A campus intelligent monitoring method and system

Abstract

The present application discloses a campus intelligent monitoring method and system, the method comprising: receiving wristband warning information; determining the area ID to which it belongs according to the wristband coordinate points in a preset virtual campus map as the target area ID; determining all cameras corresponding to the target area ID, and intercepting according to the positioning timestamp to obtain a set of image frame sequences to be analyzed; based on a pre-trained campus behavior analysis model, analyzing the set of image frame sequences to be analyzed to obtain a behavior analysis result, and if the behavior analysis result is a conflicting label, performing a stop prompt processing based on the wristband coordinate points, and if the behavior analysis result is a non-conflicting label, performing no processing. The present application uses wristband coordinate points to determine the location of students, uses the campus behavior analysis model to perform analysis to obtain the behavior analysis result, and performs a stop prompt processing for the situation where a conflicting label appears, thereby improving the accuracy and efficiency of monitoring and analyzing campus violence incidents.

Description

Technical Field

[0001] The present application belongs to the technical field of smart campus, and in particular, relates to a campus intelligent monitoring method and system. Background Art

[0002] In today's society, school violence has become a social issue that has received much attention, and the negative impact of school violence has spread widely. However, although the existing smart campus monitoring system can monitor the entire campus environment, it cannot stop school violence in time, and there is still the problem of untimely warning of campus conflicts. Therefore, there is an urgent need for a solution that can accurately identify campus conflict behaviors and stop conflict behaviors in time when campus violence occurs. Summary of the invention

[0003] In order to overcome the defects and shortcomings of the prior art, the present application provides a campus intelligent monitoring method and system.

[0004] In order to achieve the above objectives, this application adopts the following technical solutions:

[0005] According to one aspect of the present application, a campus intelligent monitoring method is provided, which is used for a cloud server, wherein the cloud server is respectively connected to at least one smart bracelet, multiple UWB base stations, multiple cameras, and multiple mobile terminals to form a smart campus system; the method comprises: receiving wristband warning information, wherein the wristband warning information comprises wristband coordinate points and positioning timestamps;

[0006] Determine the area ID of the wristband coordinate point in the preset virtual campus map as the target area ID;

[0007] Determine all cameras corresponding to the target area ID, and intercept them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed;

[0008] Based on the pre-trained campus behavior analysis model, the image frame sequence set to be analyzed is analyzed to obtain the behavior analysis result. If the behavior analysis result is a conflicting label, a stop prompt is performed based on the wristband coordinate point. If the behavior analysis result is a non-conflicting label, no processing is performed.

[0009] Preferably, the wristband warning information is triggered and generated after the smart wristband analyzes the conversation text information and determines it as a threat tag;

[0010] When analyzing the conversation text information, specifically including: using a threat analysis model to analyze the conversation text information to obtain a first threat analysis result, when the first threat analysis result is a threat label, it indicates that the bracelet user has a threat risk in the current situation, and when the first threat analysis result is a non-threat label, it indicates that the bracelet user does not have a threat risk in the current situation;

[0011] The threat analysis model is obtained by machine learning training using multiple sets of training data, the multiple sets of training data include first-category data and second-category data, each set of data in the first-category data includes: threatening dialogue training text and a threat label that identifies the threatening dialogue training text as having a threatening intent, and each set of data in the second-category data includes: non-threatening dialogue training text and a non-threatening label that identifies the non-threatening dialogue training text as having no threatening intent.

[0012] Preferably, the campus behavior analysis model is obtained by machine learning training using multiple sets of training data, the multiple sets of training data include third category data and fourth category data, each set of data in the third category data includes: violent conflict behavior training images and conflict labels indicating that violence exists in the violent conflict behavior training images, and each set of data in the fourth category data includes: non-violent conflict behavior training images and non-conflict labels indicating that violence does not exist in the non-violent conflict behavior training images.

[0013] Preferably, the process of performing the stop prompt based on the wristband coordinate point specifically includes:

[0014] Obtaining the terminal coordinate point of each mobile terminal;

[0015] The relative horizontal distance between each mobile terminal and the wristband coordinate point is calculated respectively, which is specifically expressed as:

[0016] d k =|x k -x0|+|y k -y0|

[0017] d k represents the horizontal plane relative distance between the terminal coordinate point of the kth mobile terminal and the wristband coordinate point, x k represents the x-axis coordinate value of the terminal coordinate point of the kth mobile terminal, x0 represents the x-axis coordinate value of the wristband coordinate point, y k represents the y-axis coordinate value of the terminal coordinate point of the k-th mobile terminal, and y0 represents the y-axis coordinate value of the wristband coordinate point;

[0018] After sorting in ascending order based on horizontal relative distance, the mobile terminal closest to the wristband coordinate point is used as the target prompt terminal;

[0019] Generate route navigation guidance information for the target prompt terminal based on the wristband coordinate points;

[0020] The image frame sequence set to be analyzed is sent to all mobile terminals, and the route navigation guidance information is sent to the target prompt terminal at the same time.

