Abnormal behavior detection method and device, computer equipment, storage medium and product

Abstract

The invention relates to the technical field of deep learning, in particular to an abnormal behavior detection method and device, computer equipment, a storage medium and a product. The method comprises the steps of obtaining a video frame image corresponding to a detection area; performing object feature recognition on the video frame image to obtain a target feature matrix corresponding to the video frame image; and performing abnormal behavior detection on the detection area according to the target feature matrix to obtain an abnormal behavior existing in the detection area. The abnormal behavior detection is performed on the detection area through the target feature matrix, the abnormal behavior detection is accurately performed on the to-be-detected object in the detection area, the accuracy of the abnormal behavior detection is improved, and the safety of teacher archive management is ensured.

Description

Technical Field

[0001] This application relates to the field of deep learning technology, and in particular to an abnormal behavior detection method, apparatus, computer equipment, storage medium, and product. Background Technology

[0002] In the context of intelligent secure rooms for teacher file management, real-time monitoring and abnormal behavior detection are crucial components for ensuring security and data privacy.

[0003] However, existing technologies for intelligent confidentiality rooms for teacher files rely too heavily on predefined behavior patterns when detecting abnormal behavior, resulting in low accuracy and an inability to effectively detect abnormal behavior in confidentiality rooms. Summary of the Invention

[0004] Therefore, it is necessary to provide an abnormal behavior detection method, device, computer equipment, storage medium, and product that can accurately detect abnormal behavior in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a method for detecting abnormal behavior. The method includes:

[0006] Obtain the video frame image corresponding to the detection area;

[0007] Perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image;

[0008] Abnormal behavior is detected in the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area.

[0009] In one embodiment, the step of detecting abnormal behavior in the detection region based on the target feature matrix to obtain the abnormal behavior existing in the detection region includes:

[0010] Perform state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix;

[0011] Based on the hidden state, abnormal behavior existing in the detection area is determined.

[0012] In one embodiment, the step of performing state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix includes:

[0013] The latent features that generate the target feature matrix;

[0014] Based on the latent features, state analysis is performed to obtain the hidden states corresponding to the target feature matrix.

[0015] In one embodiment, the step of performing object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image includes:

[0016] Perform object recognition on the video frame image to obtain the initial feature matrix corresponding to the video frame image;

[0017] The initial feature matrix is ​​convolved to obtain the target feature matrix.

[0018] In one embodiment, the convolution process performed on the initial feature matrix to obtain the target feature matrix includes:

[0019] The initial feature matrix is ​​used to construct an image to obtain a feature image; wherein the nodes contained in the feature image are the objects to be detected contained in the video frame image, and the edges contained in the feature image represent the spatial relationships between different objects;

[0020] The feature image is processed by a graph convolutional network to obtain the target feature matrix.

[0021] In one embodiment, the initial feature matrix includes at least one of location features, velocity features, category features, and environmental features.

[0022] Secondly, this application also provides an abnormal behavior detection device. The device includes:

[0023] The acquisition module is used to acquire video frame images corresponding to the detection area;

[0024] The recognition module is used to perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image;

[0025] The detection module is used to detect abnormal behavior in the detection area based on the target feature matrix, and to obtain the abnormal behavior existing in the detection area.

[0026] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the following steps:

[0027] Obtain the video frame image corresponding to the detection area;

[0028] Perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image;

[0029] Abnormal behavior is detected in the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area.

[0030] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps:

[0031] Obtain the video frame image corresponding to the detection area;

[0032] Perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image;

[0033] Abnormal behavior is detected in the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area.

[0034] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps:

[0035] Obtain the video frame image corresponding to the detection area;

[0036] Perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image;

[0037] Abnormal behavior is detected in the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area.

