data;

[0054] Data acquisition equipment refers to equipment used to collect and record information in the physical world and convert it into image data and audio data for further processing and analysis, such as cameras, microphones, and other equipment.

[0055] Among them, the target object refers to the specific person that the data collection device pays attention to, records and monitors. The target object can include one person or multiple people, without specific limitation.

[0056] Among them, scene data is a collection of comprehensive information about the target object and its surrounding environment captured by the data acquisition device at a specific time, place and conditions.

[0057] Image data refers to the visual information of the target object and its surrounding environment captured by the data acquisition device. It can be a static picture or a continuous video stream.

[0058] Among them, audio data refers to the sound information recorded by data acquisition equipment. It reflects the sound characteristics of the target object and its surrounding environment, such as human voice, environmental noise, specific sound events, etc.

[0059] S102, performing image recognition on the image data to obtain an image recognition result, and performing semantic recognition on the audio data to obtain a semantic recognition result;

[0060] Image recognition refers to a variety of recognition tasks performed on image data, including image behavior recognition, expression recognition, and face recognition.

[0061] In one embodiment, the image recognition result may include: behavior feature information corresponding to image behavior recognition performed on the image data, and face feature information corresponding to face recognition performed on the image data.

[0062] The behavior feature information may include target body feature information and expression feature information. The target body feature information may represent the body movements of the target person in the image data. The expression feature information may represent the expression of the target person in the image data.

[0063] Semantic recognition refers to the process of in-depth understanding and analysis of text data, aiming to extract semantic information such as the meaning, intention or emotion behind the text. For example, semantic recognition can be performed on audio data, such as converting audio data into text data, matching sensitive words on the text data based on a preset sensitive word library, and generating language attack feature information of the text data based on the word segmentation if there are word segments in the text data that match sensitive words in the sensitive word library. The language attack feature information obtained is the semantic recognition result of the audio data.

[0064] S103, determining the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result;

[0065] The image recognition result and the semantic recognition result are used as the basis for determining the severity of the event indicated by the scene data. Then, the event type is divided according to the severity, for example, into the first event type, the second event type, the third event type, etc.

[0066] In one embodiment, image behavior recognition is performed on the image data to obtain behavior feature information of the image data (such as target body feature information and expression feature information). Face recognition is performed on the image data to obtain face feature information of the image data. Language attack feature information of the audio data is extracted. If the target limb feature information matches the preset attack behavior feature information, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the first type, and the event corresponding to the first event type can be determined as a violent event; if the target limb feature information matches the attack behavior feature information, the expression feature information matches the preset emotional feature information, and the language attack feature information matches the attack feature information, then the event type of the event indicated by the scene data is determined to be the second type, and the event corresponding to the second event type can be determined as a bullying embryo event; if the target limb feature information matches the attack behavior feature information, the expression feature information matches the preset emotional feature information, the face indicated by the facial feature information is recognized at least twice within a preset time period, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the third type, and the third event type can be determined as a standard bullying event.

[0067] S104: Generate event warning information of the event based on the event type, send the event warning information to the processing terminal, and the processing terminal intervenes in the event based on the received event warning information.

[0068] Among them, event warning information refers to the warning information about the event generated and sent to the processing terminal after the event type is determined based on the analysis of event-related scene data, such as image recognition and semantic recognition. The event warning information generated by different event types is different. This information is intended to remind the processing terminal to pay attention and take necessary intervention measures to prevent further escalation or recurrence of the event.

[0069] Among them, processing terminals refer to those terminal devices that are used to receive event warning information and intervene in events based on the received information. These terminal devices may include office computers, mobile phones, and consoles of security monitoring centers of school administrators. When these processing terminals receive warning information of events, they will trigger corresponding response mechanisms, such as notifying relevant personnel to go to the scene for processing, launching emergency plans, and recording event details, to ensure that the events are handled in a timely and effective manner.

[0070] In an embodiment of the present application, by simultaneously collecting image data and audio data, and performing image recognition and semantic recognition respectively, the details of the bullying incident can be captured more comprehensively and accurately, and real-time monitoring of bullying behavior can be achieved. This multi-dimensional data fusion analysis method significantly improves the recognition rate and warning accuracy. Based on the image recognition results and semantic recognition results, a scientific bullying incident type confirmation system is constructed. Through the comprehensive analysis of scene data, the severity of the bullying incident can be accurately judged and the corresponding event warning information can be generated. This intelligent classification mechanism helps the processing terminal to take targeted intervention measures according to the event type and improve the intervention effect. The event warning information is sent to the processing terminal in a timely manner, realizing a rapid closed loop from data collection, identification analysis to warning response. After receiving the warning information, the processing terminal can quickly start the emergency plan and organize relevant personnel to intervene and deal with it, effectively shortening the response time and reducing the harm of bullying incidents to victims.

