AI-based target person abnormal behavior prediction method

Abstract

The invention relates to the technical field of IT and software development, and discloses an AI-based target personnel abnormal behavior prediction method, which comprises the following steps: S1, obtaining corrected personnel information data through a correction information module; the personnel information data comprises face information data and correction types; s2, dividing the target area into an activity area and a forbidden area through a division module; s3, acquiring image data in the corresponding moving area through an image acquisition unit; the position acquisition unit acquires position information of the corresponding correction personnel; s4, analyzing the image data and the position data through an analysis module, and judging whether the correction personnel have abnormal behaviors or not; if it is judged that the abnormal behavior exists, risk rating is carried out on the abnormal behavior, and corresponding early warning is carried out according to a rating result; whether each correction person has an abnormal behavior or not is automatically identified and analyzed, and corresponding early warning is carried out when different abnormal behaviors occur, so that a manager can effectively process the abnormal behaviors in time.

Description

Technical Field

[0001] The present invention relates to the field of IT and software development technology, and in particular to an AI-based method for predicting abnormal behavior of a target person. Background Art

[0002] Community correction is a form of criminal punishment execution that places criminals who have committed less serious crimes, have little subjective malice or cause less social harm in designated communities to receive correction and improve their behavior. The aim is to help criminals readjust to society and reduce recidivism rates.

[0003] Community correction service is a comprehensive program that requires the provision of multiple services such as supervision, education, and vocational training to ensure maximum correction results. In the correction system, monitoring and analyzing the behavior of correction personnel is crucial. Currently, the main way to supervise correction personnel is through video surveillance and manual observation.

[0004] Since there are many areas under video surveillance and few staff members to manage the video surveillance, it is impossible to view all video images at the same time. Therefore, when correctional personnel exhibit abnormal behavior, it is difficult to detect, which may cause serious safety accidents. Summary of the invention

[0005] The purpose of this invention is to provide an AI-based method for predicting abnormal behavior of target personnel to solve the following technical problems:

[0006] How to promptly detect abnormal behavior of correctional personnel and issue early warnings.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] An AI-based method for predicting abnormal behavior of a target person includes the following steps:

[0009] S1: Obtain correction personnel information data through a correction information module; the personnel information data includes face information data and correction type;

[0010] S2: Divide the target area into multiple activity areas and multiple restricted areas through the division module;

[0011] S3: obtaining real-time monitoring information of correction personnel through a monitoring module; the monitoring module includes a plurality of image acquisition units and a plurality of position acquisition units; the image acquisition units correspond one-to-one with the activity areas and are used to acquire image data in the corresponding activity areas; the position acquisition units correspond one-to-one with the correction personnel and are used to acquire position information of the corresponding correction personnel;

[0012] S4: The image data and location data are analyzed through the analysis module to determine whether the correctional personnel have abnormal behavior; if it is determined that abnormal behavior exists, the risk of the abnormal behavior is rated and corresponding warnings are issued according to the rating results.

[0013] As a further solution of the present invention: by formula:

[0014] P n =σ n f(Q np -Q0)

[0015] Calculate the third abnormal behavior judgment index P of the nth activity area n :

[0016] Where f(X) is the judgment function. When X>0, f(X)=1; when X≤0, f(X)=0; σ n is the activity time coefficient of the nth activity area; Q np is the number of correctional personnel in the nth activity area; Q0 is the preset number of correctional personnel.

[0017] As a further solution of the present invention: the process of judging whether the third abnormal behavior exists is as follows:

[0018] When P n = 0, there is no third abnormal behavior in the nth activity area;

[0019] When P n =1, the nth active area has the third abnormal behavior.

[0020] As a further solution of the present invention: by formula:

[0021] P m =σ q f(T ms -ΔT m )

[0022] Calculate the second abnormal behavior judgment index P of the mth correctional personnel m :

[0023] Among them, σ q is the sign-in status coefficient; T ms The length of time that the current time exceeds the preset check-in time; ΔT m The allowed overtime for the mth correctional officer.

