Video monitoring system and video monitoring method

Abstract

The invention discloses a video monitoring system and a video monitoring method, and belongs to the technical field of video monitoring, and the video monitoring system comprises a video monitoring module, an image processing module, a region association module, an abnormal behavior recognition module, a parameter storage module, a target determination module, a controller and an early warning module. The method has the advantages that the characteristics of different types of abnormal behaviors can be comprehensively evaluated, and a strong early warning signal or a weak early warning signal is generated; and the problem of low accuracy of the deep learning algorithm in the initial stage can be solved, and the accuracy of abnormal behavior recognition of the deep learning algorithm is gradually improved.

Description

Technical Field

[0001] The invention relates to the technical field of video monitoring. Background Art

[0002] At present, my country has basically realized video surveillance in public places with dense traffic such as shopping malls, railway stations, and subway stations. However, the current video surveillance system can only perform the functions of shooting and storage, and cannot capture and detect abnormal behavior in a timely manner. At present, there have been research reports on the application of AI in the scene of abnormal behavior recognition. However, due to the complexity of human behavior and the changeable scenes, relying solely on AI recognition technology has the problem of low accuracy and inability to comprehensively evaluate and warn specific scenes and behaviors. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a video surveillance system that can conduct a comprehensive evaluation of the characteristics of different types of abnormal behaviors and generate a strong warning signal or a weak warning signal; and can make up for the problem of low initial accuracy of deep learning algorithms and gradually improve the accuracy of deep learning algorithms in identifying abnormal behaviors.

[0004] In order to solve the above technical problems, the technical solution of the present invention is a video monitoring system, including a video monitoring module, an image processing module, a region association module, an abnormal behavior recognition module, a parameter storage module, a target determination module, a controller, and an early warning module; The video monitoring module is used to obtain real-time video streams of respective monitoring areas through cameras; The image processing module is used to split the real-time video stream into continuous frames, obtain each frame image, and perform image preprocessing; The area association module is used to associate different cameras according to a preset rule, and the mutually associated cameras form an associated area; The abnormal behavior recognition module is used to use the trained convolutional neural network model to extract and match features of each frame of the image processed by the image processing module, and determine whether there is abnormal behavior in the image. If there is abnormal behavior, the type of abnormal behavior and the probability of occurrence p are output, and the abnormal image is sent to the target determination module; The parameter storage module is used to store the risk factor k, monitoring period T, critical abnormal time t corresponding to each type of abnormal behavior cv , critical outlier value U cv and critical probability P cv ; The target determination module is used to determine whether the target person can be confirmed based on the features extracted from the abnormal image; if not, the images taken by all cameras in the associated area within the monitoring period T are retrieved according to the time and location information of the abnormal image, and matched with the abnormal image to confirm the target person, and the result is sent to the controller; The controller is used to calculate the duration t of abnormal behavior and the abnormal value U according to all abnormal images received for the target person, where U=p*k*t; and cv ,t cv and U cv The relationship between the two is used to determine whether to send a strong warning signal or a weak warning signal to the warning module.

[0005] Furthermore, if t≧t cv or p≧P cv or U≧U cv A strong warning signal is sent to the warning module; otherwise, a weak warning signal is sent to the warning module.

[0006] Furthermore, the features extracted from the abnormal image include facial features, body features, clothing features, and gender features.

[0007] Furthermore, when a weak warning signal is generated, the controller is also used to send all videos containing the target person within the monitoring period T of all cameras in the associated area to the cloud management platform, and the management personnel will determine whether the recognition results of abnormal behavior need to be corrected and feed back the recognition results to the controller.

[0008] Furthermore, it also includes a correction number statistics module; the correction number statistics module is used to count the correction number n corresponding to each abnormal behavior type, and the correction number n is counted as follows: if the recognition result received by the controller is to correct the abnormal behavior of type A to B, then the correction number for the abnormal behavior of type A is -1, and the correction number for the abnormal behavior of type B is +1; if the recognition result received by the controller is to correct the abnormal behavior of type A to none, then the correction number for the abnormal behavior of type A is -1; the controller determines whether it is necessary to adjust the critical abnormal time tcv, critical abnormal value Ucv and critical probability Pcv corresponding to each type of abnormal behavior based on the correction number n of the abnormal behavior; if adjustment is required, the correction number n of the abnormal behavior of this type is cleared after each adjustment.

