Method and device for determining target behavior, storage medium and electronic device

Abstract

The present invention discloses a method and device for determining a target behavior, a storage medium and an electronic device, wherein the method comprises: in the process of tracking the trajectory of a target object, performing bone key point detection on the first body frame of the target object, and obtaining first behavior pre-selection information when the hand and head of the target object are detected in the first body frame; obtaining the wrist detection frame and the head detection frame of the target object, and obtaining second behavior pre-selection information when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; determining whether the target object has performed a target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on the positional relationship between the hand and the chin of the target object. The above technical solution solves the problem that the traditional method cannot accurately identify the behavior of the teller holding the chin.

Description

Technical Field

[0001] The present invention relates to the field of communications, and in particular to a method and device for determining a target behavior, a storage medium and an electronic device. Background Art

[0002] Thanks to the in-depth research and widespread use of deep learning technology, artificial intelligence products have also developed rapidly. Whether in terms of heavy usage or intelligence, artificial intelligence products provide convenience for people's work and life in all aspects of the city.

[0003] In the financial industry, tellers need to provide face-to-face service to customers, and good service attitude and standardized work behavior can improve customer satisfaction and brand awareness. During the work process, bank tellers resting their chins for a long time, being in a daze, etc., are all manifestations of a negative work attitude for the bank. Through artificial intelligence solutions, real-time monitoring and warning of tellers' behavior can greatly remind tellers of their own work behavior and improve their service awareness.

[0004] In the related technology, illegal operations and behaviors can only be identified by recognizing the teller's voice and processing the captured images, but it is impossible to accurately identify the chin-lifting behavior.

[0005] Regarding related technologies, traditional methods cannot accurately identify the teller's chin-supporting behavior, and no effective solution has been proposed yet.

[0006] Therefore, it is necessary to improve the related technology to overcome the above-mentioned defects in the related technology. Summary of the invention

[0007] The embodiments of the present invention provide a method and device for determining a target behavior, a storage medium, and an electronic device, so as to at least solve the problem that the traditional method cannot accurately identify the teller's chin-supporting behavior.

[0008] According to one aspect of an embodiment of the present invention, a method for determining a target behavior is provided, comprising: in a process of tracking a trajectory of a target object, performing bone key point detection on a first body frame of the target object, and obtaining first behavior pre-selection information when the hand and head of the target object are detected in the first body frame; acquiring a wrist detection frame and a head detection frame of the target object, and obtaining second behavior pre-selection information when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; and determining whether the target object has performed a target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on the positional relationship between the hand and chin of the target object.

[0009] Further, determining whether the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information includes: obtaining a wrist key point corresponding to the first behavior pre-selection information and a wrist detection frame corresponding to the second behavior pre-selection information; determining whether the wrist key point is located in the wrist detection frame; if it is determined that the wrist key point is located in the wrist detection frame, determining that the target object has performed the target behavior; if it is determined that the wrist key point is not located in the wrist detection frame, determining that the target object has not performed the target behavior.

[0010] Furthermore, the method also includes: in a case where it is determined that the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, acquiring an image of the target object at a target moment, wherein the target moment is a moment when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; and inputting the image into a deep learning model to again determine whether the target object has performed the target behavior.

[0011] Further, determining that the hands and head of the target object are detected in the first body frame is determined in the following manner, including: acquiring wrist key points, elbow key points, and shoulder key points obtained in the process of performing bone key point detection on the first body frame; determining whether the wrist key point is located in the head-shoulder frame of the target object; when the wrist key point is located in the head-shoulder frame, acquiring an angle between a first vector formed by the elbow key point and the wrist key point and a second vector formed by the elbow key point and the shoulder key point; and determining whether the hands and head of the target object are detected in the first body frame according to a size relationship between the angle and a preset angle.

