Security protection method and device, electronic equipment and readable storage medium

Abstract

The invention provides a safety protection method and device, electronic equipment and a readable storage medium. The method comprises the following steps: acquiring voice acquisition data corresponding to a target person and scene acquisition data corresponding to the position of the target person; determining whether the target person has a dangerous behavior according to the voice acquisition data and the scene acquisition data; and if the target person has the dangerous behavior, executing a reminding operation. By acquiring the voice of the target person, the current behavior state of the target person can be known, and meanwhile, the environment condition of the position where the target person is located is determined in combination with the scene acquisition data, so that the interaction condition of the target person and the environment is comprehensively understood through multi-modal data fusion; therefore, whether the target person has the dangerous behavior or not is accurately determined, the reminding operation is executed when the dangerous behavior exists, the dangerous behavior is processed in time, and the safety problem of the target person is avoided.

Description

Technical Field

[0001] The present invention relates to the field of personnel safety, and in particular to a safety protection method, device, electronic equipment and readable storage medium. Background Art

[0002] Because vulnerable groups such as children and the elderly lack the ability to protect themselves and recognize dangers, they are prone to various safety hazards in their daily lives. Traditional safety protection measures mainly rely on adult supervision, but adults cannot always be with children or the elderly, and in some complex scenarios, even with adult supervision, they may not be able to detect children or the elderly's dangerous behaviors or help signals in time. Summary of the invention

[0003] The main purpose of the present invention is to propose a safety protection method, device, electronic device and readable storage medium, aiming to solve the problem of how to promptly discover dangerous behaviors of vulnerable groups in the prior art.

[0004] To achieve the above object, the present invention provides a safety protection method, which comprises the steps of:

[0005] Acquire voice collection data corresponding to the target person, and scene collection data corresponding to the location of the target person;

[0006] Determining whether the target person has dangerous behavior according to the voice collection data and the scene collection data;

[0007] If the target person has dangerous behavior, a reminder operation is performed.

[0008] Optionally, determining whether the target person has dangerous behavior according to the voice collection data and the scene collection data includes:

[0009] Obtain the trained dangerous behavior recognition model;

[0010] Inputting the voice collection data and the scene collection data into the trained dangerous behavior recognition model;

[0011] Running the trained dangerous behavior recognition model to obtain a recognition result;

[0012] Determine whether the target person has dangerous behavior based on the recognition result.

[0013] Optionally, the step of obtaining the trained dangerous behavior recognition model includes:

[0014] Acquiring training samples, wherein the training samples include voice samples, scene samples, and dangerous behavior identifiers having corresponding relationships;

[0015] Obtaining an initial dangerous behavior recognition model;

[0016] The associated voice sample and the scene sample are used as input samples of the initial dangerous behavior recognition model, and the dangerous behavior identifier is used as an output sample of the initial dangerous behavior recognition model to train the initial dangerous behavior recognition model to obtain the trained dangerous behavior recognition model.

[0017] Optionally, the speech sample includes speech subsamples corresponding to a plurality of consecutive speech sounds.

[0018] Optionally, the performing reminder operation includes:

[0019] Determine the risk level corresponding to the risky behavior;

[0020] Matching a target reminder operation corresponding to the danger level;

[0021] The target reminder operation is performed.

[0022] Optionally, the matching target reminder operation corresponding to the danger level includes:

[0023] Determine a target reminder object corresponding to the danger level;

[0024] The child information corresponding to the target person is obtained, and the child information, the voice collection data, the scene collection data and the danger level are sent to the target reminder object.

[0025] Optionally, the performing reminder operation includes:

[0026] Determining a target tone feature corresponding to the target person;

[0027] Obtaining a target reminder voice corresponding to the dangerous behavior;

[0028] The target reminder voice is played with the target tone feature.

[0029] To achieve the above object, the present invention also provides a safety protection device, the safety protection device comprising:

[0030] A first acquisition module is used to acquire voice acquisition data corresponding to a target person, and scene acquisition data corresponding to a location of the target person;

[0031] A first determination module, used to determine whether the target person has dangerous behavior according to the voice collection data and the scene collection data;

[0032] The first execution module is used to execute a reminder operation if the target person has dangerous behavior.

[0033] To achieve the above objectives, the present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the steps of the security protection method described above when executed by the processor.

