Abnormal behavior detection method and device, medium and electronic equipment
Through key point detection and skeleton image analysis of human body images in video streams, combined with classification models, the use of smart devices is judged, and the problem of low detection accuracy in the existing technology is solved, and more accurate abnormal behavior recognition and security management are achieved. .
Patent Information
- Application Number
- CN202411131080.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-16
- Publication Date
- 2024-11-19
Smart Images

Figure CN118982871A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of image technology, and in particular to an abnormal behavior detection method, device, medium and electronic device. Background Art
[0002] Smart devices have become an indispensable part of people's lives in modern society, such as smartphones and tablets. However, being addicted to smartphones, especially spending a lot of time browsing social media and playing games, can seriously affect normal work and study. Especially in some specific work environments, playing with smart devices in a specific work environment may cause distraction and may even lead to serious safety issues.
[0003] In the related art, although there are some automated detections of smart devices, their detection methods are not accurate and are prone to erroneous judgments. Summary of the invention
[0004] In order to overcome the problems existing in the related art, the purpose of the present disclosure is to provide an abnormal behavior detection method, device, medium and electronic device.
[0005] According to a first aspect of an embodiment of the present disclosure, a method for detecting abnormal behavior is provided, the method comprising: Acquire a human body image in an original image in a video stream, where the human body image is used to refer to an object in the original image; Performing key point detection on the human body image to obtain positions of key points in the human body image; Determine a first skeleton image according to the positions of the key points, and determine a second skeleton image according to the positions of the key points and the human body image, wherein the second skeleton image carries texture information of the original image; determining a first prediction result according to the first skeleton image and the second skeleton image, wherein the first prediction result is used to characterize whether the posture of the object is in a posture of performing an abnormal behavior; When the subject is in a posture of performing the abnormal behavior, determining a relative position between a hand of the subject and the smart device, wherein the key point includes the hand; A second prediction result is determined according to the relative position, where the second prediction result is used to characterize whether the object performs the abnormal behavior with respect to the smart device.
[0006] Optionally, the second skeleton image is determined by: Converting the human body image into a grayscale image; According to the positions of the key points, the key points are drawn in the grayscale image to obtain a second skeleton image.
[0007] Optionally, determining a first prediction result according to the first skeleton image and the second skeleton image includes: The first skeleton image and the second skeleton image are input into a trained classification model to obtain a first prediction result output by the classification model.
[0008] Optionally, when the posture of the object is in a posture of performing the abnormal behavior, determining the relative position of the hand of the object and the smart device includes: When the posture of the object is in the posture of performing the abnormal behavior, performing target detection on the human body image to determine whether the smart device exists in the human body image; In a case where the smart device exists in the human body image, a relative position between the hand of the subject and the smart device is determined.
[0009] Optionally, determining the relative position between the hand of the subject and the smart device includes: Determining the location of the smart device according to the target detection result of the human body image; Determine a target circle with the position of the hand as the point and a preset distance as the radius, wherein the preset distance is the distance from the hand to the elbow; According to the position of the smart device, it is determined whether the smart device is within the target circle, so as to determine the relative position of the hand and the smart device, wherein the relative position indicates whether the smart device is within the target circle.
[0010] Optionally, determining a second prediction result according to the relative position includes: In a case where the relative position indicates that the smart device is within the target circle, determining that the object performs the abnormal behavior on the smart device; When the relative position indicates that the smart device is not within the target circle, it is determined that the object does not perform the abnormal behavior on the smart device.
[0011] Optionally, the hands include a left hand and a right hand.
[0012] According to a second aspect of an embodiment of the present disclosure, there is provided an abnormal behavior detection device, the device comprising: An acquisition module is configured to acquire a human body image in an original image in a video stream, wherein the human body image is used to refer to an object in the original image; A first detection module is configured to perform key point detection on the human body image to obtain positions of key points in the human body image; A first determination module is configured to determine a first skeleton image according to the positions of the key points, and to determine a second skeleton image according to the positions of the key points and the human body image, wherein the second skeleton image carries texture information of the original image; a first prediction module configured to determine a first prediction result according to the first skeleton image and the second skeleton image, wherein the first prediction result is used to characterize whether the posture of the object is in a posture of performing an abnormal behavior; A second determination module is configured to determine a relative position between a hand of the object and the smart device when the object is in a posture of performing the abnormal behavior, wherein the key point includes the hand; The second prediction module is configured to determine a second prediction result according to the relative position, where the second prediction result is used to characterize whether the object performs the abnormal behavior on the smart device.
