CN112257546B - Event early warning method and device, electronic equipment and storage medium - Google Patents

Event early warning method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN112257546B
CN112257546B CN202011116823.2A CN202011116823A CN112257546B CN 112257546 B CN112257546 B CN 112257546B CN 202011116823 A CN202011116823 A CN 202011116823A CN 112257546 B CN112257546 B CN 112257546B
Authority
CN
China
Prior art keywords
event
sample
pair
result
cause
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202011116823.2A
Other languages
Chinese (zh)
Other versions
CN112257546A (en
Inventor
周迪
徐爱华
贺正方
邓黄燕
杨齐期
何斌
沈润杰
朱忠攀
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang Uniview Technologies Co Ltd
Original Assignee
Zhejiang Uniview Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang Uniview Technologies Co Ltd filed Critical Zhejiang Uniview Technologies Co Ltd
Priority to CN202011116823.2A priority Critical patent/CN112257546B/en
Publication of CN112257546A publication Critical patent/CN112257546A/en
Application granted granted Critical
Publication of CN112257546B publication Critical patent/CN112257546B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/44—Event detection

Abstract

The application discloses an event early warning method, which comprises the following steps: acquiring a monitoring picture shot by monitoring equipment, and identifying a current event corresponding to the monitoring picture; querying a predicted event associated with the current event by using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events; and if the predicted event is a target type event, generating early warning information corresponding to the predicted event. The method and the device can be used for early warning before the malignant event occurs. The application also discloses an event early warning device, electronic equipment and a storage medium, which have the beneficial effects.