[0021] Preferably, in generating route navigation guidance information for the target prompt terminal based on the wristband coordinate point, it specifically includes: taking the wristband coordinate point as the starting point and the terminal coordinate point corresponding to the target prompt terminal as the end point, performing path planning to find the shortest path, and packaging the path point set corresponding to the shortest path into route navigation guidance information in the guidance order.

[0022] Preferably, the method further comprises:

[0023] In the warning table, the wristband warning information is classified and recorded based on the area ID, a key patrol list is set based on the warning table, and patrol reminder processing is performed based on the key patrol list.

[0024] Preferably, the key patrol list is set based on the early warning table, specifically including:

[0025] Traverse the warning table in turn, classify and count according to the area ID to obtain the statistical times corresponding to different area IDs;

[0026] Sort by the number of statistics from largest to smallest;

[0027] Starting from the first sort number, filter the area IDs according to the preset filter quantity and add them to the priority patrol list;

[0028] Count the warning time frequency of different area IDs in the key patrol list respectively;

[0029] In the key patrol list, the highest warning time frequency of each area ID is set as a key patrol mark, and the preset time period corresponding to the highest warning time frequency is used as the target patrol time period.

[0030] Preferably, the patrol reminder processing based on the key patrol list specifically includes:

[0031] The area ID and the corresponding target patrol time period are selected according to the key patrol identification, and the patrol prompt information is obtained by sorting the target patrol time period in chronological order from morning to night;

[0032] The patrol reminder information is sent to the mobile terminal with the duty identification.

[0033] Preferably, in the triggering wristband warning information, at least one target participant is identified based on the target voice information, a participant list is generated based on the at least one target participant, and the participant list is added to the warning table according to the area ID corresponding to the wristband warning information;

[0034] The step of identifying at least one target participant based on the target voice information specifically includes:

[0035] When the threat analysis model is used to analyze the conversation text information to obtain a first threat analysis result, if the first threat analysis result is a threat label, the conversation voice information corresponding to the conversation text information is determined as the target voice information;

[0036] Separating the target voice information into at least one reconstructed voice signal based on a pre-trained voice separation model, wherein the reconstructed voice signal is a voice signal corresponding to a participant in the conversation voice information;

[0037] Performing voiceprint recognition on at least one reconstructed voice signal to determine a corresponding target participant;

[0038] The speech separation model is obtained by pre-training with multiple sets of data through machine learning. During training, the multiple sets of data include multiple dialogue aliasing speech samples and multiple participant speech signals corresponding to each dialogue aliasing speech sample. The dialogue aliasing speech sample is obtained by aliasing any multiple participant speech signals.

[0039] According to another aspect of the present application, a campus intelligent monitoring system is also provided, characterized in that the system comprises:

[0040] An information receiving module is used to receive wristband warning information, wherein the wristband warning information includes wristband coordinate points and positioning timestamps;

[0041] A target area determination module, used to determine the area ID to which the wristband coordinate point belongs in a preset virtual campus map as the target area ID;

[0042] An image frame preprocessing module is used to determine all cameras corresponding to the target area ID, and intercept them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed;

[0043] The campus behavior analysis module is used to analyze the image frame sequence set to be analyzed based on the pre-trained campus behavior analysis model to obtain the behavior analysis result. If the behavior analysis result is a conflicting label, a stop prompt is processed based on the wristband coordinate point. If the behavior analysis result is a non-conflicting label, no processing is performed;

[0044] The patrol reminder module is used to classify and record the wristband warning information based on the area ID in the warning table, set a key patrol list based on the warning table, and perform patrol reminder processing based on the key patrol list.

[0045] According to another aspect of the present application, a terminal is also provided, comprising: at least one memory and at least one processor; characterized in that the at least one memory is used to store program code, and the at least one processor is used to call the program code stored in the at least one memory to execute any of the above-mentioned methods.