[0009] Furthermore, the critical abnormal time t cv , critical outlier value U cv and critical probability P cv The adjustment method is: when the number of corrections n of a certain type of abnormal behavior is positive and reaches or exceeds the preset value N1, the critical abnormal time t corresponding to the abnormal behavior of this type is reduced.cv , critical outlier value U cv and critical probability P cv ; When the number of corrections n of a certain type of abnormal behavior is negative and reaches or exceeds the preset value N2, the critical abnormal time t corresponding to the abnormal behavior of this type is increased. cv , critical outlier value U cv and critical probability P cv .

[0010] Further, the critical abnormal time t is adjusted cv , critical outlier value U cv and critical probability P cv When the value is set, the absolute value of the initial adjustment amplitude is set manually. During subsequent adjustments, if the direction of this adjustment is the same as the direction of the previous adjustment, the absolute value of the adjustment amplitude remains unchanged. If the direction of this adjustment is different from the direction of the previous adjustment, the absolute value of the adjustment amplitude becomes half of the absolute value of the previous adjustment amplitude.

[0011] When the correction times n corresponding to each abnormal behavior type in the correction times statistics module has not reached or exceeded the preset N1 and N2 within a certain period of time, the manual correction behavior under the weak warning signal state can be stopped.

[0012] Furthermore, it also includes a storage module; when the recognition result of abnormal behavior needs to be corrected, the controller sends the corresponding abnormal image and the correction result to the storage module for storage, and feeds the convolutional neural network model as new data for learning.

[0013] The present invention also provides a video monitoring method, comprising the following steps: S1. Obtaining images: The image processing module splits the real-time video stream obtained by the video monitoring module into continuous frames, obtains each frame image, and pre-processes the image; S2. Identify abnormal behavior: The abnormal behavior identification module uses the trained convolutional neural network model to extract and match features of each preprocessed frame image to determine whether there is abnormal behavior in the image. If there is abnormal behavior, the type of abnormal behavior and the probability of occurrence p are output, and the abnormal image is sent to the target determination module; the features extracted from the abnormal image include facial features, body features, clothing features, and gender features; S3, confirm the target person: the target determination module determines whether the target person can be confirmed based on the features extracted from the abnormal image; if not, the images taken by all cameras in the associated area within the monitoring period T are retrieved according to the time and location information of the abnormal image, and matched with the abnormal image to confirm the target person, and the result is sent to the controller; S4, generate warning signal: the controller calculates the duration t of abnormal behavior and the abnormal value U according to all abnormal images received for the target person, where U=p*k*t; according to p, t, U and P cv ,t cv and U cv The relationship between k and P determines whether to send a strong warning signal or a weak warning signal to the warning module; cv ,t cv and U cv are the risk factor, critical probability, critical abnormal time and critical abnormal value corresponding to each type of abnormal behavior; if t≧t cv or p≧P cv or U≧U cv If the warning signal is strong, a strong warning signal is sent to the warning module; otherwise, a weak warning signal is sent to the warning module; S5. Warning: The warning module issues warnings based on the received warning signals. Furthermore, after step S5, the following steps are also included: S6, result correction: When a weak warning signal is generated, the controller sends all videos containing the target person from all cameras in the associated area within the monitoring period T to the cloud management platform, and the management personnel determine whether the recognition result of abnormal behavior needs to be corrected, and feeds the recognition result back to the controller; S7, updating the database: when the recognition result of abnormal behavior needs to be corrected, the controller sends the corresponding abnormal image and the correction result to the storage module for storage, and feeds the convolutional neural network model as new data for learning; S8. Correction count: The correction count module counts the correction count n corresponding to each abnormal behavior type. The correction count n is calculated as follows: if the recognition result received by the controller is to correct the abnormal behavior of type A to type B, then the correction count for the abnormal behavior of type A is -1, and the correction count for the abnormal behavior of type B is +1; if the recognition result received by the controller is to correct the abnormal behavior of type A to type B, then the correction count for the abnormal behavior of type A is -1; S9, adjustment parameters: the controller determines whether it is necessary to adjust the critical abnormal time tcv, critical abnormal value Ucv and critical probability Pcv corresponding to each type of abnormal behavior according to the number of corrections n of each type of abnormal behavior; if adjustment is required, the number of corrections n of this type of abnormal behavior is reset to zero after each adjustment; when the number of corrections n of a certain type of abnormal behavior is positive and reaches or exceeds the preset value N1, the critical abnormal time tcv corresponding to this type of abnormal behavior is reduced. cv , critical outlier value U cv and critical probability P cv; When the number of corrections n of a certain type of abnormal behavior is negative and reaches or exceeds the preset value N2, the critical abnormal time t corresponding to the abnormal behavior of this type is increased. cv , critical outlier value U cv and critical probability P cv .