Description

Event early warning method and device, electronic equipment and storage medium
Technical Field
The present disclosure relates to the field of security detection technologies, and in particular, to an event early warning method and apparatus, an electronic device, and a storage medium.
Background
The safety monitoring system can monitor a certain area through pictures shot by cameras arranged at various positions. With the development of artificial intelligence, machine vision technology is gradually applied to a safety monitoring system, and machine vision can identify specific events corresponding to pictures shot by a camera through a trained neural network model. For example, after detecting malignant events such as crashing, frame hitting and the like, the safety monitoring system can alarm according to the event identification result. However, the above-mentioned method can only give an alarm after a malignant event occurs, and when the malignant event is found to occur through machine vision, the loss of the property of the person is already caused, and the malignant event cannot be prevented in time.
Therefore, how to perform early warning before occurrence of a malignant event is a technical problem that a person skilled in the art needs to solve at present.
Disclosure of Invention
The application aims to provide an event early warning method, an event early warning device, electronic equipment and a storage medium, which can perform early warning before a malignant event occurs.
In order to solve the technical problems, the application provides an event early warning method, which comprises the following steps:
acquiring a monitoring picture shot by monitoring equipment, and identifying a current event corresponding to the monitoring picture;
querying a predicted event associated with the current event by using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events;
and if the predicted event is a target type event, generating early warning information corresponding to the predicted event.
Optionally, the process of constructing the event knowledge graph includes:
taking a monitoring picture shot by monitoring equipment in a preset area in a preset time period as a sample picture, and identifying a sample event corresponding to the sample picture;
selecting a plurality of sample event pairs from the sample events; each sample event pair comprises a first-occurring cause event and a later-occurring result event, the event occurrence time difference between the cause event and the result event is smaller than a first preset value, the event occurrence distance between the cause event and the result event is smaller than a second preset value, and the times of successively occurring the cause event and the result event are larger than a third preset value;
And constructing the event knowledge graph according to the association relation between the cause event and the result event in the sample event pair.
Optionally, before selecting the sample event pair from the sample events, the method further includes:
setting the third preset value according to the area size of the preset area; wherein the region size of the preset region is positively correlated with the third preset value;
and/or setting the third preset value according to the time span value of the preset time period; wherein the time span value of the preset time period is positively correlated with the third preset value.
Optionally, after constructing the event knowledge graph according to the association relationship between the cause event and the result event in the sample event pair, the method further includes:
determining an event group according to the incidence relation between the sample event pairs; the event group comprises a first sample event pair and a second sample event pair, the result event of the first sample event pair is the cause event of the second sample event pair, the times of occurrence of the cause event and the result event in the first sample event pair and the second sample event pair are both larger than a fourth preset value, and the fourth preset value is larger than the third preset value;
Generating a logic event pair according to the event group; wherein the cause event of the logical event pair is a cause event of the first sample event pair, and the result event of the logical event pair is a result event of the second sample event pair;
and adding the association relation between the cause event and the result event in the logic event pair to the event knowledge graph.
Optionally, after identifying the sample event corresponding to the sample picture, the method further includes:
setting a plurality of sample events occurring in a target time and a target area as a companion event combination;
adding association relations of any two sample events in the event knowledge graph in the event combination, and adding binding marks for the events belonging to the same event combination;
correspondingly, after determining that the predicted event is a target type event, the method further comprises:
judging whether the predicted event in the event knowledge graph is added with the accompanying mark or not;
if yes, inquiring the event knowledge graph to which the same accompanying mark as the predicted event is added;
and if the accompanying event is the target type event, generating prompt information corresponding to the accompanying event.
Optionally, after identifying the sample event corresponding to the sample picture, the method further includes:
selecting an alternative event pair from the sample events; the event occurrence time difference between the alternative cause event and the alternative result event is smaller than a first preset value, the event occurrence distance between the alternative cause event and the alternative result event is smaller than a second preset value, and the times of occurrence of the alternative cause event and the alternative result event are smaller than or equal to a third preset value and larger than a fifth preset value;
setting the alternative event pairs with the same alternative result event as a homologous event combination;
setting all the alternative causative events in the homologous event combination as reference causative events, and setting the alternative result events of the homologous event combination as reference result events;
correspondingly, after identifying the current event corresponding to the monitoring picture, the method further comprises the following steps:
judging whether the number of the reference cause events included in the current event is larger than a sixth preset value;
if yes, judging that the predicted event corresponding to the current event is the reference result event.
Optionally, generating the early warning information corresponding to the predicted event includes:
and generating early warning information corresponding to the predicted event according to the occurrence time and the occurrence position of the current event.
The application also provides an event early warning device, which comprises:
the event identification module is used for acquiring a monitoring picture shot by the monitoring equipment and identifying a current event corresponding to the monitoring picture;
the event prediction module is used for inquiring a predicted event associated with the current event by using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events;
and the event early warning module is used for generating early warning information corresponding to the predicted event if the predicted event is a target type event.
The present application also provides a storage medium having stored thereon a computer program which, when executed, implements the steps performed by the event early warning method described above.
The application also provides electronic equipment, which comprises a memory and a processor, wherein the memory stores a computer program, and the processor realizes the steps executed by the event early warning method when calling the computer program in the memory.
The application provides an event early warning method, which comprises the following steps: acquiring a monitoring picture shot by monitoring equipment, and identifying a current event corresponding to the monitoring picture; querying a predicted event associated with the current event by using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events; and if the predicted event is a target type event, generating early warning information corresponding to the predicted event.
After a monitoring picture shot by the monitoring equipment is acquired, the current event corresponding to the monitoring picture is identified. The event knowledge graph stores the association relation among a plurality of events, and after the current event is determined, the predicted event associated with the current event can be obtained by prediction according to the event relation graph. If the predicted event is an event requiring alarm, generating early warning information corresponding to the predicted time. Because a certain sequence exists in the occurrence of the events, the early warning can be carried out before the occurrence of the target type event, rather than the alarm after the occurrence of the target type event, and the damage caused by the occurrence of the target type event can be effectively reduced. The application also provides an event early warning device, an electronic device and a storage medium, which have the beneficial effects and are not described in detail herein.