[0012] Further, based on the size relationship between the angle and the preset angle, it is determined that the hands and head of the target object are detected in the first body frame, including: when the angle is smaller than the preset angle, determining that the hands and head of the target object are detected in the first body frame; when the angle is greater than the preset angle, determining that the hands and head of the target object are not detected in the first body frame.

[0013] Further, when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture, obtaining a second behavior as pre-selected information includes: when the positions of the wrist detection frame and the head detection frame are less than a preset distance and the wrist detection frame and the head detection frame overlap, obtaining the second behavior as pre-selected information.

[0014] Furthermore, after determining whether the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, the method also includes: when it is determined that the duration for the target object to perform the target behavior is greater than a preset time threshold, issuing an alarm event for the target behavior.

[0015] According to another aspect of an embodiment of the present invention, a device for determining a target behavior is also provided, including: a detection module, used to perform bone key point detection on a first body frame of the target object during trajectory tracking of the target object, and obtain first behavior pre-selection information when the hand and head of the target object are detected in the first body frame; an acquisition module, used to acquire a wrist detection frame and a head detection frame of the target object, and obtain second behavior pre-selection information when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; a determination module, used to determine whether the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on the positional relationship between the hand and chin of the target object.

[0016] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining the target behavior when running.

[0017] According to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the method for determining the target behavior through the computer program.

[0018] Through the present invention, in the process of tracking the trajectory of the target object, the target object is detected for bone key points, and when the hand and head of the target object are detected, the first behavior pre-selection information of the target behavior is obtained, and the positional relationship between the wrist detection frame and the head detection frame is further detected to obtain the second behavior pre-selection information of the target behavior. According to the first behavior pre-selection information and the second behavior pre-selection information, it is determined whether the target object has performed the target behavior, wherein the target behavior includes a behavior determined based on the positional relationship between the hand and the chin of the target object. The above technical solution solves the problem that the traditional method cannot accurately identify the teller's chin-supporting behavior and other problems. Therefore, the teller's chin-supporting behavior can be accurately detected. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0020] Figure 1 It is a hardware structure block diagram of a computer terminal of a method for determining a target behavior according to an embodiment of the present invention;

[0021] Figure 2 is a flow chart of a method for determining a target behavior according to an embodiment of the present invention (I);

[0022] Figure 3 is a flow chart of a method for determining a target behavior according to an embodiment of the present invention (II);

[0023] Figure 4 4 is a structural block diagram of a device for determining a target behavior according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or a similar computing device. Taking running on a computer terminal as an example, Figure 1 1 is a hardware structure block diagram of a computer terminal of the method for determining a target behavior according to an embodiment of the present invention. Figure 1 As shown, the computer terminal may include one or more ( Figure 1Only one is shown in the figure) processor 102 (processor 102 may include but is not limited to a microprocessor (Microprocessor Unit, referred to as MPU) or a programmable logic device (Programmable logic device, referred to as PLD)) and a memory 104 for storing data. In an exemplary embodiment, the computer terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. Figure 1 More or fewer components as shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 A different configuration with more features is shown.

[0027] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining the target behavior in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0028] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of a computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0029] In this embodiment, a method for determining a target behavior is provided. Figure 2 1 is a flow chart of a method for determining a target behavior according to an embodiment of the present invention (I), the flow chart comprising the following steps:

[0030] Step S202, in the process of tracking the trajectory of the target object, performing bone key point detection on the first body frame of the target object, and obtaining first behavior pre-selected information when the hand and the head of the target object are detected in the first body frame;

[0031] It should be noted that the target object in this embodiment can be understood as an animal with a brain and arms, such as a human.

[0032] Step S204, obtaining a wrist detection frame and a head detection frame of the target object, and obtaining a second line as pre-selected information when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture;

[0033] Step S206, determining whether the target object has performed a target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on a positional relationship between a hand and a chin of the target object.