[0034] To achieve the above objectives, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the security protection method as described above are implemented.

[0035] The present invention proposes a safety protection method, device, electronic device and readable storage medium, which obtains the voice collection data corresponding to the target person and the scene collection data corresponding to the location of the target person; determines whether the target person has dangerous behavior based on the voice collection data and the scene collection data; if the target person has dangerous behavior, performs a reminder operation. By collecting the voice of the target person, the current behavior state of the target person can be known, and the environmental conditions of the location of the target person can be determined in combination with the scene collection data, so as to fully understand the interaction between the target person and the environment through multimodal data fusion, so as to accurately determine whether the target person has dangerous behavior, and perform a reminder operation when there is dangerous behavior, so as to deal with the dangerous behavior in time and avoid safety problems for the target person. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0038] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0039] Figure 1 This is a schematic diagram of the process of the first embodiment of the safety protection method of the present invention;

[0040] Figure 2It is a schematic diagram of the overall operation of the safety protection method of the present invention;

[0041] Figure 3 It is a module structure diagram of the safety protection device of the present invention;

[0042] Figure 4 It is a schematic diagram of the module structure of the electronic device of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0044] The disclosure below provides many different embodiments or examples to implement different structures of the present invention. In order to simplify the disclosure of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present invention. In addition, the present invention can repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0045] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only embodiments of a part of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work should fall within the scope of protection of the present application.

[0046] The present invention provides a safety protection method, referring to Figure 1 , Figure 1 This is a flow chart of a first embodiment of a safety protection method of the present invention, wherein the method comprises the following steps:

[0047] Step S10, acquiring voice collection data corresponding to the target person, and scene collection data corresponding to the location of the target person;

[0048] The target person is a person who needs to be protected in the environment, such as a child, an elderly person, etc. It is understandable that there may be multiple target persons in the environment, so the safety protection method of the present application can be performed for each target person. This embodiment and subsequent embodiments are described by taking the target person as a child as an example.

[0049] The voice collection data is the data obtained by collecting the sound made by the target person; the voice collection data can be collected by a microphone set in the environment, and can also be collected by a microphone set on a device in the environment. For example, if the target person wears a children's watch, a communication connection can be established with the children's watch to obtain the voice collection data collected by the microphone of the children's watch; for example, if the parent's mobile phone is in the environment, a communication connection can be established with the mobile phone to obtain the voice collection data collected by the microphone of the mobile phone; for example, if a smart home is set up in the environment, a communication connection can be established with the smart home to obtain the voice collection data collected by the microphone of the smart home.

[0050] The scene acquisition data is data obtained by collecting information about the location of the target person; the scene acquisition data can be specifically set to contain different types of information based on actual needs; for example, the scene acquisition data may include but is not limited to images, contour images, temperature, ambient gas composition, sound, weather, time, distance from the target person, and the movement status of the target person.

[0051] The structure of the scene can be clarified through images or contour images, such as the placement of furniture, location of electrical appliances, location of doors and windows in indoor scenes, or the distribution of amusement facilities, road conditions, and pedestrian distribution in outdoor scenes.

[0052] Temperature can be used to clarify the temperature conditions in the scene and determine whether events such as fire have occurred.

[0053] The ambient gas composition can clarify the air quality in the scene, thereby determining whether flammable gas leakage occurs, determining the PM2.5 situation, etc.

[0054] Sound can clearly indicate the noise level in a scene, such as the sound of a TV, birds singing, flowing water, wind, rain, etc.

[0055] Weather can clearly define the weather conditions in the scene, such as sunny, rainy, snowy, and windy.

[0056] The time of day can specify whether the scene is during the day or at night.

[0057] The distance to the target person can determine the positional relationship between the target person and surrounding objects, thereby determining whether there is a collision risk.

[0058] The target person's movement status can determine whether the target person has a movement safety risk.

[0059] Different types of scene acquisition data can be collected based on corresponding acquisition devices, such as images can be collected through cameras, contour images can be collected through infrared cameras, temperature can be collected through temperature sensors, ambient gas composition can be collected through gas sensors, and sound can be collected through microphones; specific acquisition devices can be independently set in the scene, or they can be set on specific devices in the scene, such as smart homes, and scene acquisition data can be obtained by establishing a communication connection with the device.