[0038] The aforementioned abnormal behavior detection method, apparatus, computer equipment, storage medium, and product acquire video frame images corresponding to the detection area; perform object feature recognition on the video frame images to obtain a target feature matrix corresponding to the video frame images; and then perform abnormal behavior detection on the detection area based on the target feature matrix to obtain the abnormal behaviors existing in the detection area. As can be seen from the above, this application performs object feature recognition on the video frame images beforehand during the abnormal behavior detection process, thereby acquiring the object features within the detection area. This ensures that the target feature matrix can effectively reflect the behavioral state of the object to be detected within the detection area. Furthermore, by performing abnormal behavior detection on the detection area using the target feature matrix, accurate detection of abnormal behavior of the object to be detected within the detection area is achieved, improving the accuracy of abnormal behavior detection and ensuring the security of teacher file management. Attached Figure Description

[0039] Figure 1 An application environment diagram of an abnormal behavior detection method provided in this application embodiment;

[0040] Figure 2 A flowchart illustrating the first abnormal behavior detection method provided in this application embodiment;

[0041] Figure 3 A flowchart illustrating the second abnormal behavior detection method provided in this application embodiment;

[0042] Figure 4 A flowchart illustrating the third abnormal behavior detection method provided in this application embodiment;

[0043] Figure 5 A flowchart illustrating the fourth abnormal behavior detection method provided in this application embodiment;

[0044] Figure 6 A structural block diagram of an abnormal behavior detection device provided in an embodiment of this application;

[0045] Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0047] The abnormal behavior detection method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on a cloud or other network server. The process involves acquiring video frame images corresponding to the detection area; performing object feature recognition on the video frame images to obtain the target feature matrix corresponding to the video frame images; and then detecting abnormal behavior in the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0048] In one embodiment, such as Figure 2 As shown, an abnormal behavior detection method is provided, which can be applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0049] S201, Obtain the video frame image corresponding to the detection area.

[0050] It should be noted that when it is necessary to obtain video frame images corresponding to the detection area, a video acquisition device can be deployed in advance within the detection area to achieve the purpose of obtaining video frame images corresponding to the detection area.

[0051] S202, Perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image.

[0052] It should be noted that when it is necessary to perform object feature recognition on video frame images to obtain the target feature matrix corresponding to the video frame images, the following can be included: performing object recognition on video frame images to obtain the initial feature matrix corresponding to the video frame images; performing convolution processing on the initial feature matrix to obtain the target feature matrix.

[0053] The initial feature matrix includes at least one of the following: location features, velocity features, category features, and environmental features.

[0054] Specifically, positional features include the center coordinates of the object within the video frame, as well as the object's width and height. Velocity features include the object's velocity change between the current and previous frames. Category features characterize the object's category information, such as "person," "bag," or "electronic device," and are represented using a one-hot encoded vector. Environmental features include ambient information such as light intensity and camera angle.

[0055] In one embodiment of this application, it is assumed that a maximum of 10 objects are detected in each video frame, and the feature dimensions of each object are: position features (4-dimensional) + velocity features (2-dimensional) + category features (N-dimensional) + environment features (M-dimensional), with a total feature dimension of [missing information]. .

[0056] S203, perform abnormal behavior detection on the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area.

[0057] It should be noted that when it is necessary to detect abnormal behavior in the detection area based on the target feature matrix and obtain the abnormal behavior existing in the detection area, the following can be included: perform state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix; and determine the abnormal behavior existing in the detection area based on the hidden state.

[0058] The aforementioned abnormal behavior detection method acquires video frame images corresponding to the detection area; performs object feature recognition on the video frame images to obtain a target feature matrix corresponding to the video frame images; and then performs abnormal behavior detection on the detection area based on the target feature matrix to obtain the abnormal behaviors existing in the detection area. As can be seen from the above, this application performs object feature recognition on the video frame images beforehand during the abnormal behavior detection process, thereby acquiring the object features within the detection area. This ensures that the target feature matrix can effectively reflect the behavioral state of the object to be detected within the detection area. Furthermore, by using the target feature matrix to perform abnormal behavior detection on the detection area, accurate detection of abnormal behavior of the object to be detected within the detection area is achieved, improving the accuracy of abnormal behavior detection and ensuring the security of teacher file management.

[0059] In one embodiment, such as Figure 3 As shown, when it is necessary to detect abnormal behavior in a detection region based on the target feature matrix and obtain the abnormal behavior existing in the detection region, the following can be included:

[0060] S301, perform state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix.

[0061] It should be noted that when it is necessary to perform state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix, the following can be included: generating the latent features of the target feature matrix; performing state analysis based on the latent features to obtain the hidden state corresponding to the target feature matrix.