[0013] According to a third aspect of an embodiment of the present disclosure, there is provided a non-temporary computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.
[0014] According to a fourth aspect of an embodiment of the present disclosure, there is provided an electronic device, including: a memory having a computer program stored thereon; A processor is used to execute the computer program in the memory to implement the steps of the method in the first aspect.
[0015] Through the above technical scheme, the first skeleton image and the second skeleton image carrying the texture information of the original image are determined based on the positions of the key points to predict the posture of the object, which can improve the prediction ability of the object's posture; in addition, when the posture of the object is in a posture of performing abnormal behavior, the relative position of the hand and the smart device is further determined to evaluate whether the smart device is near the hand based on the relative position, thereby excluding the situation where the smart device is only near the object's body but not held in hand, improving the object's judgment ability on whether the smart device performs abnormal behavior against the smart device, avoiding misjudgment of abnormal behavior, and providing strong technical support for the management and safety of public places.
[0016] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings: Figure 1The figure is a flow chart of an abnormal behavior detection method according to an exemplary embodiment.
[0018] Figure 2 The figure is a block diagram of an abnormal behavior detection device according to an exemplary embodiment.
[0019] Figure 3 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION
[0020] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.
[0021] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.
[0022] Video surveillance technology plays a vital role in many fields such as public safety, education, and transportation. However, in traditional surveillance systems, the identification of individual behavior mainly relies on manual observation and analysis, which is not only inefficient but also prone to errors and omissions due to human factors. Especially in surveillance scenarios, it may be difficult for manual observers to continuously focus on multiple surveillance images, which may lead to the neglect of important events. With the popularity of smart devices such as smartphones and people's increasing dependence on smartphones, the phenomenon of individuals being addicted to smartphones has become a common problem. Playing with mobile phones in a specific working environment may cause distraction, which may lead to serious safety problems. Therefore, there is an urgent need for an automated method to detect and identify the behavior of individuals playing with mobile phones in public places.
[0023] In the related art, although some automatic methods for detecting mobile phone playing have been proposed, most of these methods are based on the simultaneous detection of people and mobile phones, and then simply judge whether an individual is playing with a mobile phone based on the presence of the detected person and mobile phone. However, even if both the person and the mobile phone are present, there may be a situation where the person is not playing with the mobile phone. Therefore, this detection method is prone to the problem of false detection.
[0024] In view of this, the embodiments of the present disclosure provide an abnormal behavior detection method, device, medium and electronic device, which improve the ability to judge whether an object performs abnormal behavior against a smart device, avoid misjudgment of abnormal behavior, and provide strong technical support for the management and safety of public places.
[0025] The embodiments of the present disclosure are further explained and illustrated below in conjunction with the accompanying drawings.
[0026] Figure 1 is a flowchart of an abnormal behavior detection method according to an exemplary embodiment. Figure 1 As shown, the method includes: Step 110: Acquire a human body image in an original image in a video stream, where the human body image refers to an object in the original image.
[0027] It is worth noting that the object in this embodiment is a user.
[0028] The video stream is a real-time video stream captured by a high-definition camera in a surveillance scene. These high-definition cameras usually have high resolution and frame rate to ensure that the quality of the captured image meets the accuracy requirements of subsequent processing. The original image is each video frame in the captured video stream, and each video frame is analyzed to continuously monitor the dynamic changes in the surveillance scene.
[0029] Among them, the target detection algorithm can be used in the original image to determine the object in the original image, and the original image can be segmented based on the target detection result to obtain a human image. As an example, the target detection algorithm can be the YOLO (You Only Look Once) algorithm. The YOLO algorithm is a real-time target detection algorithm that converts the target detection task into a regression problem by dividing the original image into a number of grids, each of which is responsible for predicting a fixed number of bounding boxes and whether there are targets in these bounding boxes. The target here is the above-mentioned object. After the target is detected, the target is segmented from the original image to obtain a human image.
[0030] Step 120, performing key point detection on the human body image to obtain the positions of the key points in the human body image.
[0031] Among them, a key point detection algorithm can be used to obtain the position of the key point in the human body image. As an example, the key point detection algorithm can be an OpenPose algorithm, which is an advanced multi-task human key point detection algorithm based on a deep learning model for accurately identifying and locating 2D and 3D human joints. It should be understood that the position of the key point can be understood as the coordinate information of the key point in the human body image.
[0032] In this embodiment, the key points in the human body image may be key points of the upper body of the object.