[0034] Through the above steps, in the process of tracking the trajectory of the target object, the target object is detected for bone key points, and when the hand and head of the target object are detected, the first behavior pre-selection information is obtained, and the positional relationship between the wrist detection frame and the head detection frame is further detected to obtain the second behavior pre-selection information. According to the first behavior pre-selection information and the second behavior pre-selection information, it is determined whether the target object has performed the target behavior, wherein the target behavior is used to indicate the positional relationship between the hand and the chin of the target object. The above technical solution solves the problem that the traditional method cannot accurately identify the teller's chin-supporting behavior and other problems. Therefore, the teller's chin-supporting behavior can be accurately detected.

[0035] It should be noted that the target behavior in the embodiment of the present invention includes a behavior determined based on the positional relationship between the hand and chin of the target object. Optionally, it can be the behavior of a person supporting the chin with his hands, specifically including supporting the chin with one hand and supporting the chin with both hands, or it can be the behavior of supporting the chin with the hands. The embodiment of the present invention is not limited to this.

[0036] Optionally, in order to better understand how to detect the hands and head of the target object in the first body frame in the above step S202, it can be implemented by the following technical solution: obtain wrist key points, elbow key points, and shoulder key points obtained in the process of performing bone key point detection on the first body frame; determine whether the wrist key point is located in the head and shoulder frame of the target object; when the wrist key point is located in the head and shoulder frame, obtain the angle between the first vector formed by the elbow key point and the wrist key point and the second vector formed by the elbow key point and the shoulder key point; determine whether the hands and head of the target object are detected in the first body frame according to the relationship between the angle and a preset angle.

[0037] That is to say, it is necessary to obtain the positions of the bone key points corresponding to the wrist, elbow, and shoulder in the first body frame, and then determine whether the wrist key point is in the detection head and shoulder frame. If so, it is necessary to determine whether the angle between the first vector formed by the elbow key point and the wrist key point and the second vector formed by the elbow key point and the shoulder key point meets the preset angle, and determine whether the hand and head of the target object are detected in the first body frame based on the relationship between the angle between the first vector and the second vector and the preset angle.

[0038] Specifically, if the angle formed by the first vector and the second vector is smaller than the preset angle, it is determined that the hands and head of the target object are detected in the first body frame; if the angle formed by the first vector and the second vector is greater than the preset angle, it is determined that the hands and head of the target object are not detected in the first body frame.

[0039] In order to better understand the above step S204, optionally, when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture, the second behavior is obtained as pre-selected information. Specifically, it is achieved in the following way: when the positions of the wrist detection frame and the head detection frame are less than a preset distance and the wrist detection frame and the head detection frame overlap, the second behavior is obtained as pre-selected information.

[0040] It should be noted that if the target object is ready to perform the target behavior, the distance between the target object's wrist and head gradually becomes smaller. That is to say, as long as the positions of the wrist detection frame and the head detection frame are detected to be less than the preset distance, and the wrist detection frame and the head detection frame overlap, it can be determined that the target object has a tendency to perform the target behavior. At this time, the second behavior is obtained as the pre-selected information.

[0041] Furthermore, the above-mentioned step S206 has multiple execution methods. In an optional embodiment, it is implemented by the following method: obtaining a wrist key point corresponding to the pre-selected information of the first behavior and a wrist detection frame corresponding to the pre-selected information of the second behavior; determining whether the wrist key point is located in the wrist detection frame; when it is determined that the wrist key point is located in the wrist detection frame, determining that the target object has executed the target behavior; when it is determined that the wrist key point is not located in the wrist detection frame, determining that the target object has not executed the target behavior.

[0042] That is to say, if the first behavior pre-selected information and the second behavior pre-selected information of the target object are obtained at the same time, it is necessary to judge whether the target object has performed the target behavior based on the first behavior pre-selected information and the second behavior pre-selected information. The first behavior pre-selected information contains the hand information of the target object, that is, the wrist key point, and the second behavior pre-selected information contains the relevant information corresponding to the wrist detection frame. If it is determined that the wrist key point is located in the wrist detection frame based on the first behavior pre-selected information and the second behavior pre-selected information, it can be determined that the target object has performed the target behavior, otherwise it has not been performed.