Drawings
For a clearer description of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described, it being apparent that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an event early warning method provided in an embodiment of the present application;
fig. 2 is an example schematic diagram of an event knowledge graph according to an embodiment of the present application;
FIG. 3 is a flowchart of a method for constructing an event knowledge graph according to an embodiment of the present application;
fig. 4 is a flowchart of an optimization method of an event knowledge graph according to an embodiment of the present application;
fig. 5 is a schematic structural diagram of an event early warning device provided in an embodiment of the present application.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
Referring to fig. 1, fig. 1 is a flowchart of an event early warning method according to an embodiment of the present application.
The specific steps may include:
s101: and acquiring a monitoring picture shot by the monitoring equipment, and identifying a current event corresponding to the monitoring picture.
The embodiment is applied to electronic equipment such as a safety monitoring device and a data analysis device, the electronic equipment can be connected with the monitoring equipment so as to receive monitoring pictures transmitted by the monitoring equipment for analysis, and the electronic equipment can also be connected with a database for storing monitoring data so as to analyze the monitoring pictures stored in the database.
The number of the monitoring devices is not limited, and monitoring pictures shot by any number of the monitoring devices can be acquired. The above-mentioned monitoring picture can be any number of frame images captured from the monitoring video, can be a video picture of a continuous monitoring video, and can also be an image captured by the monitoring device. After the monitoring picture shot by the monitoring equipment is obtained, an event identification operation can be performed on the monitoring picture to obtain a current event corresponding to the monitoring picture. The current event refers to an event that is occurring when a monitoring screen is photographed.
As a possible implementation, this step may identify the current event corresponding to the monitoring screen through machine vision technology. In the process of identifying the current event by utilizing the machine vision technology, the monitoring picture can be input into the trained neural network model, and the current event corresponding to the monitoring picture can be determined according to the output result of the neural network model.
S102: and querying the predicted event associated with the current event by using the event knowledge graph.
Before this step, there may be an operation of constructing an event knowledge graph, which is a knowledge graph for describing an event relationship. Knowledge maps are structured semantic knowledge bases that are used to rapidly describe concepts and their interrelationships in the physical world. The logic architecture in the event knowledge graph is as follows: { event 1, relationship, event 2}, event is the most basic element in the event knowledge graph, relationship refers to the relationship between events, and in the logic framework of the event knowledge graph, relationship is generally the association relationship of events, and the symbol "→". For example, the logical architecture { event 1, →, event 2} represents: the occurrence of event 1 results in the occurrence of event 2. The event knowledge graph can store the association relation among a plurality of events, so that after the current event is determined, the event knowledge graph can be used for determining the event which will happen after the current event, namely, the predicted event associated with the current event is queried.
With the deep application of the big data technology in the public safety field, the capabilities of information insight, analysis, research and judgment, investigation and hit and command management can be improved by integrating various data, constructing a multidimensional analysis model and the like. With the continuous deep data fusion and continuous integration of business modeling, the demands of public safety big data on more deep associated mining capability, more intelligent early warning prediction capability and more comprehensive analysis and research capability become more urgent. The embodiment applies the knowledge graph to the public security field, and performs relation depth mining, intelligent case reasoning and active event prediction by means of strong interconnection and reasoning capability of the knowledge graph. Since much information in public safety domain is dynamically generated, the embodiment can be to use event knowledge graph including dynamic event association relationship.
As a possible implementation, the number of predicted events associated with a current event may be any number, and the plurality of predicted events associated with the plurality of current events may be the same event.
S103: and if the predicted event is a target type event, generating early warning information corresponding to the predicted event.
Before this step, there may be an operation of determining whether the predicted event is a target type event, and if the predicted event is not a target type event, the process may be ended or a new monitoring screen may be acquired so as to reenter the operation process of this embodiment. The target type event can be an event preset by a user, and can be set according to an actual application scene. For example, in a scenario of monitoring community streets, target type events may include theft events, racking events, fire events, explosion events, and the like; in the context of monitoring traffic conditions, the target type event may include an speeding event, a rear-end event, a vehicle porcelain event, etc.; in the scenario of monitoring a banking transaction area, the target type event may include a queue event, a robbery event, and the like. If the predicted event is a target type event, each predicted event can have corresponding early warning information, and the early warning information corresponding to the predicted event can be generated so as to prevent the occurrence of the predicted time or reduce the harm caused by the occurrence of the early warning event.
As a possible implementation manner, the process of generating the early warning information may be: and generating early warning information corresponding to the predicted event according to the occurrence time and the occurrence position of the current event. For example, the predicted event of the personnel gathering event is a frame-taking event, so after the personnel gathering event is detected, early warning information corresponding to the predicted event can be generated according to the occurrence time and the occurrence position of the personnel gathering event. The embodiment determines the causal relationship between the cause event and the result event based on the event knowledge graph, and when the camera system identifies the cause event, the cause event does not need to be alarmed, but when the result event of the cause event is a target type event, the early warning is carried out, so that the alarm is not carried out after the loss of the property of the person caused by the target type event occurs.
In the embodiment, before the occurrence of the malignant event, the event in germination can be found through machine vision, and the possibility of the occurrence of the malignant event is judged through an event knowledge graph, so that early warning is performed. The traditional knowledge graph is static information association, and in the public safety field, a lot of information is dynamically generated, and the event knowledge graph comprises dynamic event association, so that early warning of safety risks is facilitated. An example of an event knowledge graph is shown in fig. 2, and fig. 2 is a schematic diagram of an example of an event knowledge graph according to an embodiment of the present application. Through monitoring and event identification of the video monitoring system, a frame-taking event is frequently generated after a quarrying event in certain occasions, an injury event is frequently generated after the frame-taking event, a hospitalization event is frequently generated after the injury event, and a treatment event, a discharge event, a return event and the like are frequently generated after the hospitalization event. When causal association occurs with high probability among events, an event knowledge graph represented by an arrow is established.