[0033] Step 130, determining a first skeleton image according to the positions of the key points, and determining a second skeleton image according to the positions of the key points and the human body image, wherein the second skeleton image carries texture information of the original image.
[0034] It should be understood that the characteristics of the key point positions are relatively abstract, and feature extraction is difficult, and it is difficult to mark when constructing samples for training classification models. Therefore, the key points can be connected to form a skeleton image for easy labeling and feature learning.
[0035] For example, the key points can be marked with different colors according to their positions, thereby obtaining a first skeleton image. In addition, for the second skeleton image, the second skeleton image can be determined in the following manner: converting the human body image into a grayscale image; drawing the key points in the grayscale image according to their positions to obtain the second skeleton image, and drawing on the basis of the grayscale image can retain the texture information in the original image, and such texture information can reflect whether the object is holding a smart device.
[0036] Step 140 , determining a first prediction result based on the first skeleton image and the second skeleton image, where the first prediction result is used to characterize whether the posture of the object is in a posture of performing an abnormal behavior.
[0037] It is worth noting that abnormal behavior refers to the subject’s behavior of playing with smart devices.
[0038] The first skeleton image and the second skeleton image can be input into a trained classification model to obtain a first prediction result output by the classification model. As an example, the classification model can be a model based on a convolutional neural network.
[0039] Among them, the classification model can be a model associated with the object. Since different objects may have different postures when playing with smart devices, corresponding classification models can be trained for different objects. When training the classification model, sample images and corresponding posture labels can be obtained. The sample images include a first sample skeleton image and a second sample skeleton image. The posture label is used to reflect whether the sample posture of the sample object is in a posture of performing abnormal behavior. The explanation and description of the first sample skeleton image and the second sample skeleton image can refer to the first skeleton image and the second skeleton image, and this embodiment will not be repeated here.
[0040] Step 150 : when the posture of the object is in a posture of performing abnormal behavior, determine the relative position of the hand of the object and the smart device, and the key point includes the hand.
[0041] Step 160: Determine a second prediction result based on the relative position, where the second prediction result is used to characterize whether the object performs abnormal behavior on the smart device.
[0042] Among them, whether the smart device is located near the hand of the object can be evaluated according to the relative position of the hand of the object and the smart device.
[0043] Through the above method, the first skeleton image and the second skeleton image carrying the texture information of the original image are determined based on the positions of the key points to predict the posture of the object, which can improve the prediction ability of the object's posture; in addition, when the posture of the object is in a posture of performing abnormal behavior, the relative position of the hand and the smart device is further determined to evaluate whether the smart device is near the hand based on the relative position, thereby excluding the situation where the smart device is only near the object's body but not held in hand, improving the object's judgment ability on whether the smart device performs abnormal behavior against the smart device, avoiding misjudgment of abnormal behavior, and providing strong technical support for the management and safety of public places.
[0044] In some embodiments, when the posture of the object is in a posture of performing abnormal behavior, the step of determining the relative position of the object's hand and the smart device can be implemented in the following way: when the posture of the object is in a posture of performing abnormal behavior, target detection is performed on the human body image to determine whether there is a smart device in the human body image; when there is a smart device in the human body image, the relative position of the object's hand and the smart device is determined.
[0045] The target detection here can be implemented by using a target detection algorithm. The explanation and description of the target detection algorithm can refer to the above-mentioned related embodiments, and this embodiment will not be described in detail here.
[0046] Through the above method, when the posture of the object is in the posture of performing the abnormal behavior, it is further determined whether there is a smart device in the human body image, and when there is a smart device in the human body image, the relative position of the object's hand and the smart device is determined. In this way, the posture of the object and the position of the smart device are comprehensively considered to achieve more accurate and low false alarm rate detection of abnormal behavior of playing smart devices.
[0047] In some embodiments, the above-mentioned step of determining the relative position of the object's hand and the smart device can be implemented in the following manner: determine the position of the smart device based on the target detection result of the human body image; determine the target circle with the position of the hand as the dot and a preset distance as the radius, and the preset distance is the distance between the hand and the elbow; determine whether the smart device is within the target circle based on the position of the smart device to determine the relative position of the hand and the smart device, and the relative position indicates whether the smart device is within the target circle.
[0048] The hand includes the left hand and the right hand. In this case, the relative position includes the relative position of the left hand and the smart device, and the relative position of the right hand and the smart device.
[0049] Among them, the step of determining the second prediction result based on the relative position can be implemented in the following manner: when the relative position represents that the smart device is within the target circle, it is determined that the object performs abnormal behavior against the smart device; when the relative position represents that the smart device is not within the target circle, it is determined that the object does not perform abnormal behavior against the smart device.