[0043] It should be noted that if in the above step S206, it is determined that the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, then an image of the target object at the target moment is obtained, wherein the target moment is the moment when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; and the image is input into the deep learning model to determine again whether the target object has performed the target behavior.

[0044] That is to say, there may be a certain error in judging whether the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, that is, there is a certain probability of judgment error. In order to more accurately determine whether the target object has performed the target behavior, an image of the target object can be obtained at the moment when the positional relationship between the wrist detection frame and the head detection frame meets the preset posture, and the image can be input into the deep learning model, and the deep learning model is used to recognize the image. If the deep learning model determines that the target object has performed the target behavior, it is finally determined that the target object has performed the target behavior.

[0045] Furthermore, if the above step S206 determines that the target object has performed the target behavior, it is also necessary to determine whether the duration for the target object to perform the target behavior is greater than a preset time threshold. If it is determined that the duration for the target object to perform the target behavior is greater than the preset time threshold, an alarm event for the target behavior is issued to alert the target object.

[0046] Obviously, the above-described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. In order to better understand the above-mentioned method for determining the target behavior, the above-mentioned process is described below in conjunction with the embodiments, but it is not intended to limit the technical solution of the embodiments of the present invention, specifically:

[0047] In an alternative embodiment, Figure 3 Flowchart (II) of a method for determining a target behavior according to an embodiment of the present invention, wherein the specific steps are as follows:

[0048] Step S302: inputting an image of a teller (equivalent to the target object in the above embodiment);

[0049] Step S304: performing target detection on the human body;

[0050] Step S306: tracking the human body, and then executing steps S308 and S312 respectively;

[0051] Step S308: Detect the wrist and palm of the human body, and then execute step S310;

[0052] Step S310: Determine whether the secondary pre-alarm logic is satisfied. If satisfied, execute step S316; if not satisfied, execute step S324;

[0053] Step S312: perform human joint point detection, and then execute step S314;

[0054] Step S314: determine whether the pre-alarm logic is satisfied. If satisfied, execute step S316; if not satisfied, execute step S324;

[0055] Step S316: judging whether the final pre-alarm logic is satisfied according to the results of step S310 and step S314, and if satisfied, executing step S318, and if not satisfied, executing step S324;

[0056] Step S318: input the image into a model (equivalent to the deep learning model in the above embodiment) for discrimination;

[0057] Step S320: Determine whether the preset time threshold is met, if so, execute step S320, if not, execute step S324;

[0058] Step S320: Alarm the target behavior;

[0059] Step S324: Output.

[0060] To better understand the above process, in an optional embodiment, the target behavior may be determined by the following steps, specifically:

[0061] Step 1: Setting a behavior time threshold (equivalent to the preset time threshold in the above embodiment) for determining whether the chin-lifting behavior (equivalent to the target behavior in the above embodiment) does not meet the requirements;

[0062] Step 2: Use the deep learning target detection algorithm to detect target frames such as the human body, head and shoulders, and head in the image, and form the teller tracking trajectory through multi-target association tracking.

[0063] Step 3: Perform bone key point detection based on the target body frame to obtain human limb information;

[0064] Step 4: Based on the acquired bone key points, the hand-head relationship information is obtained, thereby completing the pre-selected information of the hand-supporting-chin behavior (equivalent to the pre-selected information of the first behavior in the above embodiment);

[0065] Step 5: Detect wrist based on the target body frame;

[0066] Step 6: Based on the acquired wrist detection frame, the secondary chin-lifting behavior pre-selection information (equivalent to the second behavior pre-selection information in the above embodiment) is obtained through the relationship between the wrist and the target head detection frame;

[0067] Step 7: The final chin-lifting behavior pre-selection information is formed by comprehensively processing the pre-selected information in steps 4 and 6;