In this embodiment, after a monitoring picture shot by a monitoring device is acquired, a current event corresponding to the monitoring picture is identified. The event knowledge graph stores the association relation among a plurality of events, and after the current event is determined, the predicted event associated with the current event can be obtained by prediction according to the event relation graph. If the predicted event is an event requiring alarm, generating early warning information corresponding to the predicted time. Because a certain sequence exists in the occurrence of the events, the early warning can be carried out before the occurrence of the target type event, rather than the alarm after the occurrence of the target type event, and the damage caused by the occurrence of the target type event can be effectively reduced.
Referring to fig. 3, fig. 3 is a flowchart of a method for constructing an event knowledge graph provided by an embodiment of the present application, where the embodiment specifically describes a process for constructing an event knowledge graph described in an embodiment corresponding to fig. 1, and the embodiment may be combined with an embodiment corresponding to fig. 1 to obtain a further implementation, and the embodiment may include the following steps:
s201: taking a monitoring picture shot by monitoring equipment in a preset area in a preset time period as a sample picture, and identifying a sample event corresponding to the sample picture.
The embodiment provides a scheme for selecting a sample picture to construct an event knowledge graph, wherein the sample picture is a monitoring picture shot by monitoring equipment in a preset area in a preset time period. As a possible embodiment, the sample screen is a monitoring screen that has been obtained before S201 is performed. Of course, the sample picture may also be a monitoring picture obtained in the process of executing S201, that is, the monitoring picture currently shot by the monitoring device in the preset area is analyzed in real time, and the sample event corresponding to the sample picture is identified.
S202: a plurality of sample event pairs are selected from the sample events.
After obtaining the sample event, a sample event pair can be selected according to the occurrence time and the occurrence position of the sample event. Each sample event pair comprises a first-occurring cause event and a later-occurring result event, the event occurrence time difference between the cause event and the result event is smaller than a first preset value, the event occurrence distance between the cause event and the result event is smaller than a second preset value, and the times of successively occurring the cause event and the result event are larger than a third preset value. Event occurrence distance refers to the distance between two event occurrence sites.
Illustrating the number of times the causative event and the resultant event occur sequentially, if 12:00, a frame-up event occurs, 12:15, when a personnel gathering event occurs, judging that the times of occurrence of the cause event and the result event are 1; if 12:20, a frame-up event occurs, 12: and 25, judging that the times of the occurrence of the cause event and the result event are 2, and so on, and adding 1 to the times of the occurrence of the cause event and the result event each time the occurrence of the frame-noisy event and the people accumulation event are detected. The recorded number of causative events and resulting events do not participate in subsequent number of times calculations.
Therefore, if the time difference between the event occurrence time of a first event and the event occurrence time of a second event is smaller than the first preset value, the event occurrence distance is smaller than the second preset value, and the number of times of occurrence of the two events is larger than the third preset value, the first event is considered to cause the second event.
S203: and constructing an event knowledge graph according to the association relation between the cause event and the result event in the sample event pair.
After obtaining the sample event pair, an event knowledge graph can be constructed according to the association relationship between the cause event and the result event in the sample event pair, and the association relationship between the events in the event knowledge graph can be saved through the following logic structure: { cause event, →result event }. The logical framework with formal relation in the event knowledge graph can make reasoning and early warning. For example, the logical framework described above has been established and if the camera finds a racking event by machine vision, then it can be predicted that a racking event is likely to occur. The frame-knocking event does not need to be alarmed, but the frame-knocking event needs to be alarmed, and the possible frame-knocking event is predicted to occur through reasoning of a logic frame in the event knowledge graph, so that early warning information can be generated, and the content of the early warning information can be as follows: a racking event may occur.
The following describes the process of event knowledge graph by way of example in practical application: the method comprises the steps of utilizing a large number of cameras in a city to identify events through machine vision, establishing event records and storing time information and space information associated with the events. Establishing a preliminary relation between the cause event and the result event according to the time confidence and the region confidence, counting the relation, wherein the initial value is 1, and increasing the preliminary relation value by 1 every time the cause event and the result event occur successively; when the preliminary relation value reaches a certain threshold value, the preliminary relation between the cause event and the result event can be upgraded into a formal relation, and then a sample event pair comprising the cause event and the result event is obtained. When the camera system finds that the cause event occurs in a certain place, the platform gives early warning, and the content of the early warning is as follows: as a result, events may occur.
Further, the present embodiment may identify an event corresponding to the view frame by using a camera with a machine vision function, for example, the camera with the machine vision function may identify an event such as a frame strike, a running, a car accident, a fire, etc. Of course, a camera without machine vision function can transmit video or image to the back-end platform, and the background can recognize the event corresponding to the monitoring picture. After the event corresponding to the monitoring picture is identified, the event type, the occurrence time, the occurrence place, the number of people and other attributes of the event can be recorded. After identifying the event corresponding to the monitoring picture, the preliminary relationship between the events can be established in real time according to the preset time confidence (for example, 7 days) and the regional confidence (for example, radius 1 km), so as to form a logic architecture { event 1, →event 2} with the preliminary relationship. The relation points to the event with later time from the event with earlier time, if the two events happen at the same time, two logic frameworks are established, and the points of the relation are just opposite, so that two logic frameworks of { event 1, →event 2} and { event 2, →event 1} are obtained.
After obtaining the logic frame with the preliminary relationship, the preliminary relationship values of the logic frame can be counted; i.e. the same logical frame is built once, the preliminary relationship value is added to 1, the preliminary relationship value having an initial value of 1. For example, a noisy event occurs somewhere, a time period (time span is less than or equal to time confidence), a cradling event occurs in the vicinity (radius is less than or equal to region confidence of the noisy event occurrence place), and then a logical frame { noisy event, →cradling event }, in which the preliminary relationship value is 1, is established; if the initial relation value of the frames is increased by 1 to become 2, and so on, if the initial relation value reaches a threshold value (for example, 3) set by a certain administrator, the initial relation can be upgraded into a formal relation, and the events included in the logic frames with the formal relation belong to the same sample event pair. The embodiment can add the logic framework with the formal relation to the event knowledge graph so as to continue event reasoning and event early warning by using the logic framework with the formal relation.