[0014] Further, the critical abnormal time t is adjusted cv , critical outlier value U cv and critical probability P cv When the value is set, the absolute value of the initial adjustment amplitude is set manually. During subsequent adjustments, if the direction of this adjustment is the same as the direction of the previous adjustment, the absolute value of the adjustment amplitude remains unchanged. If the direction of this adjustment is different from the direction of the previous adjustment, the absolute value of the adjustment amplitude becomes half of the absolute value of the previous adjustment amplitude.

[0015] When the correction times n corresponding to each abnormal behavior type by the correction times statistics module has not reached or exceeded the preset N1 and N2 within a certain period of time, steps S6-S9 can be cancelled.

[0016] Beneficial effects of the present invention: The present invention utilizes a deep learning algorithm to identify whether there is abnormal behavior in the monitoring image, and conducts a comprehensive evaluation based on the characteristics of different types of abnormal behaviors to generate a strong warning signal or a weak warning signal; and manual correction can be performed for some abnormal behaviors, and the system is continuously optimized based on the correction results. On the one hand, it can make up for the problem of low accuracy of the deep learning algorithm in the early stage, and gradually improve the accuracy of the deep learning algorithm in identifying abnormal behaviors; on the other hand, it can also cancel manual corrections after the number of corrections stabilizes in the later stage, thereby reducing labor intensity. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a structural block diagram of Example 1. DETAILED DESCRIPTION

[0018] The specific embodiments of the present invention are further described below in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other. Example

[0019] A video surveillance system, such as Figure 1 As shown, it includes a video monitoring module, an image processing module, a region association module, an abnormal behavior recognition module, a parameter storage module, a target determination module, a controller, and an early warning module; The video monitoring module is used to obtain real-time video streams of respective monitoring areas through cameras; The image processing module is used to split the real-time video stream into continuous frames, obtain each frame image, and perform image preprocessing; in this embodiment, the image preprocessing includes image denoising and enhancement; The area association module is used to associate different cameras according to preset rules, and the mutually associated cameras form an associated area; the preset rules can be set according to geographical locations; The abnormal behavior recognition module is used to use the trained convolutional neural network model to extract and match features of each frame of the image processed by the image processing module, and determine whether there is abnormal behavior in the image. If there is abnormal behavior, the type of abnormal behavior and the probability of occurrence p are output, and the abnormal image is sent to the target determination module; The parameter storage module is used to store the risk factor k, monitoring period T, critical abnormal time t corresponding to each type of abnormal behavior cv , critical outlier value U cv and critical probability P cv ; The target determination module is used to determine whether the target person can be confirmed based on the features extracted from the abnormal image; if not, the images taken by all cameras in the associated area within the monitoring period T are retrieved according to the time and location information of the abnormal image, and matched with the abnormal image to confirm the target person, and the result is sent to the controller; it may not be sufficient to determine the target person by relying solely on the abnormal image taken by a certain camera. In this case, the target person can be confirmed by calling the images of other cameras; The controller is used to calculate the duration t of abnormal behavior and the abnormal value U according to all abnormal images received for the target person, where U=p*k*t; and cv ,t cv and U cv The relationship between the two is used to determine whether to send a strong warning signal or a weak warning signal to the warning module. The calculation method of the abnormal behavior duration t is to summarize the abnormal images captured by each camera, and the time information difference between the first and the last abnormal image is the abnormal behavior time captured by the camera; the abnormal behavior time captured by all cameras in the associated area against the target person is summed up to obtain the abnormal behavior duration t.