[0060] The above scene collection data is only for illustration. In actual application, specific types of scene collection data may be collected based on actual needs.

[0061] It is understandable that scene collection data can be obtained by collecting data from multiple scenes at the same time; and when judging dangerous behaviors of children, the main focus can be on the scene collection data of the child's location; for example, the indoor kitchen, living room, room, and bathroom can be taken as independent scenes, and when the target person is in the living room, the scene collection data collected from the living room will be used as the scene collection data that needs to be applied at this time.

[0062] Step S20, determining whether the target person has any dangerous behavior according to the voice collection data and the scene collection data;

[0063] The voice collection data can determine the sounds made by the target person, such as calling for help, shouting, requesting, crying, etc.; the scene collection data indicates the status within the scene. Under different scene states, the safety threats faced by children are different. Therefore, by combining the voice collection data reflecting the children themselves and the scene collection data reflecting the scene state, it is possible to comprehensively determine whether the target person has dangerous behavior.

[0064] Dangerous behaviors are used to indicate security threats to the target person.

[0065] Step S30: If the target person has dangerous behavior, a reminder operation is performed.

[0066] If the target person does not have any dangerous behavior, the voice collection data and the scene collection data are still collected.

[0067] When the target person has dangerous behavior, it means that the safety of the target person is threatened. Therefore, a reminder operation is performed to eliminate the dangerous behavior. The reminder operation can be set based on actual needs, such as reminding children, reminding parents or calling the police.

[0068] This embodiment can know the current behavior state of the target person by collecting the voice of the target person, and at the same time determine the environmental conditions of the target person's location by combining the scene collection data, so as to comprehensively understand the interaction between the target person and the environment through multimodal data fusion, so as to accurately determine whether the target person has dangerous behavior, and perform reminder operations when dangerous behavior exists, so as to deal with the dangerous behavior in time and avoid safety problems for the target person.

[0069] Further, see also Figure 2 In the second embodiment of the security protection method of the present invention proposed based on the first embodiment of the present invention, the step S20 includes the steps of:

[0070] Step S21, obtaining a trained dangerous behavior recognition model;

[0071] Step S22, inputting the voice collection data and the scene collection data into the trained dangerous behavior recognition model;

[0072] Step S23, running the trained dangerous behavior recognition model to obtain a recognition result;

[0073] Step S24, determining whether the target person has any dangerous behavior based on the recognition result.

[0074] The dangerous behavior recognition model is used to identify dangerous behaviors; the specific structure of the dangerous behavior recognition model can be set based on actual needs, such as the dangerous behavior recognition model can include CNN (Convolutional Neural Networks) and RNN (Recurrent Neural Network) or LSTM (Long Short-Term Memory); among them, CNN can be used to extract spectral features of speech collection data, and RNN or LSTM can process sequence information and semantic context in speech collection data. The dangerous behavior recognition model can adjust the hyperparameters of the model, such as the number of network layers, the number of neurons, the learning rate, etc., through methods such as cross-validation. At the same time, regularization technology can be used to prevent the model from overfitting.

[0075] The input of the dangerous behavior recognition model is voice collection data and scene collection data, and the output recognition result is the relevant information of the dangerous behavior. For example, the recognition result may include whether there is a dangerous behavior, the type of dangerous behavior, the danger level, etc.

[0076] After obtaining the identification results, it can be determined whether the target person has any dangerous behavior.

[0077] In the specific implementation, the voice collection data and the scene collection data are first collected; in order to obtain more accurate recognition results, the voice collection data can be enhanced in real time to remove interference components such as environmental noise and echo; the specific enhancement processing method can be set based on actual needs, such as adaptive filtering, which dynamically adjusts the filtering parameters according to the real-time environmental noise conditions to improve the clarity of the voice collection data; the voice collection data can also be processed in other ways to improve its clarity, such as spectral subtraction, Wiener filtering and other noise reduction algorithms.