[0046] Compared with the prior art, this application has the following advantages and beneficial effects:

[0047] (1) The present application receives wristband warning information as a trigger condition for processing, and then uses the wristband coordinate points to determine the student's position, and then determines the target area ID based on the position. Combining the wristband warning information and the preset virtual campus map, the area corresponding to the target area ID can be quickly located, thereby improving the efficiency of monitoring; at the same time, the campus behavior analysis model is used to analyze the image frame sequence set to be analyzed in the area where the target area ID is located to obtain the behavior analysis result, and a stop prompt is given for the situation where conflicting labels appear. By combining the data analysis of the wristband and the camera, the accuracy and efficiency of monitoring and analysis of campus violence incidents are improved, and the workload of manual monitoring is reduced. By using precise positioning, area recognition, and real-time analysis of student behavior, a stop prompt is given for the situation where conflicting labels appear, which can timely prevent possible campus violence incidents, effectively prevent the conflict from escalating further, and ensure the personal safety of students on campus.

[0048] (2) This application classifies and records the wristband warning information based on the area ID in the warning table, sets a key patrol list based on the warning table, and performs patrol reminder processing based on the key patrol list. By analyzing the key patrol list in combination with the wristband warning information, the patrol work of campus patrol personnel is assisted, and patrol reminder processing can be given priority to key areas, thereby improving the efficiency and pertinence of campus security monitoring, thereby improving the patrol efficiency of campus patrol personnel, and effectively monitoring the campus security situation, achieving accurate positioning, behavior analysis and rapid prevention, and improving the level of intelligent management of campus security.

[0049] (3) In the present application, in triggering the wristband warning information, at least one target participant is identified based on the target voice information, and participant list information is generated based on the at least one target participant, and the participant list information is added to the warning table according to the area ID corresponding to the wristband warning information; by identifying the target participant in the conversation voice information with a threat label, the participant list information is obtained, and then the auxiliary patrol personnel are screened for potential bullying students, so as to timely help students with precise education guidance, thereby improving the personalization of campus education. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic diagram of a flow chart of a campus intelligent monitoring method provided in Example 1 of the present application;

[0051] Figure 2 This is a schematic diagram of the steps for performing patrol reminder processing in Example 1 of the present application;

[0052] Figure 3This is a schematic diagram of the steps for identifying the target participant in Example 1 of the present application;

[0053] Figure 4 This is a schematic diagram of the storage structure of the warning table in Example 1 of the present application;

[0054] Figure 5 A schematic block diagram of a campus intelligent monitoring system is provided in Example 2 of the present application;

[0055] Figure 6 A schematic diagram of the structure of a terminal provided in Embodiment 3 of the present application;

[0056] Figure 7 This is a schematic diagram of the structure of a computer device provided in Example 4 of the present application. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the solution of the present application, 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 only 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 creative work should fall within the scope of protection of the present application.

[0058] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0059] In Example 1, this embodiment provides a campus intelligent monitoring method, which is applied to a cloud server, and the cloud server is respectively connected to at least one smart bracelet, multiple UWB base stations, multiple cameras, and multiple mobile terminals to form a smart campus system;

[0060] In this embodiment, the smart bracelet has GPS and UWB positioning functions to determine the location of the bracelet in time, and the smart bracelet is worn by students; cameras are set in different areas of the campus; UWB base stations are deployed in the campus, and each UWB base station is set in different areas of the campus, such as walls, ceilings, corridors, doorways, etc., so that the UWB base station can obtain the student's location in real time through UWB signals. Each camera corresponds to an area ID, and then the corresponding area ID is determined according to the student's location.

[0061] In this embodiment, the mobile terminal is carried by a teacher, for example, a mobile phone, wherein the mobile terminal has GPS and UWB positioning functions.

[0062] like Figure 1 As shown, the method includes:

[0063] Step S100: receiving wristband warning information, wherein the wristband warning information includes wristband coordinate points and positioning timestamp;

[0064] In this embodiment, the position of the wristband is located by UWB signal to determine the position of the student. For example, the wristband coordinate point is expressed in the form of spatial coordinates, such as (X1, Y1, Z1), corresponding to the distribution of the X-axis, Y-axis and Z-axis respectively, and the positioning timestamp is the timestamp corresponding to when the wristband coordinate point is located and determined.

[0065] In this embodiment, the wristband warning information is triggered and generated after the smart wristband analyzes the conversation text information and determines it as a threat tag.

