CN117649736A - Video management method and system based on AI video management platform - Google Patents
Video management method and system based on AI video management platform Download PDFInfo
- Publication number
- CN117649736A CN117649736A CN202410120796.8A CN202410120796A CN117649736A CN 117649736 A CN117649736 A CN 117649736A CN 202410120796 A CN202410120796 A CN 202410120796A CN 117649736 A CN117649736 A CN 117649736A
- Authority
- CN
- China
- Prior art keywords
- intrusion
- module
- video management
- target
- data
- 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.)
- Pending
Links
- 238000007726 management method Methods 0.000 title claims abstract description 58
- 238000012544 monitoring process Methods 0.000 claims abstract description 31
- 238000007405 data analysis Methods 0.000 claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 12
- 230000009545 invasion Effects 0.000 claims abstract description 9
- 238000013136 deep learning model Methods 0.000 claims abstract description 7
- 206010000117 Abnormal behaviour Diseases 0.000 claims description 85
- 238000001514 detection method Methods 0.000 claims description 21
- 238000004458 analytical method Methods 0.000 claims description 14
- 230000006399 behavior Effects 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 claims description 7
- 238000010801 machine learning Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 230000005856 abnormality Effects 0.000 claims description 5
- 238000007418 data mining Methods 0.000 claims description 4
- 230000033001 locomotion Effects 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 3
- 230000000877 morphologic effect Effects 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 8
- 230000006870 function Effects 0.000 abstract description 7
- 238000007689 inspection Methods 0.000 abstract description 4
- 239000000463 material Substances 0.000 abstract description 4
- 230000004044 response Effects 0.000 abstract description 4
- 230000009977 dual effect Effects 0.000 abstract description 3
- 238000011897 real-time detection Methods 0.000 abstract description 3
- 230000000052 comparative effect Effects 0.000 description 9
- 230000002159 abnormal effect Effects 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000013523 data management Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004148 unit process Methods 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Abstract
The invention relates to the technical field of AI video management, in particular to a video management method and a system based on an AI video management platform, wherein the video management method comprises the following steps: the system comprises a data acquisition module, an object identification module, a regional intrusion alarm module, an AI early warning reporting module, a data analysis module and a message notification module; a video management method comprising the steps of: s1, data acquisition; s2, object identification; s3, an area invasion alarm module; s4AI early warning report; s5, data analysis. The method is applied to an AI video management platform, and the target object is identified by utilizing a deep learning model through collecting video data, so that the functions of real-time detection of the target object, quick response of an intrusion event and timely pushing of early warning information are realized, the dual warning effect of warning and early warning is provided, the safety of a video monitoring area can be better ensured, the cost of manpower and material resources brought by manual inspection or video monitoring is reduced, and the monitoring efficiency is greatly improved.
Description
Technical Field
The invention relates to the technical field of AI video management, in particular to a video management method and a video management system based on an AI video management platform.
Background
The AI video management platform is a video monitoring platform based on artificial intelligence technology, integrates the functions of video acquisition, transmission, storage, analysis, display and the like, and realizes intelligent management and application of video data. The AI video management platform mainly comprises video data acquisition and transmission, video data storage and management, video data analysis, video display, visualization and the like. Video data analysis is even more important.
An intelligent management platform (publication No. CN116743970B, publication No. 2023-11-21) with video AI early warning analysis comprises a monitoring equipment layout module, a monitoring area real-time induction monitoring module, a monitoring picture automatic adjustment module, a shouting early warning module, an early warning reaction identification module, a management information base, a fishing road line analysis module and a fishing early warning information automatic generation module.
