Non-standard lockage behavior monitoring method and device, equipment and storage medium

Abstract

The invention relates to the technical field of intelligent monitoring, and discloses a non-standard lockage behavior monitoring method, device and equipment and a storage medium, and the method comprises the steps: carrying out the detection of an image frame, and obtaining the human body key point data of a lockage person; acquiring human body posture data according to the human body key point data; and non-standard lockage behavior judgment is carried out on lockage personnel according to the human body posture data. The invention discloses a non-standard lockage behavior monitoring technology based on human body key point analysis, which can more accurately identify a non-standard lockage behavior and reduce misjudgment and missing detection by analyzing a motion track and a behavior mode of a human body key point. Under the condition of dense people flow, effective information can still be extracted, the accuracy and reliability of non-standard lockage behavior detection are improved, and an innovative solution is provided for passing and operation management of public places.

Description

Technical Field

[0001] The present application relates to the field of intelligent monitoring technology, and in particular to a monitoring method, device, equipment and storage medium for non-standard gate-passing behavior. Background Art

[0002] With the rapid development of urban services such as rail transit and office buildings, setting up gates in public places to allow authorized personnel or verified personnel to pass has become an important means of effectively managing the safety of public areas. However, irregular gate entry and exit behaviors, such as subway fare evasion, illegal gate entry, illegal gate entry, etc., are common, causing significant economic losses and management challenges to site management operators. Traditional irregular gate entry behaviors such as fare evasion mainly rely on video surveillance motion recognition technology. Although it can monitor fare evasion to a certain extent, it is prone to misjudgment and missed detection during peak hours when there are dense crowds, and accuracy cannot be guaranteed. Summary of the invention

[0003] The main purpose of the present application is to provide a monitoring method, device, equipment and storage medium for non-standard gate-passing behavior, aiming to solve the technical problem of insufficient accuracy in monitoring non-standard gate-passing behavior in the prior art.

[0004] The first aspect of the present application provides a monitoring method for non-standard gate-passing behavior, and the monitoring method for non-standard gate-passing behavior includes: Detecting the image frames to obtain the key point data of the human body of the person passing through the gate, wherein the image frames are extracted from the video stream of the gate area of ​​the target site, and the key point data of the human body includes the coordinates of each key point of the human body; According to the key point data of the human body, obtain the human body posture data; The non-standard gate-crossing behavior of people passing through the gate is judged based on the height information and limb angle information in the human posture data.

[0005] The second aspect of the present application further provides a monitoring device for non-standard gate-passing behavior, the monitoring device for non-standard gate-passing behavior comprising: The target detection module is used to detect the image frame and obtain the key point data of the human body of the person passing through the gate, wherein the image frame is extracted from the video stream of the gate area of ​​the target site, and the key point data of the human body includes the coordinates of each key point of the human body; A posture data acquisition module is used to acquire human posture data based on human key point data; The behavior judgment module is used to judge the non-standard gate-crossing behavior of people passing through the gate based on the height information and limb angle information in the human posture data.

[0006] The third aspect of the present application provides a computer device, comprising: a memory and at least one processor, wherein instructions are stored in the memory; and at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned monitoring method for non-standard gate-passing behavior.

[0007] A fourth aspect of the present application provides a computer-readable storage medium, in which instructions are stored, and when the computer-readable storage medium is run on a computer, the computer executes the above-mentioned method for monitoring non-standard gate-passing behavior.

[0008] This application discloses a non-standard gate-passing behavior monitoring technology based on human key point analysis. By analyzing the motion trajectory and behavior pattern of human key points, it can more accurately identify non-standard gate-passing behaviors and reduce misjudgments and missed detections. In the case of dense crowds, the solution of this application can still extract effective information, improve the accuracy and reliability of non-standard gate-passing behavior detection, and provide an innovative solution for the passage and operation management of public places, which has important application value and market prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a flow chart of a first embodiment of a method for monitoring non-standard gate-passing behavior in an embodiment of the present application; Figure 2 This is a schematic diagram of key point monitoring in a dense crowd; Figure 3 This is a schematic diagram of key points of the human body in the embodiment of this application; Figure 4 A schematic diagram of the behavior of jumping over the gate from above in an embodiment of the present application; Figure 5 A schematic diagram of the behavior of crossing the gate obliquely from above in an embodiment of the present application; Figure 6 A schematic diagram of the behavior of passing through the gate from below in an embodiment of the present application; Figure 7 This is a functional module diagram of an embodiment of a monitoring device for non-standard gate-passing behavior in an embodiment of the present application; Figure 8 It is a schematic diagram of an embodiment of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application 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 described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "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.

[0011] With the rapid development of urban rail transit, subways have become an important means of daily travel for people. However, the increase in subway fare evasion has brought significant economic losses and management challenges to operators. Although traditional fare evasion detection technology can monitor fare evasion to a certain extent, it is prone to misjudgment and missed detection during peak hours when there are dense crowds, reducing the accuracy of the system. In addition, existing technologies lack in-depth analysis of fare evaders' behavior and cannot provide effective fare evasion path and behavior feature information, further affecting management and tracking effects.

[0012] Based on this, the present application provides a monitoring solution for non-standard gate-crossing behavior.

[0013] refer to Figure 1 In one embodiment of the present application, a monitoring method for non-standard gate-passing behavior is provided, and the monitoring method for non-standard gate-passing behavior includes: S100: Detecting the image frame to obtain the human body key point data of the person passing through the gate, wherein the image frame is extracted from the video stream of the gate area of ​​the target site, and the human body key point data includes the coordinates of each human body key point.