[0066] In this embodiment, when analyzing the conversation text information, it specifically includes: using a threat analysis model to analyze the conversation text information to obtain a first threat analysis result, when the first threat analysis result is a threat label, it indicates that the bracelet user has a threat risk in the current situation, and when the first threat analysis result is a non-threat label, it indicates that the bracelet user does not have a threat risk in the current situation; in actual application, a pre-trained threat analysis model is built into the smart bracelet, and the threat analysis model is obtained by machine learning training using multiple sets of training data, and the multiple sets of training data include a first category of data and a second category of data, each set of data in the first category of data includes: a threat conversation training text and a threat label that identifies that the threat conversation training text has a threat intention, and each set of data in the second category of data includes: a non-threat conversation training text and a non-threat label that identifies that the non-threat conversation training text does not have a threat intention; exemplarily, the threat analysis model can use an existing deep learning model, such as a recurrent neural network (RNN) or a long short-term memory network (LSTM).

[0067] In this embodiment, the analysis of the dialogue text information can also adopt a text matching method, which specifically includes: successively performing stop word removal processing, sentence splitting processing, and word segmentation processing on the dialogue text information to obtain a set of dialogue keywords, successively traversing the set of dialogue keywords and using a preset threat keyword library for matching to calculate the threat matching degree, determining the second threat analysis result according to the threat matching degree, setting a non-threat label if the threat matching degree is less than the preset threat matching threshold, and otherwise setting a threat label;

[0068] In this embodiment, the threat matching degree is expressed as

[0069]

[0070] represents the threat matching degree corresponding to the dialogue text information, θ(i) represents the preset score corresponding to the i-th successfully matched threat keyword, and M represents the number of successfully matched in the set of dialogue keywords;

[0071] The preset threat keyword library includes multiple preset threat keywords, and each threat keyword has a corresponding preset score. Exemplarily, in the form of a full score of 10 points, the preset score corresponding to "protection fee" is 7 points, the preset scores corresponding to "beat" and "punch" are both 8 points, the preset scores corresponding to "do", "fix", and "exclude" are both 5 points, the expected scores corresponding to "humiliate" and "siege" are both 9 points, the preset scores corresponding to "carry a knife" and "carry a stick" are both 10 points, and the preset scores corresponding to "take revenge", "settle accounts", and "fight an appointment" are 8 points.

[0072] In this embodiment, the stop word removal processing, sentence splitting processing, and word segmentation processing are all existing technologies, and their specific implementations are not within the scope of protection of this application. Exemplarily, stop words refer to words that often appear in natural language but have no actual contribution to the content, such as "of", "already", "is", etc. When extracting text keywords, these stop words need to be removed to reduce interference. Exemplarily, sentence splitting means splitting the entire text into several sentences, and word segmentation means splitting each sentence into several words, for example, using a dedicated natural language processing library to complete.

[0073] In this embodiment, the dialogue text information can be obtained by collecting dialogue voice information through a built-in microphone in the smart bracelet and then converting it using existing speech-to-text technology.

[0074] Step S200: Determine the area ID to which it belongs in the preset virtual campus map according to the bracelet coordinate point as the target area ID;

[0075] In this embodiment, the preset virtual campus map is a three-dimensional model, and each scene point in the campus has a corresponding area ID, so that each area ID has a corresponding preset coordinate range. By finding the preset coordinate range to which the bracelet coordinate point belongs, the area ID to which it belongs is used as the target area ID.

[0076] For example, the corridor on the first floor of the library corresponds to area ID 100, the corridor on the second floor of the library corresponds to area ID 200, the corridor on the first floor of teaching building A corresponds to area ID 300... and so on. Different scene points in the campus have corresponding area IDs.

[0077] In this embodiment, establishing a preset virtual campus map is a prior art. Exemplarily, the preset virtual campus map is a three-dimensional campus model established by combining a campus plane base map and campus elevation data, wherein the campus plane base map can be a file in the format of a scene CAD drawing, Shapefile, etc. about the campus, and the campus elevation data is a digital elevation model (DEM).

[0078] In actual application, the target area ID is determined according to the coordinate points of the bracelet, achieving the effect of precise positioning and providing positioning support for subsequent matching of the corresponding camera.

[0079] Step S300: determining all cameras corresponding to the target area ID, and intercepting them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed;

[0080] In this embodiment, the set of image frame sequences to be analyzed includes at least one image frame sequence to be analyzed, and each image frame sequence to be analyzed corresponds to a camera. Each camera has a corresponding area ID, and the cameras with the same ID as the target area are screened out, and then the video data collected by the camera is intercepted according to the positioning timestamp. For example, a preset interception time is set, such as 3 minutes, 5 minutes, etc., and the video data is intercepted from the positioning timestamp forward according to the preset interception time, so as to obtain the corresponding image frame sequence to be analyzed, and all the image frame sequences to be analyzed with the same ID as the target area are aggregated to obtain the set of image frame sequences to be analyzed.