[0062] In one embodiment of this application, a variational autoencoder (VAE) can be introduced and integrated into an LSTM network, enabling the model to generate richer feature representations. Then, the latent features of the target feature matrix are generated using the following calculation formula:

[0063] ;

[0064] in, This refers to the latent features of the target feature matrix; This refers to the mean of VAE; This refers to the standard deviation of VAE; and Used for parameterizing latent variables .

[0065] Furthermore, the latent features generated by VAE By incorporating this into an LSTM network, sequence-to-sequence learning is performed, enabling state analysis based on latent features to obtain the hidden states corresponding to the target feature matrix.

[0066] Specifically, ;

[0067] in, This refers to the hidden state corresponding to the target feature matrix. This refers to the mean of VAE.

[0068] S302, Based on the hidden state, determine the abnormal behavior present in the detection area.

[0069] In one embodiment of this application, the behavior at each time point is classified based on the hidden state corresponding to the target feature matrix. An alarm is triggered when a specific abnormal behavior (such as "abnormal stay" or "illegal items") is detected.

[0070] The aforementioned abnormal behavior detection method determines the hidden states corresponding to the target feature matrix, and then identifies abnormal behaviors within the detection area based on these hidden states. This achieves accurate detection of abnormal behaviors in the target object within the detection area, improving the accuracy of abnormal behavior detection and ensuring the security of teacher file management.

[0071] In one embodiment, such as Figure 4 As shown, when it is necessary to perform object feature recognition on video frame images to obtain the target feature matrix corresponding to the video frame images, the following can be included:

[0072] S401, Perform object recognition on the video frame image to obtain the initial feature matrix corresponding to the video frame image.

[0073] The initial feature matrix includes at least one of the following: location features, velocity features, category features, and environmental features.

[0074] S402, perform convolution on the initial feature matrix to obtain the target feature matrix.

[0075] When it is necessary to perform convolution processing on the initial feature matrix to obtain the target feature matrix, the following can be included: constructing an image from the initial feature matrix to obtain a feature image; wherein the nodes contained in the feature image are the objects to be detected contained in the video frame image, and the edges contained in the feature image represent the spatial relationships between different objects; and performing graph convolution processing on the feature image using a graph convolution network to obtain the target feature matrix.

[0076] In one embodiment of this application, object recognition processing of video frame images can be achieved through the GCN core algorithm model.

[0077] The GCN algorithm model design is shown below:

[0078] ;

[0079] It is time The graph convolution outputs a feature matrix, which represents the new feature representation of each node after convolution. It is a node In time Features It is a node The neighbors. It is a node The set of neighboring nodes. It is a normalization constant, usually and The number of common neighbors. It is the weight matrix of the graph convolutional layer. It is a bias term.

[0080] By applying Gaussian Networks (GCNs) to the initial feature matrix of a video frame image, complex spatial relationships between all objects in the detection region can be captured. For example, multiple nodes represent proximity relationships in a set of adjacency matrices. Through GCNs, this adjacency information is integrated into a high-level feature representation, thereby improving the accuracy of abnormal behavior recognition.

[0081] The aforementioned abnormal behavior detection method obtains an initial feature matrix corresponding to a video frame image by performing object recognition on the video frame image; then, it performs convolution processing on the initial feature matrix to obtain the target feature matrix. This achieves accurate detection of abnormal behavior of the object to be detected within the detection area, improves the accuracy of abnormal behavior detection, and ensures the security of teacher file management.

[0082] As an example: In a smart secure room environment, assuming four wide-angle ultra-high-definition cameras (30 frames per second) are used, each camera can detect up to 10 objects. We set a monitoring cycle of 10 seconds (i.e., 300 video frames) to capture abnormal behaviors that may occur during teacher file management. The monitoring equipment produces 300 video frames, each represented by a 10x13 feature matrix. The data generation process is as follows: First step, normal behavior data generation: In the first 200 frames, the normal patrol behavior of the teacher in the file room is simulated. The speed and position characteristics of all objects change smoothly, and the lighting and camera angles are stable.

[0083] Sampling data:

[0084] Object 1 (person): ((x,y)=(120,300),(w,h)=(60,150),(v_x,v_y)=(0.2,0.1),Category=[1,0,0],L=0.8,A=45).

[0085] Object 2 (package): ((x,y)=(500,200),(w,h)=(30,30),(v_x,v_y)=(0,0),class=[0,1,0],L=0.8,A=45).