[0020] Furthermore, if t≧t cv or p≧P cv or U≧U cv A strong warning signal is sent to the warning module; otherwise, a weak warning signal is sent to the warning module.

[0021] Furthermore, the features extracted from the abnormal image include facial features, body features, clothing features, and gender features.

[0022] Furthermore, when a weak warning signal is generated, the controller is also used to send all videos containing the target person within the monitoring period T of all cameras in the associated area to the cloud management platform, and the management personnel will determine whether the recognition results of abnormal behavior need to be corrected and feed back the recognition results to the controller.

[0023] Furthermore, it also includes a correction number statistics module; the correction number statistics module is used to count the correction number n corresponding to each abnormal behavior type, and the correction number n is counted as follows: if the recognition result received by the controller is to correct the abnormal behavior of type A to B, then the correction number for the abnormal behavior of type A is -1, and the correction number for the abnormal behavior of type B is +1; if the recognition result received by the controller is to correct the abnormal behavior of type A to none, then the correction number for the abnormal behavior of type A is -1; the controller determines whether it is necessary to adjust the critical abnormal time tcv, critical abnormal value Ucv and critical probability Pcv corresponding to each type of abnormal behavior based on the correction number n of the abnormal behavior; if adjustment is required, the correction number n of the abnormal behavior of this type is cleared after each adjustment.

[0024] Furthermore, the critical abnormal time t cv , critical outlier value U cv and critical probability P cv The adjustment method is: when the number of corrections n of a certain type of abnormal behavior is positive and reaches or exceeds the preset value N1, the critical abnormal time t corresponding to the abnormal behavior of this type is reduced. cv , critical outlier value U cv and critical probability P cv ; When the number of corrections n of a certain type of abnormal behavior is negative and reaches or exceeds the preset value N2, the critical abnormal time t corresponding to the abnormal behavior of this type is increased. cv , critical outlier value U cv and critical probability P cv .

[0025] Further, the critical abnormal time t is adjusted cv , critical outlier value U cv and critical probability P cv When the value is set, the absolute value of the initial adjustment amplitude is set manually. During subsequent adjustments, if the direction of this adjustment is the same as the direction of the previous adjustment, the absolute value of the adjustment amplitude remains unchanged. If the direction of this adjustment is different from the direction of the previous adjustment, the absolute value of the adjustment amplitude becomes half of the absolute value of the previous adjustment amplitude.

[0026] In one embodiment, there are five types of abnormal behaviors, A, B, C, D, and E. The value of N1 is set to 5, and the value of N2 is set to -5. The critical abnormal time t for a given type of abnormal behavior is cv , critical outlier value U cv and critical probability P cv The absolute values ​​of the adjustment range are Ht, Hu, and Hp respectively. When the number of corrections for type A abnormal behavior in the correction number statistics module reaches 5, the critical abnormal time t is reduced. cv , critical outlier value U cv and critical probability P cv The value of t cv =t cv -Ht,U cv =U cv -Hu, P cv =P cv -Hp; if after a period of time, the number of corrections for type A abnormal behavior in the correction count module reaches 5 again, continue to reduce the critical abnormal time t cv , critical outlier value U cv and critical probability P cv The value of t cv =t cv -Ht,U cv =U cv -Hu, P cv =P cv -Hp; if after a period of time, the number of corrections for type A abnormal behavior in the correction count module reaches -5, the critical abnormal time t is increased. cv , critical outlier value U cv and critical probability P cv The value of t cv =t cv +Ht / 2,U cv =U cv +Hu / 2,P cv =P cv +Hp / 2.

[0027] When the correction times statistics module for each type of abnormal behavior has not reached or exceeded the preset N1 and N2 within a certain period of time, it means that the convolutional neural network model's recognition result for abnormal behavior is accurate enough, and manual correction behavior under weak warning signal state can be stopped at this time.