[0050] It is worth noting that, from the above content, it can be seen that the relative position includes the relative position of the left hand and the smart device, and the relative position of the right hand and the smart device. Therefore, when any one of the two relative positions indicates that the smart device is within the target circle, it is determined that the subject has performed an abnormal behavior on the smart device. When all the relative positions of the two relative positions indicate that the smart device is not within the target circle, it is determined that the subject has not performed an abnormal behavior on the smart device.
[0051] Through the above method, the encircled target circle is used as the minimum area to judge whether the smart device is close to the hand. If the smart device is within this minimum area, it is judged that the subject is playing with the smart device. If the smart device is not within this minimum area, it is judged that the subject is not playing with the smart device. That is, through geometric analysis, the situation where the smart device is only near the body but not used is excluded, which effectively improves the subject's ability to judge whether it performs abnormal behavior against the smart device, avoids misjudgment of abnormal behavior, and provides strong technical support for the management and safety of public places.
[0052] Based on the same concept, Figure 2 is a block diagram of an abnormal behavior detection device according to an exemplary embodiment of the present disclosure. Figure 2 , the abnormal behavior detection device 200 comprises: An acquisition module 201 is configured to acquire a human body image in an original image in a video stream, where the human body image is used to refer to an object in the original image; The first detection module 202 is configured to perform key point detection on the human body image to obtain the positions of key points in the human body image; A first determination module 203 is configured to determine a first skeleton image according to the positions of the key points, and to determine a second skeleton image according to the positions of the key points and the human body image, wherein the second skeleton image carries texture information of the original image; A first prediction module 204 is configured to determine a first prediction result based on the first skeleton image and the second skeleton image, wherein the first prediction result is used to characterize whether the posture of the object is in a posture of performing an abnormal behavior; The second determination module 205 is configured to determine the relative position of the hand of the object and the smart device when the posture of the object is in the posture of performing the abnormal behavior, and the key point includes the hand; The second prediction module 206 is configured to determine a second prediction result according to the relative position, where the second prediction result is used to characterize whether the object performs the abnormal behavior on the smart device.
[0053] Optionally, the second skeleton image is determined by: Converting the human body image into a grayscale image; According to the positions of the key points, the key points are drawn in the grayscale image to obtain a second skeleton image.
[0054] Optionally, the first prediction module 204 is further configured to input the first skeleton image and the second skeleton image into a trained classification model to obtain a first prediction result output by the classification model.
[0055] Optionally, the second determining module 205 includes: A first determination submodule is configured to perform target detection on the human image when the posture of the object is in a posture of performing the abnormal behavior, so as to determine whether the smart device exists in the human image; The second determination submodule is configured to determine the relative position between the hand of the subject and the smart device when the smart device exists in the human body image.
[0056] Optionally, the second determination submodule is further configured to, when the smart device exists in the human image, determine the position of the smart device according to the target detection result of the human image; determine a target circle with the position of the hand as the dot and a preset distance as the radius, wherein the preset distance is the distance between the hand and the elbow; determine whether the smart device is within the target circle according to the position of the smart device, so as to determine the relative position of the hand and the smart device, wherein the relative position indicates whether the smart device is within the target circle.
[0057] Optionally, the second prediction module 206 is further configured to: In a case where the relative position indicates that the smart device is within the target circle, determining that the object performs the abnormal behavior on the smart device; When the relative position indicates that the smart device is not within the target circle, it is determined that the object does not perform the abnormal behavior on the smart device.
[0058] Optionally, the hands include a left hand and a right hand.
[0059] Regarding the abnormal behavior detection device 200 in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the abnormal behavior detection method, and will not be elaborated here. Based on the same concept, the embodiment of the present disclosure provides a non-temporary computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.
[0060] Based on the same concept, an embodiment of the present disclosure provides an electronic device, including: a memory having a computer program stored thereon; A processor is used to execute the computer program in the memory to implement the steps of the above method.
[0061] Figure 3 FIG. 1 is a block diagram of an electronic device according to an exemplary embodiment. Figure 3 As shown, the electronic device 300 may include: a processor 301 , a memory 302 . The electronic device 300 may also include one or more of a multimedia component 303 , an input / output (I / O) interface 304 , and a communication component 305 .