[0014] Specifically, the monitoring method of non-standard gate-passing behavior in this embodiment can be applied to scenarios such as the detection of fare evasion by rail transit passengers (for example, fare evasion detection in public areas such as subways and train stations), the detection of non-standard gate-passing behavior by people outside shopping malls or office buildings, the detection of illegal entry and exit of gates, and the monitoring of illegal gate passage.

[0015] The monitoring method for non-standard gate-crossing behavior in this embodiment is a behavior monitoring method based on key point analysis, which can monitor both the people leaving the gate and the people entering the gate.

[0016] Obtain real-time video stream data of the gate area of ​​the target site; the target site can be a rail transit area such as a subway, railway station, high-speed railway station, etc., or it can be a shopping mall, office building, etc. where gates are installed as access control areas. This application does not impose any restrictions on this.

[0017] By extracting frames from the video stream data, image frames of the gate area of ​​the target site can be obtained, wherein one image frame can be extracted every m image frames, or image frames can be extracted frame by frame, which is not limited in the present application.

[0018] The human body key point detection algorithm or the human body posture estimation algorithm can be used to extract the human body key point data of the person passing through the gate from the image frame.

[0019] More specifically, image frames are extracted from the video stream data, and the obtained image frames are input into the trained target detection model; the trained target detection model is used to extract the human key point data of the people passing through the gate in the image frame, wherein the human key point data includes the coordinates of each human key point; the human key point is the key bone point, for example, one or more parts or bone points such as ankles, knees, hips, shoulders, elbows, wrists, ears, eyes, head, nose, etc., and this application does not limit this. People passing through the gate can be people leaving the station, people entering the station, passengers, staff, etc., and this application does not limit this.

[0020] Among them, the target detection model can be a YOLO series model such as YOLO-Pose model, YOLO V3, YOLO V5, etc., and this application does not limit this.

[0021] The YOLO-POSE model is an algorithm for real-time human pose estimation. It can run on a variety of computer platforms, including mobile devices and desktop computers. The basic principle of the YOLO-POSE framework is to use deep neural networks to achieve human pose estimation. It uses a special neural network structure called YOLOv3-tiny. This structure can quickly calculate the position of all human key points in the image, thereby achieving pose estimation.

[0022] The YOLO-POSE model is characterized by its high efficiency and real-time performance. By optimizing the model structure and calculation process, YOLO-POSE can achieve fast human pose estimation with lower hardware configuration. In addition, the YOLO-POSE model also has good scalability and can be easily integrated with other computer vision algorithms and models to achieve more complex functions and applications.

[0023] In one specific embodiment, the video stream is used as input for system analysis to ensure that people passing through the gate can be monitored in real time. Subsequently, the system loads a trained YOLO-Pose deep learning model that has the ability to identify the location information of each key point.

[0024] During the processing, the system analyzes the input video stream frame by frame, and each frame is sent to the YOLO-Pose model for processing to extract the key point information of the person. YOLO-Pose can identify 17 key points including the head, shoulders, arms, torso, legs and ankles, and record the coordinates of each key point. The system will then perform posture estimation based on the detected key points to fully understand the person's movements. This process is achieved by calculating the coordinates of the key points to ensure the accuracy of action recognition.

[0025] S200: Acquire human body posture data according to human body key point data.

[0026] Specifically, in order to facilitate subsequent analysis, the system can connect the detected key points according to the coordinates of the key points of the human body, such as Figure 3 As shown in the figure, a skeleton diagram or posture diagram is formed. The intuitive display of recognition results helps analysts better understand the data and provides reliable support for the identification of non-standard gate-passing behaviors such as fare evasion. Finally, the system outputs the detected key point information and its corresponding coordinates to ensure the consistency and accuracy of the data, providing a basis for the subsequent non-standard gate-passing behavior identification module such as fare evasion.

[0027] The trajectories of key bone points of human body movement can be identified based on the skeleton diagram, and the human body posture data can be calculated based on the human skeleton diagram and the coordinates of the key points of the human body.

[0028] Human body posture data may include, for example, height information of certain key points relative to a baseline or reference point, angle information between certain parts of the human body, or one or more of the above data.

[0029] or, The skeleton graph or posture graph is used as human body posture data.

[0030] S300: judging the non-standard gate-crossing behavior of the person passing through the gate according to the height information and body angle information in the human body posture data.

[0031] Specifically, Figure 2 This is a schematic diagram of key point monitoring in a dense crowd; Figure 3 This is a schematic diagram of the key points of the human body in an embodiment of the present application; the key points of the human body include but are not limited to: nose, left eye, right eye, left ear, right ear, left shoulder, right shoulder, left elbow, right elbow, left wrist, right wrist, left hip, right hip (crotch), left knee, right knee, left ankle (left ankle joint), right ankle (right ankle joint), etc. One or more thereof.

[0032] pass Figure 2 and Figure 3It can be seen that when people enter and exit the gate normally, they usually walk upright or occasionally bend their upper body and lean, etc., and usually have certain behavioral norms. In addition, through the key point monitoring of the human body, multiple people can be detected in crowded conditions.

[0033] After completing the key point detection in this embodiment, the system can identify non-standard gate-passing behaviors such as fare evasion, which can mainly include entering and exiting the gate from above and entering and exiting the gate from below.

[0034] More specifically, in the case of illegal entry and exit of the gate, the behavior of the gate passers-by will be different from the normal behavior norms. For example, jumping over the gate, crossing the gate with one foot diagonally, crawling through the gate from the bottom, etc. are all non-standard gate-crossing behaviors and need to be monitored and identified.