[0068] Step 8: Use deep learning recognition technology to finally confirm whether the chin-lifting behavior is true for the result of step 6;

[0069] Step 9: If the duration of the result of step 7 is greater than the preset time threshold, an alarm for chin-lifting violation behavior is finally given;

[0070] It should be noted that in step 4, the main steps are as follows:

[0071] Step (1): Obtain the positions of the corresponding bone key points of the wrist, elbow, and shoulder; determine whether the wrist key point is within the detection target head and shoulder frame. If so, execute step (2); otherwise, return directly;

[0072] Step (2): Determine the angle between the vector formed by the elbow to the wrist (equivalent to the first vector in the above embodiment) and the vector formed by the elbow to the shoulder (equivalent to the second vector in the above embodiment); if the angle between the two vectors is less than a threshold value (equivalent to the preset angle in the above embodiment), it is considered to be the chin-lifting preselected information; otherwise, return directly;

[0073] It should be noted that in step 6, the main steps are as follows:

[0074] Step (1): Obtain the target wrist detection frame and head detection frame;

[0075] Step (2): Determine the positional relationship between the wrist detection frame and the head detection frame. If the two detection frames are close to each other and overlap, it is considered to be the secondary pre-selection information of chin-lifting. Otherwise, it is returned directly.

[0076] It should be noted that in step 7, the main steps are as follows:

[0077] Step (1): Obtain the first chin-lift pre-selection information in step 4 and the second chin-lift pre-selection information in step 6; if both pre-selection information exist, execute step (2); otherwise, return directly;

[0078] Step (2): Determine whether the wrist key point in the pre-selected information corresponding to step 4 is within the wrist detection frame of the pre-selected information corresponding to step 6. If so, output the final chin-lifting pre-alarm information, otherwise return directly;

[0079] It should be noted that in step 8, the main steps are as follows:

[0080] Step (1): obtaining a large number of pre-selected result images of step 7 and manually annotating them; training a corresponding recognition model (equivalent to the deep learning model in the above embodiment);

[0081] Step (2): Input the result of step 7 into the recognition model, and finally output the result of the recognition model.

[0082] In addition, the above technical solution of the embodiment of the present invention integrates multiple technologies such as human target detection, joint point detection, trajectory judgment, wrist detection, target pattern recognition and judgment to determine whether the teller has made an illegal action of supporting the chin. In the process of fusion of each key point, the position relationship and angle relationship of the detection results are used to complete the alarm pre-selection, and finally the alarm is filtered through the recognition module to reduce the possibility of false alarms. At the same time, time domain accumulation is used to exclude false alarms caused by detection errors of the recognition model in a certain state.

[0083] It should be noted that, in this embodiment, human joints are used as pre-selection for judging the chin-lifting action, which can make the action judgment accuracy high; the integration of wrist detection and joint point technology solves the problem of poor detection effect of joint point model in some occasions; recognition model recognition is used as the final confirmation of the alarm, which enhances the accuracy of action judgment; and time domain accumulation can ensure the accuracy of the alarm.

[0084] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention.

[0085] In the present embodiment, a device for determining a target behavior is also provided, and the device is used to implement the above-mentioned embodiment and the preferred implementation mode, and the description thereof will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the equipment described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware is also possible and conceived.

[0086] Figure 4 is a structural block diagram of a device for determining a target behavior according to an embodiment of the present invention, the device comprising:

[0087] A detection module 42 is used to perform bone key point detection on a first body frame of the target object during trajectory tracking of the target object, and obtain first behavior pre-selected information when a hand and a head of the target object are detected in the first body frame;

[0088] An acquisition module 44 is used to acquire a wrist detection frame and a head detection frame of the target object, and obtain a second line as pre-selected information when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture;

[0089] The determination module 46 is used to determine whether the target object has performed a target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on the positional relationship between the hand and chin of the target object.