As a further introduction to the corresponding embodiment of fig. 3, there may also be an operation of setting a third preset value before selecting a sample event pair from the sample events. As a possible implementation manner, the present embodiment may set the third preset value according to the area size of the preset area; wherein the area size of the preset area is positively correlated with the third preset value. For example, if only the monitoring screen shot by the camera in beijing is used as the sample screen for constructing the event knowledge graph, the third preset value may be set to 100; if the monitoring picture shot by the cameras in Beijing and Hangzhou is taken as a sample picture for constructing the event knowledge graph, the third preset value can be set to be 200. As another possible implementation manner, the present embodiment may further set the third preset value according to a time span value of the preset time period; wherein the time span value of the preset time period is positively correlated with the third preset value. For example, if a monitoring screen shot by the camera within one month is used as a sample screen for constructing an event knowledge graph, the third preset value may be set to 50; if the monitoring picture shot by the camera in three months is taken as a sample picture for constructing the event knowledge graph, the third preset value can be set to 150. Of course, the present embodiment may determine the third preset value by combining the area size of the preset area and the time span value of the preset time period.
Referring to fig. 4, fig. 4 is a flowchart of an event knowledge graph optimization method provided by an embodiment of the present application, where after the event knowledge graph is constructed, the embodiment describes a scheme for optimizing a sample event pair in the event knowledge graph, and the embodiment may be combined with an embodiment corresponding to fig. 3 to obtain a further implementation, and the embodiment may include the following steps:
s301: determining an event group according to the incidence relation between the sample event pairs;
based on the obtained sample event pairs, the method can combine a plurality of sample event pairs so as to improve timeliness and accuracy of early warning. The present embodiment may determine the event group according to the following rule: the result event of the first sample event pair is a cause event of the second sample event pair, the times of occurrence of the cause event and the result event in the first sample event pair and the second sample event pair are both larger than a fourth preset value, and the fourth preset value is larger than the third preset value. If there are a first sample event pair and a second sample event pair that meet the above rule, an event group including the first sample event pair and the second sample event pair may be constructed. Illustrating the above procedure, if there are two sample event pairs: { noisy events }, personnel gathering events }, { personnel gathering events }, }, and the occurrence times of the two sample event pairs meet the preset requirement, the { noisy events }, the personnel gathering events }, { personnel gathering events, }, and the cradling events } can be considered to belong to the same event group.
S302: logical event pairs are generated from the event clusters.
The cause event of the logic event pair is the cause event of the first sample event pair, and the result event of the logic event pair is the result event of the second sample event pair. Taking the same event group as an example, the two sample event pairs including { noisy events, →people gathering events } and { people gathering events, →getting-frame events }, the logical event pair generated in this step is { noisy events, →getting-frame events }.
S303: and adding the association relation between the cause event and the result event in the logic event pair to the event knowledge graph.
The above embodiment describes a scheme for optimizing a logic framework in an event knowledge graph by using big data, and proposes a file for establishing the logic framework by using logic between events besides time confidence and region confidence. For example, the following logical framework is established based on time confidence and region confidence: the logical frames of the noisy events, the personnel gathering events and the cradling events are not established due to the limitation of the time confidence and/or the region confidence, so that the cradling events are found in the camera system and the cradling events are not early-warned. However, in practice, this logic is generally true, so this embodiment may further mine the association between the logic frames by generating the logic event pairs as described above, and may establish the logic frames of { noisy events, →noisy events }, by "shorting". Thus, a frame-up event is found in the camera system, and the frame-up event can be early-warned. Of course, the generation of the logic event pairs needs to take big data as a basis, that is, the times of occurrence of the cause event and the result event in the first sample event pair and the second sample event pair are both larger than a fourth preset value, the fourth preset value is larger than the third preset value, and the time confidence and the region confidence can be shortened and the accuracy of the event knowledge graph is improved by the generation mode of the logic event pairs.
As a further introduction to the corresponding embodiment of fig. 3, after identifying the sample event corresponding to the sample screen, a plurality of sample events occurring in the target time and the target area may also be set as a companion event combination; and adding the association relation of any two sample events in the event knowledge graph in the event combination, and adding binding marks for the events belonging to the same event combination. Correspondingly, after the predicted event is judged to be the target type event, whether the accompanying mark is added to the predicted event in the event knowledge graph or not can be judged; if yes, inquiring the event knowledge graph to which the same accompanying mark as the predicted event is added; and if the accompanying event is the target type event, generating prompt information corresponding to the accompanying event.
The present embodiment constructs two events occurring within a period of time before and after (even at the same time as) the same camera as a companion event combination, and if any event in the companion event combination is detected to occur, it can be considered that all events in the companion event combination may occur. For example, if a camera finds that a fire event is accompanied by an explosion event multiple times, a logical framework { fire event, →, explosion event } and { explosion event, →, fire event } may be established. If the early warning information of the fire event is generated, the early warning information of the explosion event can be generated; otherwise, if the early warning information event of the explosion event is generated, the early warning information of the fire event can be generated at the same time. Since no explosion event occurs when a fire event occurs sometimes, it is likely that an explosion event eventually occurs as the fire propagates, such an early warning as described above can effectively alert relevant personnel to the process.
As a further introduction to the corresponding embodiment of fig. 3, after identifying the sample event corresponding to the sample picture, before constructing the event knowledge graph, the predicted event corresponding to the current event may also be determined by:
step 1: selecting an alternative event pair from the sample events;
the event occurrence time difference between the alternative cause event and the alternative result event is smaller than a first preset value, the event occurrence distance between the alternative cause event and the alternative result event is smaller than a second preset value, and the times of occurrence of the alternative cause event and the alternative result event are smaller than or equal to a third preset value and larger than a fifth preset value;
step 2: setting the alternative event pairs with the same alternative result event as a homologous event combination;
step 3: setting all the alternative causative events in the homologous event combination as reference causative events, and setting the alternative result events of the homologous event combination as reference result events;
step 4: after the current event corresponding to the monitoring picture is identified, judging whether the number of the reference caused events included in the current event is larger than a sixth preset value or not; if yes, judging that the predicted event corresponding to the current event is the reference result event.
In the process of constructing the event knowledge graph, the event combination which does not accord with the sample event pair standard can be used for selecting an alternative event pair, and the alternative event pair is used as a reference for determining the predicted event. When multiple causative events point to the same result event through alternative event pairs, the result event can be used as a predicted event, and early warning is carried out when the predicted time is a target type event. For example, the following alternative event pairs are included in the causal event combination: { A event, →X event }, { B event, →X event }, { C event, { X event }, { D event, { X event }, { E event, { X event }; if the current event corresponding to the monitoring picture comprises an event A, an event B and an event C and the sixth preset value is 2, the predicted event can be judged to be an event X. And if the X event is a target type event, generating early warning information corresponding to the X event. That is, when a plurality of simultaneous causative events point to the same resultant event through the association relationship recorded in the homologous event group, the corresponding frames may be counted by the preliminary relationship value, and when the sum of the counts is greater than a sixth preset value, it is determined that the resultant event is about to occur.