[0035] This embodiment can accurately determine whether the person passing through the gate has non-standard gate-passing behavior based on the difference in human body posture data between non-standard gate-passing and normal gate-passing.

[0036] If the human body posture data is a skeleton graph or a posture graph, the skeleton graph or the posture graph is input into the trained behavior recognition model for behavior recognition to obtain a behavior recognition result, which can specifically indicate that the gate-passing person has non-standard gate-passing behavior, or that the gate-passing person has no non-standard gate-passing behavior. In addition, the behavior recognition result can also specifically indicate which non-standard gate-passing behavior it is, for example, it can be one of jumping over the gate, crossing the gate with one foot diagonally, crawling from the bottom of the gate, etc.

[0037] If the human body posture data includes one or more data such as the horizontal height of certain key points relative to a certain baseline, the angle between certain parts of the human body, etc., it can be judged based on these calculated heights or angles whether the person entering the gate has non-standard gate-crossing behavior and what kind of non-standard gate-crossing behavior it is.

[0038] The solution of this embodiment can be used to detect behaviors such as evading tickets from above gates or jumping from above gates in rail transit areas such as subways and train stations, and can also be used to detect illegal behaviors such as non-office personnel in office buildings illegally jumping from above gates to enter office areas. The solution of this embodiment can be applied to various application scenarios, and this application does not limit this.

[0039] The non-standard gate-passing behavior monitoring solution of this embodiment performs real-time monitoring of gate-passing personnel based on key points of the human body, and accurately judges ticket evasion behavior by analyzing the relative positions and angles between key points. This accuracy is reflected in the algorithm's sensitivity to movement posture, allowing the system to maintain stable recognition performance in different environments and reduce misjudgments caused by external factors.

[0040] This embodiment discloses a non-standard gate-passing behavior monitoring technology based on human key point analysis. By analyzing the motion trajectory and behavior pattern of human key points, it can more accurately identify non-standard gate-passing behaviors and reduce misjudgments and missed detections. In the case of dense crowds, the solution of this embodiment can still extract effective information, improve the accuracy and reliability of non-standard gate-passing behavior detection, and provide an innovative solution for the passage and operation management of public places, which has important application value and market prospects.

[0041] In one embodiment, the key points of the human body include ankles and knees. In step S300, the non-standard gate-crossing behavior of the gate-crossing person is judged according to the height information and limb angle information in the human posture data, including: In response to the leg key points of the human body posture data being higher than the top horizontal height of the gate, it is determined that the person passing through the gate has non-standard gate-passing behavior; wherein the leg key points include at least one of the ankle joint and the knee.

[0042] Specifically, the top horizontal height of the gate in the image frame in the same coordinate system is obtained.

[0043] If the key points of the legs include the ankle joints, the human body posture data includes the horizontal height of the ankle joints. If the horizontal height of the ankle joints is higher than the horizontal height of the top of the gate, it is determined that the person passing through the gate has jumped over the gate.

[0044] In a specific embodiment, if the ankle joints of both feet are higher than the top horizontal height of the gate, it is determined that the person passing through the gate has jumped over the gate.

[0045] If the key points of the human body include the knees, the human body posture data includes the horizontal height of the knees. If the horizontal height of the knees is higher than the horizontal height of the top of the gate, it is determined that the person passing through the gate has jumped over the gate.

[0046] In a specific embodiment, if the knees of both legs are higher than the top level of the gate, it is determined that the person passing through the gate has jumped over the gate.

[0047] In addition, if the person crossing the gate crosses the gate with one leg diagonally, one leg of the person crossing the gate is on the ground and the other leg is in the air. Based on this, according to the height difference between the ankle joints of the two feet and the top level of the gate, it can be judged whether the person crossing the gate crosses the gate diagonally from above.

[0048] In one embodiment, in response to the leg key points of the human body posture data being higher than the top horizontal height of the gate, it is determined that the gate-passing person has non-standard gate-passing behavior, including: In response to the fact that the left and right sets of leg key points are both higher than the top level of the gate, it is determined that the person passing through the gate has jumped over the gate; and / or, In response to one group of leg key points being higher than the top horizontal height of the gate and the other group of leg key points being lower than the top horizontal height of the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above.

[0049] Specifically, in one embodiment, the key point of the human body includes the ankle joint, and the human body posture data includes the horizontal height of the ankle joint. In response to the fact that both the left and right sets of leg key points are above the top level of the gate at the same time, it is determined that the person passing through the gate has jumped over the gate, including: Get the top horizontal height of the gate in the image frame in the same coordinate system, If the level of the ankle joint is higher than the top level of the gate, it is determined that the person passing through the gate has jumped over the gate; or, The key points of the human body include the knees, and the human body posture data includes the horizontal height of the knees. In response to the fact that both the left and right sets of leg key points are above the top level of the gate at the same time, it is determined that the person passing through the gate has jumped over the gate, including: Get the top horizontal height of the gate set in the image frame in the same coordinate system, If the level of the knee is higher than the level of the top of the gate, it is determined that the person passing through the gate has jumped over the gate.

[0050] More specifically, the top horizontal height of the gate and the horizontal height of the ankle joint in the image frame in the same coordinate system are obtained.

[0051] In this embodiment, the top horizontal height of the gate and the horizontal height of the ankle joint in the image frame are both based on the height of the same reference line or the same reference point in the image frame, such as the height relative to the ground.

[0052] In a specific embodiment, the detected coordinates of the ankle joint are obtained by detecting the input image frame. The horizontal height of the top of the gate in the original image frame of the input image frame relative to the top of a reference line is obtained.