[0028] Furthermore, it also includes a storage module; when the recognition result of abnormal behavior needs to be corrected, the controller sends the corresponding abnormal image and the correction result to the storage module for storage, and feeds the convolutional neural network model as new data for learning. By manually judging the abnormal behavior that presents a warning signal, and judging whether it needs to be corrected according to the judgment result, the accuracy of the convolutional neural network model recognition can be gradually improved. Example

[0029] The present invention also provides a video monitoring method, comprising the following steps: S1. Obtaining images: The image processing module splits the real-time video stream obtained by the video monitoring module into continuous frames, obtains each frame image, and pre-processes the image; S2. Identify abnormal behavior: The abnormal behavior identification module uses the trained convolutional neural network model to extract and match features of each preprocessed frame image to determine whether there is abnormal behavior in the image. If there is abnormal behavior, the type of abnormal behavior and the probability of occurrence p are output, and the abnormal image is sent to the target determination module; the features extracted from the abnormal image include facial features, body features, clothing features, and gender features; S3, confirm the target person: the target determination module determines whether the target person can be confirmed based on the features extracted from the abnormal image; if not, the images taken by all cameras in the associated area within the monitoring period T are retrieved according to the time and location information of the abnormal image, and matched with the abnormal image to confirm the target person, and the result is sent to the controller; S4, generate warning signal: the controller calculates the duration t of abnormal behavior and the abnormal value U according to all abnormal images received for the target person, where U=p*k*t; according to p, t, U and P cv ,t cv and U cv The relationship between k and P determines whether to send a strong warning signal or a weak warning signal to the warning module; cv ,t cv and U cv are the risk factor, critical probability, critical abnormal time and critical abnormal value corresponding to each type of abnormal behavior; if t≧t cv or p≧P cv or U≧U cv If the warning signal is strong, a strong warning signal is sent to the warning module; otherwise, a weak warning signal is sent to the warning module; S5, early warning: the early warning module issues an early warning according to the received early warning signal; S6, result correction: When a weak warning signal is generated, the controller sends all videos containing the target person from all cameras in the associated area within the monitoring period T to the cloud management platform, and the management personnel determine whether the recognition result of abnormal behavior needs to be corrected, and feeds the recognition result back to the controller; S7, updating the database: when the recognition result of abnormal behavior needs to be corrected, the controller sends the corresponding abnormal image and the correction result to the storage module for storage, and feeds the convolutional neural network model as new data for learning; S8. Correction count: The correction count module counts the correction count n corresponding to each abnormal behavior type. The correction count n is calculated as follows: if the recognition result received by the controller is to correct the abnormal behavior of type A to type B, then the correction count for the abnormal behavior of type A is -1, and the correction count for the abnormal behavior of type B is +1; if the recognition result received by the controller is to correct the abnormal behavior of type A to type B, then the correction count for the abnormal behavior of type A is -1; S9, adjustment parameters: the controller determines whether it is necessary to adjust the critical abnormal time tcv, critical abnormal value Ucv and critical probability Pcv corresponding to each type of abnormal behavior according to the number of corrections n of each type of abnormal behavior; if adjustment is required, the number of corrections n of this type of abnormal behavior is reset to zero after each adjustment; when the number of corrections n of a certain type of abnormal behavior is positive and reaches or exceeds the preset value N1, the critical abnormal time tcv corresponding to this type of abnormal behavior is reduced. cv , critical outlier value U cv and critical probability P cv ; When the number of corrections n of a certain type of abnormal behavior is negative and reaches or exceeds the preset value N2, the critical abnormal time t corresponding to the abnormal behavior of this type is increased. cv , critical outlier value U cv and critical probability P cv .

[0030] When the correction times n corresponding to each abnormal behavior type by the correction times statistics module has not reached or exceeded the preset N1 and N2 within a certain period of time, steps S6-S9 can be cancelled.