[0071] See also Figure 3 , which is a flow chart of an event warning method provided in an embodiment of the present application. Figure 3 As shown, the method of the embodiment of the present application may include the following steps S201-S209.

[0072] S201, acquiring scene data for a target object sent by a data acquisition device, where the scene data includes image data and audio data;

[0073] For details, please refer to step S101 in the above-mentioned embodiment of the specification, which will not be described in detail here.

[0074] S202, performing image behavior recognition on the image data to obtain behavior feature information of the image data, performing face recognition on the image data to obtain face feature information, and determining the behavior feature information and the face feature information as the image recognition result;

[0075] Among them, the behavior feature information refers to the information about the behavior pattern of the target object identified and analyzed from the image data. The behavior feature information may include: target limb feature information and expression feature information. The target limb feature information may include: action type: identifying the action being performed by the target object in the image, such as walking, running, jumping, waving, hugging, etc.; motion trajectory: the motion trajectory of the target object's action can be further analyzed to understand the direction, speed and acceleration of the action; posture analysis: locating the joints and parts of the target object's body, analyzing the posture and position of the body, such as standing, sitting, lying down, etc.; interactive behavior: identifying the interactive behavior between the characters in the target object in the image, such as shaking hands, talking, passing objects, fighting, etc. Expression feature information may include: feature information representing happiness, feature information representing sadness, feature information representing anger, feature information representing surprise, feature information representing fear, feature information representing pain, feature information representing disgust, etc.

[0076] In one embodiment, target limb feature information matching preset limb feature information can be extracted from the image data. Face expression feature extraction is performed on the image data to obtain expression feature information of the image data. The target limb feature information and expression feature information are determined as behavior feature information.

[0077] In one embodiment, the behavioral features in the image data can be automatically learned by training the Long Short-Term Memory networks (LSTM), and classified and identified to obtain the behavioral feature information of the image data. The key points (such as joints) of the characters in the image data can also be identified by the posture estimation algorithm, and then the behavior of the characters can be analyzed based on the motion trajectory of these key points to obtain the behavioral feature information of the image data. Image behavior recognition can capture and analyze the specific behaviors of the characters in the image data, such as pushing, beating, snatching, etc., which are typical behaviors in bullying incidents. By extracting the behavioral feature information, the system can more accurately determine whether bullying has occurred in the scene, as well as the nature and intensity of the behavior.

[0078] Among them, expression feature information refers to the emotional state reflected by the facial expressions of people identified and analyzed from image data. This information can include: basic emotions: such as happiness, sadness, anger, surprise, pain, fear, disgust, etc.; emotional intensity: indicates the intensity of emotional expression; emotional combination: sometimes facial expressions may contain a mixture of multiple emotions at the same time, and it is necessary to identify and quantify the combination of these emotions; micro-expressions: short and imperceptible changes in facial expressions, which may reveal deeper emotions or intentions.

[0079] In one embodiment, a convolutional neural network can be used to automatically learn the features of facial expressions through training with a large number of labeled expression data sets to obtain emotional feature information of image data. Expression recognition provides emotional state information of the characters in the image, such as anger, fear, sadness, pain, etc. In bullying incidents, the emotional state of the victim and the perpetrator is often an important basis for judging the nature and severity of the incident. Through the analysis of emotional feature information, the background of the incident can be more fully understood, providing a more scientific basis for the classification of incident levels.

[0080] Among them, facial feature information refers to specific biometric information about a person's face extracted from image data. This information usually includes: facial contour: describes the overall shape and contour of the face; facial features: including the position, size, shape, etc. of the eyes, nose, mouth, ears, etc.; texture features: such as skin texture, wrinkles, moles and other detailed features; key points: key points on the face, such as the corners of the eyes, corners of the mouth, nose tip, etc., for precise positioning and identification; three-dimensional information: including the depth information and geometric structure of the face.

[0081] In one embodiment, the image data including the face of a person can be matched with a pre-stored face template to determine the identity. Face recognition technology can quickly identify the identity information of a person from an image, which is of great significance for subsequent intervention and accountability. At the same time, combined with face recognition technology, key figures in bullying incidents can also be continuously tracked to ensure the continuity and effectiveness of incident handling.

[0082] In an embodiment of the present application, after performing more detailed image behavior recognition and face recognition on the image data, the obtained behavior feature information and face feature information can further improve the accuracy and comprehensiveness of event monitoring, providing more comprehensive and powerful support for subsequent intervention processing.