[0078] After the voice collection data and the scene collection data are input into the trained dangerous behavior recognition model, the voice collection data is first converted into text information, and the text information is semantically understood through natural language processing technology, and the semantics of the text information is analyzed using methods such as lexical analysis, syntactic analysis and semantic role labeling; it should be noted that the voice collection data may contain multiple consecutive voice corresponding voice sub-data, so as to obtain the sound made by the target person in a short time, and the trained dangerous behavior recognition model can determine the meaning of the voice by combining the context information composed of multiple voice sub-data and the scene collection data; specifically, a pattern library can be established in advance, in which the correspondence between various common help sentences and dangerous behavior expressions is set, and the corresponding similarity threshold is set; the identified semantics is matched with the pre-defined help sentences and dangerous behavior patterns; if the content similarity between the identified semantics and the help sentence in the pattern library exceeds the threshold, it is considered to be a help signal, and the corresponding dangerous behavior is determined.

[0079] On this basis, the factors of scene acquisition data are further combined to determine dangerous behaviors; specifically, a correspondence is established between scene acquisition data, help signals and dangerous behaviors; for example, if the scene indicated by the scene acquisition data is at the stairwell, and the semantics indicate that the target person says "I want to run down", and at the same time, the sensor data in the scene acquisition data shows that the target person is approaching the stairs quickly, then it can be determined that this is a dangerous behavior; the specific correspondence between scene acquisition data, help signals and dangerous behaviors can be set based on actual needs.

[0080] After determining the dangerous behavior, the degree of the dangerous behavior, that is, the danger level, can also be determined; the danger level can also be determined based on the voice collection data and the scene collection data; for example, the danger level of children climbing high places may be higher than the danger level of running on flat ground; when the semantics corresponding to the children's voice collection data are sentences such as "help" and "I am injured and bleeding a lot", the corresponding danger level can be an emergency help level; when the semantics corresponding to the children's voice collection data are sentences such as "I can't find my toy", the corresponding danger level can be a general help level. The correspondence between the specific voice collection data, scene collection data and danger level can be set based on actual needs.

[0081] Furthermore, the step S21 includes the following steps:

[0082] Step S25, obtaining training samples, wherein the training samples include voice samples, scene samples and dangerous behavior identifiers having corresponding relationships;

[0083] Step S26, obtaining an initial dangerous behavior recognition model;

[0084] Step S27, using the associated voice sample and the scene sample as input samples of the initial dangerous behavior recognition model, and using the dangerous behavior identifier as an output sample of the initial dangerous behavior recognition model to train the initial dangerous behavior recognition model, so as to obtain the trained dangerous behavior recognition model.

[0085] In order to obtain a trained dangerous behavior recognition model, the initial dangerous behavior recognition model needs to be trained.

[0086] The training samples are used to train the initial dangerous behavior recognition model.

[0087] The speech samples indicate the sounds made by the children, corresponding to the speech collection data in the model application; the scene samples indicate the information within the scene, corresponding to the scene collection data in the model application.

[0088] Voice samples can be obtained by collecting children's voices in different states in different scenarios; for example, in different places such as kindergartens, schools, and homes, special recording equipment can be set up to collect children's voices; voice samples of children in various activities such as playing, studying, and resting can be collected in different places; in addition to voice in normal scenarios, it is also necessary to focus on collecting the voices made by children when they encounter danger or need help, such as "I fell down and it hurts", "Someone bullied me", "I'm lost"; at the same time, voice samples of children in different emotional states, such as fear, anxiety, and crying, can also be collected.

[0089] Furthermore, voice samples can be collected for different types of children, such as classification based on age, gender, and region, while ensuring that different types of children have a sufficient number of voice samples to improve the generalization ability of the model.

[0090] The scene sample can collect relevant information about the specific scene where the child is located; the details can be referred to the aforementioned sample collection data, which will not be repeated here. It should be noted that the voice sample corresponds to the scene sample one-to-one, and the voice sample is collected in the scene where the associated scene sample is located.

[0091] The dangerous behavior identification is the relevant information of the dangerous behavior determined by marking; the dangerous behavior identification may include whether it is a dangerous behavior, the type of dangerous behavior, and the danger level.