[0031] Adjust the critical abnormal time t cv , critical outlier value U cv and critical probability P cv When the value is set, the absolute value of the initial adjustment amplitude is set manually. During subsequent adjustments, if the direction of this adjustment is the same as the direction of the previous adjustment, the absolute value of the adjustment amplitude remains unchanged. If the direction of this adjustment is different from the direction of the previous adjustment, the absolute value of the adjustment amplitude becomes half of the absolute value of the previous adjustment amplitude.

[0032] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions and variations of these embodiments are made without departing from the principles and spirit of the present invention, and still fall within the scope of protection of the present invention.

Claims

1. A video surveillance system, characterized in that: It includes video monitoring module, image processing module, area association module, abnormal behavior recognition module, parameter storage module, target determination module, controller and early warning module; The video monitoring module is used to obtain real-time video streams of respective monitoring areas through cameras; The image processing module is used to split the real-time video stream into continuous frames, obtain each frame image, and perform image preprocessing; The area association module is used to associate different cameras according to a preset rule, and the mutually associated cameras form an associated area; The abnormal behavior recognition module is used to use the trained convolutional neural network model to extract and match features of each frame of the image processed by the image processing module, and determine whether there is abnormal behavior in the image. If there is abnormal behavior, the type of abnormal behavior and the probability of occurrence p are output, and the abnormal image is sent to the target determination module; The parameter storage module is used to store the risk factor k, monitoring period T, critical abnormal time t corresponding to each type of abnormal behavior cv , critical outlier value U cv and critical probability P cv ; The target determination module is used to determine whether the target person can be confirmed based on the features extracted from the abnormal image; if not, the images taken by all cameras in the associated area within the monitoring period T are retrieved according to the time and location information of the abnormal image, and matched with the abnormal image to confirm the target person, and the result is sent to the controller; The controller is used to calculate the duration t of abnormal behavior and the abnormal value U according to all abnormal images received for the target person, where U=p*k*t; and cv ,t cv and U cv The relationship between the two is used to determine whether to send a strong warning signal or a weak warning signal to the warning module.

2. The video surveillance system according to claim 1, characterized in that: If t≧t cv or p≧P cv or U≧U cv A strong warning signal is sent to the warning module; otherwise, a weak warning signal is sent to the warning module.

3. The video surveillance system according to claim 1, characterized in that: The features extracted from the abnormal image include facial features, body features, clothing features, and gender features.

4. The video surveillance system according to claim 1, characterized in that: When a weak warning signal is generated, the controller is also used to send all videos containing the target person from all cameras in the associated area within the monitoring period T to the cloud management platform, and the management personnel will determine whether the recognition results of abnormal behavior need to be corrected and feed back the recognition results to the controller.

5. The video surveillance system according to claim 4, characterized in that: It also includes a correction number statistics module; the correction number statistics module is used to count the correction number n corresponding to each abnormal behavior type, and the correction number n is counted as follows: if the recognition result received by the controller is to correct the abnormal behavior of type A to B, then the correction number for the abnormal behavior of type A is -1, and the correction number for the abnormal behavior of type B is +1; if the recognition result received by the controller is to correct the abnormal behavior of type A to none, then the correction number for the abnormal behavior of type A is -1; the controller determines whether it is necessary to adjust the critical abnormal time tcv, critical abnormal value Ucv and critical probability Pcv corresponding to each type of abnormal behavior according to the correction number n of each type of abnormal behavior; If adjustment is required, the number of corrections n for this type of abnormal behavior will be reset to zero after each adjustment.

6. The video surveillance system according to claim 5, characterized in that: The critical abnormal time t cv , critical outlier value U cv and critical probability P cv The adjustment method is: when the number of corrections n of a certain type of abnormal behavior is positive and reaches or exceeds the preset value N1, the critical abnormal time t corresponding to the abnormal behavior of this type is reduced. cv , critical outlier value U cv and critical probability P cv ; When the number of corrections n of a certain type of abnormal behavior is negative and reaches or exceeds the preset value N2, the critical abnormal time t corresponding to the abnormal behavior of this type is increased. cv , critical outlier value U cv and critical probability P cv .