[0062] The processor 301 is used to control the overall operation of the electronic device 300 to complete all or part of the steps in the above-mentioned abnormal behavior detection method. The memory 302 is used to store various types of data to support the operation of the electronic device 300. For example, these data may include instructions for any application or method used to operate on the electronic device 300, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 303 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 302 or sent through the communication component 305. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 304 provides an interface between the processor 301 and other interface modules, and the other interface modules may be keyboards, mice, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 305 is used for wired or wireless communication between the electronic device 300 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, is not limited here. Therefore, the corresponding communication component 305 may include: Wi-Fi module, Bluetooth module, NFC module, etc.
[0063] In an exemplary embodiment, the electronic device 300 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned abnormal behavior detection method.
[0064] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned abnormal behavior detection method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 302 including program instructions, and the above-mentioned program instructions can be executed by the processor 301 of the electronic device 300 to complete the above-mentioned abnormal behavior detection method.
[0065] In another exemplary embodiment, a computer program product is further provided. The computer program product includes a computer program that can be executed by a programmable device. The computer program has a code portion for executing the above abnormal behavior detection method when executed by the programmable device.
[0066] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, a variety of simple modifications can be made to the technical solution of the present disclosure, and these simple modifications all fall within the protection scope of the present disclosure.
[0067] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.
[0068] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.
Claims
1. A method for detecting abnormal behavior, characterized in that: The method comprises: Acquire a human body image in an original image in a video stream, where the human body image is used to refer to an object in the original image; Performing key point detection on the human body image to obtain positions of key points in the human body image; Determine a first skeleton image according to the positions of the key points, and determine a second skeleton image according to the positions of the key points and the human body image, wherein the second skeleton image carries texture information of the original image; determining a first prediction result according to the first skeleton image and the second skeleton image, wherein the first prediction result is used to characterize whether the posture of the object is in a posture of performing an abnormal behavior; When the subject is in a posture of performing the abnormal behavior, determining a relative position between a hand of the subject and the smart device, wherein the key point includes the hand; A second prediction result is determined according to the relative position, where the second prediction result is used to characterize whether the object performs the abnormal behavior with respect to the smart device.
2. The method according to claim 1, characterized in that The second skeleton image is determined by: Converting the human body image into a grayscale image; According to the positions of the key points, the key points are drawn in the grayscale image to obtain a second skeleton image.
3. The method according to claim 1 or 2, characterized in that: The determining a first prediction result according to the first skeleton image and the second skeleton image includes: The first skeleton image and the second skeleton image are input into a trained classification model to obtain a first prediction result output by the classification model.
4. The method according to claim 1, characterized in that When the posture of the subject is in a posture of performing the abnormal behavior, determining the relative position of the hand of the subject and the smart device includes: When the posture of the object is in the posture of performing the abnormal behavior, performing target detection on the human body image to determine whether the smart device exists in the human body image; In a case where the smart device exists in the human body image, a relative position between the hand of the subject and the smart device is determined.
5. The method according to claim 4, characterized in that The determining the relative position between the hand of the subject and the smart device includes: Determining the location of the smart device according to the target detection result of the human body image; Determine a target circle with the position of the hand as the point and a preset distance as the radius, wherein the preset distance is the distance from the hand to the elbow; According to the position of the smart device, it is determined whether the smart device is within the target circle, so as to determine the relative position of the hand and the smart device, wherein the relative position indicates whether the smart device is within the target circle.
6. The method according to claim 5, characterized in that Determining a second prediction result according to the relative position includes: In a case where the relative position indicates that the smart device is within the target circle, determining that the object performs the abnormal behavior on the smart device; When the relative position indicates that the smart device is not within the target circle, it is determined that the object does not perform the abnormal behavior on the smart device.
7. The method according to claim 1, characterized in that The hands include a left hand and a right hand.
8. An abnormal behavior detection device, characterized in that: The device comprises: An acquisition module is configured to acquire a human body image in an original image in a video stream, wherein the human body image is used to refer to an object in the original image; A first detection module is configured to perform key point detection on the human body image to obtain positions of key points in the human body image; A first determination module is configured to determine a first skeleton image according to the positions of the key points, and to determine a second skeleton image according to the positions of the key points and the human body image, wherein the second skeleton image carries texture information of the original image; a first prediction module configured to determine a first prediction result according to the first skeleton image and the second skeleton image, wherein the first prediction result is used to characterize whether the posture of the object is in a posture of performing an abnormal behavior; A second determination module is configured to determine a relative position between a hand of the object and the smart device when the object is in a posture of performing the abnormal behavior, wherein the key point includes the hand; The second prediction module is configured to determine a second prediction result according to the relative position, where the second prediction result is used to characterize whether the object performs the abnormal behavior on the smart device.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
10. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 7.