[0090] Through the above module, in the process of tracking the trajectory of the target object, the target object is detected for bone key points, and when the hand and head of the target object are detected, the first behavior pre-selection information is obtained, and the positional relationship between the wrist detection frame and the head detection frame is further detected to obtain the second behavior pre-selection information of the target behavior. According to the first behavior pre-selection information and the second behavior pre-selection information, it is determined whether the target object has performed the target behavior, wherein the target behavior is used to indicate the positional relationship between the hand and the chin of the target object. The above technical solution solves the problem that the traditional method cannot accurately identify the teller's chin-supporting behavior and other problems. Therefore, the teller's chin-supporting behavior can be accurately detected.

[0091] It should be noted that the target behavior in the embodiment of the present invention includes a behavior determined based on the positional relationship between the hand and chin of the target object. Optionally, it can be the behavior of a person supporting the chin with his hands, specifically including supporting the chin with one hand and supporting the chin with both hands, or it can be the behavior of supporting the chin with the hands. The embodiment of the present invention is not limited to this.

[0092] Optionally, the determination module 46 is further used to obtain a wrist key point corresponding to the pre-selected information of the first behavior and a wrist detection frame corresponding to the pre-selected information of the second behavior; determine whether the wrist key point is located in the wrist detection frame; if it is determined that the wrist key point is located in the wrist detection frame, determine that the target object has performed the target behavior; if it is determined that the wrist key point is not located in the wrist detection frame, determine that the target object has not performed the target behavior.

[0093] Optionally, the determination module 46 is also used to obtain an image of the target object at a target moment when it is determined that the target object has performed the target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target moment is the moment when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; and input the image into the deep learning model to determine again whether the target object has performed the target behavior.

[0094] Optionally, the detection module 42 is further used to obtain wrist key points, elbow key points, and shoulder key points obtained in the process of performing bone key point detection on the first body frame; determine whether the wrist key point is located in the head and shoulder frame of the target object; when the wrist key point is located in the head and shoulder frame, obtain the angle between the first vector formed by the elbow key point and the wrist key point and the second vector formed by the elbow key point and the shoulder key point; and determine whether the hand and head of the target object are detected in the first body frame based on the relationship between the angle and a preset angle.

[0095] Optionally, the detection module 42 is also used to determine that the hands and head of the target object are detected in the first body frame when the angle is smaller than the preset angle; and to determine that the hands and head of the target object are not detected in the first body frame when the angle is greater than the preset angle.

[0096] Optionally, the acquisition module 44 is further configured to obtain the second behavior as pre-selected information when the positions of the wrist detection frame and the head detection frame are less than a preset distance and the wrist detection frame and the head detection frame overlap.

[0097] Optionally, the determination module 46 is further configured to issue an alarm event of the target behavior when it is determined that the duration for which the target object performs the target behavior is greater than a preset time threshold.

[0098] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0099] S1, in a process of tracking a trajectory of a target object, performing bone key point detection on a first body frame of the target object, and obtaining first behavior pre-selected information when a hand and a head of the target object are detected in the first body frame;

[0100] S2, obtaining a wrist detection frame and a head detection frame of the target object, and obtaining a second line as pre-selected information when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture;

[0101] S3. Determine whether the target object has performed a target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on a positional relationship between a hand and a chin of the target object.

[0102] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0103] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0104] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0105] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0106] S1, in the process of tracking the trajectory of the target object, performing bone key point detection on a first body frame of the target object, and obtaining first behavior pre-selected information of the target behavior when a hand and a head of the target object are detected in the first body frame;

[0107] S2, obtaining a wrist detection frame and a head detection frame of the target object, and obtaining pre-selected information of a second behavior of the target behavior when the positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture;

[0108] S3. Determine whether the target object has performed a target behavior based on the first behavior pre-selection information and the second behavior pre-selection information, wherein the target behavior includes a behavior determined based on a positional relationship between a hand and a chin of the target object.

