CN111597970A - Abnormal behavior identification method and device - Google Patents
Abnormal behavior identification method and device Download PDFInfo
- Publication number
- CN111597970A CN111597970A CN202010406965.6A CN202010406965A CN111597970A CN 111597970 A CN111597970 A CN 111597970A CN 202010406965 A CN202010406965 A CN 202010406965A CN 111597970 A CN111597970 A CN 111597970A
- Authority
- CN
- China
- Prior art keywords
- cameras
- abnormal behavior
- confidence
- determining
- abnormal
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
Abstract
The invention discloses a method and a device for identifying abnormal behaviors, wherein the method comprises the following steps: respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and determining at least two cameras acquiring the abnormal behavior and the corresponding abnormal behavior confidence coefficient of each camera when the abnormal behavior is identified; determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras in the at least two cameras which acquire the abnormal behavior; determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera; determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras; when the confidence coefficient of the abnormal behavior is greater than or equal to the preset confidence coefficient threshold, the abnormal behavior is determined as the recognition result, and the accuracy of the abnormal behavior recognition can be improved.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a method and an apparatus for identifying an abnormal behavior.
Background
The bank outlets belong to financial transaction places, have high requirements on security, and are very sensitive to the actions of crowd charging, fighting and the like which occur on site. With the prevalence of artificial intelligence technology, the bank adopts a fighting behavior recognition model, carries out video monitoring on bank outlets based on the principle of limb interaction and superposition in images, can quickly find and recognize abnormal behaviors such as fighting, charging and the like in the bank outlets, and reminds workers to handle the abnormal behaviors in time.
However, because the density of the population at the website is high, when a video is shot at a single angle, due to the problem of linear vision, the depth in the image cannot be identified, local shielding can occur, and target vision overlapping can be caused when people cross walk, so that the crowd with a certain distance is often misjudged as an abnormal behavior when position exchange or cross movement occurs, and particularly when the crowd lines up, the wrong identification of the abnormal behavior often occurs.
Most of the prior art is based on deep learning, image data of fighting behaviors are collected, and fighting behavior recognition models are optimized, but due to the physical limitation of two-dimensional images of a single-orientation camera, model optimization progress is difficult, and accuracy of abnormal behavior recognition is low.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the invention provides a method for identifying abnormal behaviors, which is used for improving the accuracy of identifying the abnormal behaviors and comprises the following steps:
respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and when abnormal behaviors are identified, determining at least two cameras acquiring the abnormal behaviors and an abnormal behavior confidence coefficient corresponding to each camera, wherein the plurality of cameras are arranged at different positions of a bank outlet, and the abnormal behavior identification model is obtained by training according to historical abnormal behavior data;
acquiring angle parameters of at least two cameras acquiring the abnormal behavior, and determining an included angle parameter between every two cameras;
determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras;
determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera;
determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras;
and when the confidence coefficient of the abnormal behavior is greater than or equal to a preset confidence coefficient threshold value, determining the abnormal behavior as a recognition result.
The embodiment of the invention provides a device for identifying abnormal behaviors, which is used for improving the accuracy of identifying the abnormal behaviors and comprises the following components:
the parameter obtaining module is used for respectively inputting the customer behavior data acquired by the multiple cameras into an abnormal behavior identification model, and when the abnormal behavior is identified, determining at least two cameras acquiring the abnormal behavior and the abnormal behavior confidence corresponding to each camera, wherein the multiple cameras are arranged at different positions of a bank website, and the abnormal behavior identification model is obtained by training according to historical abnormal behavior data;
the included angle determining module is used for acquiring the angle parameters of at least two cameras acquiring the abnormal behavior and determining the included angle parameter between every two cameras;
the accuracy rate determining module is used for determining the identification accuracy rate of each two cameras according to the included angle parameter between each two cameras;
the confidence coefficient determining module is used for determining the abnormal behavior confidence coefficient of each two cameras according to the recognition accuracy rate of each two cameras and the abnormal behavior confidence coefficient corresponding to each camera;
the abnormal behavior confidence determining module is used for determining the confidence of the abnormal behavior according to the abnormal behavior confidence of every two cameras;
and the identification result determining module is used for determining the abnormal behavior as an identification result when the confidence coefficient of the abnormal behavior is greater than or equal to a preset confidence coefficient threshold value.