[0081] In actual application, by determining all cameras corresponding to the target area ID and intercepting them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed, the effect of area recognition is achieved, providing data support for subsequent real-time analysis of student behavior.

[0082] Step S400: Based on the pre-trained campus behavior analysis model, the image frame sequence set to be analyzed is analyzed to obtain the behavior analysis result. If the behavior analysis result is a conflicting label, a stop prompt is processed based on the wristband coordinate point. If the behavior analysis result is a non-conflicting label, no processing is performed;

[0083] In this embodiment, the campus behavior analysis model is obtained by machine learning training using multiple sets of training data, and the multiple sets of training data include third-category data and fourth-category data, and each set of data in the third-category data includes: violent conflict behavior training images and conflict labels indicating that the violent conflict behavior training images have violence, and each set of data in the fourth-category data includes: non-violent conflict behavior training images and non-conflict labels indicating that the non-violent conflict behavior training images do not have violence; illustratively, the campus behavior analysis model can use existing deep learning models, such as recurrent neural networks (RNN). Exemplarily, the violent conflict behavior training images use training images corresponding to violent behavior characteristics such as fighting, brawling, pushing, kicking, etc.

[0084] In the stop prompt processing based on the wristband coordinate points, it specifically includes:

[0085] Step S410: Acquire the terminal coordinate point of each mobile terminal; in this step, the terminal coordinate point of each mobile terminal is determined by the UWB base stations around the mobile terminal, and then the terminal coordinate point of each mobile terminal is sent to the cloud server.

[0086] Step S420: Calculate the horizontal plane relative distance between each mobile terminal and the wristband coordinate point, which is specifically expressed as:

[0087] d k =|x k -x0|+|y k -y0|

[0088] d k represents the horizontal plane relative distance between the terminal coordinate point of the kth mobile terminal and the wristband coordinate point, x k represents the x-axis coordinate value of the terminal coordinate point of the kth mobile terminal, x0 represents the x-axis coordinate value of the wristband coordinate point, y k represents the y-axis coordinate value of the terminal coordinate point of the k-th mobile terminal, and y0 represents the y-axis coordinate value of the wristband coordinate point;

[0089] Step S430: sorting in ascending order based on horizontal relative distances and taking the mobile terminal closest to the wristband coordinate point as the target prompt terminal;

[0090] Step S440: Generate route navigation guidance information for the target prompt terminal based on the wristband coordinate point; in this step, specifically including: taking the wristband coordinate point as the starting point and the terminal coordinate point corresponding to the target prompt terminal as the end point, performing path planning to find the shortest path, and packaging the path point set corresponding to the shortest path into route navigation guidance information in the guidance order;

[0091] Step S450: sending the image frame sequence set to be analyzed to all mobile terminals, and sending route navigation guidance information to the target prompt terminal at the same time.

[0092] In actual application, the campus behavior analysis model is used to conduct real-time analysis of student behavior to obtain the behavior analysis results, and then in the stop prompt processing based on the wristband coordinate points, the target prompt terminal is found in time, and route navigation guidance information is sent, which reduces the response time of campus conflicts, promptly stops possible campus violence incidents, effectively prevents the conflict from escalating further, and ensures the personal safety of students on campus.

[0093] In this embodiment, in order to further improve the patrol efficiency of the campus and effectively stop campus violence in a timely manner. Figure 2 As shown, the method also includes:

[0094] Step S500: Classify and record the wristband warning information based on the area ID in the warning table, set a key patrol list based on the warning table, and perform patrol prompt processing based on the key patrol list;

[0095] The key patrol list set up based on the early warning table includes:

[0096] Step S510: traverse the warning table in sequence, and classify and count the number of times corresponding to different area IDs according to the area IDs;

[0097] In this embodiment, each wristband warning information in the warning table can be used as a statistical number.

[0098] Step S520: sorting in descending order based on the statistical times;

[0099] Step S530: Starting from the first sort number, filter the area ID according to the preset filter quantity and add it to the key patrol list; technical personnel in this field can set the preset filter quantity according to actual needs, such as 1, 2, 3, etc., and no specific limitation is made here.