[0053] If there is a scaling ratio between the detected input image frame and its original image frame, the detected coordinates of the ankle joint are converted into actual coordinates in the coordinate system of the original image frame according to the scaling ratio. The horizontal height of the ankle joint relative to the reference line in the original image frame is calculated according to the actual coordinates of the ankle joint.

[0054] Figure 4Schematic diagram of the behavior of jumping over the gate from above in the embodiment of the present application; Figure 4 If the person who passes through the gate jumps over the gate from above, the horizontal height of the ankle joint is higher than the horizontal height of the top of the gate. Based on this, the horizontal height of the ankle joint and the horizontal height of the top of the gate can accurately determine whether the person who passes through the gate has non-standard gate-passing behavior, that is, illegal gate entry and exit behavior.

[0055] In order to ensure the accuracy of detection and prevent misjudgment and missed judgment, in a specific embodiment, if it is detected in multiple consecutive image frames that the horizontal heights of the ankle joints of both feet are higher than the horizontal height of the top of the gate, it is determined that the person passing through the gate has jumped over the gate.

[0056] In addition, if there are multiple people passing through the gate in the image frame, each person passing through the gate will be accurately identified, distinguished and continuously monitored.

[0057] or, Get the top horizontal height of the gate and the horizontal height of the knee in the image frame in the same coordinate system. Figure 4 It can be seen that if the horizontal height of both knees is higher than the horizontal height of the top of the gate, it is judged that the person passing through the gate has jumped over the gate.

[0058] This height judgment function effectively filters out normal traffic situations, thereby reducing the occurrence of false alarms.

[0059] The solution of this embodiment can be used to detect behaviors such as evading tickets from above gates or jumping from above gates in rail transit areas such as subways and train stations, and can also be used to detect illegal behaviors such as non-office personnel in office buildings illegally jumping from above gates to enter office areas. The solution of this embodiment can be applied to various application scenarios, and this application does not limit this.

[0060] This embodiment can monitor non-standard behaviors using only its own skeleton points, reduce the judgment of other objects, and refine the behaviors of non-standard gate-passing postures, conduct a specific and comprehensive analysis of the action postures, and at the same time refine the non-standard gate-passing behaviors of various different postures.

[0061] This embodiment is based on the behavioral characteristics of jumping over the gate. According to the horizontal height of the knees or ankles of the people passing through the gate, the behavior of jumping over the gate can be accurately identified, and normal people passing through can be filtered out, which can effectively reduce misjudgments and missed detections.

[0062] In one embodiment, the leg key points include ankle joints of left and right feet; In response to one set of leg key points being higher than the top horizontal height of the gate, and the other set of leg key points being lower than the top horizontal height of the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above, including: Get the top horizontal height of the gate in the image frame in the same coordinate system; If it is detected in multiple consecutive image frames that the ankle joint of one foot is lower than the top horizontal height of the gate, and the ankle joint of the other foot is higher than the top horizontal height of the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above; or, Key leg points include the knees of the left and right legs; In response to one set of leg key points being higher than the top horizontal height of the gate, and the other set of leg key points being lower than the top horizontal height of the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above, including: Get the top horizontal height of the gate in the image frame in the same coordinate system; If it is detected in multiple consecutive image frames that the knee of one leg is lower than the top horizontal height of the gate, and the knee of the other leg is higher than the top horizontal height of the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above.

[0063] Specifically, in order to more comprehensively identify the behaviors of fare evasion or diagonally crossing the gate from above, the system will also consider the diagonal crossing. If one leg or one foot is above the top of the gate, and the other leg or the other foot is below the gate or below the top of the gate, the system will identify the behavior as diagonally crossing the gate. This usually happens when fare evaders or people passing through the gate try to pass through the gate quickly, and they may adopt a sideways movement. By detecting the relative position of the key points of the legs (for example, the ankles or knees) on the height line, the system can accurately determine this crossing behavior, further improving the accuracy of identifying illegal gate crossing behaviors.

[0064] The ankle joints of the left and right feet Figure 3 The left ankle joint (key point 15) and the right ankle joint (key point 16) in the figure. The human body posture data includes the horizontal height of the ankle joint.

[0065] Obtain the top horizontal height of the gate and the horizontal height of the ankle joint in the image frame in the same coordinate system.

[0066] In this embodiment, the top horizontal height of the gate and the horizontal height of the ankle joint in the image frame are both based on the height of the same reference line or reference point in the image frame, such as the height relative to the ground.

[0067] In a specific embodiment, the detected coordinates of the ankle joint are obtained by detecting the input image frame. The horizontal height of the top of the gate in the original image frame of the input image frame relative to the top of a reference line is obtained.

[0068] If there is a scaling ratio between the detected input image frame and its original image frame, the detected coordinates of the ankle joint are converted into actual coordinates in the coordinate system of the original image frame according to the scaling ratio. The horizontal height of the ankle joint relative to the reference line in the original image frame is calculated according to the actual coordinates of the ankle joint. The coordinates of the reference line in the original image frame can be obtained from the original image frame.

[0069] Figure 5 Schematic diagram of the behavior of crossing the gate obliquely from above in the embodiment of the present application; Figure 5 It can be seen that if the person passing through the gate crosses the gate from above, it can be detected in multiple consecutive image frames that at the same moment, the horizontal height of the ankle joint of one foot of the person passing through the gate is lower than the horizontal height of the top of the gate, and the horizontal height of the ankle joint of the other foot is higher than the horizontal height of the top of the gate, that is, it is determined that the person passing through the gate has crossed the gate diagonally from above.