[0109] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0110] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0111] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0112] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for determining a target behavior, characterized in that: include: In the process of tracking the trajectory of the target object, bone key point detection is performed on a first body frame of the target object, and when a hand and a head of the target object are detected in the first body frame, pre-selected information of a first behavior is obtained; Acquire a wrist detection frame and a head detection frame of the target object, and obtain a second line as pre-selected information when a positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; Acquire the wrist key point corresponding to the pre-selected information in the first behavior and the wrist detection frame corresponding to the pre-selected information in the second behavior; Determining whether the wrist key point is located in the wrist detection frame; In a case where it is determined that the wrist key point is located in the wrist detection frame, determining that the target object has performed the target behavior; When it is determined that the wrist key point is not located in the wrist detection frame, it is determined that the target object does not perform the target behavior, wherein the target behavior includes a behavior determined based on a positional relationship between a hand and a chin of the target object.

2. The method for determining a target behavior according to claim 1, characterized in that: The method further comprises: When it is determined that the target object has performed the target behavior according to the first behavior pre-selection information and the second behavior pre-selection information, acquiring an image of the target object at a target time, wherein the target time is a time when the positional relationship between the wrist detection frame and the head detection frame meets a preset posture; The image is input into a deep learning model to again determine whether the target object has performed the target behavior.

3. The method for determining a target behavior according to claim 1, characterized in that: Determining that the hand and the head of the target object are detected in the first body frame in the following manner includes: Acquire wrist key points, elbow key points, and shoulder key points obtained in the process of performing bone key point detection on the first body frame; Determining whether the wrist key point is located within the head and shoulder frame of the target object; When the wrist key point is located in the head-shoulder frame, obtaining an angle between a first vector formed by the elbow key point and the wrist key point and a second vector formed by the elbow key point and the shoulder key point; According to the magnitude relationship between the angle and a preset angle, it is determined whether the hand and the head of the target object are detected in the first body frame.

4. The method for determining a target behavior according to claim 3, characterized in that: Determining, according to a magnitude relationship between the angle and a preset angle, that a hand and a head of the target object are detected in the first body frame includes: When the angle is smaller than the preset angle, determining that the hand and the head of the target object are detected in the first body frame; When the angle is greater than the preset angle, it is determined that the hand and the head of the target object are not detected in the first body frame.

5. The method for determining a target behavior according to claim 1, characterized in that: When the positional relationship between the wrist detection frame and the head detection frame conforms to the preset posture, obtaining the second line as pre-selected information includes: When the positions of the wrist detection frame and the head detection frame are smaller than a preset distance and the wrist detection frame and the head detection frame overlap, the second line is obtained as pre-selected information.

6. The method for determining a target behavior according to claim 1, characterized in that: After determining whether the target object has performed a target behavior according to the first behavior pre-selection information and the second behavior pre-selection information, the method further includes: When it is determined that the duration of the target object performing the target behavior is greater than a preset time threshold, an alarm event of the target behavior is issued.

7. A device for determining a target behavior, characterized in that: include: a detection module, configured to perform bone key point detection on a first body frame of the target object during trajectory tracking of the target object, and obtain first behavior pre-selected information when a hand and a head of the target object are detected in the first body frame; an acquisition module, configured to acquire a wrist detection frame and a head detection frame of the target object, and obtain a second line as pre-selected information when a positional relationship between the wrist detection frame and the head detection frame conforms to a preset posture; A determination module, configured to obtain a wrist key point corresponding to the pre-selected information of the first behavior and a wrist detection frame corresponding to the pre-selected information of the second behavior; Determining whether the wrist key point is located in the wrist detection frame; In a case where it is determined that the wrist key point is located in the wrist detection frame, determining that the target object has performed the target behavior; When it is determined that the wrist key point is not located in the wrist detection frame, it is determined that the target object does not perform the target behavior, wherein the target behavior includes a behavior determined based on a positional relationship between a hand and a chin of the target object.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 6 when executed.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

Images