[0100] Step S540: Count the frequency of warning times for different area IDs in the key patrol list respectively;

[0101] Step S550: setting a key patrol flag for the highest warning time frequency of each area ID in the key patrol list, and taking the preset time period corresponding to the highest warning time frequency as the target patrol time period;

[0102] In this embodiment, the warning time frequency situation includes frequency values ​​corresponding to multiple preset time periods. For example, the preset time periods are obtained by dividing the school time into intervals of 1 hour, such as 6:30-7:29 time period, 7:30-8:29 time period, 8:30-9:29 time period, 9:30-10:29 time period, 10:30-11:29 time period, and so on, until the end of the school's evening self-study, 21:30-22:29 time period.

[0103] In the patrol reminder processing based on the key patrol list, it specifically includes:

[0104] Step S560: Filter the area ID and the corresponding target patrol time period according to the key patrol identifier, and sort the target patrol time period in chronological order from morning to night to obtain patrol prompt information;

[0105] Step S570: Send patrol reminder information to the mobile terminal with the duty identification.

[0106] In this embodiment, the duty identification is updated by triggering a timer, and the update is completed by marking the duty identification on the mobile terminal used by the user in the pre-stored patrol duty list. The patrol duty list is uploaded to the cloud server by the user in advance.

[0107] In actual application, by combining the wristband warning information to analyze the key patrol list to assist the patrol work of campus patrol personnel, patrol reminders can be given priority in key areas, improving the efficiency and pertinence of campus security monitoring, thereby improving the patrol efficiency of campus patrol personnel, and through effective monitoring of campus safety, accurate positioning, behavior analysis and rapid prevention can be achieved, thereby improving the level of intelligent management of campus safety.

[0108] In this embodiment, in order to further effectively discover potential campus violence situations. Figure 3 As shown, the method also includes:

[0109] Step S600: in the triggering wristband warning information, at least one target participant is identified based on the target voice information, participant list information is generated based on the at least one target participant, and the participant list information is added to the warning table according to the area ID corresponding to the wristband warning information;

[0110] In actual application, by timely identifying the target participants to generate participation list information, patrol personnel are assisted in screening out potential bullying students, reducing school violence, thereby helping students to receive precise education guidance in a timely manner and improving the personalization of campus education.

[0111] The step of identifying at least one target participant based on the target voice information specifically includes:

[0112] Step S610: when the first threat analysis result is obtained by analyzing the conversation text information using the threat analysis model, if the first threat analysis result is a threat label, determining the conversation voice information corresponding to the conversation text information as the target voice information;

[0113] Step S620: Separating the target voice information into at least one reconstructed voice signal based on a pre-trained voice separation model, where the reconstructed voice signal is a voice signal corresponding to a participant in the conversation voice information;

[0114] In this embodiment, the speech separation model is obtained by pre-training using multiple sets of data through machine learning. During training, the multiple sets of data include multiple dialogue aliasing speech samples and multiple participant object speech signals corresponding to each dialogue aliasing speech sample. The dialogue aliasing speech sample is obtained by aliasing any multiple participant object speech signals.

[0115] Step S630: Perform voiceprint recognition on at least one reconstructed voice signal to determine the corresponding target participant.

[0116] The step of performing voiceprint recognition on at least one reconstructed voice signal to determine a corresponding target participant specifically includes:

[0117] Step S631: extracting corresponding voiceprint information to be recognized from each reconstructed speech signal;

[0118] Step S632: based on the pre-stored voiceprint database, traverse each voiceprint information to be identified in turn and calculate the similarity between each voiceprint information to be identified and different voiceprint reference information respectively;

[0119] Step S633: for each voiceprint information to be identified, the voiceprint reference information with the greatest similarity and greater than a preset similarity threshold is screened out, and identity identification information is obtained from the identity mapping relationship according to the voiceprint reference information to determine the target participant;

[0120] In this embodiment, the voiceprint database is obtained by pre-recording the voice of each student to extract the corresponding voiceprint reference information, and the identity mapping relationship is used to form a matching mapping between the identity information and the corresponding voiceprint reference information. Exemplarily, in the identity mapping relationship, each student number corresponds to a name and a voiceprint reference information, and each student corresponds to a unique student number, so the name and student number can be packaged as the identity information corresponding to the voiceprint reference information.

[0121] In generating the participant list information based on at least one target participant, it specifically includes: creating the participant list information; adding the identity identification information corresponding to each target participant to the participant list information in turn, thereby completing the generation process of the participant list information.