[0083] S203, converting the audio data into text data, performing sensitive word matching on the text data based on a preset sensitive word library, and if there is a word segmentation in the text data that matches a sensitive word in the sensitive word library, generating language attack feature information of the text data based on the word segmentation, and determining the language attack feature information as a semantic recognition result;

[0084] Among them, the conversion of audio data into text data can be achieved using speech recognition (Automatic Speech Recognition, ASR) technology. Then, a series of rules are defined to identify the language attack features in the text. This method requires the manual construction of a rule base, for example, building a preset sensitive word library and constantly updating it to adapt to new forms of language attacks. By using a large amount of annotated language attack data to train a machine learning model (such as a support vector machine, a neural network, etc.), it can automatically learn and identify the language attack features in the text, so that when the text data is input into the machine learning model, the language attack feature information of the text data can be output. Text data is easier to process and analyze automatically than audio data. Through semantic recognition technology, the system can deeply analyze the text content and identify language attack features such as insults, threats, and ridicule. This text-based analysis method is usually more accurate and reliable than directly analyzing the audio.

[0085] In the embodiment of the present application, the audio data contains rich communication information. By converting it into text data and matching sensitive words, it is possible to capture verbal attacks that may not be directly reflected by the image data. This broadens the monitoring scope and enables the system to more comprehensively identify various manifestations of bullying incidents.

[0086] S204: if the target body feature information matches the preset attack behavior feature information, and the language attack feature information matches the preset attack feature information, then determining that the event type of the event indicated by the scene data is the first type;

[0087] If the target's limb feature information matches the preset attack behavior feature information, it means that the target person in the image data has violent, threatening, aggressive or provocative actions. For example, these actions may include: limb movements: such as clenching fists, swinging fists, kicking legs, pushing and other violent or aggressive actions; postures: certain postures may imply attack intentions, such as bending over to prepare for a collision, leaning forward when standing, etc.; object use: using weapons or tools (such as knives, sticks) to threaten or attack; multi-person interaction: in group scenes, aggressive behaviors such as siege and suppression may be identified.

[0088] Among them, if the language attack feature information matches the preset attack feature information, it means that the person in the audio data contains insulting, abusive, and malicious and aggressive words.

[0089] S205: if the target body feature information matches the attack behavior feature information, the expression feature information matches the preset emotion feature information, and the language attack feature information matches the attack feature information, then the event type of the event indicated by the scene data is determined to be the second type;

[0090] The preset emotional characteristic information is a predefined characteristic information. The preset emotional characteristic information can represent the characteristics of the character experiencing a negative emotional state, such as pain, fear and other emotion-related characteristics.

[0091] In one embodiment, the target's limb feature information, emotional feature information, and verbal attack feature information can be integrated to confirm that these feature information do contain attack behavior feature information, emotional feature information, and attack feature information. According to the preset rules, the event corresponding to this combination is determined to be the second event type, indicating the occurrence of a more urgent situation. The events corresponding to the second event type are more urgent than the events corresponding to the first event type. Next, early warning information for the event corresponding to the event type can be constructed, such as clearly and accurately describing the content of the event, including the people involved, time, place, and key information such as the detected attack behavior, painful emotions, and abusive words. The event type of the event is clearly marked as the second event type, indicating that it has a higher urgency and importance. Preliminary response measures or suggestions can be provided based on the type and severity of the event.

[0092] S206: If the target body feature information matches the attack behavior feature information, the expression feature information matches the preset emotion feature information, the face indicated by the face feature information is recognized at least twice within the preset time period, and the language attack feature information matches the preset attack feature information, then determine that the event type of the event indicated by the scene data is the third type;

[0093] The preset time period refers to a predefined time length, which indicates a time period in the past. Recognizing the repeated appearance of a face within the preset time period can more accurately determine that the event may be continuous or repetitive.

[0094] In one embodiment, it is confirmed that the scene data contains target limb feature information, emotional feature information, and the face indicated by the facial feature information is recognized at least twice within a preset time period, and it is confirmed that the scene data contains verbal attack feature information. Combined with the preset rules, the event corresponding to this combination can be determined to be the third event type. The content of the event can be clearly and accurately described, including the time, location, people involved, specific manifestations of the attack behavior, emotional state, facial features, and abusive words. The event type of the event is clearly marked as the third event type, emphasizing its urgency and importance. The events corresponding to the third event type are more urgent than the events corresponding to the second event type. According to the type and severity of the event, specific response measures or suggestions are provided.

[0095] In the embodiment of the present application, by combining information from multiple dimensions (body language, facial expressions, language, and face), the system can more comprehensively assess the nature of the event and reduce misjudgments caused by insufficient or incorrect single information sources. Accurate classification: According to different combination conditions, events are subdivided into different types (e.g., first event type, second event type, third event type), which helps to process and respond more accurately in the future.