7. The video surveillance system according to claim 6, characterized in that: Adjust the critical abnormal time t cv , critical outlier value U cv and critical probability P cv When the value is set, the absolute value of the initial adjustment amplitude is set manually. During subsequent adjustments, if the direction of this adjustment is the same as the direction of the previous adjustment, the absolute value of the adjustment amplitude remains unchanged. If the direction of this adjustment is different from the direction of the previous adjustment, the absolute value of the adjustment amplitude becomes half of the absolute value of the previous adjustment amplitude.

8. The video surveillance system according to claim 4, characterized in that: It also includes a storage module; when the recognition result of abnormal behavior needs to be corrected, the controller sends the corresponding abnormal image and the correction result to the storage module for storage, and feeds the convolutional neural network model as new data for learning.

9. A monitoring method applicable to the video monitoring system according to claims 1-8, characterized in that: The steps include: S1. Obtaining images: The image processing module splits the real-time video stream obtained by the video monitoring module into continuous frames, obtains each frame image, and pre-processes the image; S2. Identify abnormal behavior: The abnormal behavior identification module uses the trained convolutional neural network model to extract and match features of each preprocessed frame image to determine whether there is abnormal behavior in the image. If there is abnormal behavior, the type of abnormal behavior and the probability of occurrence p are output, and the abnormal image is sent to the target determination module; the features extracted from the abnormal image include facial features, body features, clothing features, and gender features; S3, confirm the target person: the target determination module determines whether the target person can be confirmed based on the features extracted from the abnormal image; if not, the images taken by all cameras in the associated area within the monitoring period T are retrieved according to the time and location information of the abnormal image, and matched with the abnormal image to confirm the target person, and the result is sent to the controller; S4, generate warning signal: the controller calculates the duration t of abnormal behavior and the abnormal value U according to all abnormal images received for the target person, where U=p*k*t; according to p, t, U and P cv ,t cv and U cv The relationship between k and P determines whether to send a strong warning signal or a weak warning signal to the warning module; cv ,t cv and U cv are the risk factor, critical probability, critical abnormal time and critical abnormal value corresponding to each type of abnormal behavior; if t≧t cv or p≧P cv or U≧U cv Then a strong warning signal is sent to the warning module; Otherwise, a weak warning signal is sent to the warning module; S5. Early warning: The early warning module issues an early warning based on the received early warning signal.

10. The monitoring method of the video monitoring system according to claim 9, characterized in that: The following steps are also included after step S5: S 6. Result correction: When a weak warning signal is generated, the controller sends all videos containing the target person from all cameras in the associated area within the monitoring period T to the cloud management platform, and the management personnel determine whether the recognition result of abnormal behavior needs to be corrected and feed back the recognition result to the controller; S7, updating the database: when the recognition result of abnormal behavior needs to be corrected, the controller sends the corresponding abnormal image and the correction result to the storage module for storage, and feeds the convolutional neural network model as new data for learning; S8. Correction count: The correction count module counts the correction count n corresponding to each abnormal behavior type. The correction count n is calculated as follows: if the recognition result received by the controller is to correct the abnormal behavior of type A to type B, then the correction count for the abnormal behavior of type A is -1, and the correction count for the abnormal behavior of type B is +1; if the recognition result received by the controller is to correct the abnormal behavior of type A to type B, then the correction count for the abnormal behavior of type A is -1; S9, adjusting parameters: the controller determines whether it is necessary to adjust the critical abnormal time tcv, critical abnormal value Ucv and critical probability Pcv corresponding to each type of abnormal behavior according to the number of corrections n of the abnormal behavior; If adjustment is required, the correction times n of this type of abnormal behavior will be reset to zero after each adjustment; when the correction times n of a certain type of abnormal behavior is positive and reaches or exceeds the preset value N1, the critical abnormal time t corresponding to this type of abnormal behavior will be reduced. cv , critical outlier value U cv and critical probability P cv ; When the number of corrections n of a certain type of abnormal behavior is negative and reaches or exceeds the preset value N2, the critical abnormal time t corresponding to the abnormal behavior of this type is increased. cv , critical outlier value U cv and critical probability P cv .

Images