[0092] After obtaining the voice samples and scene samples, the samples can be labeled; when labeling the voice samples, the samples can be labeled as help signals, and "Help" and "Come and help me" can be labeled as emergency help according to the semantics; "I feel uncomfortable" and "I am injured" can be labeled as the general danger level of help, and the dangerous behavior type of "injury" can be labeled; the voice features when asking for help can also be labeled, such as the increase in tone, the increase in speaking speed, etc.; for example, the samples can be labeled as danger signals, such as "I want to climb a tree" and "I want to climb over this railing" can be labeled as dangerous danger levels according to the semantics, and the dangerous behavior type of "climbing" can be labeled, and the meaning of these voices in the corresponding scene samples can be labeled; the voice features such as hesitation and curiosity that may appear in children before performing dangerous behaviors can also be labeled; when labeling the scene samples, the type of scene, specific place, dangerous area, and common sound features and object features can be labeled; types such as indoor and outdoor, specific places such as kindergarten playgrounds and family kitchens, and dangerous areas such as pool sides and stairways.

[0093] Through labeling, we can eventually obtain unit samples containing labeled information, such as speech samples, scene samples, and dangerous behavior signs; a combination of multiple unit samples constitutes a training sample.

[0094] Furthermore, the speech sample includes speech subsamples corresponding to a plurality of consecutive speech sounds.

[0095] By using the speech sub-samples corresponding to multiple consecutive speech occurrences, the dangerous behavior recognition model can combine the contextual information composed of the multiple speech sub-samples and the scene collection data to determine the meaning of the speech.

[0096] When training with training samples, the speech samples can be processed first, such as speech denoising, speech segmentation, data normalization, etc.; speech denoising can specifically adopt speech denoising algorithms, such as spectral subtraction, Wiener filtering, etc., to remove background noise in speech data; corresponding noise models can also be established for noise in different scenarios for targeted noise reduction; speech segmentation divides the continuous speech stream into individual speech segments for subsequent feature extraction and analysis, and can also be segmented according to speech pauses, changes in intonation, etc.; data normalization normalizes the speech data, and adjusts the volume, pitch and other features of different speech samples to a unified range to reduce data differences; at the same time, the scene data is standardized so that it can be better read and analyzed by the model.

[0097] After that, speech feature extraction is performed on the processed speech samples, which may specifically include acoustic feature extraction, language feature analysis, and emotional feature extraction. In addition to common features such as pitch, duration, and intonation, acoustic feature extraction can also extract more complex acoustic features, such as formant, LPC (Linear Predictive Coding), etc. Specifically, the STFT (Short-Time Fourier Transform) and other methods can be used to convert speech samples from the time domain to the frequency domain, and then acoustic feature extraction is realized based on the spectrum features. Language feature analysis analyzes the vocabulary and grammatical structure of children's speech, and a children's language vocabulary library can be established to count the commonly used vocabulary and expressions of children of different age groups. At the same time, the language habits and semantic understanding methods of children in different scenarios can be analyzed. Emotional feature extraction can extract children's emotional features, such as happiness, fear, anger, etc., through the results of acoustic feature extraction and language feature analysis. Specifically, the emotional features can be classified and modeled through machine learning algorithms.

[0098] After the voice feature extraction is completed, the scene samples are subjected to feature fusion, which specifically includes multimodal data fusion, scene dynamic modeling, and context information utilization. Multimodal data fusion is the fusion of image data and sensor data collected by the camera with voice features. For example, the objects and people in the scene are identified through image recognition technology, and they are associated with voice features. The sensor data is used to determine the movement direction and speed of the child, and the child’s behavioral intention is inferred in combination with the voice features. Scene dynamic modeling requires the establishment of a dynamic model of the scene, taking into account the changes in objects in the scene and the activity trajectory of the child. For example, when a child approaches a dangerous area, such as a pool, the weight of the safety judgment is dynamically adjusted according to the child’s voice features and movement status. After that, the contextual information of the voice and the historical data of the scene are fully utilized. If a child mentions a dangerous behavior or expresses curiosity about an object multiple times over a period of time, the system can pay more attention to its subsequent voice and behavior.

[0099] The dangerous behavior recognition model obtains a recognition result after the above processing, and the model iteration is achieved by comparing the recognition result with the dangerous behavior recognition identifier; after the model training meets the training completion conditions, a trained dangerous behavior recognition model is obtained.

[0100] The specific training method can be set based on actual needs. For example, a multi-stage training method is adopted. First, pre-training is performed on a large-scale general children's speech dataset to obtain a general speech feature representation; then, fine-tuning is performed on labeled specific scenarios and dangerous behavior datasets to enable the model to better adapt to actual application scenarios.