[0070] More specifically, when the person passing through the gate just crosses the gate diagonally, the first foot is on the ground, so the level of the ankle joint of the first foot is lower than the top level of the gate; at the same time, the second foot will be lifted up and pass over the gate, so the level of the ankle joint of the second foot is higher than the top level of the gate.

[0071] After the second foot lands, the level of the ankle joint of the second foot is lower than the top level of the gate; at the same time, the first foot is lifted up and passes over the gate, so the level of the ankle joint of the first foot is higher than the top level of the gate.

[0072] Of course, in another embodiment, the top horizontal height of the gate in the original image frame may be converted to the top horizontal height of the gate in the input image frame according to the scaling ratio, and then the top horizontal height of the gate in the input image frame is compared with the horizontal heights of the ankle joints of the two feet in the input image frame.

[0073] This embodiment can monitor non-standard behaviors using only its own skeleton points, reduce the judgment of other objects, and refine the behaviors of non-standard gate-passing postures, conduct a specific and comprehensive analysis of the action postures, and at the same time refine the non-standard gate-passing behaviors of various different postures.

[0074] This embodiment can accurately detect the gate-crossing behavior of people passing through the gate based on the action characteristics of single-leg diagonal crossing of the gate through key points of the human body, further improve the accuracy and comprehensiveness of the identification of non-standard gate-crossing behavior, and effectively ensure the safety and reliability of the supervision area where the gate is located.

[0075] In one embodiment, the human body key points include ankles, knees, hips, and shoulders; In step S200, human body posture data is obtained according to human body key point data, including: According to the coordinates of the key points of the human body, the first angle between the torso and the thigh of the person passing through the gate and the second angle between the thigh and the calf are calculated; In step S300, the non-standard gate-crossing behavior of the gate-crossing person is judged according to the height information and limb angle information in the human posture data, including: If both the first angle and the second angle are smaller than the corresponding angle thresholds, it is determined that the person passing through the gate has jumped over the gate.

[0076] or, In one embodiment, the key points of the human body also include the hips and shoulders; In step S200, human body posture data is obtained according to human body key point data, including: According to the coordinates of the key points of the human body, the first angle between the torso and the thigh of the person passing through the gate and the second angle between the thigh and the calf are calculated; In response to both the left and right groups of leg key points being higher than the top level of the gate at the same time, further comprising: In response to the first angle and the second angle being both smaller than the corresponding angle thresholds, it is determined that the person passing through the gate has jumped over the gate.

[0077] Specifically, the ankle joint Figure 3 Left ankle and / or right ankle, knee Figure 3 left knee and / or right knee, hip, i.e. Figure 3 Left and / or right hip, shoulder Figure 3 Middle left shoulder and / or right shoulder.

[0078] The torso is defined by the line connecting the hips and shoulders, the thigh is defined by the line connecting the hips and knees, and the calf is defined by the line connecting the ankle joint and the knee.

[0079] Based on the coordinates of the ankle, knee, hip and shoulder, the first angle between the trunk and thigh of the person passing through the gate and the second angle between the thigh and calf can be calculated. Figure 3 As shown, based on the coordinates of the left ankle joint (key point 15), the left knee (key point 13), the left hip (key point 11) and the left shoulder (key point 5), the first angle between the torso and the left thigh of the person passing through the gate and the second angle between the left thigh and the left calf can be calculated.

[0080] Among them, the line connecting the left ankle joint (key point 15) and the left knee (key point 13) represents the left calf, the line connecting the left knee (key point 13) and the left hip (key point 11) represents the left thigh, and the line connecting the left hip (key point 11) and the left shoulder (key point 5) represents the torso.

[0081] or, Based on the coordinates of the right ankle joint (key point 16), right knee (key point 14), right hip (key point 12) and right shoulder (key point 6), the first angle between the torso and the right thigh of the person passing through the gate and the second angle between the right thigh and the right calf can be calculated.

[0082] Among them, the line connecting the right ankle joint (key point 16) and the right knee (key point 14) represents the right calf, the line connecting the right knee (key point 14) and the right hip (key point 12) represents the right thigh, and the line connecting the right hip (key point 12) and the right shoulder (key point 6) represents the torso.

[0083] Figure 4 Schematic diagram of the behavior of jumping over the gate from above in the embodiment of the present application; Figure 4 It can be seen that if the person passing through the gate jumps from above the gate to enter or exit the gate, the first angle between the torso and the thigh of the person passing through the gate and the second angle between the thigh and the calf are both less than the corresponding angle threshold. The first angle and the second angle correspond to an angle threshold respectively, and the two angle thresholds can be the same or different. For example, the angle threshold corresponding to the first angle and / or the second angle is a value between 30 degrees and 45 degrees, which is specifically configured according to actual conditions, and this application does not limit this.

[0084] In a specific embodiment, if the first angle between the torso and the left thigh of the person passing through the gate and the second angle between the left thigh and the left calf are both smaller than the corresponding angle thresholds, and / or the first angle between the torso and the right thigh of the person passing through the gate and the second angle between the right thigh and the right calf are both smaller than the corresponding angle thresholds, it is determined that the person passing through the gate has jumped over the gate.

[0085] In a specific embodiment, if the left and right groups of leg key points are both higher than the top horizontal height of the gate, and the first angle and the second angle are both smaller than the corresponding angle thresholds, it is determined that the person passing through the gate has jumped over the gate.