Referring to fig. 5, fig. 5 is a schematic structural diagram of an event early warning device provided in an embodiment of the present application;
the apparatus may include:
the event identification module 100 is configured to obtain a monitoring picture shot by a monitoring device, and identify a current event corresponding to the monitoring picture;
an event prediction module 200, configured to query a predicted event associated with the current event using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events;
the event early warning module 300 is configured to generate early warning information corresponding to the predicted event if the predicted event is a target type event.
In this embodiment, after a monitoring picture shot by a monitoring device is acquired, a current event corresponding to the monitoring picture is identified. The event knowledge graph stores the association relation among a plurality of events, and after the current event is determined, the predicted event associated with the current event can be obtained by prediction according to the event relation graph. If the predicted event is an event requiring alarm, generating early warning information corresponding to the predicted time. Because a certain sequence exists in the occurrence of the events, the early warning can be carried out before the occurrence of the target type event, rather than the alarm after the occurrence of the target type event, and the damage caused by the occurrence of the target type event can be effectively reduced.
Optionally, the method further comprises:
the sample time identifying module is used for taking a monitoring picture shot by monitoring equipment in a preset area in a preset time period as a sample picture and identifying a sample event corresponding to the sample picture;
a sample event pair selecting module, configured to select a plurality of sample event pairs from the sample events; each sample event pair comprises a first-occurring cause event and a later-occurring result event, the event occurrence time difference between the cause event and the result event is smaller than a first preset value, the event occurrence distance between the cause event and the result event is smaller than a second preset value, and the times of successively occurring the cause event and the result event are larger than a third preset value;
and the event knowledge graph construction module is used for constructing the event knowledge graph according to the association relation between the cause event and the result event in the sample event pair.
Further, the method further comprises the following steps:
the first parameter setting module is used for setting the third preset value according to the area size of the preset area; wherein the region size of the preset region is positively correlated with the third preset value;
and/or a second parameter setting module, configured to set the third preset value according to a time span value of the preset time period; wherein the time span value of the preset time period is positively correlated with the third preset value.
Further, the method further comprises the following steps:
the event group construction module is used for determining an event group according to the incidence relation between the sample event pairs after constructing the event knowledge graph according to the incidence relation between the cause event and the result event in the sample event pairs; the event group comprises a first sample event pair and a second sample event pair, the result event of the first sample event pair is the cause event of the second sample event pair, the times of occurrence of the cause event and the result event in the first sample event pair and the second sample event pair are both larger than a fourth preset value, and the fourth preset value is larger than the third preset value;
the logic event pair generating module is used for generating logic event pairs according to the event group; wherein the cause event of the logical event pair is a cause event of the first sample event pair, and the result event of the logical event pair is a result event of the second sample event pair;
and the event knowledge graph optimization module is used for adding the association relation between the cause event and the result event in the logic event pair to the event knowledge graph.
Further, the method further comprises the following steps:
The accompanying event setting module is used for setting a plurality of sample events which occur in a target time and a target area as an accompanying event combination after the sample events corresponding to the sample pictures are identified;
the marking module is used for adding the association relation of any two sample events in the event combination in the event knowledge graph and adding binding marks for the events belonging to the same event combination;
the accompanying event early warning module is used for judging whether the predicted event is added with the accompanying mark in the event knowledge graph after judging that the predicted event is a target type event; if yes, inquiring the event knowledge graph to which the same accompanying mark as the predicted event is added; and if the accompanying event is the target type event, generating prompt information corresponding to the accompanying event.
Further, the method further comprises the following steps:
the homologous event combination construction module is used for selecting an alternative event pair from the sample events after identifying the sample event corresponding to the sample picture; the event occurrence time difference between the alternative cause event and the alternative result event is smaller than a first preset value, the event occurrence distance between the alternative cause event and the alternative result event is smaller than a second preset value, and the times of occurrence of the alternative cause event and the alternative result event are smaller than or equal to a third preset value and larger than a fifth preset value; the method is also used for setting the alternative event pairs with the same alternative result event as a homologous event combination; the method is also used for setting all the alternative causative events in the homologous event combination as reference causative events and setting the alternative result events of the homologous event combination as reference result events;
The joint judgment module is used for judging whether the number of the reference caused events included in the current event is larger than a sixth preset value or not; if yes, judging that the predicted event corresponding to the current event is the reference result event.
Further, the event early warning module is specifically a module for generating early warning information corresponding to the predicted event according to the occurrence time and the occurrence position of the current event.
Since the embodiments of the apparatus portion and the embodiments of the method portion correspond to each other, the embodiments of the apparatus portion are referred to the description of the embodiments of the method portion, and are not repeated herein.
The present application also provides a storage medium having stored thereon a computer program which, when executed, performs the steps provided by the above embodiments. The storage medium may include: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The application also provides an electronic device, which may include a memory and a processor, where the memory stores a computer program, and the processor may implement the steps provided in the foregoing embodiments when calling the computer program in the memory. Of course the electronic device may also include various network interfaces, power supplies, etc.
In the description, each embodiment is described in a progressive manner, and each embodiment is mainly described by the differences from other embodiments, so that the same similar parts among the embodiments are mutually referred. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section. It should be noted that it would be obvious to those skilled in the art that various improvements and modifications can be made to the present application without departing from the principles of the present application, and such improvements and modifications fall within the scope of the claims of the present application.
It should also be noted that in this specification, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