The embodiment of the invention also provides computer equipment which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor realizes the identification method of the abnormal behaviors when executing the computer program.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program for executing the above-mentioned method for identifying an abnormal behavior is stored in the computer-readable storage medium.
The embodiment of the invention comprises the following steps: respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and determining at least two cameras acquiring the abnormal behavior and the corresponding abnormal behavior confidence coefficient of each camera when the abnormal behavior is identified; acquiring angle parameters of at least two cameras acquiring the abnormal behavior, and determining an included angle parameter between every two cameras; determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras; determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera; determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras; when the confidence of the abnormal behavior is larger than or equal to the preset confidence threshold, the abnormal behavior is determined as the recognition result, the method can carry out comprehensive analysis and secondary optimization on the abnormal behavior recognition result under multiple angles based on the multiple cameras and the abnormal behavior recognition model, solves the limitation that the image depth cannot be recognized at a single angle, avoids the misjudgment of the abnormal behavior caused by the problems of personnel dislocation, shielding and the like, and improves the accuracy of the abnormal behavior recognition.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts. In the drawings:
FIG. 1 is a schematic diagram illustrating a flow of a method for identifying abnormal behavior according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a plurality of cameras in an embodiment of the invention;
FIG. 3 is a diagram showing an association relationship between an included angle of a camera and an identification accuracy rate in an embodiment of the present invention;
FIG. 4 is a diagram illustrating an abnormal behavior recognition apparatus according to an embodiment of the present invention;
fig. 5 is a schematic diagram of another structure of an abnormal behavior recognition apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
In order to solve the technical problem that identification accuracy is low due to the fact that identification of abnormal behaviors is performed based on a single-orientation camera in the prior art, an embodiment of the present invention provides an identification method of abnormal behaviors, so as to improve the identification accuracy of the abnormal behaviors, fig. 1 is a schematic diagram of a flow of the identification method of abnormal behaviors in the embodiment of the present invention, and as shown in fig. 1, the method includes:
step 101: respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and when abnormal behaviors are identified, determining at least two cameras acquiring the abnormal behaviors and an abnormal behavior confidence coefficient corresponding to each camera, wherein the plurality of cameras are arranged at different positions of a bank outlet, and the abnormal behavior identification model is obtained by training according to historical abnormal behavior data;
step 102: acquiring angle parameters of at least two cameras acquiring the abnormal behavior, and determining an included angle parameter between every two cameras;
step 103: determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras;
step 104: determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera;
step 105: determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras;
step 106: and when the confidence coefficient of the abnormal behavior is greater than or equal to a preset confidence coefficient threshold value, determining the abnormal behavior as a recognition result.
As shown in fig. 1, an embodiment of the present invention is implemented by: respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and determining at least two cameras acquiring the abnormal behavior and the corresponding abnormal behavior confidence coefficient of each camera when the abnormal behavior is identified; acquiring angle parameters of at least two cameras acquiring the abnormal behavior, and determining an included angle parameter between every two cameras; determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras; determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera; determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras; when the confidence of the abnormal behavior is larger than or equal to the preset confidence threshold, the abnormal behavior is determined as the recognition result, the method can carry out comprehensive analysis and secondary optimization on the abnormal behavior recognition result under multiple angles based on the multiple cameras and the abnormal behavior recognition model, solves the limitation that the image depth cannot be recognized at a single angle, avoids the misjudgment of the abnormal behavior caused by the problems of personnel dislocation, shielding and the like, and improves the accuracy of the abnormal behavior recognition.
In specific implementation, in step 101, fig. 2 is a schematic diagram of a plurality of cameras in the embodiment of the present invention, as shown in fig. 2, a plurality of cameras may be installed at a bank outlet centering on a customer activity area, customer behavior images may be collected from different angles, and then the customer behavior images collected by the cameras are respectively input into an abnormal behavior recognition model, where the abnormal behavior recognition model may be a fighting behavior recognition model, the fighting behavior recognition model may be trained by collecting sample images of historical fighting behaviors, and the abnormal behavior recognition model may also be other behavior recognition models, which is not limited in this respect. When the abnormal behavior is identified, determining at least two cameras which acquire the abnormal behavior and the confidence coefficient of the abnormal behavior corresponding to each camera, for example, the cameras a and B acquire a fighting event together, and the confidence coefficient of the abnormal behavior corresponding to the camera a is X1And the confidence coefficient of the abnormal behavior corresponding to the camera B is X2Confidence of abnormal behavior X1And X2Can be directly obtained through an abnormal behavior recognition model.
In one embodiment, before obtaining the angle parameters of the at least two cameras acquiring the abnormal behavior in step 102, the method may further include:
configuring angle parameters of a plurality of cameras based on 0-359 degrees by taking a client as a circle center and taking the angle of any camera as a reference angle; establishing an angle parameter configuration table of each camera according to the angle parameter of each camera;
and obtaining the angle parameters of the at least two cameras from the angle parameter configuration table corresponding to the at least two cameras which acquire the abnormal behavior.
In specific implementation, as shown in fig. 2, in step 102, a plurality of cameras may be installed at a banking outlet with a customer activity area as a center, an orientation angle of any camera may be taken as a reference angle of 0 degrees, or an angle of any angle is taken as a reference angle of 0 degrees, angle parameters of each camera are configured based on 0 degrees to 359 degrees, an angle parameter configuration table of each camera may also be established, and the angle parameter configuration table may include a number and an angle parameter of each camera. When abnormal behaviors are identified, the angle parameters of at least two cameras can be read from the angle parameter configuration table corresponding to the at least two cameras collecting the abnormal behaviors, and the included angle parameter between every two cameras is determined, for example, the cameras A, B and C simultaneously collect fighting events together, the angle parameter of the camera A is 0 degrees, the angle parameter of the camera B is 30 degrees, the angle parameter of the camera C is 90 degrees, the included angle parameter between the cameras A, B is 30 degrees, the included angle parameter between the cameras A, C is 90 degrees, and the included angle parameter between the cameras B, C is 60 degrees.
In one embodiment, before determining the recognition accuracy of each two cameras according to the angle parameter between each two cameras in step 103, the method further includes:
determining an incidence relation between an included angle parameter between two cameras and identification accuracy according to historical abnormal behavior data of the cameras;
and determining the identification accuracy of every two cameras according to the included angle parameter between every two cameras and the incidence relation between the included angle parameter between every two cameras and the identification accuracy.
In one embodiment, the correlation between the angle parameter between the two cameras and the recognition accuracy is a sinusoidal relationship.
In specific implementation, the inventor has conducted a great deal of research on abnormal behavior images acquired by a plurality of cameras in a historical database and actually occurring abnormal behaviors, and finds that the identification accuracy of the abnormal behaviors by the plurality of cameras and the incidence relation between the included angles of the two cameras follow a sinusoidal regular distribution, fig. 3 is a graph of the incidence relation between the included angles of the cameras and the identification accuracy in the embodiment of the present invention, as shown in fig. 3, for example: the camera A gathers together fighting incident with B simultaneously, and when the contained angle parameter between camera A and B was 90 or 270, the discernment rate of accuracy was the highest, and when the contained angle parameter between camera A and B was 180 or 360, the discernment rate of accuracy was the lowest. In step 103, the identification accuracy of each two cameras can be determined according to the included angle parameter between each two cameras and the sinusoidal relationship between the included angle parameter between the two cameras and the identification accuracy, for example, when the included angle parameter between the cameras a and B is 90 °, the identification accuracy of the cameras a and B is 1, and when the included angle parameter between the cameras a and B is 180 °, the identification accuracy of the cameras a and B is 0.
In one embodiment, in step 104, the confidence of the abnormal behavior of each two cameras can be determined according to the following formula (1), including:
Y＝|sin(θ1-θ2)×(X1+X2)|
(1)
wherein Y is the confidence of abnormal behaviors of the two cameras, and theta1And theta2Respectively, angle parameter, X, of each camera1And X2And respectively corresponding abnormal behavior confidence degrees of each camera.
In specific implementation, the abnormal behavior confidence of each two cameras can be determined according to the identification accuracy of each two cameras and the abnormal behavior confidence corresponding to each camera according to the formula (1), for example, the cameras a and B simultaneously collect fighting events together, and the abnormal behavior confidence corresponding to the camera a is X1Angle parameter θ of camera A1Is 60 degrees, the confidence coefficient of the abnormal behavior corresponding to the camera B is X2Angle parameter θ of camera B1Is 135 deg. sin (theta)1-θ2) Substituting the parameters into the formula (1) for the identification accuracy of the cameras A and B can obtain the confidence coefficients of the abnormal behaviors of the cameras A and B.
In one embodiment, step 105 may comprise:
when the number of the cameras for acquiring the abnormal behaviors is two, determining the confidence coefficients of the abnormal behaviors of the two cameras as the confidence coefficients of the abnormal behaviors;
and when the number of the cameras for acquiring the abnormal behaviors is more than two, determining the maximum confidence coefficient of the confidence coefficients of the abnormal behaviors of every two cameras as the confidence coefficient of the abnormal behaviors.
In specific implementation, in step 105, if the number of the cameras acquiring the abnormal behavior is two, for example: the cameras acquiring the abnormal behavior are cameras a and B, the confidence levels of the abnormal behavior of the cameras a and B obtained in step 104 may be determined as the confidence levels of the abnormal behavior, if the number of the cameras acquiring the abnormal behavior is greater than two, the confidence levels of the abnormal behavior of every two cameras may be respectively calculated according to formula (1) in step 104, for example, when the cameras acquiring the abnormal behavior are cameras A, B and C, the confidence levels of the abnormal behavior of the cameras a and B, the confidence levels of the abnormal behavior of the cameras a and C, and the confidence levels of the abnormal behavior of the cameras B and C may be respectively calculated according to formula (1), and the maximum confidence level is determined as the confidence level of the abnormal behavior.
In one embodiment, further comprising: and carrying out abnormal behavior alarm on the identification result.
In specific implementation, in step 106, a confidence threshold may be preset, and when the confidence of the abnormal behavior is greater than or equal to the preset confidence threshold, the abnormal behavior is determined as an identification result, and an alarm is given to the identification result to remind a worker to handle the abnormality in time. It should be noted that when only one camera that acquires an abnormal behavior is provided, and none of the other cameras acquires the abnormal behavior, the overall confidence level is considered to be low, and the abnormal behavior may be filtered according to a preset confidence level threshold, and only recording is performed, and no alarm is performed.
The following is a specific example to facilitate an understanding of how the invention may be practiced.
Firstly, taking a customer activity area as a center, installing a plurality of cameras at a bank outlet, taking the orientation angle of any camera as a reference angle of 0 degree, configuring the angle parameters of each camera based on 0 to 359 degrees, establishing an angle parameter configuration table of each camera, and then executing the following steps:
the first step is as follows: respectively inputting the customer behavior images acquired by the cameras into an abnormal behavior identification model, and when the abnormal behavior is identified, determining the cameras A and B acquiring the abnormal behavior and the confidence X of the abnormal behavior corresponding to the camera A1And the confidence coefficient X of the abnormal behavior corresponding to the camera B2；
The second step is that: reading the angle parameter of the camera A to be 60 degrees and the angle parameter of the camera B to be 135 degrees from the angle parameter configuration table corresponding to the cameras A and B which acquire the abnormal behavior, and calculating the included angle parameter between the cameras A, B to be 75 degrees;
the third step: determining the identification accuracy of the cameras A and B to be sin75 degrees according to the fact that the included angle parameter between the cameras A, B is 75 degrees and the sine relationship between the included angle parameter between the two cameras and the identification accuracy;
the fourth step: according to the recognition accuracy rate of the cameras A and B being sin75 degrees and the corresponding abnormal behavior confidence degree X of the camera A1And the confidence coefficient X of the abnormal behavior corresponding to the camera B2And calculating the confidence degrees of the abnormal behaviors of the cameras A and B according to the formula (1).
The fifth step: determining the confidence coefficients of the abnormal behaviors of the cameras A and B as the confidence coefficients of the abnormal behaviors, respectively calculating the confidence coefficients of the abnormal behaviors of every two cameras according to the formula (1) when the number of the cameras for collecting the abnormal behaviors is more than two, and determining the maximum confidence coefficient of the confidence coefficients of the abnormal behaviors of every two cameras as the confidence coefficient of the abnormal behaviors;
and a sixth step: and when the confidence of the abnormal behavior is greater than or equal to a preset confidence threshold, determining the abnormal behavior as an identification result, and giving an alarm on the abnormal behavior on the identification result to remind a worker to deal with the abnormality in time. And when only one camera for collecting the abnormal behaviors is available, filtering the abnormal behaviors according to a preset confidence level threshold value without alarming.
Based on the same inventive concept, the embodiment of the present invention further provides an abnormal behavior recognition apparatus, such as the following embodiments. Because the principle of solving the problem of the abnormal behavior recognition device is similar to that of the abnormal behavior recognition method, the implementation of the device can be referred to the implementation of the method, and repeated details are not repeated. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
An embodiment of the present invention provides an abnormal behavior recognition apparatus, so as to improve accuracy of abnormal behavior recognition, where fig. 4 is a schematic diagram of a structure of the abnormal behavior recognition apparatus in the embodiment of the present invention, as shown in fig. 4, the apparatus includes:
the parameter obtaining module 01 is used for respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and when an abnormal behavior is identified, determining at least two cameras acquiring the abnormal behavior and an abnormal behavior confidence coefficient corresponding to each camera, wherein the plurality of cameras are arranged at different positions of a bank website, and the abnormal behavior identification model is obtained by training according to historical abnormal behavior data;
the included angle determining module 02 is used for acquiring the angle parameters of at least two cameras acquiring the abnormal behavior and determining the included angle parameter between every two cameras;
the accuracy determining module 03 is configured to determine the identification accuracy of each two cameras according to an included angle parameter between each two cameras;
the confidence coefficient determining module 04 is configured to determine the confidence coefficient of the abnormal behavior of each two cameras according to the recognition accuracy of each two cameras and the confidence coefficient of the abnormal behavior corresponding to each camera;
the abnormal behavior confidence determining module 05 is configured to determine a confidence of the abnormal behavior according to the abnormal behavior confidence of every two cameras;
and the recognition result determining module 06 is configured to determine the abnormal behavior as the recognition result when the confidence of the abnormal behavior is greater than or equal to the preset confidence threshold.
Fig. 5 is a schematic diagram of another structure of an abnormal behavior recognition apparatus in an embodiment of the present invention, as shown in fig. 5, in an embodiment, the apparatus may further include: an alarm module 07, configured to: and carrying out abnormal behavior alarm on the identification result.
As shown in fig. 5, in one embodiment, the apparatus may further include: an angle parameter configuration table establishing module 08, configured to:
configuring angle parameters of a plurality of cameras based on 0-359 degrees by taking a client as a circle center and taking the angle of any camera as a reference angle;
establishing an angle parameter configuration table of each camera according to the angle parameter of each camera;
the included angle determining module 02 is specifically configured to: and obtaining the angle parameters of the at least two cameras from the angle parameter configuration table corresponding to the at least two cameras which acquire the abnormal behavior.
As shown in fig. 5, in one embodiment, the apparatus may further include: an association determination module 09 configured to:
determining an incidence relation between an included angle parameter between two cameras and identification accuracy according to historical abnormal behavior data of the cameras;
the accuracy determining module 03 is specifically configured to: and determining the identification accuracy of every two cameras according to the included angle parameter between every two cameras and the incidence relation between the included angle parameter between every two cameras and the identification accuracy.
In one embodiment, the confidence determination module 04 is specifically configured to:
determining the confidence of the abnormal behaviors of every two cameras according to the following modes, including:
Y＝|sin(θ1-θ2)×(X1+X2)|；
wherein Y is the confidence of abnormal behaviors of the two cameras, and theta1And theta2Respectively, angle parameter, X, of each camera1And X2Is respectively for eachAnd (4) the corresponding abnormal behavior confidence of the camera.
In one embodiment, the abnormal behavior confidence determination module 05 is specifically configured to:
when the number of the cameras for acquiring the abnormal behaviors is two, determining the confidence coefficients of the abnormal behaviors of the two cameras as the confidence coefficients of the abnormal behaviors;
and when the number of the cameras for acquiring the abnormal behaviors is more than two, determining the maximum confidence coefficient of the confidence coefficients of the abnormal behaviors of every two cameras as the confidence coefficient of the abnormal behaviors.
The embodiment of the invention also provides computer equipment which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor realizes the identification method of the abnormal behaviors when executing the computer program.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program for executing the above-mentioned method for identifying an abnormal behavior is stored in the computer-readable storage medium.
In summary, the embodiment of the present invention provides: respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and determining at least two cameras acquiring the abnormal behavior and the corresponding abnormal behavior confidence coefficient of each camera when the abnormal behavior is identified; acquiring angle parameters of at least two cameras acquiring the abnormal behavior, and determining an included angle parameter between every two cameras; determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras; determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera; determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras; when the confidence of the abnormal behavior is larger than or equal to the preset confidence threshold, the abnormal behavior is determined as the recognition result, the method can carry out comprehensive analysis and secondary optimization on the abnormal behavior recognition result under multiple angles based on the multiple cameras and the abnormal behavior recognition model, solves the limitation that the image depth cannot be recognized at a single angle, avoids the misjudgment of the abnormal behavior caused by the problems of personnel dislocation, shielding and the like, and improves the accuracy of the abnormal behavior recognition.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and variations of the embodiment of the present invention may occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for identifying abnormal behavior, comprising:
respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and when abnormal behaviors are identified, determining at least two cameras acquiring the abnormal behaviors and an abnormal behavior confidence coefficient corresponding to each camera, wherein the plurality of cameras are arranged at different positions of a bank outlet, and the abnormal behavior identification model is obtained by training according to historical abnormal behavior data;
acquiring angle parameters of at least two cameras acquiring the abnormal behavior, and determining an included angle parameter between every two cameras;
determining the identification accuracy of each two cameras according to the included angle parameter between each two cameras;
determining the abnormal behavior confidence of each two cameras according to the identification accuracy of each two cameras and the corresponding abnormal behavior confidence of each camera;
determining the confidence coefficient of the abnormal behavior according to the confidence coefficients of the abnormal behaviors of every two cameras;
and when the confidence coefficient of the abnormal behavior is greater than or equal to a preset confidence coefficient threshold value, determining the abnormal behavior as a recognition result.
2. The method of claim 1, further comprising:
and carrying out abnormal behavior alarm on the identification result.
3. The method of claim 1, wherein prior to obtaining the angular parameters of the at least two cameras that acquired the abnormal behavior, the method further comprises:
configuring angle parameters of a plurality of cameras based on 0-359 degrees by taking a client as a circle center and taking the angle of any camera as a reference angle;
establishing an angle parameter configuration table of each camera according to the angle parameter of each camera;
obtaining angle parameters of at least two cameras acquiring the abnormal behavior, comprising:
and obtaining the angle parameters of the at least two cameras from the angle parameter configuration table corresponding to the at least two cameras which acquire the abnormal behavior.
4. The method of claim 1, wherein before determining the recognition accuracy of each two cameras based on the angle parameter between each two cameras, the method further comprises:
determining an incidence relation between an included angle parameter between two cameras and identification accuracy according to historical abnormal behavior data of the cameras;
according to the included angle parameter between every two cameras, confirm the discernment rate of accuracy of every two cameras, include:
and determining the identification accuracy of every two cameras according to the included angle parameter between every two cameras and the incidence relation between the included angle parameter between every two cameras and the identification accuracy.
5. The method of claim 4, wherein the correlation between the angle parameter between two cameras and the recognition accuracy is sinusoidal.
6. The method of claim 1, wherein determining the confidence of the abnormal behavior of each two cameras comprises:
Y＝|sin(θ1-θ2)×(X1+X2)|；
wherein Y is the confidence of abnormal behaviors of the two cameras, and theta1And theta2Respectively, angle parameter, X, of each camera1And X2And respectively corresponding abnormal behavior confidence degrees of each camera.
7. The method of claim 1, wherein determining the confidence level of the abnormal behavior according to the confidence levels of the abnormal behavior of every two cameras comprises:
when the number of the cameras for acquiring the abnormal behaviors is two, determining the confidence coefficients of the abnormal behaviors of the two cameras as the confidence coefficients of the abnormal behaviors;
and when the number of the cameras for acquiring the abnormal behaviors is more than two, determining the maximum confidence coefficient of the confidence coefficients of the abnormal behaviors of every two cameras as the confidence coefficient of the abnormal behaviors.
8. An apparatus for identifying abnormal behavior, comprising:
the parameter obtaining module is used for respectively inputting customer behavior data acquired by a plurality of cameras into an abnormal behavior identification model, and when abnormal behaviors are identified, determining at least two cameras acquiring the abnormal behaviors and an abnormal behavior confidence coefficient corresponding to each camera, wherein the plurality of cameras are arranged at different positions of a bank website, and the abnormal behavior identification model is obtained by training according to historical abnormal behavior data;
the included angle determining module is used for acquiring the angle parameters of at least two cameras acquiring the abnormal behavior and determining the included angle parameter between every two cameras;
the accuracy rate determining module is used for determining the identification accuracy rate of each two cameras according to the included angle parameter between each two cameras;
the confidence coefficient determining module is used for determining the abnormal behavior confidence coefficient of each two cameras according to the recognition accuracy rate of each two cameras and the abnormal behavior confidence coefficient corresponding to each camera;
the abnormal behavior confidence determining module is used for determining the confidence of the abnormal behavior according to the abnormal behavior confidence of every two cameras;
and the identification result determining module is used for determining the abnormal behavior as an identification result when the confidence coefficient of the abnormal behavior is greater than or equal to a preset confidence coefficient threshold value.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program for executing the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010406965.6A CN111597970B (en) | 2020-05-14 | 2020-05-14 | Abnormal behavior identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010406965.6A CN111597970B (en) | 2020-05-14 | 2020-05-14 | Abnormal behavior identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111597970A true CN111597970A (en) | 2020-08-28 |
| CN111597970B CN111597970B (en) | 2023-05-02 |
Family
ID=72183719
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010406965.6A Active CN111597970B (en) | 2020-05-14 | 2020-05-14 | Abnormal behavior identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111597970B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112215150A (en) * | 2020-10-13 | 2021-01-12 | 中国银行股份有限公司 | Customer behavior identification method and device |
| CN112349150A (en) * | 2020-11-19 | 2021-02-09 | 飞友科技有限公司 | Video acquisition method and system for airport flight guarantee time node |
| CN113380038A (en) * | 2021-07-06 | 2021-09-10 | 深圳市城市交通规划设计研究中心股份有限公司 | Vehicle dangerous behavior detection method, device and system |
| CN113992829A (en) * | 2021-10-29 | 2022-01-28 | 华清科盛(北京)信息技术有限公司 | Intelligent sorting system and method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355581A (en) * | 2016-09-29 | 2017-01-25 | 上海智驾电子科技有限公司 | Device and method for realizing online calibration of vehicle-mounted camera by vehicle detection |
| CN108189626A (en) * | 2017-12-28 | 2018-06-22 | 深圳市灵动飞扬科技有限公司 | A kind of tire pressure detection method, device, storage medium and automobile |
| CN109460702A (en) * | 2018-09-14 | 2019-03-12 | 华南理工大学 | Passenger's abnormal behaviour recognition methods based on human skeleton sequence |
| CN110020618A (en) * | 2019-03-27 | 2019-07-16 | 江南大学 | A kind of crowd's abnormal behaviour monitoring method can be used for more shooting angle |
| CN110443109A (en) * | 2019-06-11 | 2019-11-12 | 万翼科技有限公司 | Abnormal behaviour monitor processing method, device, computer equipment and storage medium |
| CN110532857A (en) * | 2019-07-16 | 2019-12-03 | 杭州电子科技大学 | Based on the Activity recognition image analysis system under multi-cam |
| WO2019235776A1 (en) * | 2018-06-07 | 2019-12-12 | 엘지이노텍 주식회사 | Device and method for determining abnormal object |
| CN110738078A (en) * | 2018-07-19 | 2020-01-31 | 青岛海信移动通信技术股份有限公司 | face recognition method and terminal equipment |
-
2020
- 2020-05-14 CN CN202010406965.6A patent/CN111597970B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355581A (en) * | 2016-09-29 | 2017-01-25 | 上海智驾电子科技有限公司 | Device and method for realizing online calibration of vehicle-mounted camera by vehicle detection |
| CN108189626A (en) * | 2017-12-28 | 2018-06-22 | 深圳市灵动飞扬科技有限公司 | A kind of tire pressure detection method, device, storage medium and automobile |
| WO2019235776A1 (en) * | 2018-06-07 | 2019-12-12 | 엘지이노텍 주식회사 | Device and method for determining abnormal object |
| CN110738078A (en) * | 2018-07-19 | 2020-01-31 | 青岛海信移动通信技术股份有限公司 | face recognition method and terminal equipment |
| CN109460702A (en) * | 2018-09-14 | 2019-03-12 | 华南理工大学 | Passenger's abnormal behaviour recognition methods based on human skeleton sequence |
| CN110020618A (en) * | 2019-03-27 | 2019-07-16 | 江南大学 | A kind of crowd's abnormal behaviour monitoring method can be used for more shooting angle |
| CN110443109A (en) * | 2019-06-11 | 2019-11-12 | 万翼科技有限公司 | Abnormal behaviour monitor processing method, device, computer equipment and storage medium |
| CN110532857A (en) * | 2019-07-16 | 2019-12-03 | 杭州电子科技大学 | Based on the Activity recognition image analysis system under multi-cam |
Non-Patent Citations (1)
| Title |
|---|
| 于明学;金鑫;李晓东;吴亚明;: "基于3D卷积双路神经网络的考场行为异常识别" * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112215150A (en) * | 2020-10-13 | 2021-01-12 | 中国银行股份有限公司 | Customer behavior identification method and device |
| CN112215150B (en) * | 2020-10-13 | 2023-10-24 | 中国银行股份有限公司 | Customer behavior recognition method and device |
| CN112349150A (en) * | 2020-11-19 | 2021-02-09 | 飞友科技有限公司 | Video acquisition method and system for airport flight guarantee time node |
| CN113380038A (en) * | 2021-07-06 | 2021-09-10 | 深圳市城市交通规划设计研究中心股份有限公司 | Vehicle dangerous behavior detection method, device and system |
| CN113992829A (en) * | 2021-10-29 | 2022-01-28 | 华清科盛(北京)信息技术有限公司 | Intelligent sorting system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111597970B (en) | 2023-05-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111597970A (en) | Abnormal behavior identification method and device | |
| CN109858244B (en) | Method and system for detecting abnormal behaviors of processes in container | |
| CN104794136A (en) | Fault analysis method and device | |
| US11336672B2 (en) | Detecting behavioral anomaly in machine learned rule sets | |
| CN105809448B (en) | Clustering method and system for account transactions | |
| CN106104496A (en) | The abnormality detection not being subjected to supervision for arbitrary sequence | |
| CN107273924A (en) | The Fault Analysis of Power Plants method of multi-data fusion based on fuzzy cluster analysis | |
| CN116308366B (en) | Payment security monitoring processing method, system and storage medium | |
| CN116668039A (en) | Computer remote login identification system and method based on artificial intelligence | |
| CN108696486A (en) | A kind of abnormal operation behavioral value processing method and processing device | |
| WO2020167155A1 (en) | Method and system for detecting troubling events during interaction with a self-service device | |
| CN113780342A (en) | Intelligent detection method and device based on self-supervision pre-training and robot | |
| CN112416979A (en) | Anti-fraud method, device and equipment based on geographic position and storage medium | |
| CN111598753A (en) | Suspect recommendation method and device, electronic equipment and storage medium | |
| Hwang et al. | Cut and continuous paste towards real-time deep fall detection | |
| CN107609330A (en) | Access log mining-based internal threat abnormal behavior analysis method | |
| Luca et al. | Anomaly detection using the Poisson process limit for extremes | |
| Nisa et al. | Implementation of Personal Protective Equipment Detection Using Django and Yolo Web at Paiton Steam Power Plant (PLTU) | |
| CN111626193A (en) | Face recognition method, face recognition device and readable storage medium | |
| CN116311029B (en) | Construction operation violation identification system and method based on big data | |
| KR102111136B1 (en) | Method, device and program for generating respond directions against attack event | |
| Artikis et al. | Industry paper: a prototype for credit card fraud management | |
| CN110813977B (en) | On-site recycling and retreatment process for bottom materials | |
| CN115809404B (en) | Detection threshold calculation method and device, storage medium and electronic equipment | |
| CN109446265B (en) | Complex abnormity identification method and identification system based on workflow |
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 |