[0096] S207, obtaining warning suggestion information matching the event type from a preset warning template library, and generating event warning information corresponding to the event based on the image recognition result, the semantic recognition result and the warning suggestion information;

[0097] The preset warning template library refers to a database or collection that pre-defines and stores a variety of different types of events and their corresponding warning advice information. This library is designed to quickly respond to and generate warning information for specific types of events to improve the efficiency and accuracy of emergency response.

[0098] In one embodiment, event warning information corresponding to events of each event type can be generated according to the following steps. Data integration: Integrate image recognition results (e.g., target limb feature information, expression feature information, face feature information) and audio recognition results (e.g., language attack feature information) to confirm that the two occur at the same timestamp or within a similar time range; Event determination: According to preset rules, determine that preset combinations constitute different event types; Information construction: Briefly describe the content of the event, such as "aggressive behavior and abusive language detected." Record the time and place of the event. Clearly mark it as a first-level event, indicating urgency. Provide preliminary response measures or suggestions based on the event type.

[0099] In the embodiment of the present application, through the preset warning template library, the system can quickly match the warning suggestion information corresponding to the event type, reduce the time of manual judgment, and improve the response speed.

[0100] S208, acquiring facial data and a terminal identifier of a processing terminal, wherein the face indicated by the facial data includes the face in the image data, and establishing an association relationship between the facial data and the corresponding processing terminal based on the terminal identifier;

[0101] The facial data may be image data uploaded manually, such as facial images of students and teachers uploaded by a school. The facial data may be facial images collected by a data collection device, etc.

[0102] In one embodiment, the terminal identification of the processing terminal may be collected, such as the ID, MAC address, serial number or other information that can uniquely identify the terminal. A database or appropriate data structure, such as a hash table, a relational database table, etc., is used to store and manage the facial data and the identification of the corresponding processing terminal, so as to establish an association relationship between the facial data and the corresponding processing terminal.

[0103] S209: Send event warning information to a corresponding processing terminal based on the association relationship and facial feature information.

[0104] In one embodiment, when an event (e.g., an event corresponding to the first event type, an event corresponding to the second event type, or an event corresponding to the third event type) is detected and event warning information needs to be sent, facial feature information can be used as a query condition to search for a processing terminal that has an association relationship with the facial data indicated by the facial feature information in an established database or data structure. After the processing terminal is determined, the event warning information is sent to the processing terminal. This allows the processing terminal to respond and process in a timely manner. The sending time and receiving status of the event warning information can also be recorded, and the subsequent processing process and results can be tracked to ensure that the event is properly resolved.

[0105] In the embodiment of the present application, by establishing an association relationship between face data and a processing terminal, it can be ensured that when an event related to a specific face is detected, the warning information can be accurately sent to the processing terminal associated with the face. This avoids the mis-sending or missing of the warning information and improves the accuracy and reliability of the warning system.

[0106] See also Figure 4 , is a flow chart of an event warning method provided in an embodiment of the present application, such as Figure 4 As shown, in the embodiment of the present application, S202, image behavior recognition is performed on the image data to obtain behavior feature information of the image data, which may specifically include the following steps S301-S302.

[0107] S301, extracting target limb feature information matching preset limb feature information from the image data;

[0108] In one embodiment, a series of limb feature information may be defined or preset, and the feature information may include specific action types (such as walking, running, jumping, etc.), motion trajectories (such as the direction, speed and acceleration of the action), posture analysis (such as standing, sitting, lying down, etc.), and interactive behaviors (such as shaking hands, talking, passing objects, fighting, etc.). The input image data may be processed, and image recognition technology (such as deep learning, computer vision algorithms, etc.) may be used to detect human objects in the image, and based on a series of preset limb feature information, their limb feature information may be extracted to obtain target limb feature information to determine whether the human object in the image is performing a preset action or showing a certain posture.

[0109] S302, extracting facial expression features from the image data to obtain expression feature information of the image data, and determining the target limb feature information and expression feature information as behavior feature information.

[0110] In one embodiment, face detection and recognition technology can be used to locate and identify faces from image data. Expression recognition algorithms (such as deep learning-based methods) can be used to analyze facial expression features, such as the position of eyebrows, the degree of eye opening and closing, the shape of the corners of the mouth, etc., which can reflect the emotional state of humans. By analyzing these expression features, expression feature information can be generated, which describes the emotional expression of the person in the image, such as happiness, sadness, anger, pain, etc.

[0111] In the embodiment of the present application, by combining body features and facial features, the behavior of the target object in the image can be more comprehensively understood. This comprehensive judgment method is more accurate than relying on body features or facial features alone.

[0112] The following will be combined Figure 5 The event warning method provided in the embodiment of the present application is described in detail with reference to a specific example.