[0122] like Figure 4 As shown, in the storage structure diagram of the warning table, the warning table includes multiple rows of record data, each row of record data includes a wristband warning information, the area ID corresponding to the wristband warning information, and the participant list information; illustratively, the first row of data in the warning table: wristband warning information 1, the area ID corresponding to the wristband warning information 1, and the participant list information; the second row of data in the warning table: wristband warning information 2, the area ID corresponding to the wristband warning information 2, and the participant list information... The nth row of data in the warning table: wristband warning information n, the area ID corresponding to the wristband warning information n, and the participant list information;

[0123] In Example 2, this embodiment provides a campus intelligent monitoring system, which corresponds to a campus intelligent monitoring method in the above-mentioned Example 1, and the same contents are not repeated here.

[0124] like Figure 5 As shown, the system includes: an information receiving module, which is used to receive wristband warning information, wherein the wristband warning information includes wristband coordinate points and positioning timestamps;

[0125] A target area determination module is used to determine the area ID according to the wristband coordinate point in the preset virtual campus map as the target area ID;

[0126] The image frame preprocessing module is used to determine all cameras corresponding to the target area ID and intercept them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed;

[0127] The campus behavior analysis module is used to analyze the image frame sequence set to be analyzed based on the pre-trained campus behavior analysis model to obtain the behavior analysis result. If the behavior analysis result is a conflicting label, a stop prompt is processed based on the wristband coordinate point. If the behavior analysis result is a non-conflicting label, no processing is performed;

[0128] The patrol reminder module is used to classify and record the wristband warning information based on the area ID in the warning table, set a key patrol list based on the warning table, and perform patrol reminder processing based on the key patrol list.

[0129] In Example 3, Figure 6As shown, this embodiment provides a terminal, including: at least one memory and at least one processor; wherein, at least one memory is used to store program code, and at least one processor is used to call the program code stored in at least one memory to execute any one of the campus intelligent monitoring methods in the above embodiments.

[0130] In embodiment 4, this embodiment provides a computer device, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is a physical layer for storing various databases. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a campus intelligent monitoring method is implemented.

[0131] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0132] In Example 5, this embodiment provides a storage medium, which is used to store program code, and the program code is used to execute the above-mentioned campus intelligent monitoring method.

[0133] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0134] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0135] The above embodiments are preferred implementation modes of the present application, but the implementation modes of the present application are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present application should be equivalent replacement methods and are included in the protection scope of the present application.

Claims

1. A campus intelligent monitoring method, characterized in that: Used for a cloud server, the cloud server is connected to at least one smart bracelet, multiple UWB base stations, multiple cameras, and multiple mobile terminals respectively; The method comprises: Receive wristband warning information, where the wristband warning information includes wristband coordinate points and positioning timestamp; Determine the area ID of the wristband coordinate point in the preset virtual campus map as the target area ID; Determine all cameras corresponding to the target area ID, and intercept them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed; Based on the pre-trained campus behavior analysis model, the image frame sequence set to be analyzed is analyzed to obtain the behavior analysis result. If the behavior analysis result is a conflicting label, a stop prompt is processed based on the wristband coordinate point. If the behavior analysis result is a non-conflicting label, no processing is performed; The method further comprises: Classify and record wristband warning information based on area ID in the warning table, set a key patrol list based on the warning table, and perform patrol reminder processing based on the key patrol list; In the triggered wristband warning information, at least one target participant is identified based on the target voice information, a participant list is generated based on the at least one target participant, and the participant list is added to the warning table according to the area ID corresponding to the wristband warning information; The step of identifying at least one target participant based on the target voice information specifically includes: When the threat analysis model is used to analyze the conversation text information to obtain a first threat analysis result, if the first threat analysis result is a threat label, the conversation voice information corresponding to the conversation text information is determined as the target voice information; Separating the target voice information into at least one reconstructed voice signal based on a pre-trained voice separation model, wherein the reconstructed voice signal is a voice signal corresponding to a participant in the conversation voice information; Performing voiceprint recognition on at least one reconstructed voice signal to determine a corresponding target participant; The speech separation model is obtained by pre-training with multiple sets of data through machine learning. During training, the multiple sets of data include multiple dialogue aliasing speech samples and multiple participant speech signals corresponding to each dialogue aliasing speech sample. The dialogue aliasing speech sample is obtained by aliasing any multiple participant speech signals.