[0024] As a further solution of the present invention: the process of judging whether the second abnormal behavior exists is as follows:

[0025] When P m =0, the mth correctional personnel does not have the second abnormal behavior and no warning is needed;

[0026] When P m =1, the mth correctional personnel has a second abnormal behavior and an early warning is issued.

[0027] As a further solution of the present invention: by formula:

[0028]

[0029] Calculate the allowed overtime ΔT for the mth correctional personnel m ;

[0030] Among them, Q alln is the number of times the mth correctional officer should sign in; Q qm is the number of times the mth correctional personnel have signed in; ρ0 is the preset ratio; ΔT0 is the preset allowed overtime; θ m is the adjustment coefficient for the mth correctional officer.

[0031] As a further solution of the present invention: by formula:

[0032]

[0033] Calculate the adjustment coefficient θ for the mth correctional personnel m ;

[0034] Among them, Q jz Q is the number of the third abnormal behavior of the mth correctional personnel so far; cx is the number of the fourth abnormal behavior of the mth correctional personnel so far; Q cj is the number of abnormal behaviors of the mth correction personnel so far; ω is the first preset coefficient; μ a is the correction level coefficient; γ1 is the first weight coefficient; γ2 is the second weight coefficient; γ3 is the third weight coefficient.

[0035] As a further solution of the present invention: the abnormal behavior includes a first abnormal behavior, a second abnormal behavior, a third abnormal behavior and a fourth abnormal behavior;

[0036] The process for rating abnormal behavior is:

[0037] If the mth correctional personnel exhibit the first abnormal behavior or the fourth abnormal behavior, the abnormal behavior is rated as high risk and a first-level warning is issued;

[0038] If the mth correctional personnel exhibits the third abnormal behavior, the abnormal behavior is rated as medium risk and a second-level warning is issued;

[0039] If the mth correctional officer exhibits a second abnormal behavior, the abnormal behavior is rated as low risk and a third-level warning is issued.

[0040] Beneficial effects of the present invention:

[0041] (1) The present invention first obtains correction personnel information data through a correction information module; the personnel information data includes facial information data and correction type; then the target area is divided into multiple activity areas and multiple restricted areas through a division module; then the real-time monitoring information of the correction personnel is obtained through a monitoring module; the monitoring module includes multiple image acquisition units and multiple position acquisition units; the image acquisition unit acquires image data in the corresponding activity area; the position acquisition unit acquires the position information of the corresponding correction personnel; finally, the image data and position data are analyzed through an analysis module to determine whether the correction personnel have abnormal behavior; if it is determined that there is abnormal behavior, the abnormal behavior is rated for risk, and corresponding warnings are issued according to the rating results; the analysis module is used to automatically identify and analyze whether each correction personnel has abnormal behavior, and corresponding warnings are issued when different abnormal behaviors occur, so that management personnel can effectively deal with abnormal behaviors in a timely manner;

[0042] (2) Correction level coefficient μ in this embodiment a Determined according to the correction level of the correction personnel, the correction level coefficient μ a The larger the value, the greater the risk of abnormal behavior. The adjustment coefficient θ of the mth correctional officer m The larger the value, the longer the allowed time for the mth correctional personnel to exceed the limit ΔT m The smaller the number, the number of times Q the third abnormal behavior of the mth correctional personnel has been so far jz The larger the value, the greater the risk of abnormal behavior. The adjustment coefficient θ of the mth correctional officer m The larger the value, the longer the allowed time for the mth correctional personnel to exceed the limit ΔT m The smaller the number, the number of the fourth abnormal behavior of the mth correctional personnel so far is Q cj The larger the value, the greater the risk of abnormal behavior. The adjustment coefficient θ of the mth correctional officer m The larger the value, the longer the allowed time for the mth correctional personnel to exceed the limit ΔT m The smaller it is, the greater the number of first abnormal behaviors of the mth correctional personnel so far, the greater the risk of abnormal behavior, and the adjustment coefficient θ of the mth correctional personnel m The larger the value, the longer the allowed time for the mth correctional personnel to exceed the limit ΔT m The smaller;

[0043] (3) After the present invention determines that abnormal behavior has occurred, if the mth correctional personnel has abnormal behavior of carrying weapons or breaking into restricted areas, the abnormal behavior is rated as high risk; a first-level warning is issued, a dispatch alarm is issued, and all management personnel quickly arrive at the scene; if the mth correctional personnel has abnormal behavior of gathering a crowd, the abnormal behavior is rated as medium risk, and a second-level warning is issued; a warning alarm is issued, and the gathering correctional personnel are guided to evacuate through broadcasting; if the mth correctional personnel has abnormal behavior of not signing in within the specified time, the abnormal behavior is rated as low risk; a third-level warning is issued; the location information of the correctional personnel who have not signed in within the specified time is sent to the management personnel, and the management personnel handle it; it realizes automatic identification and analysis of whether each correctional personnel has abnormal behavior, and issues corresponding warnings when different abnormal behaviors occur, so that the management personnel can effectively handle the abnormal behavior in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The present invention will be further described below in conjunction with the accompanying drawings.

[0045] Figure 1 The present invention is a method flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] See also Figure 1 As shown, in one embodiment, a method for predicting abnormal behavior of a target person based on AI is provided, comprising the following steps:

[0048] S1: Obtain correction personnel information data through a correction information module; the personnel information data includes face information data and correction type;

[0049] S2: Divide the target area into multiple activity areas and multiple restricted areas through the division module;

[0050] S3: obtaining real-time monitoring information of correction personnel through a monitoring module; the monitoring module includes a plurality of image acquisition units and a plurality of position acquisition units; the image acquisition units correspond one-to-one with the activity areas and are used to acquire image data in the corresponding activity areas; the position acquisition units correspond one-to-one with the correction personnel and are used to acquire position information of the corresponding correction personnel;

[0051] S4: Analyze the image data and location data through the analysis module to determine whether the correctional personnel have abnormal behavior; if it is determined that there is abnormal behavior, the abnormal behavior is rated for risk, and a corresponding warning is issued according to the rating result;

[0052] Through the above technical scheme, this embodiment first obtains the correction personnel information data through the correction information module; the personnel information data includes facial information data and correction type; then the target area is divided into multiple activity areas and multiple restricted areas through the division module; then the real-time monitoring information of the correction personnel is obtained through the monitoring module; the monitoring module includes multiple image acquisition units and multiple position acquisition units; the image acquisition unit corresponds to the activity area one by one, and is used to collect image data in the corresponding activity area; the position acquisition unit corresponds to the correction personnel one by one, and is used to collect the position information of the corresponding correction personnel; finally, the image data and position data are analyzed by the analysis module to determine whether the correction personnel have abnormal behavior; if it is determined that there is abnormal behavior, the abnormal behavior is risk rated, and corresponding warnings are issued according to the rating results; the analysis module is used to automatically identify and analyze whether each correction personnel has abnormal behavior, and corresponding warnings are issued when different abnormal behaviors occur, so that management personnel can effectively deal with abnormal behaviors in a timely manner.

[0053] As an implementation mode of the present invention, the abnormal behavior includes a first abnormal behavior, a second abnormal behavior, a third abnormal behavior and a fourth abnormal behavior; the first abnormal behavior is entering a restricted area; the second abnormal behavior is failing to sign in within the specified time; the third abnormal behavior is gathering; and the fourth abnormal behavior is carrying a weapon.

[0054] Through the above technical scheme, this embodiment determines whether the correctional personnel is located in the restricted area by obtaining the position information of the correctional personnel, and whether the correctional personnel has entered the restricted area based on the position information of the correctional personnel. This is the existing technology and will not be described in detail; by obtaining the position information of the correctional personnel, the activity area where the correctional personnel is located is determined; the image information of the activity area is obtained through the image information of the image acquisition unit of the activity area, and the image information is performed by face recognition to determine whether the position information of the correctional personnel is correct; by performing hand recognition on the image information, it is determined whether the correctional personnel is holding a weapon; this is the existing technology and will not be described in detail here.

[0055] As an implementation mode of the present invention, by formula:

[0056] P n =σ n f(Q np -Q0)

[0057] Calculate the third abnormal behavior judgment index P of the nth activity area n :

[0058] Where f(X) is the judgment function. When X>0, f(X)=1; when X≤0, f(X)=0; σ n is the activity time coefficient of the nth activity area; Q np is the number of correctional personnel in the nth activity area; Q0 is the preset number of correctional personnel;

[0059] When P n = 0, there is no third abnormal behavior in the nth activity area;

[0060] When P n =1, the third abnormal behavior exists in the nth activity area;

[0061] Through the above technical solution, this embodiment sets the group activity time and the individual activity time according to the work arrangement of the nth activity area; determines the activity time type of the current time; determines the activity time coefficient σ according to the activity time type n ; When the activity time is group activity time, the activity time coefficient σ n =0; when the activity time is the single activity time, the activity time coefficient σ n =1; in the formula σ n f(Q np -Q0), the X in the judgment function f(X) refers to Q np -Q0;Q np -Q0 is the difference between the number of correctional personnel in the nth activity area and the preset number of correctional personnel; when Q np -Q0≤0, it means that the number of correctional personnel in the nth activity area does not exceed the preset number of correctional personnel; f(Q np -Q0)=0; when Q np - When Q0>0, it means that the number of correctional personnel in the nth activity area exceeds the preset number of correctional personnel; f(Q np -Q0)=1; when σ n =1 and f(Q np -Q0)=1, that is, when P n =1, it means that the number of correctional personnel in the nth activity area during the individual activity time exceeds the preset number of correctional personnel, so the correctional personnel in the nth activity area at this time all have the third abnormal behavior;

[0062] It should be noted that the preset number of correctional personnel Q0 is a preset value, which is obtained based on experience and will not be elaborated here.

[0063] As an implementation mode of the present invention, by formula:

[0064] P m =σ q f(T ms -ΔTm )

[0065] Calculate the second abnormal behavior judgment index P of the mth correctional personnel m :

[0066] Among them, σ q is the sign-in status coefficient; T ms The length of time that the current time exceeds the preset check-in time; ΔT m The allowed overtime for the mth correctional officer;

[0067] When P m =0, the mth correctional personnel has signed in within the specified time, there is no second abnormal behavior, and no warning is required;

[0068] When P m =1, the mth correctional personnel fails to sign in within the specified time, and there is a second abnormal behavior, and an early warning is issued;

[0069] Through the above technical solution, the check-in status of this embodiment includes checked-in and not checked-in; the check-in status coefficient σ q is 0; the check-in status coefficient σ for those who have not checked in q is 1; in the formula σ q f(T ms -ΔT m ), the X in the judgment function f(X) refers to T ms -ΔT m ; T ms -ΔT m is the difference between the time the current time exceeds the preset check-in time and the allowed time of the mth correctional personnel; when T ms -ΔT m ≤0, it means that the current time exceeds the preset check-in time but does not exceed the allowed overtime of the mth correctional personnel; f(T ms -ΔT m )=0; when T ms -ΔT m >0, it means that the current time exceeds the preset check-in time and exceeds the allowed overtime of the mth correctional personnel; f(T ms -ΔT m )=1; when σ q =1 and f(T ms -ΔT m )=1, that is, when P m =1, it means that the current time exceeds the preset sign-in time and exceeds the allowed overtime for the mth correctional personnel, and the mth correctional personnel has not signed in. Therefore, the mth correctional personnel has not signed in within the specified time, and an early warning is required.

[0070] As an implementation mode of the present invention, by formula:

[0071]

[0072] Calculate the allowed overtime ΔT for the mth correctional personnel m ;

[0073] Among them, Q alln is the number of times the mth correctional officer should sign in; Q qm is the number of times the mth correctional personnel have signed in; ρ0 is the preset ratio; ΔT0 is the preset allowed overtime; θ m is the adjustment coefficient of the mth correctional officer;

[0074] Through the above technical solution, this embodiment is the ratio of the number of times the mth correctional officer has signed in to the number of times he should sign in; θ m ρ0 is the preset ratio of the mth correction personnel; θ m ≥1; adjustment coefficient θ of the mth correctional personnel m The larger the value, the greater the abnormal risk of the correction personnel; therefore, the preset ratio θ of the mth correction personnel m The larger ρ0 is, is the difference between the ratio of the number of times the mth correctional personnel have signed in to the number of times they should sign in and the preset ratio of the mth correctional personnel; When , it means that the ratio of the number of times the mth correctional personnel have signed in to the number of times they should sign in is less than the preset ratio of the mth correctional personnel; therefore, the risk of abnormality of the mth correctional personnel is relatively high. Therefore, ΔT m <ΔT0; when , it means that the ratio of the number of times the mth correctional personnel have signed in to the number of times they should sign in is greater than the preset ratio of the mth correctional personnel; therefore, the risk of the mth correctional personnel being abnormal is small, so ΔT m >ΔT0;

[0075] It should be noted that the preset ratio ρ0 and the preset allowed exceeding time ΔT0 are preset values, which are obtained based on experience and will not be described in detail here.

[0076] As an implementation mode of the present invention, by formula:

[0077]

[0078] Calculate the adjustment coefficient θ for the mth correctional personnel m ;

[0079] Among them, Q jz Q is the number of the third abnormal behavior of the mth correctional personnel so far; cxis the number of the fourth abnormal behavior of the mth correctional personnel so far; Q cj is the number of abnormal behaviors of the mth correction personnel so far; ω is the first preset coefficient; μ a is the correction level coefficient; γ1 is the first weight coefficient; γ2 is the second weight coefficient; γ3 is the third weight coefficient;

[0080] Through the above technical solution, the correction level coefficient μ in this embodiment is a Determined according to the correction level of the correction personnel, the correction level coefficient μ a The larger the value, the greater the risk of abnormal behavior. The adjustment coefficient θ of the mth correctional officer m The larger the number, the number of the third abnormal behavior of the mth correctional personnel so far is Q jz The larger the value, the greater the risk of abnormal behavior. The adjustment coefficient θ of the mth correctional officer m The larger the number, the number of the fourth abnormal behavior of the mth correctional personnel so far is Q cj The larger the value, the greater the risk of abnormal behavior. The adjustment coefficient θ of the mth correctional officer m The larger the number of first abnormal behaviors of the mth correctional personnel so far, the greater the risk of abnormal behavior, and the adjustment coefficient θ of the mth correctional personnel m The bigger;

[0081] It should be noted that the correction level coefficient μ a , the first preset coefficient ω, the first weight coefficient γ1, the second weight coefficient γ2 and the third weight coefficient γ3 are preset values, which are obtained based on experience and will not be described in detail here.

[0082] As an implementation mode of the present invention, the process of rating abnormal behavior is as follows:

[0083] If the mth correctional officer shows abnormal behavior of carrying a weapon or entering a restricted area, the abnormal behavior will be rated as high risk and a level 1 warning will be issued;

[0084] If the mth correctional personnel exhibit abnormal behavior of gathering in a crowd, the abnormal behavior is rated as medium risk and a level 2 warning is issued;

[0085] If the mth correctional personnel exhibits abnormal behavior of not signing in within the specified time, the abnormal behavior is rated as low risk and a level 3 warning is issued;

[0086] Through the above technical scheme, in this embodiment, if the mth correctional personnel exhibits abnormal behavior of carrying weapons or breaking into restricted areas, the abnormal behavior is rated as high risk; a first-level warning is issued, a dispatch alarm is issued, and all management personnel quickly arrive at the scene; if the mth correctional personnel exhibits abnormal behavior of gathering a crowd, the abnormal behavior is rated as medium risk, and a second-level warning is issued; a warning alarm is issued, and the gathering correctional personnel are guided to evacuate through broadcasting; if the mth correctional personnel exhibits abnormal behavior of not signing in within the specified time, the abnormal behavior is rated as low risk; a third-level warning is issued; the location information of the correctional personnel who failed to sign in within the specified time is sent to the management personnel, and the management personnel handle it.

[0087] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for predicting abnormal behavior of target personnel based on AI, characterized in that: The following steps are involved: S1: Obtain correction personnel information data through a correction information module; the personnel information data includes face information data and correction type; S2: Divide the target area into multiple activity areas and multiple restricted areas through the division module; S3: obtaining real-time monitoring information of the correction personnel through a monitoring module; the monitoring module includes a plurality of image acquisition units and a plurality of position acquisition units; the image acquisition unit is used to acquire image data in the corresponding activity area; the position acquisition unit is used to acquire the position information of the corresponding correction personnel; S4: Analyze the image data and location data through the analysis module to determine whether the correction personnel have abnormal behavior; If abnormal behavior is determined to exist, a risk rating will be conducted on the abnormal behavior, and corresponding warnings will be issued based on the rating results.

2. According to claim 1, a method for predicting abnormal behavior of a target person based on AI is characterized in that: By formula: P n =σ n f(Q np -Q0 Calculate the third abnormal behavior judgment index P of the nth activity area n : Where f(X) is the judgment function. When X>0, f(X)=1; when X≤0, f(X)=0; σ n is the activity time coefficient of the nth activity area; Q np is the number of correctional personnel in the nth activity area; Q0 is the preset number of correctional personnel.

3. The method for predicting abnormal behavior of a target person based on AI according to claim 2, characterized in that: The process of determining whether there is a third abnormal behavior is as follows: When P n = 0, there is no third abnormal behavior in the nth activity area; When P n =1, the nth active area has the third abnormal behavior.

4. The method for predicting abnormal behavior of a target person based on AI according to claim 3 is characterized in that: By formula: P m =σ q f(T ms -ΔT m ) Calculate the second abnormal behavior judgment index P of the mth correctional personnel m : Among them, σ q is the sign-in status coefficient; T ms The length of time that the current time exceeds the preset check-in time; ΔT m The allowed overtime for the mth correctional officer.

5. The method for predicting abnormal behavior of a target person based on AI according to claim 4, characterized in that: The process of determining whether there is a second abnormal behavior is as follows: When P m =0, the mth correctional personnel does not have the second abnormal behavior and no warning is needed; When P m =1, the mth correctional personnel has a second abnormal behavior and an early warning is issued.

6. The method for predicting abnormal behavior of a target person based on AI according to claim 5, characterized in that: By formula: Calculate the allowed overtime ΔT for the mth correctional personnel m ; Among them, Q alln is the number of times the mth correctional officer should sign in; Q qm is the number of times the mth correctional personnel have signed in; ρ0 is the preset ratio; ΔT0 is the preset allowed overtime; θ m is the adjustment coefficient for the mth correctional officer.

7. The method for predicting abnormal behavior of a target person based on AI according to claim 6, characterized in that: By formula: Calculate the adjustment coefficient θ for the mth correctional personnel m ; Among them, Q jz Q is the number of the third abnormal behavior of the mth correctional personnel so far; cx is the number of the fourth abnormal behavior of the mth correctional personnel so far; Q cj is the number of abnormal behaviors of the mth correction personnel so far; ω is the first preset coefficient; μ a is the correction level coefficient; γ1 is the first weight coefficient; γ2 is the second weight coefficient; γ3 is the third weight coefficient.

8. The method for predicting abnormal behavior of a target person based on AI according to claim 7, characterized in that: The abnormal behavior includes a first abnormal behavior, a second abnormal behavior, a third abnormal behavior and a fourth abnormal behavior; The process for rating abnormal behavior is: If the mth correctional personnel exhibit the first abnormal behavior or the fourth abnormal behavior, the abnormal behavior is rated as high risk and a first-level warning is issued; If the mth correctional personnel exhibits the third abnormal behavior, the abnormal behavior is rated as medium risk and a second-level warning is issued; If the mth correctional officer exhibits a second abnormal behavior, the abnormal behavior is rated as low risk and a third-level warning is issued.

Images