[0101] In the actual application of the trained dangerous behavior recognition model, the relevant data of each recognition time can be regularly analyzed to generate new training samples, and the dangerous behavior recognition model can be updated through the new training samples.

[0102] This embodiment can realize the training of the dangerous behavior recognition model.

[0103] Further, in a third embodiment of the security protection method of the present invention proposed based on the first embodiment of the present invention, step S30 includes the steps of:

[0104] Step S31, determining the risk level corresponding to the dangerous behavior;

[0105] Step S32, matching a target reminder operation corresponding to the danger level;

[0106] Step S33, executing the target reminder operation.

[0107] The danger level indicates the degree to which the safety of the target person is or will be threatened; it is understandable that the higher the danger level, the higher the degree of safety threat to the target person.

[0108] The target reminder operation is a reminder operation that needs to be performed under a specific danger level; it is understandable that the types of operations that need to be performed are different under different danger levels; for example, when the danger level is low, the target person or guardian can be reminded to be careful, and when the danger level is high, the alarm can be directly sounded; therefore, in this embodiment, the corresponding target reminder operation is performed based on the danger level, so that the executed reminder operation can meet the needs of the actual situation.

[0109] Furthermore, the step S32 includes the steps of:

[0110] Step S321, determining a target reminder object corresponding to the danger level;

[0111] Step S322, obtaining the child information corresponding to the target person, and sending the child information, the voice collection data, the scene collection data and the danger level to the target reminder object.

[0112] The target reminder object is the object that needs to be reminded under a specific danger level.

[0113] Children's information is used to indicate relevant information of the target person; children's information includes but is not limited to name, age, gender, guardian contact information, health status, and allergy history.

[0114] The correspondence between specific danger levels and reminder objects can be set based on actual needs; for example, danger levels can include general help-seeking level, danger level, and emergency level; for example, situations where the target person says "I can't find my toy" and are not very dangerous can be classified as general help-seeking level; situations where the target person is climbing at a height and a safety accident may occur can be classified as a danger level; situations where a safety accident has occurred such as the target person shouting "Help" or "A lot of blood is flowing" can be classified as an emergency level.

[0115] The target reminder object corresponding to the general help level can be the target person or the guardian; the target reminder object corresponding to the danger level can be the guardian or teacher; the target reminder object corresponding to the emergency level can be the guardian, teacher, emergency center, police or other rescue departments; it can be understood that there can be multiple target reminder objects corresponding to the same danger level.

[0116] Different target reminder objects can perform reminder operations in different ways; if the target reminder object is a target person, the target person can be reminded by voice, light, device vibration, etc.; such as playing reminder voice through the speaker set in the scene, and controlling the bracelet or smart device worn by the target person to vibrate and display reminder information. For example, if the target reminder object is a guardian or teacher, a reminder message or software push can be sent to the guardian or teacher's mobile phone, which can include child information, voice collection data, scene collection data and danger level; so that the guardian or teacher can understand the situation of the target person in time. For example, if the target reminder object is a rescue department, the alarm can be automatically realized through the pre-established linkage mechanism. At the same time, the alarm information will contain information such as child information, voice collection data, scene collection data and danger level, so that the rescue department can quickly allocate resources based on this information for rescue.

[0117] When the target person continues to engage in dangerous behavior, the target person's relevant situation can be monitored in real time, and updated and pushed to the target reminder object in real time.

[0118] When the danger level is an emergency level, after the target reminder object arrives in the scene, rescue measures can be generated based on the current status of the target person, and the rescue error can be provided to the target reminder object through voice or information push for preliminary rescue.

[0119] This embodiment can remind the target reminder object at different danger levels to solve the current dangerous behavior of the target person.

[0120] Further, in a fourth embodiment of the security protection method of the present invention proposed based on the first embodiment of the present invention, step S30 includes the steps of:

[0121] Step S34, determining the target tone feature corresponding to the target person;

[0122] Step S35, obtaining a target reminder voice corresponding to the dangerous behavior;

[0123] Step S36, playing the target reminder voice with the target tone feature.

[0124] It is understandable that different children have different personalities and behavioral habits, so different children can accept different reminder methods; in order to enable the target person to obey the reminder content when the target reminder object is a target person, different expected features can be used to play the target reminder voice.