[0086] The second step is the generation of abnormal behavior data: In the last 100 frames, two types of abnormal behavior are simulated:

[0087] Abnormal Behavior 1 (Excessive Personnel Stay Time): Person 1 stays in a certain area (such as in front of a filing cabinet) for more than the prescribed time. This is manifested by a decrease in position coordinate changes and a velocity approaching 0.

[0088] Sampling data:

[0089] Object 1 (person): ((x,y)=(130,310),(w,h)=(60,150),(v_x,v_y)=(0.01,0.02),Category=[1,0,0],L=0.75,A=45).

[0090] Abnormal behavior 2 (carrying unauthorized items): A new object (electronic device) appears in the filing cabinet area.

[0091] Sampling data:

[0092] Object 3 (electronic device): ((x,y)=(150,320),(w,h)=(10,20),(v_x,v_y)=(0,0),Category=[0,0,1],L=0.75,A=45).

[0093] In the algorithm model validation process, we input the above simulation data into an LSTM model combining GCN and VAE, and perform the following steps:

[0094] Step 1, GCN model application:

[0095] The initial feature matrix of each video frame image is used to construct a feature image, where nodes represent detected objects and edges represent spatial relationships between objects (such as physical distance and category similarity).

[0096] The second step is to integrate VAE with LSTM applications:

[0097] For a time series of length 10 First, latent feature representations are generated using a VAE model:

[0098] ;

[0099] latent features generated by VAE Integrating it into an LSTM network for sequence-to-sequence learning:

[0100] ;

[0101] Step 3, Abnormal Behavior Detection:

[0102] The behavior at each time point is classified using the hidden state output by the LSTM. An alarm is triggered when a specific behavior (such as "abnormal stay" or "illegal item") is detected.

[0103] Step 4: Experimental verification and result analysis

[0104] Experimental verification results:

[0105] Recognition accuracy: Tested on 200 frames of normal behavior data and 100 frames of abnormal behavior data, the improved model (GCN+VAE+LSTM) achieved a recognition accuracy of 95%, which is significantly higher than the baseline model (CNN+LSTM) of 83%.

[0106] Sensitivity of abnormal behavior detection: The improved model has a sensitivity of 92% in detecting abnormal behavior (excessive dwell time and carrying prohibited items), while the sensitivity of the baseline model is 75%.

[0107] False alarm rate: The improved model has a false alarm rate of 7%, which is 5 percentage points lower than the baseline model's false alarm rate of 12%.

[0108] In one embodiment, such as Figure 5 As shown, when it is necessary to detect abnormal behavior in the detection area, the following can be included:

[0109] S501, acquire the video frame image corresponding to the detection area.

[0110] S502, perform object recognition on the video frame image to obtain the initial feature matrix corresponding to the video frame image.

[0111] S503, construct an image from the initial feature matrix to obtain a feature image; wherein, the nodes contained in the feature image are the objects to be detected contained in the video frame image, and the edges contained in the feature image represent the spatial relationships between different objects.

[0112] S504 uses a graph convolutional network to perform graph convolution processing on the feature image to obtain the target feature matrix.

[0113] S505, generates latent features of the target feature matrix.

[0114] S506, perform state analysis based on latent features to obtain the hidden state corresponding to the target feature matrix.

[0115] S507, based on the hidden state, determine the abnormal behavior present in the detection area.

[0116] The aforementioned abnormal behavior detection method acquires video frame images corresponding to the detection area; performs object feature recognition on the video frame images to obtain a target feature matrix corresponding to the video frame images; and then performs abnormal behavior detection on the detection area based on the target feature matrix to obtain the abnormal behaviors existing in the detection area. As can be seen from the above, this application performs object feature recognition on the video frame images beforehand during the abnormal behavior detection process, thereby acquiring the object features within the detection area. This ensures that the target feature matrix can effectively reflect the behavioral state of the object to be detected within the detection area. Furthermore, by using the target feature matrix to perform abnormal behavior detection on the detection area, accurate detection of abnormal behavior of the object to be detected within the detection area is achieved, improving the accuracy of abnormal behavior detection and ensuring the security of teacher file management.