[0086] The solution of this embodiment can be used to detect behaviors such as fare evasion from above the gate or fare evasion from above the gate in rail transit areas such as subways and railway stations, and can also be used to detect illegal behaviors such as non-office personnel in office buildings illegally entering the office area from above the gate. The solution of this embodiment can be applied to various application scenarios, and this application does not limit this.

[0087] This embodiment can monitor non-standard behaviors using only its own skeleton points, reduce the judgment of other objects, and refine the behaviors of non-standard gate-passing postures, conduct a specific and comprehensive analysis of the action postures, and at the same time refine the non-standard gate-passing behaviors of various different postures.

[0088] This embodiment is based on the behavioral characteristics of jumping over the gate. According to the first angle between the torso and thigh of the person passing through the gate and the second angle between the thigh and calf, it can significantly enhance the accuracy of identifying the behavior of jumping over the gate and effectively reduce misjudgments and missed detections.

[0089] In one embodiment, the key points of the human body include ankles, knees, hips and shoulders. In step S200, human body posture data is obtained according to human body key point data, including: According to the coordinates of the key points of the human body, the first angle between the torso and the thigh of the person passing through the gate, the second angle between the thigh and the calf, and the horizontal height of the ankle joint are calculated; In step S300, the non-standard gate-crossing behavior of the gate-crossing person is judged according to the height information and limb angle information in the human posture data, including: Get the top horizontal height of the gate in the image frame in the same coordinate system, If the horizontal height of the ankle joint is higher than the horizontal height of the top of the gate, and both the first angle and the second angle are smaller than the corresponding angle thresholds, it is determined that the gate-passing person has jumped over the gate; or, Key points of the human body include ankles, knees, hips and shoulders. In step S200, human body posture data is obtained according to human body key point data, including: According to the coordinates of the key points of the human body, the first angle between the trunk and the thigh of the person passing through the gate, the second angle between the thigh and the calf, and the horizontal height of the knee are calculated; In step S300, the non-standard gate-crossing behavior of the gate-crossing person is judged according to the height information and limb angle information in the human posture data, including: Get the top horizontal height of the gate set in the image frame in the same coordinate system, If the level of the knee is higher than the level of the top of the gate, and both the first angle and the second angle are smaller than the corresponding angle thresholds, it is determined that the person passing through the gate has jumped over the gate.

[0090] In one embodiment, the human body key points also include elbows and wrists; In step S200, the human body posture data is obtained according to the human body key point data, and further includes: According to the coordinates of the key points of the human body, the third angle between the arm and torso of the person passing through the gate is calculated; In response to the left and right groups of leg key points being higher than the top level of the gate at the same time, the method further includes: In response to the third angle being greater than the corresponding angle threshold, it is determined that the person passing through the gate has jumped over the gate.

[0091] Specifically, the third angle between the arm and the torso can also be detected. If the person passing through the gate jumps over the gate, he or she will usually use both hands to support the body, which will cause a specific angle and posture to be formed between the arms and the torso. If the third angles between the two arms and the torso are greater than the corresponding angle threshold, it can also be used as a basis for determining that the person passing through the gate jumped from over the gate. In addition, if the left and right sets of leg key points are both higher than the horizontal height of the top of the gate, and the third angle is greater than the corresponding angle threshold, it is determined that the person passing through the gate has jumped from over the gate. Combining the two judgment conditions, such as Figure 4 This analysis step significantly enhances the accuracy of the recognition.

[0092] Of course, the judgment basis of this embodiment can also be combined with the other judgment basis mentioned above to judge the behavior of the person passing through the gate, so as to increase the reliability and accuracy of the judgment.

[0093] In one embodiment, the key points of the human body further include the crotch; In response to one set of leg key points being above the top level of the gate and the other set of leg key points being below the top level of the gate, the method further includes: In response to the crotch being in the top area of ​​the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above.

[0094] Specifically, the non-standard gate-crossing behavior of the gate-crossing personnel is judged based on the height information and limb angle information in the human posture data, including: Obtain the top horizontal height of the gate in the image frame under the same coordinate system, as well as the distance between the hips of the person passing through the gate and the gate; If a group of leg key points are detected to be higher than the top horizontal height of the gate in multiple consecutive image frames, and another group of leg key points are detected to be lower than the top horizontal height of the gate, and the distance between the hips of the person passing through the gate and the gate is within the preset distance (that is, the hips are in the top area of ​​the gate), then it is determined that the person passing through the gate has crossed the gate diagonally from above.

[0095] In order to reduce the false alarm rate, the algorithm also takes into account the position of the person's hips. The system will confirm whether the line segment where the hips are located is close to the gate. If the hips are near the gate, the system will enhance the credibility of the judgment to ensure the accuracy of the recognition. This detection is based on the characteristics that when a fare evader or a person who crosses the gate diagonally, the hips are relatively low and close to the gate. By taking these factors into consideration, the system can more accurately identify and judge the behavior of the person who crosses the gate diagonally, thereby ensuring the effective monitoring of the safety of the gate area.

[0096] The top horizontal height of the gate and the horizontal height of the ankle joint or the knee in the image frame under the same coordinate system, as well as the distance between the hip and the gate, are obtained. The distance between the hip and the gate may be the distance from the hip to the gate in the vertical direction.

[0097] In this embodiment, the top horizontal height of the gate and the horizontal height of the ankle joint or knee in the image frame are based on the height of the same reference line in the image frame, such as the height relative to the ground or the height relative to the same reference point or the y-axis coordinate.