[0113] like Figure 5As shown, a communication connection between a data acquisition device (such as a camera, a microphone, etc.) and a student bullying prevention platform can be established. Before collecting scene data of student activities within the school scope through the data acquisition device, the student file and the terminal identification of the corresponding responsible teacher can be obtained first. An association relationship between the face image in the student file and the terminal identification of the processing terminal of the corresponding responsible teacher is established to generate a face base library. The responsible teacher can be the head teacher, subject teacher, etc. of the relevant students. Through the establishment of the face base library, when a certain event (for example, an event corresponding to the first event type, an event corresponding to the second event type, and an event corresponding to the third event type) is detected and an event warning information needs to be sent, the corresponding processing terminal can be found according to the association relationship, so that the event can be intervened by the corresponding responsible teacher. When collecting scene data of student activities within the school scope through the data acquisition device, the scene data may include image data and audio data. The data acquisition device sends the collected image data and audio data to the access layer of the student bullying prevention platform. Image behavior recognition, expression recognition and face recognition are performed on the image data at the access layer to obtain behavior feature information, emotional feature information and face feature information respectively. Perform semantic recognition on audio data at the access layer to obtain language attack feature information.

[0114] If the behavior characteristic information includes attack behavior characteristic information (the attack behavior characteristic information indicates that the students in the collected image data have attack behavior), and the language attack characteristic information includes insult characteristic information (the insult characteristic information indicates that the students in the collected audio data have insult behavior), then the event type of the event indicated by the scene data is determined to be the first event type (that is, the combination of attack behavior and insult behavior will be determined as a violent event). Event warning information corresponding to the event of the first event type can be generated. According to the association relationship established previously, the event warning information is sent to the corresponding processing terminal. After receiving the event warning information, the processing terminal intervenes in the event corresponding to the first event type, for example, pays attention to confirmation and focuses on prevention.

[0115] If the behavior characteristic information includes attack behavior characteristic information (the attack behavior characteristic information indicates that the students in the collected image data have attack behavior), the emotional characteristic information includes preset characteristic information (the preset characteristic information indicates that the students in the collected image data have painful expressions on their faces), and the verbal attack characteristic information includes insult characteristic information (the insult characteristic information indicates that the students in the collected audio data have insulting behavior), then the event type of the event indicated by the scene data is determined to be the second event type (that is, the combination of attack behavior, insult behavior, and facial painful expression will be determined as a bullying budding event). An event warning level corresponding to the event of the second event type can be generated. According to the previously established association relationship, the event warning information is sent to the corresponding processing terminal. After receiving the event warning information, the processing terminal intervenes in the event corresponding to the second event type, such as targeted intervention.

[0116] If the behavior feature information includes attack behavior feature information (the attack behavior feature information indicates that the student in the collected image data has an attack behavior), the emotion feature information includes preset feature information (the preset feature information indicates that the student in the collected image data has a painful expression on his face), and the language attack feature information includes insult feature information (the insult feature information indicates that the student in the collected audio data has an insulting behavior), and the face indicated by the facial feature information is recognized at least twice within a preset time period, then the event type of the event indicated by the scene data is determined to be a third event type (that is, the combination of attack behavior, insulting behavior, facial painful expression, and the same face being recognized multiple times will be determined as a standard bullying event). An event warning level corresponding to the event of the third event type can be generated. According to the previously established association relationship, the event warning information is sent to the corresponding processing terminal. After receiving the event warning information, the processing terminal intervenes in the event corresponding to the third event type, such as preventing deterioration and providing follow-up counseling.

[0117] In an embodiment of the present application, by simultaneously collecting image data and audio data, and performing image recognition and semantic recognition respectively, the details of the bullying incident can be captured more comprehensively and accurately, and real-time monitoring of bullying behavior can be achieved. This multi-dimensional data fusion analysis method significantly improves the recognition rate and warning accuracy. Based on the image recognition results and semantic recognition results, a scientific bullying incident type confirmation system is constructed. Through the comprehensive analysis of scene data, the severity of the bullying incident can be accurately judged and the corresponding event warning information can be generated. This intelligent classification mechanism helps the processing terminal to take targeted intervention measures according to the event type and improve the intervention effect. The event warning information is sent to the processing terminal in a timely manner, realizing a rapid closed loop from data collection, identification analysis to warning response. After receiving the warning information, the processing terminal can quickly start the emergency plan and organize relevant personnel to intervene and deal with it, effectively shortening the response time and reducing the harm of bullying incidents to victims.

[0118] based on Figure 1 The following is a schematic diagram of the scene. Figure 6 , the event warning device provided in the embodiment of the present application is introduced in detail. It should be noted that, Figure 6 The event warning device in the present application is used to execute Figure 2-Figure 5 For the convenience of explanation, only the part related to the embodiment of the present application is shown. For the specific technical details not disclosed, please refer to the present application. Figure 2-Figure 5 The embodiment shown.