2. The method according to claim 1, characterized in that The wristband warning information is triggered and generated when the smart wristband analyzes the conversation text information and determines it as a threat tag; When analyzing the conversation text information, specifically including: using a threat analysis model to analyze the conversation text information to obtain a first threat analysis result, when the first threat analysis result is a threat label, it indicates that the bracelet user has a threat risk in the current situation, and when the first threat analysis result is a non-threat label, it indicates that the bracelet user does not have a threat risk in the current situation; The threat analysis model is obtained by machine learning training using multiple sets of training data, the multiple sets of training data include first-category data and second-category data, each set of data in the first-category data includes: threatening dialogue training text and a threat label that identifies the threatening dialogue training text as having a threatening intent, and each set of data in the second-category data includes: non-threatening dialogue training text and a non-threatening label that identifies the non-threatening dialogue training text as having no threatening intent.

3. The method according to claim 1, characterized in that The campus behavior analysis model is obtained by machine learning training using multiple sets of training data, the multiple sets of training data include third-category data and fourth-category data, each set of data in the third-category data includes: violent conflict behavior training images and conflict labels indicating that violence exists in the violent conflict behavior training images, and each set of data in the fourth-category data includes: non-violent conflict behavior training images and non-conflict labels indicating that violence does not exist in the non-violent conflict behavior training images.

4. The method according to claim 3, characterized in that The process of stopping the prompt based on the wristband coordinate point specifically includes: Obtaining the terminal coordinate point of each mobile terminal; The relative horizontal distance between each mobile terminal and the wristband coordinate point is calculated respectively, which is specifically expressed as: d k =|x k -x0|+|y k -y0| d k represents the horizontal plane relative distance between the terminal coordinate point of the kth mobile terminal and the wristband coordinate point, x k represents the x-axis coordinate value of the terminal coordinate point of the kth mobile terminal, x0 represents the x-axis coordinate value of the wristband coordinate point, y k represents the y-axis coordinate value of the terminal coordinate point of the k-th mobile terminal, and y0 represents the y-axis coordinate value of the wristband coordinate point; After sorting in ascending order based on horizontal relative distance, the mobile terminal closest to the wristband coordinate point is used as the target prompt terminal; Generate route navigation guidance information for the target prompt terminal based on the wristband coordinate points; The image frame sequence set to be analyzed is sent to all mobile terminals, and the route navigation guidance information is sent to the target prompt terminal at the same time.

5. The method according to claim 4, characterized in that In generating route navigation guidance information for the target prompt terminal based on the wristband coordinate point, it specifically includes: taking the wristband coordinate point as the starting point and the terminal coordinate point corresponding to the target prompt terminal as the end point, performing path planning to find the shortest path, and packaging the path point set corresponding to the shortest path into route navigation guidance information in the guidance order.

6. The method according to claim 1, characterized in that The key patrol list set up based on the early warning table includes: Traverse the warning table in turn, classify and count according to the area ID to obtain the statistical times corresponding to different area IDs; Sort by the number of statistics from largest to smallest; Starting from the first sort number, filter the area IDs according to the preset filter quantity and add them to the priority patrol list; Count the warning time frequency of different area IDs in the key patrol list respectively; In the key patrol list, the highest warning time frequency of each area ID is set as a key patrol mark, and the preset time period corresponding to the highest warning time frequency is used as the target patrol time period.

7. The method according to claim 6, characterized in that In the patrol reminder processing based on the key patrol list, it specifically includes: The area ID and the corresponding target patrol time period are selected according to the key patrol identification, and the patrol prompt information is obtained by sorting the target patrol time period in chronological order from morning to night; The patrol reminder information is sent to the mobile terminal with the duty identification.

8. A campus intelligent monitoring system, characterized in that: For executing the method according to claim 1, the system comprises: An information receiving module is used to receive wristband warning information, wherein the wristband warning information includes wristband coordinate points and positioning timestamps; A target area determination module, used to determine the area ID to which the wristband coordinate point belongs in a preset virtual campus map as the target area ID; An image frame preprocessing module is used to determine all cameras corresponding to the target area ID, and intercept them according to the positioning timestamp to obtain a set of image frame sequences to be analyzed; The campus behavior analysis module is used to analyze the image frame sequence set to be analyzed based on the pre-trained campus behavior analysis model to obtain the behavior analysis result. If the behavior analysis result is a conflicting label, a stop prompt is processed based on the wristband coordinate point. If the behavior analysis result is a non-conflicting label, no processing is performed; The patrol reminder module is used to classify and record the wristband warning information based on the area ID in the warning table, set a key patrol list based on the warning table, and perform patrol reminder processing based on the key patrol list.

Images