[0098] Figure 5 Schematic diagram of the behavior of crossing the gate obliquely from above in the embodiment of the present application; Figure 5 It can be seen that if the person passing through the gate crosses over the gate from above, it can be detected in multiple consecutive image frames that at the same moment, the horizontal height of the ankle joint of one foot of the person passing through the gate is lower than the horizontal height of the top of the gate, and the horizontal height of the ankle joint of the other foot is higher than the horizontal height of the top of the gate, and the distance between the hips of the person passing through the gate and the gate is within the preset distance, that is, it is determined that the person passing through the gate has crossed the gate diagonally from above.

[0099] More specifically, when the person passing through the gate just crosses the gate diagonally, the first foot is on the ground, so the level of the ankle joint of the first foot is lower than the top level of the gate; at the same time, the second foot will be lifted up and pass over the gate, so the level of the ankle joint of the second foot is higher than the top level of the gate.

[0100] After the second foot lands, the level of the ankle joint of the second foot is lower than the top level of the gate; at the same time, the first foot is lifted up and passes over the gate, so the level of the ankle joint of the first foot is higher than the top level of the gate.

[0101] During this period, the distance between the person passing through the gate and the gate is very close. Therefore, in order to prevent misjudgment and missed detection, this embodiment further determines whether the person passing through the gate is crossing the gate diagonally based on the distance between the hips of the person passing through the gate and the gate.

[0102] Of course, in another embodiment, the top horizontal height of the gate in the original image frame may be converted to the top horizontal height of the gate in the input image frame according to the scaling ratio, and then the top horizontal height of the gate in the input image frame is compared with the horizontal heights of the ankle joints of the two feet in the input image frame.

[0103] The distance between the hips of the person passing through the gate and the gate in the original image frame is obtained, and according to the scaling ratio, the distance between the hips of the person passing through the gate and the gate in the original image frame is converted into the distance between the hips of the person passing through the gate and the gate in the input image frame. The distance between the hips of the person passing through the gate and the gate is compared with the corresponding preset distance.

[0104] It should be noted that the preset distances in the original image frame coordinate system and the input image frame coordinate system are different, and also need to be converted accordingly according to the scaling ratio.

[0105] In order to more comprehensively identify the behavior of fare evasion from above, this embodiment also considers the diagonal crossing. If one leg is above the gate and the other leg is below the gate, the system identifies the behavior as diagonal crossing. This usually happens when the fare evader tries to pass through the gate quickly, and may adopt a side-crossing action. By detecting the relative position of the key points of the legs on the height line, the system can accurately determine this crossing behavior, further improving the accuracy of fare evasion identification.

[0106] To reduce the false alarm rate, the algorithm also takes into account the position of the person's hips. The system will confirm whether the line segment where the hips are located is close to the gate. If the hips are near the gate, the system will enhance the credibility of the judgment to ensure the accuracy of the recognition. This detection is based on the characteristics that when the fare evader crosses the gate diagonally, the hips are relatively low and close to the gate. By taking these factors into consideration, the system can more accurately identify and judge the behavior of fare evaders, thereby ensuring the effective monitoring of the safety of the subway gate area.

[0107] This embodiment can accurately detect the behavior of people crossing the gate based on the action characteristics of one leg diagonally crossing the gate. It can detect the height difference between the key points of the human body and the bottom of the gate and the distance between the buttocks and the gate, reduce the false alarm rate, further improve the accuracy and comprehensiveness of the identification of non-standard gate-crossing behavior, and effectively ensure the safety and reliability of the supervision area where the gate is located.

[0108] In one embodiment, in step S300, judging the non-standard gate-crossing behavior of the gate-crossing person according to the height information and the body angle information in the human posture data includes: If it is detected that the horizontal height of all key points of the person passing through the gate is lower than the bottom horizontal height of the gate, it is determined that the person passing through the gate is passing from below the gate; or, In step S300, the non-standard gate-crossing behavior of the gate-crossing person is judged according to the height information and limb angle information in the human posture data, including: If it is detected that the horizontal heights of all key points of the person passing through the gate are lower than the bottom horizontal height of the gate, and the horizontal height of the head is lower than the horizontal height of the shoulders, it is determined that the person passing through the gate is passing through from below the gate.

[0109] Specifically, for situations where someone is diving under a gate or evading a fare, the system also relies on key point detection. The system requires that the key points of all limbs appear under the gate to ensure that the action of the fare evader or gate diver is through the bottom of the gate. Through real-time monitoring, the system can quickly identify potential fare evasion or gate diver behavior.

[0110] In this case, the system can also check whether the person's head fulcrum is lower than the shoulder position. This judgment is based on the characteristics that fare evaders or gate crawlers usually need to lower their heads to pass through the gate. If the head fulcrum is lower than the shoulder, the system will further confirm that the person is evading the ticket from below or crawling through the gate. This step ensures that low-profile fare evasion or gate crawling behaviors are captured, improving the comprehensiveness of recognition.

[0111] Obtain the bottom horizontal height of the gate and the horizontal height of all key points of the human body of the people passing through the gate in the image frame in the same coordinate system.

[0112] In this embodiment, the bottom horizontal height of the gate in the image frame is based on the height of the same reference line in the image frame, such as the height relative to the ground or the height relative to the same reference point.

[0113] In a specific embodiment, the detection coordinates of each key point obtained by detecting the input image frame are obtained. The bottom horizontal height of the gate in the original image frame of the input image frame relative to a reference line is obtained.

[0114] If there is a scaling ratio between the detected input image frame and its original image frame, the detected coordinates of the key points are converted into actual coordinates in the coordinate system of the original image frame according to the scaling ratio. The horizontal height of the key points relative to the reference line in the original image frame is calculated according to the actual coordinates of the key points. The coordinates of the reference line in the original image frame can be obtained from the original image frame.