[0119] See also Figure 6 , which is a schematic diagram of the structure of an event warning device provided in an embodiment of the present application. Figure 6 As shown, the event warning device 1 of the embodiment of the present application may include: a data acquisition unit 11, an identification unit 12, an event type determination unit 13 and an information generation unit 14.

[0120] The data acquisition unit 11 is used to acquire scene data for the target object sent by the data acquisition device, and the scene data includes image data and audio data;

[0121] The recognition unit 12 is used to perform image recognition on the image data to obtain an image recognition result, and perform semantic recognition on the audio data to obtain a semantic recognition result;

[0122] An event type determination unit 13, configured to determine the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result;

[0123] The information generating unit 14 is used to generate event warning information of the event based on the event type, and send the event warning information to the processing terminal, and the processing terminal intervenes in the event based on the received event warning information.

[0124] Optionally, the identification unit 12 includes:

[0125] The information recognition subunit 121 is used to perform image behavior recognition on the image data to obtain behavior feature information of the image data, and perform face recognition on the image data to obtain face feature information;

[0126] The result determination subunit 122 is used to determine the behavior feature information and the face feature information as the image recognition result.

[0127] Optionally, the recognition unit 12 is also used to extract target limb feature information that matches preset limb feature information from the image data; extract facial expression features from the image data to obtain expression feature information of the image data; and determine the target limb feature information and expression feature information as behavioral feature information.

[0128] Optionally, the identification unit 12 includes:

[0129] The data conversion subunit 123 is used to convert the audio data into text data, and perform sensitive word matching on the text data based on a preset sensitive word library;

[0130] The information generating subunit 124 is used to generate language attack feature information of the text data based on the word segmentation if there is a word segmentation matching the sensitive word in the sensitive word library in the text data, and determine the language attack feature information as the semantic recognition result.

[0131] Optionally, the event type determination unit 13 includes:

[0132] A first event type determination subunit 131, configured to determine that the event type of the event indicated by the scene data is a first type if the target body feature information matches the preset attack behavior feature information, and the language attack feature information matches the preset attack feature information;

[0133] The second event type determination subunit 132 is used to determine that the event type of the event indicated by the scene data is a second type if the target body feature information matches the attack behavior feature information, the expression feature information matches the preset emotion feature information, and the language attack feature information matches the attack feature information;

[0134] The third event type determination subunit 133 is used to determine that the event type of the event indicated by the scene data is the third type if the target limb feature information matches the attack behavior feature information, the expression feature information matches the preset emotion feature information, the face indicated by the face feature information is recognized at least twice within a preset time period, and the language attack feature information matches the preset attack feature information.

[0135] Optionally, the information generating unit 14 includes:

[0136] The information acquisition subunit 141 is used to acquire the warning suggestion information matching the event type from the preset warning template library;

[0137] The information generation subunit 142 is used to generate event warning information corresponding to the event based on the image recognition result, the semantic recognition result and the warning suggestion information.

[0138] Optionally, the information generating unit 14 is further used to obtain facial data and a terminal identifier of a processing terminal, where the face indicated by the facial data includes the face in the image data; and based on the terminal identifier, establish an association relationship between the facial data and the corresponding processing terminal.

[0139] Optionally, the information generating unit 14 includes:

[0140] The information sending subunit 143 is used to send the event warning information to the corresponding processing terminal based on the association relationship and facial feature information.

[0141] In the embodiments of the present application, the beneficial effects of the solutions are described in the introduction of the method embodiments and will not be repeated here.

[0142] See also Figure 7 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 7 As shown, the electronic device 500 includes a processor 501 and a memory 502. The processor 501 is electrically connected to the memory 502.

[0143] The processor 501 is the control center of the electronic device 500 and may include one or more processing cores. The processor 501 uses various interfaces and lines to connect various parts of the entire electronic device 500, and executes various functions and processes data of the electronic device 500 by running or calling computer programs stored in the memory 502, and calling data stored in the memory 502, so as to control the electronic device 500 as a whole. Optionally, the processor 501 can be implemented in at least one hardware form of digital signal processing (DSP), field programmable gate array (FPGA), and programmable logic array (PLA). The processor 501 can integrate one or a combination of CPU, graphics processing unit (GPU), modem, etc. Among them, the CPU mainly processes the operating system, user pages, and applications; the GPU is responsible for rendering and drawing display content; and the modem is used to process wireless communications. It is understandable that the above-mentioned modem may not be integrated into the processor 501, but may be implemented separately through a communication chip.

[0144] The memory 502 can be used to store software programs and modules, and the processor 501 executes various functional applications and data processing by running the computer programs and modules stored in the memory 502. The memory 502 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, a computer program required for at least one function, etc.; the data storage area can store data created according to the use of the electronic device 500, etc.