[0125] The target reminder voice is the voice that needs to be played for the current dangerous behavior of the target person. For example, if the target person is climbing at a high place, the corresponding target reminder voice can be "XXX, don't climb the tree, come down quickly", where XXX is the name of the target person.

[0126] The target tone is the tone used for voice playback; for example, for timid children, a gentle tone can be used for voice reminders; for naughty children, a serious tone can be used for reminders. Furthermore, the timbre of the guardian of the target person can be collected in advance, and then the target reminder voice can be played by combining the guardian's timbre with the target tone. The correspondence between the specific target person and the target tone characteristics can be set according to actual needs.

[0127] This embodiment sets different target voice tone features for different target persons, so that when the target reminder object is the target person, the target person can listen to the reminder content.

[0128] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0129] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0130] The present application also provides a safety protection device for implementing the above safety protection method, see Figure 3 , the safety protection devices include:

[0131] A first acquisition module is used to acquire voice acquisition data corresponding to a target person, and scene acquisition data corresponding to a location of the target person;

[0132] A first determination module, used to determine whether the target person has dangerous behavior according to the voice collection data and the scene collection data;

[0133] The first execution module is used to execute a reminder operation if the target person has dangerous behavior.

[0134] This safety protection device can know the current behavior status of the target person by collecting the voice of the target person, and at the same time determine the environmental conditions of the target person's location by combining the scene collection data, so as to fully understand the interaction between the target person and the environment through multimodal data fusion, so as to accurately determine whether the target person has dangerous behavior, and execute reminder operations when dangerous behavior exists, so as to deal with the dangerous behavior in time and avoid safety problems for the target person.

[0135] It should be noted that the first acquisition module in this embodiment can be used to execute step S10 in the embodiment of the present application, the first determination module in this embodiment can be used to execute step S20 in the embodiment of the present application, and the first execution module in this embodiment can be used to execute step S30 in the embodiment of the present application.

[0136] Furthermore, the first determining module includes:

[0137] A first acquisition unit is used to acquire a trained dangerous behavior recognition model;

[0138] A first input unit, used to input the voice collection data and the scene collection data into the trained dangerous behavior recognition model;

[0139] A first operation unit is used to operate the trained dangerous behavior recognition model to obtain a recognition result;

[0140] The first determining unit is used to determine whether the target person has dangerous behavior according to the recognition result.

[0141] Furthermore, the first determining module further includes:

[0142] A second acquisition unit is used to acquire training samples, wherein the training samples include voice samples, scene samples and dangerous behavior identifiers having corresponding relationships;

[0143] A third acquisition unit is used to acquire an initial dangerous behavior recognition model;

[0144] The first execution unit is used to use the associated voice sample and the scene sample as input samples of the initial dangerous behavior recognition model and the dangerous behavior identifier as the output sample of the initial dangerous behavior recognition model to train the initial dangerous behavior recognition model to obtain the trained dangerous behavior recognition model.

[0145] Furthermore, the speech sample includes speech subsamples corresponding to a plurality of consecutive speech sounds.

[0146] Furthermore, the first execution module includes:

[0147] A second determining unit, used to determine a danger level corresponding to the dangerous behavior;

[0148] A first matching unit, used for matching a target reminder operation corresponding to the danger level;

[0149] The second execution unit is used to execute the target reminder operation.

[0150] Furthermore, the first matching unit includes:

[0151] A first determining subunit is used to determine a target reminder object corresponding to the danger level;

[0152] The first sending subunit is used to obtain the child information corresponding to the target person, and send the child information, the voice collection data, the scene collection data and the danger level to the target reminder object.

[0153] Furthermore, the first execution module includes:

[0154] A third determining unit, configured to determine a target tone feature corresponding to the target person;

[0155] A fourth acquisition unit, configured to acquire a target reminder voice corresponding to the dangerous behavior;

[0156] The first playing unit is used to play the target reminder voice with the target tone feature.

[0157] Reference Figure 4 In terms of hardware structure, the electronic device may include components such as a communication module 10, a memory 20, and a processor 30. In the electronic device, the processor 30 is connected to the memory 20 and the communication module 10 respectively, and a computer program is stored in the memory 20. The computer program is executed by the processor 30 at the same time, and the steps of the above method embodiment are implemented when the computer program is executed.