Claims (9)

1. An event early warning method is characterized by comprising the following steps:
acquiring a monitoring picture shot by monitoring equipment, and identifying a current event corresponding to the monitoring picture;
querying a predicted event associated with the current event by using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events;
if the predicted event is a target type event, generating early warning information corresponding to the predicted event;
after the event knowledge graph is constructed, the process of optimizing the sample event pairs in the event knowledge graph comprises the following steps:
determining an event group according to the incidence relation between the sample event pairs; the sample event pair comprises a first-occurring cause event and a later-occurring result event, the event occurrence time difference between the cause event and the result event is smaller than a first preset value, the event occurrence distance between the cause event and the result event is smaller than a second preset value, and the times of successively occurring the cause event and the result event are larger than a third preset value; the event group comprises a first sample event pair and a second sample event pair, the result event of the first sample event pair is a cause event of the second sample event pair, the times of occurrence of the cause event and the result event in the first sample event pair and the second sample event pair are both larger than a fourth preset value, and the fourth preset value is larger than the third preset value;
Generating a logic event pair according to the event group; wherein the cause event of the logical event pair is a cause event of the first sample event pair, and the result event of the logical event pair is a result event of the second sample event pair;
and adding the association relation between the cause event and the result event in the logic event pair to the event knowledge graph.
2. The event pre-warning method according to claim 1, wherein the process of constructing the event knowledge graph comprises:
taking a monitoring picture shot by monitoring equipment in a preset area in a preset time period as a sample picture, and identifying a sample event corresponding to the sample picture;
selecting a plurality of sample event pairs from the sample events;
and constructing the event knowledge graph according to the association relation between the cause event and the result event in all the sample event pairs.
3. The event pre-warning method of claim 2, further comprising, prior to selecting a sample event pair from the sample events:
setting the third preset value according to the area size of the preset area; wherein the region size of the preset region is positively correlated with the third preset value;
And/or setting the third preset value according to the time span value of the preset time period; wherein the time span value of the preset time period is positively correlated with the third preset value.
4. The event early warning method according to claim 2, further comprising, after identifying a sample event corresponding to the sample screen:
setting a plurality of sample events occurring in a target time and a target area as a companion event combination;
adding association relations of any two sample events in the event knowledge graph in the event combination, and adding binding marks for the events belonging to the same event combination;
correspondingly, after determining that the predicted event is a target type event, the method further comprises:
judging whether the predicted event in the event knowledge graph is added with the binding mark or not;
if yes, inquiring the event knowledge graph to which the same binding mark as the predicted event is added;
and if the accompanying event is the target type event, generating prompt information corresponding to the accompanying event.
5. The event early warning method according to claim 2, further comprising, after identifying a sample event corresponding to the sample screen:
Selecting an alternative event pair from the sample events; the event occurrence time difference between the alternative cause event and the alternative result event is smaller than a first preset value, the event occurrence distance between the alternative cause event and the alternative result event is smaller than a second preset value, and the times of occurrence of the alternative cause event and the alternative result event are smaller than or equal to a third preset value and larger than a fifth preset value;
setting the alternative event pairs with the same alternative result event as a homologous event combination;
setting all the alternative causative events in the homologous event combination as reference causative events, and setting the alternative result events of the homologous event combination as reference result events;
correspondingly, after identifying the current event corresponding to the monitoring picture, the method further comprises the following steps:
judging whether the number of the reference cause events included in the current event is larger than a sixth preset value;
if yes, judging that the predicted event corresponding to the current event is the reference result event.
6. The event early warning method according to any one of claims 1 to 5, wherein generating early warning information corresponding to the predicted event includes:
And generating early warning information corresponding to the predicted event according to the occurrence time and the occurrence position of the current event.
7. An event early warning device, characterized by comprising:
the event identification module is used for acquiring a monitoring picture shot by the monitoring equipment and identifying a current event corresponding to the monitoring picture;
the event prediction module is used for inquiring a predicted event associated with the current event by using an event knowledge graph; wherein, the event knowledge graph stores the association relation between events;
the event early warning module is used for generating early warning information corresponding to the predicted event if the predicted event is a target type event;
the event group construction module is used for determining an event group according to the incidence relation between the sample event pairs after constructing the event knowledge graph according to the incidence relation between the cause event and the result event in the sample event pairs; the sample event pair comprises a first-occurring cause event and a later-occurring result event, the event occurrence time difference between the cause event and the result event is smaller than a first preset value, the event occurrence distance between the cause event and the result event is smaller than a second preset value, and the times of successively occurring the cause event and the result event are larger than a third preset value; the event group comprises a first sample event pair and a second sample event pair, the result event of the first sample event pair is a cause event of the second sample event pair, the times of occurrence of the cause event and the result event in the first sample event pair and the second sample event pair are both larger than a fourth preset value, and the fourth preset value is larger than the third preset value;
The logic event pair generating module is used for generating logic event pairs according to the event group; wherein the cause event of the logical event pair is a cause event of the first sample event pair, and the result event of the logical event pair is a result event of the second sample event pair;
and the event knowledge graph optimization module is used for adding the association relation between the cause event and the result event in the logic event pair to the event knowledge graph.
8. An electronic device comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the steps of the event early warning method of any one of claims 1 to 6 when the computer program in the memory is invoked by the processor.
9. A storage medium having stored therein computer executable instructions which when loaded and executed by a processor implement the steps of the event early warning method of any one of claims 1 to 6.
CN202011116823.2A 2020-10-19 2020-10-19 Event early warning method and device, electronic equipment and storage medium Active CN112257546B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011116823.2A CN112257546B (en) 2020-10-19 2020-10-19 Event early warning method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011116823.2A CN112257546B (en) 2020-10-19 2020-10-19 Event early warning method and device, electronic equipment and storage medium