[0145] In addition, the memory 502 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory 502 may also include a memory controller to provide the processor 501 with access to the memory 502.

[0146] In this embodiment, the processor 501 in the electronic device 500 loads instructions corresponding to the processes of one or more computer programs into the memory 502 according to the following steps, and the processor 501 runs the computer program stored in the memory 502 to implement various functions, as follows:

[0147] Acquire scene data for a target object sent by a data acquisition device, the scene data including image data and audio data; perform image recognition on the image data to obtain an image recognition result, and perform semantic recognition on the audio data to obtain a semantic recognition result; determine the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result; generate event warning information of the event based on the event type, and send the event warning information to a processing terminal, which intervenes in the event based on the received event warning information.

[0148] Optionally, when the processor 501 performs image recognition on the image data and obtains the image recognition result, it specifically executes:

[0149] Perform image behavior recognition on the image data to obtain behavior feature information of the image data, perform face recognition on the image data to obtain face feature information; determine the behavior feature information and face feature information as the image recognition result.

[0150] Optionally, when the processor 501 performs semantic recognition on the audio data and obtains a semantic recognition result, the processor 501 specifically performs:

[0151] Target limb feature information matching preset limb feature information is extracted from the image data; facial expression features are extracted from the image data to obtain expression feature information of the image data; and the target limb feature information and expression feature information are determined as behavior feature information.

[0152] Optionally, when the processor 501 performs semantic recognition on the audio data and obtains a semantic recognition result, the processor 501 specifically performs:

[0153] The audio data is converted into text data, and sensitive words are matched on the text data based on a preset sensitive word library; if there are word segments in the text data that match sensitive words in the sensitive word library, language attack feature information of the text data is generated based on the word segmentation, and the language attack feature information is determined as a semantic recognition result.

[0154] Optionally, when the processor 501 determines the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result, the processor 501 specifically performs:

[0155] If the target body feature information matches the preset attack behavior feature information, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the first type; if the target body feature information matches the attack behavior feature information, the facial expression feature information matches the preset emotional feature information, and the language attack feature information matches the attack feature information, then the event type of the event indicated by the scene data is determined to be the second type; if the target body feature information matches the attack behavior feature information, the facial expression feature information matches the preset emotional feature information, the face indicated by the facial feature information is recognized at least twice within a preset time period, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the third type.

[0156] Optionally, when the processor 501 generates event warning information of an event based on the event type, it specifically performs:

[0157] Obtain warning suggestion information that matches the event type from the preset warning template library; generate event warning information corresponding to the event based on the image recognition results, semantic recognition results and warning suggestion information.

[0158] Optionally, before executing the step of sending the event warning information to the processing terminal, the processor 501 may further execute:

[0159] Based on the association relationship and facial feature information, the event warning information is sent to the corresponding processing terminal.

[0160] In the embodiments of the present application, the beneficial effects of the solutions are described in the introduction of the method embodiments and will not be repeated here.

[0161] See also Figure 8 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 8 As shown, the electronic device 500 includes: a processor 501, a memory 502, a display screen 503, a camera component 504, an audio circuit 505, a sensor 506 and a power supply 507. The processor 501 is electrically connected to the display 503, the camera component 504, the audio circuit 505, the sensor 506 and the power supply 507 respectively.

[0162] The display screen 503 may be used to display information input by a user or information provided to a user and various graphical user interfaces of the electronic device 500 . These graphical user interfaces may be composed of images, texts, icons, videos and any combination thereof.

[0163] The camera assembly 504 may include an image processing circuit, which may be implemented using hardware and / or software components and may include various processing units that define an image signal processing pipeline. The image processing circuit may include at least: multiple cameras, an image signal processor (ISP processor), a control logic, and an image memory, etc. Each camera may include at least one or more lenses and an image sensor. The image sensor may include a color filter array (such as a Bayer filter). The image sensor may obtain light intensity and wavelength information captured by each imaging pixel of the image sensor, and provide a set of raw image data that can be processed by an image signal processor.

[0164] The audio circuit 505 can be used to provide an audio interface between the user and the electronic device 500 through a speaker and a microphone. The audio circuit 505 includes a microphone. The microphone is electrically connected to the processor 501. The microphone is used to receive voice information input by the user.

[0165] The sensor 506 is used to collect information of the electronic device 500 itself, information of the user, or external environment information. For example, the sensor 506 may include one or more sensors such as a vibration sensor, a temperature sensor, a distance sensor, a magnetic field sensor, a light sensor, an acceleration sensor, a fingerprint sensor, a Hall sensor, a position sensor, a gyroscope, an inertial sensor, a posture sensor, a barometer, and a heart rate sensor.

[0166] The power supply 507 is used to supply power to various components of the electronic device 500. In some embodiments, the power supply 507 may be logically connected to the processor 501 through a power management system, so that the power management system can manage charging, discharging, power consumption, and other functions.