[0158] The communication module 10 can be connected to an external communication device through a network. The communication module 10 can receive requests from an external communication device, and can also send requests, instructions and information to the external communication device, which can be other electronic devices, servers or Internet of Things devices, such as televisions, etc.

[0159] The memory 20 can be used to store software programs and various data. The memory 20 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function (such as obtaining voice acquisition data corresponding to a target person, and scene acquisition data corresponding to the location of the target person), etc.; the data storage area can include a database, and the data storage area can store data or information created according to the use of the system, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0160] The processor 30 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It executes various functions of the electronic device and processes data by running or executing software programs and / or modules stored in the memory 20, and calling data stored in the memory 20, so as to monitor the electronic device as a whole. The processor 30 may include one or more processing units; optionally, the processor 30 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 30.

[0161] although Figure 4 Although not shown, the electronic device may further include a circuit control module, which is used to connect to a power source to ensure the normal operation of other components. Figure 4 The electronic device structure shown in the figure does not constitute a limitation of the electronic device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0162] The present invention also provides a computer-readable storage medium on which a computer program is stored. The computer-readable storage medium may be Figure 4 The memory 20 in the electronic device may also be at least one of a ROM (Read-Only Memory) / RAM (Random Access Memory), a magnetic disk, and an optical disk. The computer-readable storage medium includes a number of instructions for enabling a terminal device with a processor (which may be a television, a car, a mobile phone, a computer, a server, a terminal, or a network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0163] In the present invention, the terms "first", "second", "third", "fourth" and "fifth" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0164] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0165] Although the embodiments of the present invention have been shown and described above, the scope of protection of the present invention is not limited thereto. It is understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. A person of ordinary skill in the art can change, modify and replace the above embodiments within the scope of the present invention, and these changes, modifications and replacements should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A child safety protection method, characterized in that: The safety protection method comprises: Acquire voice collection data corresponding to the target person, and scene collection data corresponding to the location of the target child; Determining whether the target person has dangerous behavior according to the voice collection data and the scene collection data; If the target person has dangerous behavior, a reminder operation is performed.

2. The safety protection method according to claim 1, characterized in that: Determining whether the target person has dangerous behavior according to the voice collection data and the scene collection data includes: Obtain the trained dangerous behavior recognition model; Inputting the voice collection data and the scene collection data into the trained dangerous behavior recognition model; Running the trained dangerous behavior recognition model to obtain a recognition result; Determine whether the target person has dangerous behavior based on the recognition result.

3. The safety protection method according to claim 2, characterized in that: The step of obtaining the trained dangerous behavior recognition model includes: Acquiring training samples, wherein the training samples include voice samples, scene samples, and dangerous behavior identifiers having corresponding relationships; Obtaining an initial dangerous behavior recognition model; The associated voice sample and the scene sample are used as input samples of the initial dangerous behavior recognition model, and the dangerous behavior identifier is used as an output sample of the initial dangerous behavior recognition model to train the initial dangerous behavior recognition model to obtain the trained dangerous behavior recognition model.

4. The safety protection method according to claim 3, characterized in that: The speech sample includes speech subsamples corresponding to a plurality of consecutive speech sounds.

5. The safety protection method according to claim 1, characterized in that: The execution reminder operation includes: Determine the risk level corresponding to the risky behavior; Matching a target reminder operation corresponding to the danger level; The target reminder operation is performed.

6. The safety protection method according to claim 5, characterized in that: The matching target reminder operation corresponding to the danger level includes: Determine a target reminder object corresponding to the danger level; The child information corresponding to the target person is obtained, and the child information, the voice collection data, the scene collection data and the danger level are sent to the target reminder object.

7. The safety protection method according to claim 1, characterized in that: The execution reminder operation includes: Determining a target tone feature corresponding to the target person; Obtaining a target reminder voice corresponding to the dangerous behavior; The target reminder voice is played with the target tone feature.

8. A safety protection device, characterized in that: The safety protection device comprises: A first acquisition module is used to acquire voice acquisition data corresponding to a target person, and scene acquisition data corresponding to a location of the target person; A first determination module, used to determine whether the target person has dangerous behavior according to the voice collection data and the scene collection data; The first execution module is used to execute a reminder operation if the target person has dangerous behavior.

9. An electronic device, characterized in that: The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the security protection method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the security protection method according to any one of claims 1 to 7 are implemented.

Images