[0117] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0118] Based on the same inventive concept, this application also provides an abnormal behavior detection device for implementing the abnormal behavior detection method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the abnormal behavior detection device provided below can be found in the limitations of the abnormal behavior detection method described above, and will not be repeated here.

[0119] In one embodiment, such as Figure 6 As shown, an abnormal behavior detection device is provided, including: an acquisition module 10, an identification module 20, and a detection module 30, wherein:

[0120] The acquisition module 10 is used to acquire the video frame image corresponding to the detection area.

[0121] The recognition module 20 is used to perform object feature recognition on video frame images to obtain the target feature matrix corresponding to the video frame images.

[0122] The detection module 30 is used to detect abnormal behavior in the detection area based on the target feature matrix, and to obtain the abnormal behavior existing in the detection area.

[0123] In one embodiment, state analysis is performed on the target feature matrix to obtain the hidden state corresponding to the target feature matrix;

[0124] Based on the hidden state, identify abnormal behaviors within the detection area.

[0125] In one embodiment, latent features of the target feature matrix are generated;

[0126] State analysis is performed based on latent features to obtain the hidden states corresponding to the target feature matrix.

[0127] In one embodiment, object recognition is performed on the video frame image to obtain an initial feature matrix corresponding to the video frame image;

[0128] The initial feature matrix is ​​convolved to obtain the target feature matrix.

[0129] In one embodiment, an image is constructed from the initial feature matrix to obtain a feature image; wherein the nodes contained in the feature image are the objects to be detected contained in the video frame image, and the edges contained in the feature image represent the spatial relationships between different objects;

[0130] A graph convolutional network is used to perform graph convolution processing on the feature image to obtain the target feature matrix.

[0131] In one embodiment, the initial feature matrix includes at least one of location features, velocity features, category features, and environmental features.

[0132] The aforementioned abnormal behavior detection method acquires video frame images corresponding to the detection area; performs object feature recognition on the video frame images to obtain a target feature matrix corresponding to the video frame images; and then performs abnormal behavior detection on the detection area based on the target feature matrix to obtain the abnormal behaviors existing in the detection area. As can be seen from the above, this application performs object feature recognition on the video frame images beforehand during the abnormal behavior detection process, thereby acquiring the object features within the detection area. This ensures that the target feature matrix can effectively reflect the behavioral state of the object to be detected within the detection area. Furthermore, by using the target feature matrix to perform abnormal behavior detection on the detection area, accurate detection of abnormal behavior of the object to be detected within the detection area is achieved, improving the accuracy of abnormal behavior detection and ensuring the security of teacher file management.

[0133] Each module in the aforementioned abnormal behavior detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0134] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements an abnormal behavior detection method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0135] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0136] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0137] Obtain the video frame image corresponding to the detection area;

[0138] Object feature recognition is performed on video frame images to obtain the target feature matrix corresponding to the video frame images;

[0139] Abnormal behavior is detected in the detection area based on the target feature matrix, and the abnormal behavior existing in the detection area is obtained.

[0140] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0141] State analysis is performed on the target feature matrix to obtain the hidden state corresponding to the target feature matrix;

[0142] Based on the hidden state, identify abnormal behaviors within the detection area.

[0143] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0144] Latent features for generating the target feature matrix;

[0145] State analysis is performed based on latent features to obtain the hidden states corresponding to the target feature matrix.

[0146] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0147] Object recognition is performed on video frame images to obtain the initial feature matrix corresponding to the video frame images;

[0148] The initial feature matrix is ​​convolved to obtain the target feature matrix.

[0149] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0150] The initial feature matrix is ​​used to construct an image to obtain a feature image; the nodes in the feature image are the objects to be detected contained in the video frame image, and the edges in the feature image represent the spatial relationships between different objects.

[0151] A graph convolutional network is used to perform graph convolution processing on the feature image to obtain the target feature matrix.

[0152] In one embodiment, the processor, when executing a computer program, also performs the following steps:

[0153] The initial feature matrix includes at least one of the following: location features, velocity features, category features, and environmental features.

[0154] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0155] Obtain the video frame image corresponding to the detection area;

[0156] Object feature recognition is performed on video frame images to obtain the target feature matrix corresponding to the video frame images;

[0157] Abnormal behavior is detected in the detection area based on the target feature matrix, and the abnormal behavior existing in the detection area is obtained.