[0167] It should be understood that the device provided in the embodiment of the present application is used to execute the above-mentioned event warning method, and thus can achieve the same effect as the above-mentioned implementation method.

[0168] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to an electronic device, the processing module may be used to control and manage the actions of the electronic device. The storage module may be used to support the electronic device in executing related program codes, etc.

[0169] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.

[0170] In addition, the device provided in the embodiment of the present application can specifically be a chip, component or module, and the chip may include a connected processor and memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute an event warning method provided in the above embodiment.

[0171] An embodiment of the present application also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is executed on a computer, the computer executes the above-mentioned related method steps to implement an event warning method provided by the above-mentioned embodiment.

[0172] This embodiment also provides a computer program product. When the computer program product runs on a computer, it enables the computer to execute the above-mentioned related steps to implement an event warning method provided by the above embodiment.

[0173] Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment is used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here.

[0174] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0175] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0176] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An event early warning method, characterized in that: The method comprises: Acquire scene data for a target object sent by a data acquisition device, wherein the scene data includes image data and audio data; Performing image recognition on the image data to obtain an image recognition result, and performing semantic recognition on the audio data to obtain a semantic recognition result; Determining, based on the image recognition result and the semantic recognition result, an event type to which the event indicated by the scene data belongs; Based on the event type, event warning information of the event is generated, and the event warning information is sent to a processing terminal, and the processing terminal intervenes in the event based on the received event warning information.

2. The method according to claim 1, characterized in that The performing image recognition on the image data to obtain an image recognition result includes: Performing image behavior recognition on the image data to obtain behavior feature information of the image data, and performing face recognition on the image data to obtain face feature information; The behavior feature information and the face feature information are determined as image recognition results.

3. The method according to claim 2, characterized in that The performing image behavior recognition on the image data to obtain behavior feature information of the image data includes: Extracting target limb feature information matching preset limb feature information from the image data; Extracting facial expression features from the image data to obtain expression feature information of the image data; The target limb feature information and the facial expression feature information are determined as behavior feature information.

4. The method according to claim 2, characterized in that: The performing semantic recognition on the audio data to obtain a semantic recognition result includes: Convert the audio data into text data, and perform sensitive word matching on the text data based on a preset sensitive word library; If there are word segments in the text data that match the sensitive words in the sensitive word library, language attack feature information of the text data is generated based on the word segments, and the language attack feature information is determined as a semantic recognition result.

5. The method according to claim 3, characterized in that: The determining, based on the image recognition result and the semantic recognition result, the event type to which the event indicated by the scene data belongs includes: If the target body feature information matches the preset attack behavior feature information, and the language attack feature information matches the preset attack feature information, then determining that the event type of the event indicated by the scene data is the first type; If the target body feature information matches the attack behavior feature information, the expression feature information matches the preset emotion feature information, and the language attack feature information matches the attack feature information, then determining that the event type of the event indicated by the scene data is the second type; If the target limb feature information matches the attack behavior feature information, the expression feature information matches the preset emotion feature information, the face indicated by the facial feature information is recognized at least twice within a preset time period, and the language attack feature information matches the preset attack feature information, then the event type of the event indicated by the scene data is determined to be the third type.

6. The method according to claim 1, characterized in that The generating event warning information of the event based on the event type includes: Obtaining warning suggestion information matching the event type from a preset warning template library; Based on the image recognition result, the semantic recognition result and the warning suggestion information, event warning information corresponding to the event is generated.

7. The method according to claim 1, characterized in that The method further comprises: Acquire facial data and a terminal identifier of a processing terminal, wherein the face indicated by the facial data includes the face in the image data; Based on the terminal identifier, establishing an association relationship between the facial data and the corresponding processing terminal; The sending the event warning information to the processing terminal includes: Based on the association relationship and the facial feature information, the event warning information is sent to the corresponding processing terminal.

8. An event warning device, characterized in that: The device comprises: A data acquisition unit, used to acquire scene data for a target object sent by a data acquisition device, wherein the scene data includes image data and audio data; A recognition unit, configured to perform image recognition on the image data to obtain an image recognition result, and perform semantic recognition on the audio data to obtain a semantic recognition result; An event type determination unit, configured to determine the event type to which the event indicated by the scene data belongs based on the image recognition result and the semantic recognition result; The information generating unit is used to generate event warning information of the event based on the event type, and send the event warning information to the processing terminal, and the processing terminal intervenes in the event based on the received event warning information.

9. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable program codes; A processor, configured to call and run the executable program code from the memory, so that the electronic device executes the method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that: The computer storage medium stores a plurality of instructions, which are suitable for being loaded by a processor and executing the steps of the method according to any one of claims 1 to 7.

Images