Publications (2)

Publication Number Publication Date
CN112257546A CN112257546A (en) 2021-01-22
CN112257546B true CN112257546B (en) 2024-04-09

Family

ID=74245391

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011116823.2A Active CN112257546B (en) 2020-10-19 2020-10-19 Event early warning method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN112257546B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113095306B (en) * 2021-06-09 2021-09-24 神威超算(北京)科技有限公司 Security alarm method and chip
CN113449116B (en) * 2021-06-22 2022-12-20 青岛海信网络科技股份有限公司 Map construction and early warning method, device and medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000339329A (en) * 1999-05-27 2000-12-08 Nippon Telegr & Teleph Corp <Ntt> Method and device for monitoring and alarming image and storage medium with image monitoring and alarm program stored therein
CN110473402A (en) * 2019-08-20 2019-11-19 河北德冠隆电子科技有限公司 A kind of accident detection early warning system based on target abnormal behaviour trajectory analysis
CN111027376A (en) * 2019-10-28 2020-04-17 中国科学院上海微系统与信息技术研究所 Method and device for determining event map, electronic equipment and storage medium
JP2020160537A (en) * 2019-03-25 2020-10-01 株式会社フジタ Sediment disaster prediction device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000339329A (en) * 1999-05-27 2000-12-08 Nippon Telegr & Teleph Corp <Ntt> Method and device for monitoring and alarming image and storage medium with image monitoring and alarm program stored therein
JP2020160537A (en) * 2019-03-25 2020-10-01 株式会社フジタ Sediment disaster prediction device
CN110473402A (en) * 2019-08-20 2019-11-19 河北德冠隆电子科技有限公司 A kind of accident detection early warning system based on target abnormal behaviour trajectory analysis
CN111027376A (en) * 2019-10-28 2020-04-17 中国科学院上海微系统与信息技术研究所 Method and device for determining event map, electronic equipment and storage medium

Also Published As

Publication number Publication date
CN112257546A (en) 2021-01-22

Similar Documents

Publication Publication Date Title
CN110380896B (en) Network security situation awareness system and method based on attack graph
CN108494810B (en) Attack-oriented network security situation prediction method, device and system
CN102946319B (en) Networks congestion control information analysis system and analytical method thereof
CN112257546B (en) Event early warning method and device, electronic equipment and storage medium
CN104836781A (en) Method distinguishing identities of access users, and device
CN103885990B (en) Searching method and system
CN104881483B (en) Automatic detection evidence collecting method for the attack of Hadoop platform leaking data
CN108092985A (en) Network safety situation analysis method, device, equipment and computer storage media
CN113486345B (en) Supervision early warning method and system with risk identification function
CN112637548B (en) Information association early warning method and device based on camera
CN112968796A (en) Network security situation awareness method and device and computer equipment
CN114124564B (en) Method and device for detecting counterfeit website, electronic equipment and storage medium
CN113726779B (en) Rule false alarm testing method and device, electronic equipment and computer storage medium
CN116248346A (en) Smart city-oriented CPS network security situation awareness establishing method and system
CN115659351A (en) Information security analysis method, system and equipment based on big data office
CN115208679A (en) Attack IP defense method and system based on honeypool cooperation
CN112733170B (en) Active trust evaluation method based on evidence sequence extraction
CN112289028B (en) Method, device, equipment and medium for discovering fake license plate related to vehicle
CN114841705A (en) Anti-fraud monitoring method based on scene recognition
CN113691524A (en) Alarm information processing method, system, electronic equipment and storage medium
CN112417007A (en) Data analysis method and device, electronic equipment and storage medium
CN109308673A (en) A kind of source of houses monitoring and managing method and device
CN116994338B (en) Site paperless auditing management system based on behavior recognition
CN112699814B (en) Violence identification method, violence identification device, and storage medium
CN114021032B (en) Network crime information mining method, system and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant