Abnormal behavior detection reminder method and system

Abstract

The present invention provides an abnormal behavior detection reminder method and system, the abnormal behavior detection reminder method comprising the following steps: extracting video frames from a video according to a preset sampling cycle to perform person detection; if the presence of a person is detected from the current video frame, obtaining a video with the previous sampling cycle as the start time from the video as a behavior detection video segment and analyzing it, and judging whether the person in the detection video segment has abnormal behavior; if there is abnormal behavior, marking the corresponding behavior detection video segment as a first abnormal video segment, and marking the person with abnormal behavior as an abnormal person, performing feature recognition on the abnormal person, obtaining and generating a feature text of the abnormal person, associating the first abnormal video segment, the location information of the camera and the feature text of the abnormal person, and saving them to the data to be sent and sending them to the administrator. The present invention has low detection cost and reduces the cost of human monitoring at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of behavior monitoring, and in particular to an abnormal behavior detection and reminder method and system. Background Art

[0002] Abnormal behaviors such as theft in supermarkets every year cause great losses to retailers. Among the existing abnormal behavior detection and reminder solutions, the full-store no-dead-angle monitoring method mainly based on Amazon, the RFID radio frequency method, and the smart shelf weighing method are relatively expensive; the method based on weight scales is difficult to control the accuracy; the existing AI smart cashier on the market has a certain loss prevention effect, and most of the losses caused by abnormal behaviors do not occur at the cashier end, but during the shopping process. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings and deficiencies in the prior art and to provide an abnormal behavior detection and reminder method and system, which has low detection cost and reduces the cost of human monitoring.

[0004] An embodiment of the present invention provides an abnormal behavior detection and reminder method, comprising the following steps:

[0005] Obtain video of the detection area through the camera;

[0006] Extract video frames from the video according to a preset sampling cycle to perform person detection;

[0007] If the presence of a person is detected from the current video frame, a video with the previous sampling cycle as the start time is obtained from the video as the behavior detection video segment;

[0008] Analyze the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior;

[0009] If there is abnormal behavior, the corresponding behavior detection video segment is marked as the first abnormal video segment, and the person with abnormal behavior is marked as the abnormal person, and the abnormal person is identified, and the abnormal person feature text is obtained and generated, and the first abnormal video segment, the location information of the camera and the abnormal person feature text are associated, and saved to the data to be sent;

[0010] The data to be sent is sent to the administrator.

[0011] Compared with the prior art, the abnormal behavior detection and reminder method of the present invention can utilize existing surveillance cameras in places such as supermarkets to obtain videos, and detect and analyze the people and their behaviors in the videos, which can reduce the detection cost and the cost of manual monitoring, and can also retain the video content when the abnormal behavior of the characters occurs.

[0012] Further, extracting video frames from the video according to a preset sampling period to perform person detection includes the following steps:

[0013] Get the sampling cycle entered by the administrator;

[0014] Extracting a video frame from the video as the current video frame according to a sampling period;

[0015] Identify the outline of a static object in the current video frame. If the outline of the static object meets the preset human body motion outline, further identify the head features of the static object. If the head features of the static object are identified, determine that there is a person in the current video frame.

[0016] Further, the analyzing the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior includes the following steps:

[0017] Acquire the body behavior of the person in the behavior detection video segment;

[0018] If the character's physical behavior includes placing the product into the character's clothes or carry-on luggage, it is determined that the corresponding character has abnormal behavior;

[0019] If the character's physical behavior includes acts that would destroy the product, it is determined that the corresponding character has abnormal behavior;

[0020] The head movement of the person in the behavior detection video segment is obtained. If the head movement of the person includes multiple looking movements, and the number of the looking movements exceeds a preset number threshold, it is determined that the corresponding person has abnormal behavior.

[0021] Further, if there is abnormal behavior, the corresponding behavior detection video segment is marked as a first abnormal video segment, and the person with abnormal behavior is marked as an abnormal person, the abnormal person is identified, and a feature text of the abnormal person is obtained and generated, the first abnormal video segment, the location information of the camera and the feature text of the abnormal person are associated, and saved in the data to be sent, and the following steps are also included:

[0022] Arrange the plurality of first abnormal video segments shot by the same camera in chronological order;

[0023] The first abnormal video segment ranked first is set as the current first abnormal video segment;

[0024] Obtaining the end time of the current first abnormal video segment as the first comparison time, and obtaining the start time of the next first abnormal video segment as the second comparison time;

[0025] If the difference between the first comparison time and the second comparison time is within a preset difference range, obtain the abnormal behavior person feature text associated with the current first abnormal video segment and compare it with the abnormal behavior person feature text of the subsequent first abnormal video segment. If they are the same, merge the subsequent first abnormal video segment into the current first abnormal video segment; otherwise, save the current first abnormal video segment and the location information of the associated camera, and the abnormal behavior person feature text into the data to be sent, and set the subsequent first abnormal video segment as the current first abnormal video segment.

[0026] Furthermore, the actor feature text includes the actor's facial features, hairstyle, hair color and clothing features.

[0027] Another embodiment of the present invention further provides an abnormal behavior detection reminder system, comprising: a video acquisition module, a person detection module, a behavior detection video segment acquisition module, an abnormal behavior analysis module, a data processing module and a data sending module;

[0028] The video acquisition module is used to acquire the video of the detection area through a camera;

[0029] The person detection module is used to extract video frames from the video according to a preset sampling period to perform person detection;

[0030] The behavior detection video segment acquisition module is used to acquire a video with the previous sampling cycle as the start time from the video as the behavior detection video segment if the presence of a person is detected from the current video frame;

[0031] The abnormal behavior analysis module is used to analyze the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior;

[0032] The data processing module is used to mark the corresponding behavior detection video segment as the first abnormal video segment if there is abnormal behavior, mark the person with abnormal behavior as the abnormal behavior person, perform feature recognition on the abnormal behavior person, obtain and generate the abnormal behavior person feature text, associate the first abnormal video segment, the location information of the camera and the abnormal behavior person feature text, and save them to the data to be sent;

[0033] The data sending module is used to send the data to be sent to the administrator.

[0034] Compared with the prior art, the abnormal behavior detection and reminder system of the present invention can utilize existing surveillance cameras in places such as supermarkets to obtain videos, and detect and analyze the people and their behaviors in the videos, which can reduce the detection cost and the cost of manual monitoring, and can also retain the video content when the abnormal behavior of the characters occurs.

[0035] Furthermore, the person detection module is used to perform the following steps:

[0036] Get the sampling cycle entered by the administrator;

[0037] Extracting a video frame from the video as the current video frame according to a sampling period;

[0038] Identify the outline of a static object in the current video frame. If the outline of the static object meets the preset human body motion outline, further identify the head features of the static object. If the head features of the static object are identified, determine that there is a person in the current video frame.

[0039] Further, the abnormal behavior analysis module is used to perform the following steps:

[0040] Acquire the body behavior of the person in the behavior detection video segment;

[0041] If the character's physical behavior includes placing the product into the character's clothes or carry-on luggage, it is determined that the corresponding character has abnormal behavior;

[0042] If the character's physical behavior includes acts that would destroy the product, it is determined that the corresponding character has abnormal behavior;

[0043] The head movement of the person in the behavior detection video segment is obtained. If the head movement of the person includes multiple looking movements, and the number of the looking movements exceeds a preset number threshold, it is determined that the corresponding person has abnormal behavior.

[0044] Furthermore, the data processing module is used to perform the following steps:

[0045] Arrange the plurality of first abnormal video segments shot by the same camera in chronological order;

[0046] The first abnormal video segment ranked first is set as the current first abnormal video segment;

[0047] Obtaining the end time of the current first abnormal video segment as the first comparison time, and obtaining the start time of the next first abnormal video segment as the second comparison time;

[0048] If the difference between the first comparison time and the second comparison time is within a preset difference range, obtain the abnormal behavior person feature text associated with the current first abnormal video segment and compare it with the abnormal behavior person feature text of the subsequent first abnormal video segment. If they are the same, merge the subsequent first abnormal video segment into the current first abnormal video segment; otherwise, save the current first abnormal video segment and the location information of the associated camera, and the abnormal behavior person feature text into the data to be sent, and set the subsequent first abnormal video segment as the current first abnormal video segment.

[0049] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the steps of the abnormal behavior detection and reminder method described above when executing the computer program.

[0050] In order to more clearly understand the present invention, the specific implementation of the present invention will be described below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flowchart of an abnormal behavior detection and reminder method according to an embodiment of the present invention;

[0052] Figure 2 This is a module connection diagram of an abnormal behavior detection and reminder system according to an embodiment of the present invention.

[0053] 1. Video acquisition module; 2. Person detection module; 3. Behavior detection video segment acquisition module; 4. Abnormal behavior analysis module; 5. Data processing module; 6. Data sending module. DETAILED DESCRIPTION

[0054] 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.

[0055] See also Figure 1 , which is a flow chart of an abnormal behavior detection and reminder method according to an embodiment of the present invention, comprising the following steps:

[0056] S1: Obtain the video of the detection area through the camera;

[0057] S2: extracting video frames from the video for person detection according to a preset sampling period;

[0058] Wherein, the step S2 comprises the following steps:

[0059] Get the sampling cycle entered by the administrator;

[0060] Extracting a video frame from the video as the current video frame according to a sampling period;

[0061] Identify the outline of a static object in the current video frame. If the outline of the static object meets the preset human action outline, further identify the head features of the static object. If the head features of the static object are identified, determine that there is a person in the current video frame. Search for head features based on the outline of the static object to improve the accuracy of person detection.

[0062] S3: If the presence of a person is detected from the current video frame, a video with the previous sampling cycle as the start time is obtained from the video as a behavior detection video segment;

[0063] S4: Analyze the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior;

[0064] Wherein, the step S4 comprises the following steps:

[0065] Acquire the body behavior of the person in the behavior detection video segment;

[0066] If the character's physical behavior includes placing the product into the character's clothes or carry-on luggage, it is determined that the corresponding character has abnormal behavior;

[0067] If the character's physical behavior includes acts that would destroy the product, it is determined that the corresponding character has abnormal behavior;

[0068] The head movement of the person in the behavior detection video segment is obtained. If the head movement of the person includes multiple looking movements, and the number of the looking movements exceeds a preset number threshold, it is determined that the corresponding person has abnormal behavior.

[0069] S5: If there is abnormal behavior, mark the corresponding behavior detection video segment as the first abnormal video segment, and mark the person with abnormal behavior as the abnormal person, perform feature recognition on the abnormal person, obtain and generate the abnormal person feature text, associate the first abnormal video segment, the camera location information and the abnormal person feature text, and save them to the data to be sent;

[0070] Preferably, the actor feature text includes the actor's facial features, hairstyle, hair color and clothing features.

[0071] S6: Send the data to be sent to the administrator.

[0072] Optionally, a warning light at the same position as the camera is also driven to start, which helps the administrator to quickly know the location and direction of the detection area.

[0073] Compared with the prior art, the abnormal behavior detection and reminder method of the present invention can utilize existing surveillance cameras in places such as supermarkets to obtain videos, and detect and analyze the people and their behaviors in the videos, which can reduce the detection cost and the cost of manual monitoring, and can also retain the video content when the abnormal behavior of the characters occurs.

[0074] In a feasible embodiment, if there is abnormal behavior, the corresponding behavior detection video segment is marked as a first abnormal video segment, and the person with abnormal behavior is marked as an abnormal person, the abnormal person is identified, and a feature text of the abnormal person is obtained and generated, the first abnormal video segment, the location information of the camera and the feature text of the abnormal person are associated, and saved in the data to be sent, and the following steps are also included:

[0075] Arrange the plurality of first abnormal video segments shot by the same camera in chronological order;

[0076] The first abnormal video segment ranked first is set as the current first abnormal video segment;

[0077] Obtaining the end time of the current first abnormal video segment as the first comparison time, and obtaining the start time of the next first abnormal video segment as the second comparison time;

[0078] If the difference between the first comparison time and the second comparison time is within a preset difference range, obtain the abnormal behavior person feature text associated with the current first abnormal video segment and compare it with the abnormal behavior person feature text of the subsequent first abnormal video segment. If they are the same, merge the subsequent first abnormal video segment into the current first abnormal video segment; otherwise, save the current first abnormal video segment and the location information of the associated camera, and the abnormal behavior person feature text into the data to be sent, and set the subsequent first abnormal video segment as the current first abnormal video segment.

[0079] In this embodiment, the associated first abnormal video segments may be merged, which helps the administrator to quickly obtain the entire process of the person performing the abnormal behavior.

[0080] Another embodiment of the present invention further provides an abnormal behavior detection reminder system, comprising: a video acquisition module 1, a person detection module 2, a behavior detection video segment acquisition module 3, an abnormal behavior analysis module 4, a data processing module 5 and a data sending module 6;

[0081] The video acquisition module 1 is used to acquire the video of the detection area through a camera;

[0082] The person detection module 2 is used to extract video frames from the video according to a preset sampling cycle to perform person detection;

[0083] The behavior detection video segment acquisition module 3 is used to acquire a video with the previous sampling cycle as the start time from the video as the behavior detection video segment if the presence of a person is detected from the current video frame;

[0084] The abnormal behavior analysis module 4 is used to analyze the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior;

[0085] The data processing module 5 is used to mark the corresponding behavior detection video segment as the first abnormal video segment if there is abnormal behavior, and mark the person with abnormal behavior as the abnormal behavior person, perform feature recognition on the abnormal behavior person, obtain and generate the abnormal behavior person feature text, associate the first abnormal video segment, the location information of the camera and the abnormal behavior person feature text, and save them to the data to be sent;

[0086] The data sending module 6 is used to send the data to be sent to the administrator.

[0087] Compared with the prior art, the abnormal behavior detection and reminder system of the present invention can utilize existing surveillance cameras in places such as supermarkets to obtain videos, and detect and analyze the people and their behaviors in the videos, which can reduce the detection cost and the cost of manual monitoring, and can also retain the video content when the abnormal behavior of the characters occurs.

[0088] In a feasible embodiment, the person detection module 2 is used to perform the following steps:

[0089] Get the sampling cycle entered by the administrator;

[0090] Extracting a video frame from the video as the current video frame according to a sampling period;

[0091] Identify the outline of a static object in the current video frame. If the outline of the static object meets the preset human action outline, further identify the head features of the static object. If the head features of the static object are identified, determine that there is a person in the current video frame. Search for head features based on the outline of the static object to improve the accuracy of person detection.

[0092] In a feasible embodiment, the abnormal behavior analysis module 4 is used to perform the following steps:

[0093] Acquire the body behavior of the person in the behavior detection video segment;

[0094] If the character's physical behavior includes placing the product into the character's clothes or carry-on luggage, it is determined that the corresponding character has abnormal behavior;

[0095] If the character's physical behavior includes acts that would destroy the product, it is determined that the corresponding character has abnormal behavior;

[0096] The head movement of the person in the behavior detection video segment is obtained. If the head movement of the person includes multiple looking movements, and the number of the looking movements exceeds a preset number threshold, it is determined that the corresponding person has abnormal behavior.

[0097] In a feasible embodiment, the data processing module 5 is used to perform the following steps:

[0098] Arrange the plurality of first abnormal video segments shot by the same camera in chronological order;

[0099] The first abnormal video segment ranked first is set as the current first abnormal video segment;

[0100] Obtaining the end time of the current first abnormal video segment as the first comparison time, and obtaining the start time of the next first abnormal video segment as the second comparison time;

[0101] If the difference between the first comparison time and the second comparison time is within a preset difference range, obtain the abnormal behavior person feature text associated with the current first abnormal video segment and compare it with the abnormal behavior person feature text of the subsequent first abnormal video segment. If they are the same, merge the subsequent first abnormal video segment into the current first abnormal video segment; otherwise, save the current first abnormal video segment and the location information of the associated camera, and the abnormal behavior person feature text into the data to be sent, and set the subsequent first abnormal video segment as the current first abnormal video segment.

[0102] In this embodiment, the associated first abnormal video segments may be merged, which helps the administrator to quickly obtain the entire process of the person performing the abnormal behavior.

[0103] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the steps of the abnormal behavior detection and reminder method described above when executing the computer program.

[0104] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting and reminding abnormal behavior, characterized in that: The following steps are involved: Obtain video of the detection area through the camera; Extract video frames from the video according to a preset sampling cycle to perform person detection; If the presence of a person is detected from the current video frame, a video with the previous sampling cycle as the start time is obtained from the video as the behavior detection video segment; Analyze the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior; If there is abnormal behavior, the corresponding behavior detection video segment is marked as the first abnormal video segment, and the person with abnormal behavior is marked as the abnormal person, and the abnormal person is identified, and the abnormal person feature text is obtained and generated, and the first abnormal video segment, the location information of the camera and the abnormal person feature text are associated, and saved to the data to be sent; Sending the data to be sent to an administrator; If there is abnormal behavior, the corresponding behavior detection video segment is marked as a first abnormal video segment, and the person with abnormal behavior is marked as an abnormal person, the abnormal person is identified, and a feature text of the abnormal person is obtained and generated, the first abnormal video segment, the location information of the camera and the feature text of the abnormal person are associated, and saved in the data to be sent, and the following steps are also included: Arrange the plurality of first abnormal video segments shot by the same camera in chronological order; The first abnormal video segment ranked first is set as the current first abnormal video segment; Obtaining the end time of the current first abnormal video segment as the first comparison time, and obtaining the start time of the next first abnormal video segment as the second comparison time; If the difference between the first comparison time and the second comparison time is within a preset difference range, obtain the abnormal behavior person feature text associated with the current first abnormal video segment and compare it with the abnormal behavior person feature text of the subsequent first abnormal video segment. If they are the same, merge the subsequent first abnormal video segment into the current first abnormal video segment; otherwise, save the current first abnormal video segment and the location information of the associated camera, and the abnormal behavior person feature text into the data to be sent, and set the subsequent first abnormal video segment as the current first abnormal video segment.

2. The abnormal behavior detection and reminder method according to claim 1, characterized in that: The method of extracting video frames from a video according to a preset sampling period for person detection includes the following steps: Get the sampling cycle entered by the administrator; Extracting a video frame from the video as the current video frame according to a sampling period; Identify the outline of a static object in the current video frame. If the outline of the static object meets the preset human body motion outline, further identify the head features of the static object. If the head features of the static object are identified, determine that there is a person in the current video frame.

3. The abnormal behavior detection and reminder method according to claim 1, characterized in that: The step of analyzing the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior includes the following steps: Acquire the body behavior of the person in the behavior detection video segment; If the character's physical behavior includes placing the product into the character's clothes or carry-on luggage, it is determined that the corresponding character has abnormal behavior; If the character's physical behavior includes acts that would destroy the product, it is determined that the corresponding character has abnormal behavior; The head movement of the person in the behavior detection video segment is obtained. If the head movement of the person includes multiple looking movements, and the number of the looking movements exceeds a preset number threshold, it is determined that the corresponding person has abnormal behavior.

4. The abnormal behavior detection and reminder method according to claim 1, characterized in that: The actor feature text includes the actor's facial features, hairstyle, hair color and clothing features.

5. An abnormal behavior detection and reminder system, characterized in that: include: Video acquisition module, person detection module, behavior detection video segment acquisition module, abnormal behavior analysis module, data processing module and data sending module; The video acquisition module is used to acquire the video of the detection area through a camera; The person detection module is used to extract video frames from the video according to a preset sampling period to perform person detection; The behavior detection video segment acquisition module is used to acquire a video with the previous sampling cycle as the start time from the video as the behavior detection video segment if the presence of a person is detected from the current video frame; The abnormal behavior analysis module is used to analyze the behavior detection video segment to determine whether the person in the detection video segment has abnormal behavior; The data processing module is used to mark the corresponding behavior detection video segment as the first abnormal video segment if there is abnormal behavior, and mark the person with abnormal behavior as the abnormal behavior person, perform feature recognition on the abnormal behavior person, obtain and generate the abnormal behavior person feature text, associate the first abnormal video segment, the camera's location information and the abnormal behavior person feature text, and save them to the data to be sent, including arranging a number of the first abnormal video segments taken by the same camera in chronological order; setting the first abnormal video segment arranged in the first position as the current first abnormal video segment; obtaining the end time of the current first abnormal video segment as as the first comparison time, obtaining the start time of the next first abnormal video segment as the second comparison time; if the difference between the first comparison time and the second comparison time is within a preset difference range, obtaining the abnormal behavior person feature text associated with the current first abnormal video segment and comparing it with the abnormal behavior person feature text of the next first abnormal video segment, if they are the same, merging the next first abnormal video segment into the current first abnormal video segment; otherwise, saving the current first abnormal video segment and the location information of the associated camera and the abnormal behavior person feature text into the data to be sent, and setting the next first abnormal video segment as the current first abnormal video segment; The data sending module is used to send the data to be sent to the administrator.

6. The abnormal behavior detection and reminder system according to claim 5, characterized in that: The person detection module is used to perform the following steps: Get the sampling cycle entered by the administrator; Extracting a video frame from the video as the current video frame according to a sampling period; Identify the outline of a static object in the current video frame. If the outline of the static object meets the preset human body motion outline, further identify the head features of the static object. If the head features of the static object are identified, determine that there is a person in the current video frame.

7. The abnormal behavior detection and reminder system according to claim 5, characterized in that: The abnormal behavior analysis module is used to perform the following steps: Acquire the body behavior of the person in the behavior detection video segment; If the character's physical behavior includes placing the product into the character's clothes or carry-on luggage, it is determined that the corresponding character has abnormal behavior; If the character's physical behavior includes acts that would destroy the product, it is determined that the corresponding character has abnormal behavior; The head movement of the person in the behavior detection video segment is obtained. If the head movement of the person includes multiple looking movements, and the number of the looking movements exceeds a preset number threshold, it is determined that the corresponding person has abnormal behavior.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

Images