According to the technology, when abnormal personnel are sensed by the monitoring camera to the water area where the abnormal personnel are prohibited to enter and exit, the focal length and the shooting angle of the monitoring camera are automatically adjusted based on the display position and the display proportion of the abnormal personnel in the current monitoring picture, so that the abnormal personnel are in a central amplification state in the display of the monitoring picture, the recognition efficiency and the recognition accuracy of the fishing behavior can be improved to the greatest extent, the reasonable effectiveness of the output of the prohibited early warning can be ensured, and the occurrence rate of ineffective early warning and delayed early warning can be reduced greatly. However, the technology does not provide a specific method for AI video management, and does not provide a process of identifying objects by videos so as to judge abnormality and send out early warning.
Disclosure of Invention
Aiming at the technical defects in the background art, the invention provides a video management method and a video management system based on an AI video management platform, which solve the technical problems and meet the actual demands, and the specific technical scheme is as follows:
a video management system based on an AI video management platform, comprising: the system comprises a data acquisition module, an object identification module, a regional intrusion alarm module, an AI early warning reporting module, a data analysis module and a message notification module;
the data acquisition module is used for acquiring video data;
the object recognition module is used for recognizing a target object of the video data;
the regional intrusion alarm module is used for judging whether the target object is an intrusion target or abnormal behavior and judging whether alarm information is sent out;
the AI early warning reporting module is used for analyzing the rule of the intrusion target to predict and judging whether to send out early warning information;
the data analysis module is used for collecting the data of the alarm information and carrying out statistics and analysis;
and the message notification module is used for pushing the alarm information and the early warning information to the user.
A video management method based on an AI video management platform comprises the following steps:
s1, data acquisition: the monitoring area is monitored in real time through a data acquisition module, and video data are acquired;
s2, object identification: identifying a target object in the video data picture by using a deep learning model;
s3, an area invasion alarm module: judging whether the target object is an intrusion target or abnormal behavior;
s3.1, setting a standard: presetting a judgment standard of an intrusion target and a judgment standard of abnormal behavior, wherein the abnormal behavior triggers a warning standard;
s3.2 intrusion alert: judging whether the target object is an intrusion target or not according to the identification result of the S2, if yes, triggering an alarm, and pushing alarm information to a user through a message notification module;
s3.3, abnormality detection: according to the actual situation, a reasonable dynamic area is set, the motion trail of the target object in the dynamic area is analyzed, and whether abnormal behaviors exist or not is judged;
s3.4, alarming judgment: judging whether the abnormal behavior triggers an alarm or not according to the abnormal behavior judging standard preset in S4.1, and if the abnormal behavior triggers the alarm, pushing alarm information to a user through a message notification module;
s4AI early warning reporting: analyzing the rule of the invasion target to predict;
s4.1, collecting historical data: collecting historical intrusion target data and historical abnormal behavior data;
s4.2 analysis rules: analyzing the data collected in S5.1 through a data mining and machine learning algorithm to obtain intrusion rules of different target types of the intrusion targets;
s4.3, data early warning: predicting the intrusion law obtained in the step S5.2 in combination with the current condition, and pushing early warning information to a user through a message notification module when the possible intrusion is predicted;
s5, data analysis: and continuously collecting and analyzing the data of the intrusion target and the abnormal behavior, optimizing algorithm parameters, and improving the accuracy and instantaneity of detection of the intrusion target and the abnormal behavior.
As a further technical scheme of the present invention, the step of S2 is as follows:
s2.1, preprocessing an image: the image of the video data is subjected to graying, filtering and binarization operation, so that noise is reduced, and identification accuracy is improved;
s2.2 target detection: identifying a target object in the image processed by the S2.1 method by adopting an image segmentation technology, optimizing the outline of the target object by combining morphological operation, and primarily judging the target type of the target object by the size and shape characteristics of the target object;
s2.3, extracting characteristics: extracting key features such as edges and corner points from the target object detected in the step S2.2, and selecting proper features as classification basis by combining the target type of the target object;
s2.4, classification and identification: selecting a proper machine learning model according to task requirements, classifying and identifying different target types, inputting the features extracted in the step S2.3 into the selected model, and carrying out parameter adjustment and training;
s2.5 model evaluation: evaluating the trained model in S2.4 by using a verification set or a test set, and calculating various indexes;
s2.6 model optimization: and (3) optimizing the model according to the evaluation result of the step (S2.5), and deploying the trained model into practical application.
As a further technical solution of the present invention, the following formula is included in S3.2: target object detection accuracy = number of correctly detected targets/total number of targets, the number of correctly detected targets being the correct number of intrusion targets determined in S3.2 for the target object, the total number of targets being the determined total number of targets.
As a further technical solution of the present invention, the following formula is included in S3.3: abnormal behavior occurrence probability = abnormal behavior number/total behavior number, the abnormal behavior number being the correct number of abnormal behaviors judged in S3.3, and the total number of judgment in S3.3.
As a further technical solution of the present invention, the following formula is included in S3.4: alarm accuracy = number of actual occurrence of abnormal behavior and correctly alerted/total number of alarms, the number of actual occurrence of abnormal behavior and correctly alerted being the number of abnormal behavior determined to trigger a warning in S3.4, the total number of alarms being the total number of determinations in S3.4.
As a further technical solution of the present invention, the video management method includes the following formula: system performance index = target object detection accuracy × abnormal behavior determination accuracy × alarm accuracy.
As a further technical scheme of the present invention, the S2.3 adopts HOG characteristics.
The invention has the beneficial effects that:
the invention is applied to an AI video management platform, and the target object is identified by collecting video data and utilizing a deep learning model, so that the functions of real-time detection of the target object, rapid response of an intrusion event, timely pushing of early warning information and the like are realized, thereby providing a dual warning effect of warning and early warning, better guaranteeing the safety of a video monitoring area, reducing the cost of manpower and material resources brought by manual inspection or video monitoring, and greatly improving the monitoring efficiency.
Detailed Description
The following description of the embodiments of the present invention is given in connection with the examples, which are not intended to limit the embodiments of the present invention, and the present invention relates to the relevant essential parts in the art, and should be construed as being known and understood by those skilled in the art.
The invention is applied to an AI video management platform, in particular to a video management system based on the AI video management platform, which comprises: the system comprises a data acquisition module, an object identification module, a regional intrusion alarm module, an AI early warning reporting module, a data analysis module and a message notification module.
The data acquisition module is used for acquiring video data; the object recognition module is used for recognizing a target object of the video data; the regional intrusion alarm module is used for judging whether a target object is an intrusion target or abnormal behavior and judging whether alarm information is sent out; the AI early warning reporting module is used for analyzing the rule of the intrusion target to predict and judging whether to send out early warning information; the data analysis module is used for collecting the data of the alarm information and carrying out statistics and analysis; and the message notification module is used for pushing the alarm information and the early warning information to the user. The invention has the double warning effects of warning and early warning, can better ensure the safety of the video monitoring area, reduces the cost of manpower and material resources brought by manual inspection or video monitoring, and greatly improves the monitoring efficiency.
The invention provides a video management method based on an AI video management platform, which comprises the following steps:
s1, data acquisition: the monitoring area is monitored in real time through a data acquisition module, and video data are acquired;
s2, object identification: identifying a target object in the video data picture by using the deep learning model;
s3, an area invasion alarm module: judging whether the target object is an intrusion target or abnormal behavior;
s3.1, setting a standard: presetting a judgment standard of an intrusion target and a judgment standard of abnormal behavior, wherein the abnormal behavior triggers a warning standard;
s3.2 intrusion alert: judging whether the target object is an intrusion target according to the identification result of the S2, if yes, triggering an alarm, and pushing alarm information to a user through a message notification module;
s3.3, abnormality detection: according to the actual situation, a reasonable dynamic area is set, the motion trail of the target object in the dynamic area is analyzed, and whether abnormal behaviors exist or not is judged;
s3.4, alarming judgment: judging whether the abnormal behavior triggers an alarm or not according to the abnormal behavior judging standard preset in S4.1, and if the abnormal behavior triggers the alarm, pushing alarm information to a user through a message notification module;
s4AI early warning reporting: analyzing rules of an intrusion target to predict;
s4.1, collecting historical data: collecting historical intrusion target data and historical abnormal behavior data;
s4.2 analysis rules: analyzing the data collected in S5.1 through a data mining and machine learning algorithm to obtain intrusion rules of different target types of intrusion targets;
s4.3, data early warning: predicting the intrusion law obtained in the step S5.2 in combination with the current condition, and pushing early warning information to a user through a message notification module when the possible intrusion is predicted;
s5, data analysis: and continuously collecting and analyzing data of the intrusion target and the abnormal behavior, optimizing algorithm parameters, and improving the accuracy and instantaneity of detection of the intrusion target and the abnormal behavior.
The method is applied to an AI video management platform, and the target object is identified by collecting video data and utilizing a deep learning model, so that the functions of real-time detection of the target object, rapid response of an intrusion event, timely pushing of early warning information and the like are realized, the dual warning effect of warning and early warning is provided, and the safety of a video monitoring area can be better ensured.
The following are examples of the present invention and comparative examples.
Examples
A video management system based on an AI video management platform, comprising: the system comprises a data acquisition module, an object identification module, a regional intrusion alarm module, an AI early warning reporting module, a data analysis module and a message notification module.
The data acquisition module is used for acquiring video data; the object recognition module is used for recognizing a target object of the video data; the regional intrusion alarm module is used for judging whether a target object is an intrusion target or abnormal behavior and judging whether alarm information is sent out; the AI early warning reporting module is used for analyzing the rule of the intrusion target to predict and judging whether to send out early warning information; the data analysis module is used for collecting the data of the alarm information and carrying out statistics and analysis; and the message notification module is used for pushing the alarm information and the early warning information to the user. The invention has the double warning effect of warning and early warning, and can better ensure the safety of the video monitoring area.
A video management method based on an AI video management platform comprises the following steps:
s1, data acquisition: the monitoring area is monitored in real time through a data acquisition module, and video data are acquired;
s2, object identification: identifying a target object in the video data picture by using the deep learning model;
s2.1, preprocessing an image: the image of the video data is subjected to operations such as graying, filtering, binarization and the like, so that noise is reduced, and identification accuracy is improved;
s2.2 target detection: identifying a target object in the image processed by the S2.1 method by adopting an image segmentation technology, optimizing the outline of the target object by combining morphological operation, and primarily judging the target type of the target object by the characteristics of the size, the shape and the like of the target object;
s2.3, extracting characteristics: extracting key features such as edges, corner points and the like from the target object detected in the step S2.2, and selecting proper features as classification basis by combining the target type of the target object;
wherein HOG (Histogram of Oriented Gradients, directional gradient histogram) is employed.
S2.4, classification and identification: selecting a proper machine learning model according to task requirements, classifying and identifying different target types, inputting the features extracted in the step S2.3 into the selected model, and carrying out parameter adjustment and training;
s2.5 model evaluation: evaluating the trained model in S2.4 by using a verification set or a test set, and calculating various indexes;
s2.6 model optimization: and (3) optimizing the model according to the evaluation result of S2.5, and deploying the trained model into practical application.
S3, an area invasion alarm module: judging whether the target object is an intrusion target or abnormal behavior;
s3.1, setting a standard: presetting a judgment standard of an intrusion target and a judgment standard of abnormal behavior, wherein the abnormal behavior triggers a warning standard;
s3.2 intrusion alert: judging whether the target object is an intrusion target according to the identification result of the S2, if yes, triggering an alarm, and pushing alarm information to a user through a message notification module;
the method comprises the following steps: target object detection accuracy = number of correctly detected targets/total number of targets, the number of correctly detected targets being the correct number of intrusion targets determined in S3.2, the total number of targets being the determined total number of targets.
First, an image or video is processed using a target detection algorithm (e.g., YOLO, fast R-CNN, etc.), outputting a detected target object. And then comparing the detection result with the actual target number, and counting the number of the correctly detected targets, thereby obtaining the target object detection accuracy.
S3.3, abnormality detection: according to the actual situation, a reasonable dynamic area is set, the motion trail of the target object in the dynamic area is analyzed, and whether abnormal behaviors exist or not is judged;
the method comprises the following steps: probability of occurrence of abnormal behavior = number of abnormal behaviors/total number of behaviors, the number of abnormal behaviors being the correct number of abnormal behaviors judged in S3.3, total number of judgment in total behaviors S3.3.
First, the characteristics of abnormal behavior, such as speed, angle, etc., are defined. And then, carrying out behavior analysis on the detected target object to judge whether abnormal behaviors exist. And finally, counting the times of abnormal behaviors, and dividing the times with the total times of the behaviors to obtain the occurrence probability of the abnormal behaviors.
S3.4, alarming judgment: judging whether the abnormal behavior triggers an alarm or not according to the abnormal behavior judging standard preset in S4.1, and if the abnormal behavior triggers the alarm, pushing alarm information to a user through a message notification module;
including the following formulas: alarm accuracy = number of actual abnormal behavior and correctly alerted/total number of alarms, the number of actual abnormal behavior and correctly alerted is the number of abnormal behavior determined to trigger a warning in S3.4, and the total number of alarms is the total number of determinations in S3.4.
Firstly, according to the abnormal behavior judgment result, the actually-occurring abnormal behavior is determined. Then, the number of times that abnormal behavior actually occurs and is correctly alerted is counted. And finally, dividing the frequency with the total alarming frequency to obtain alarming accuracy.
S4AI early warning reporting: analyzing rules of an intrusion target to predict;
s4.1, collecting historical data: collecting historical intrusion target data and historical abnormal behavior data;
s4.2 analysis rules: analyzing the data collected in S5.1 through a data mining and machine learning algorithm to obtain intrusion rules of different target types of intrusion targets;
s4.3, data early warning: predicting the intrusion law obtained in the step S5.2 in combination with the current condition, and pushing early warning information to a user through a message notification module when the possible intrusion is predicted;
s5, data analysis: and continuously collecting and analyzing data of the intrusion target and the abnormal behavior, optimizing algorithm parameters, and improving the accuracy and instantaneity of detection of the intrusion target and the abnormal behavior.
Performance of the overall video management system: system performance index = target object detection accuracy × abnormal behavior determination accuracy × alarm accuracy. And respectively calculating the detection accuracy, the abnormal behavior judgment accuracy and the alarm accuracy of the target object according to the three indexes. Then, these three accuracies are multiplied to obtain the system performance index. This index can be used to evaluate the performance of the system, the higher the index, the better the system performance.
Comparative example 1
An intelligent recognition and early warning method for abnormal behavior (publication number: CN115311735A, publication date: 2022-11-08) comprises the following steps:
s1: by applying leading edge information technologies such as artificial intelligence, internet of things, big data, 5G mobile interconnection and the like, through a high-altitude camera network which widely covers a target area, an edge computing visual perception platform is built, and abnormal behavior places of students are monitored in real time and accurately positioned;
s2: the abnormal agent is automatically identified and positioned by the video AI intelligent identification and early warning management system, the system automatically sends out warning information after identification and positioning in the video image, and the information is sent to the safety manager in real time, so that the aim of automatically discovering the abnormal behavior of the student by the system is fulfilled.
Comparative example 2
An early warning method, equipment and system (publication number: CN112466077A, publication date: 2021-03-09) includes the following steps:
setting an alert area range of early warning and alarming, comprising: the early warning area and the alarm area;
the method comprises the steps of setting RFID equipment, a security camera and a security radar which fully cover a warning area, and setting a main control unit and an early warning alarm unit, wherein the main control unit is respectively connected with the RFID equipment, the security camera, the security radar and the early warning alarm unit to realize data interaction;
the security camera and security radar collect the movable object entering the warning area range, and the main control unit identifies the identity and the number of the movable object; the RFID equipment collects staff information entering the warning area range and counts the number of staff;
the main control unit processes the collected information to judge whether illegal invasion exists, and if the number of the movable objects is greater than the number of staff, the security camera collects video information of the warning area range, and the early warning alarm unit sends out early warning or alarm signals.
An early warning alert device comprising: the RFID device is used for identifying staff information entering a set warning area and counting staff quantity information; the security cameras acquire image information in the warning area by adopting double cameras; the security radar is used for monitoring whether an object enters and moves in a warning area covered by the radar and identifying an intrusion object;
the RFID equipment, the security camera and the security radar are respectively connected with the main control unit, and the main control unit judges whether illegal invasion exists or not according to information acquired by the RFID equipment, the security camera and the security radar;
the main control unit is respectively connected with the audible and visual alarm and the prompt sound, and sends control signals to the audible and visual alarm and the prompt sound to start or stop early warning or alarming;
the early warning and alarming equipment adopts a double-battery power supply system.
Combining example with comparative example 1:
the system for identifying and positioning abnormal behaviors automatically through the video AI intelligent identification and early warning management system by collecting the abnormal behavior event data of the students can judge the normal behaviors and quickly early warn the response, so that the safety of the monitoring area or the students is ensured.
However, the difference is that the early warning indicated in comparative example 1 is warning information sent out for judging the abnormal behavior of the target object, while the early warning indicated in this embodiment is that the action rule of the target object is found out for prediction through statistics and analysis of historical data, and early warning information is sent out when prediction possibly occurs. Further, the implementation effect in comparative example 1 is similar to the abnormal behavior judgment alarm in this embodiment, but this embodiment has two alarm effects of intrusion target, abnormal behavior alarm and rule prediction and early warning, so that the safety of the monitoring area is further ensured.
Combining examples with comparative example 2:
the comparative example 2 has the intrusion target monitoring function as in the example, and after the intrusion target is monitored, the intelligent alarm is given, so that the cost of manpower and material resources brought by manual inspection or video monitoring is reduced, and the monitoring efficiency is greatly improved.
But different from comparative example 2, the embodiment has the functions of abnormal behavior alarm judgment and early warning of intrusion target rule analysis and prediction, and combines the alarm judgment and the early warning function to realize double warning effects, so that the safety of a monitoring area is further ensured.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (8)
1. A video management system based on an AI video management platform, comprising: the system comprises a data acquisition module, an object identification module, a regional intrusion alarm module, an AI early warning reporting module, a data analysis module and a message notification module;
the data acquisition module is used for acquiring video data;
the object recognition module is used for recognizing a target object of the video data;
the regional intrusion alarm module is used for judging whether the target object is an intrusion target or abnormal behavior and judging whether alarm information is sent out;
the AI early warning reporting module is used for analyzing the rule of the intrusion target to predict and judging whether to send out early warning information;
the data analysis module is used for collecting the data of the alarm information and carrying out statistics and analysis;
and the message notification module is used for pushing the alarm information and the early warning information to the user.
2. The video management method based on the AI video management platform is characterized by comprising the following steps:
s1, data acquisition: the monitoring area is monitored in real time through a data acquisition module, and video data are acquired;
s2, object identification: identifying a target object in the video data picture by using a deep learning model;
s3, an area invasion alarm module: judging whether the target object is an intrusion target or abnormal behavior;
s3.1, setting a standard: presetting a judgment standard of an intrusion target and a judgment standard of abnormal behavior, wherein the abnormal behavior triggers a warning standard;
s3.2 intrusion alert: judging whether the target object is an intrusion target or not according to the identification result of the S2, if yes, triggering an alarm, and pushing alarm information to a user through a message notification module;
s3.3, abnormality detection: according to the actual situation, a reasonable dynamic area is set, the motion trail of the target object in the dynamic area is analyzed, and whether abnormal behaviors exist or not is judged;
s3.4, alarming judgment: judging whether the abnormal behavior triggers an alarm or not according to the abnormal behavior judging standard preset in S4.1, and if the abnormal behavior triggers the alarm, pushing alarm information to a user through a message notification module;
s4AI early warning reporting: analyzing the rule of the invasion target to predict;
s4.1, collecting historical data: collecting historical intrusion target data and historical abnormal behavior data;
s4.2 analysis rules: analyzing the data collected in S5.1 through a data mining and machine learning algorithm to obtain intrusion rules of different target types of the intrusion targets;
s4.3, data early warning: predicting the intrusion law obtained in the step S5.2 in combination with the current condition, and pushing early warning information to a user through a message notification module when the possible intrusion is predicted;
s5, data analysis: and continuously collecting and analyzing the data of the intrusion target and the abnormal behavior, optimizing algorithm parameters, and improving the accuracy and instantaneity of detection of the intrusion target and the abnormal behavior.
3. The AI video management platform-based video management method of claim 2, wherein the step of S2 is as follows:
s2.1, preprocessing an image: the image of the video data is subjected to graying, filtering and binarization operation, so that noise is reduced, and identification accuracy is improved;
s2.2 target detection: identifying a target object in the image processed by the S2.1 method by adopting an image segmentation technology, optimizing the outline of the target object by combining morphological operation, and primarily judging the target type of the target object by the size and shape characteristics of the target object;
s2.3, extracting characteristics: extracting key features such as edges and corner points from the target object detected in the step S2.2, and selecting proper features as classification basis by combining the target type of the target object;
s2.4, classification and identification: selecting a proper machine learning model according to task requirements, classifying and identifying different target types, inputting the features extracted in the step S2.3 into the selected model, and carrying out parameter adjustment and training;
s2.5 model evaluation: evaluating the trained model in S2.4 by using a verification set or a test set, and calculating various indexes;
s2.6 model optimization: and (3) optimizing the model according to the evaluation result of the step (S2.5), and deploying the trained model into practical application.
4. The AI video management platform-based video management method of claim 2, wherein the following formula is included in S3.2: target object detection accuracy = number of correctly detected targets/total number of targets, the number of correctly detected targets being the correct number of intrusion targets determined in S3.2 for the target object, the total number of targets being the determined total number of targets.
5. The AI video management platform-based video management method of claim 2, wherein the following formula is included in S3.3: abnormal behavior occurrence probability = abnormal behavior number/total behavior number, the abnormal behavior number being the correct number of abnormal behaviors judged in S3.3, and the total number of judgment in S3.3.
6. The AI video management platform-based video management method of claim 2, wherein S3.4 includes the following formula: alarm accuracy = number of actual occurrence of abnormal behavior and correctly alerted/total number of alarms, the number of actual occurrence of abnormal behavior and correctly alerted being the number of abnormal behavior determined to trigger a warning in S3.4, the total number of alarms being the total number of determinations in S3.4.
7. The AI video management platform-based video management method of claim 2, wherein the video management method includes the following formula: system performance index = target object detection accuracy × abnormal behavior determination accuracy × alarm accuracy.
8. The AI-video-management-platform-based video management method of claim 3, wherein S2.3 employs a HOG feature.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410120796.8A CN117649736A (en) | 2024-01-29 | 2024-01-29 | Video management method and system based on AI video management platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410120796.8A CN117649736A (en) | 2024-01-29 | 2024-01-29 | Video management method and system based on AI video management platform |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117649736A true CN117649736A (en) | 2024-03-05 |
Family
ID=90045480
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410120796.8A Pending CN117649736A (en) | 2024-01-29 | 2024-01-29 | Video management method and system based on AI video management platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117649736A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103065409A (en) * | 2012-12-14 | 2013-04-24 | 广州供电局有限公司 | Power transmission line monitoring and early warning system |
CN110363146A (en) * | 2019-07-16 | 2019-10-22 | 杭州睿琪软件有限公司 | A kind of object identification method, device, electronic equipment and storage medium |
CN111783744A (en) * | 2020-07-31 | 2020-10-16 | 上海仁童电子科技有限公司 | Operation site safety protection detection method and device |
CN114758236A (en) * | 2022-04-13 | 2022-07-15 | 华中科技大学 | Non-specific shape object identification, positioning and manipulator grabbing system and method |
CN116416281A (en) * | 2023-04-28 | 2023-07-11 | 云观智慧科技(无锡)有限公司 | Grain depot AI video supervision and analysis method and system |
CN116778657A (en) * | 2023-05-24 | 2023-09-19 | 武汉众智数字技术有限公司 | Method and system for intelligently identifying intrusion behavior |
-
2024
- 2024-01-29 CN CN202410120796.8A patent/CN117649736A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103065409A (en) * | 2012-12-14 | 2013-04-24 | 广州供电局有限公司 | Power transmission line monitoring and early warning system |
CN110363146A (en) * | 2019-07-16 | 2019-10-22 | 杭州睿琪软件有限公司 | A kind of object identification method, device, electronic equipment and storage medium |
CN111783744A (en) * | 2020-07-31 | 2020-10-16 | 上海仁童电子科技有限公司 | Operation site safety protection detection method and device |
CN114758236A (en) * | 2022-04-13 | 2022-07-15 | 华中科技大学 | Non-specific shape object identification, positioning and manipulator grabbing system and method |
CN116416281A (en) * | 2023-04-28 | 2023-07-11 | 云观智慧科技(无锡)有限公司 | Grain depot AI video supervision and analysis method and system |
CN116778657A (en) * | 2023-05-24 | 2023-09-19 | 武汉众智数字技术有限公司 | Method and system for intelligently identifying intrusion behavior |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108053427B (en) | Improved multi-target tracking method, system and device based on KCF and Kalman | |
CN111144291B (en) | Video monitoring area personnel intrusion discrimination method and device based on target detection | |
CN108052859B (en) | Abnormal behavior detection method, system and device based on clustering optical flow characteristics | |
CN112785798B (en) | Behavior analysis method for constructors of power substation engineering construction project | |
CN103065121B (en) | The engine driver's method for monitoring state analyzed based on video human face and device | |
CN102164270A (en) | Intelligent video monitoring method and system capable of exploring abnormal events | |
CN111629181B (en) | Fire-fighting life passage monitoring system and method | |
KR102356666B1 (en) | Method and apparatus for risk detection, prediction, and its correspondence for public safety based on multiple complex information | |
CN114973140A (en) | Dangerous area personnel intrusion monitoring method and system based on machine vision | |
CN111783530A (en) | Safety system and method for monitoring and identifying behaviors in restricted area | |
CN112329691A (en) | Monitoring video analysis method and device, electronic equipment and storage medium | |
CN113484858A (en) | Intrusion detection method and system | |
CN113887445A (en) | Method and system for identifying standing and loitering behaviors in video | |
CN112163572A (en) | Method and device for identifying object | |
CN112288975A (en) | Event early warning method and device | |
CN115311735A (en) | Intelligent recognition early warning method for abnormal behaviors | |
CN114333273A (en) | Intelligent analysis system and method for nuclear power construction site | |
CN111652128B (en) | High-altitude power operation safety monitoring method, system and storage device | |
CN113052125A (en) | Construction site violation image recognition and alarm method | |
CN110300280A (en) | A kind of scenic spot passenger flow monitoring system | |
CN117649736A (en) | Video management method and system based on AI video management platform | |
CN202904792U (en) | Intelligent visualized alarm system | |
CN115294712A (en) | Intrusion early warning method, early warning management system, electronic device and storage medium | |
CN111223335B (en) | Ship superelevation early warning method, device and system | |
CN113191273A (en) | Oil field well site video target detection and identification method and system based on neural network |
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 |