[0158] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0159] State analysis is performed on the target feature matrix to obtain the hidden state corresponding to the target feature matrix;

[0160] Based on the hidden state, identify abnormal behaviors within the detection area.

[0161] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0162] Latent features for generating the target feature matrix;

[0163] State analysis is performed based on latent features to obtain the hidden states corresponding to the target feature matrix.

[0164] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0165] Object recognition is performed on video frame images to obtain the initial feature matrix corresponding to the video frame images;

[0166] The initial feature matrix is ​​convolved to obtain the target feature matrix.

[0167] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0168] The initial feature matrix is ​​used to construct an image to obtain a feature image; the nodes in the feature image are the objects to be detected contained in the video frame image, and the edges in the feature image represent the spatial relationships between different objects.

[0169] A graph convolutional network is used to perform graph convolution processing on the feature image to obtain the target feature matrix.

[0170] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0171] The initial feature matrix includes at least one of the following: location features, velocity features, category features, and environmental features.

[0172] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0173] Obtain the video frame image corresponding to the detection area;

[0174] Object feature recognition is performed on video frame images to obtain the target feature matrix corresponding to the video frame images;

[0175] Abnormal behavior is detected in the detection area based on the target feature matrix, and the abnormal behavior existing in the detection area is obtained.

[0176] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0177] State analysis is performed on the target feature matrix to obtain the hidden state corresponding to the target feature matrix;

[0178] Based on the hidden state, identify abnormal behaviors within the detection area.

[0179] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0180] Latent features for generating the target feature matrix;

[0181] State analysis is performed based on latent features to obtain the hidden states corresponding to the target feature matrix.

[0182] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0183] Object recognition is performed on video frame images to obtain the initial feature matrix corresponding to the video frame images;

[0184] The initial feature matrix is ​​convolved to obtain the target feature matrix.

[0185] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0186] The initial feature matrix is ​​used to construct an image to obtain a feature image; the nodes in the feature image are the objects to be detected contained in the video frame image, and the edges in the feature image represent the spatial relationships between different objects.

[0187] A graph convolutional network is used to perform graph convolution processing on the feature image to obtain the target feature matrix.

[0188] In one embodiment, when the computer program is executed by a processor, it also performs the following steps:

[0189] The initial feature matrix includes at least one of the following: location features, velocity features, category features, and environmental features.

[0190] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0191] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0192] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0193] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting abnormal behavior, characterized in that, The method includes: Obtain the video frame image corresponding to the detection area; Perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image; Abnormal behavior is detected in the detection area based on the target feature matrix to obtain the abnormal behavior existing in the detection area.

2. The method according to claim 1, characterized in that, The step of detecting abnormal behavior in the detection region based on the target feature matrix to obtain the abnormal behavior existing in the detection region includes: Perform state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix; Based on the hidden state, abnormal behavior existing in the detection area is determined.

3. The method according to claim 2, characterized in that, The step of performing state analysis on the target feature matrix to obtain the hidden state corresponding to the target feature matrix includes: The latent features that generate the target feature matrix; Based on the latent features, state analysis is performed to obtain the hidden states corresponding to the target feature matrix.

4. The method according to any one of claims 1-3, characterized in that, The step of performing object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image includes: Perform object recognition on the video frame image to obtain the initial feature matrix corresponding to the video frame image; The initial feature matrix is ​​convolved to obtain the target feature matrix.

5. The method according to claim 4, characterized in that, The step of performing convolution processing on the initial feature matrix to obtain the target feature matrix includes: The initial feature matrix is ​​used to construct an image to obtain a feature image; wherein the nodes contained in the feature image are the objects to be detected contained in the video frame image, and the edges contained in the feature image represent the spatial relationships between different objects; The feature image is processed by a graph convolutional network to obtain the target feature matrix.

6. The method according to claim 4, characterized in that, The initial feature matrix includes at least one of the following: location features, velocity features, category features, and environmental features.

7. An abnormal behavior detection device, characterized in that, The device includes: The acquisition module is used to acquire video frame images corresponding to the detection area; The recognition module is used to perform object feature recognition on the video frame image to obtain the target feature matrix corresponding to the video frame image; The detection module is used to detect abnormal behavior in the detection area based on the target feature matrix, and to obtain the abnormal behavior existing in the detection area.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images