[0115] Of course, in another embodiment, the bottom horizontal height of the gate in the original image frame may be converted to the bottom horizontal height of the gate in the input image frame according to the scaling ratio, and then the bottom horizontal height of the gate in the input image frame is compared with the horizontal heights of each key point in the input image frame.

[0116] Figure 6Schematic diagram of the behavior of passing through the gate from below in the embodiment of the present application; Figure 6 It can be seen that if the person passing through the gate passes from below the gate, the horizontal height of all the key points of the person passing through the gate is lower than the bottom horizontal height of the gate; or, the horizontal height of all the key points of the person passing through the gate is lower than the bottom horizontal height of the gate, and the horizontal height of the head is lower than the horizontal height of the shoulders. The horizontal height of the head can be Figure 3 The horizontal height of any one or more key points among the eyes, ears, and nose.

[0117] In a specific embodiment, to prevent misjudgment and missed detection, if it is detected in at least n consecutive image frames that the horizontal heights of all the key points of the human body of the person passing through the gate are lower than the bottom horizontal height of the gate, it is determined that the person passing through the gate has passed through from below the gate; or, if it is detected in at least n consecutive image frames that the horizontal heights of all the key points of the human body of the person passing through the gate are lower than the bottom horizontal height of the gate, and the horizontal height of the head is lower than the horizontal height of the shoulders, it is determined that the person passing through the gate has passed through from below the gate.

[0118] In the case of fare evasion from under the gate, this embodiment performs key point detection. When the key points of all limbs appear under the gate, it is determined that the person passing through the gate passes through the bottom of the gate. Through real-time monitoring, the system can quickly identify potential non-standard gate-passing behaviors such as fare evasion.

[0119] In this case, the system can also check whether the person's head fulcrum is lower than the shoulder position. This judgment is based on the characteristic that people passing through the gate usually need to lower their heads to pass through the gate. If the head fulcrum is lower than the shoulder, the system will further confirm that the person passed through the gate from below. This step ensures that non-standard gate-passing behaviors such as fare evasion with a low posture are captured, improving the comprehensiveness of recognition.

[0120] This embodiment can accurately capture low-profile gate-passing behavior, judge non-standard gate-passing behavior from below the bottom of the gate, and further comprehensively identify non-standard gate-passing behavior by comparing the horizontal height between the key points of the body of the person passing through the gate and the bottom of the gate.

[0121] In one embodiment, the image frame is detected in step S100 to obtain key point data of the human body of the person passing through the gate, including: The human body posture estimation algorithm is used to detect the zoomed image frame and obtain the key point detection data of the human body of the person passing through the gate; According to the scaling ratio of the scaled image frame to the corresponding original image frame, the human key point detection data is converted into human key point data in the original image frame coordinate system.

[0122] Specifically, the scaled image frame is an input image frame after scaling the original image frame. The scaled image frame can be detected using a human key point detection algorithm or a human posture estimation algorithm to obtain human key point detection data of the person passing through the gate; the human key point detection data includes the detection coordinates of each human key point.

[0123] According to the scaling ratio, the detected coordinates of each key point of the human body can be converted into coordinates in the original image frame coordinate system to achieve coordinate alignment and unification, which is convenient for subsequent height comparison or distance comparison in the same coordinate system and ensures the accuracy of monitoring non-standard gate-crossing behavior.

[0124] It should be noted that the coordinates obtained from the image frame in the present application are all coordinates relative to the same reference point or origin, and the y-axis coordinate can be used as the horizontal height.

[0125] This application can not only identify the basic status of non-standard gate-passing behaviors such as fare evasion, but also deeply analyze the details of the posture. A complete non-standard gate-passing behavior posture must have some specific features such as leg height, angle, etc. The system will only identify it as a non-standard gate-passing behavior after comprehensively judging these factors. This multi-dimensional analysis method improves the accuracy of the judgment, and allows the system to be subdivided into multiple situations such as single-leg fare evasion, double-leg fare evasion, fare evasion from above, and non-standard gate-passing from below, making the identification of non-standard gate-passing behavior more comprehensive and accurate. This comprehensiveness is not limited to identifying the presence or absence of non-standard gate-passing behavior, but can also provide security personnel with specific non-standard gate-passing methods and characteristics, so that they can take targeted countermeasures.

[0126] pass Figure 2 , Figure 4-Figure 6 It can be seen that in Figure 2 All the people passing through the gate are considered to be passing through the gate normally. No one is suspected of any non-standard passing behavior, such as jumping over the gate, crossing the gate diagonally from above, or passing through the gate from below.

[0127] exist Figure 4 The person who is passing through the gate jumps over the gate with both legs, so Figure 4 There is a behavior of jumping over the gate, but there is no behavior of crossing the gate diagonally from above or passing the gate from below the gate (that is, normal).

[0128] exist Figure 5 The person passing through the gate has one leg diagonally across the gate, so Figure 5 There is a behavior of crossing the gate diagonally from above, but there is no behavior of jumping over the gate or passing the gate from below the gate (that is, normal).

[0129] exist Figure 6 The personnel passing through the gate crawl under the gate, so Figure 6There is a behavior of passing through the gate from below, but there is no behavior of jumping over the gate or crossing the gate diagonally from above (that is, normal).

[0130] The monitoring scheme for non-standard gate-passing behaviors based on key point analysis proposed in this application can be applied to subway fare evasion monitoring, office building personnel verification entry and exit monitoring, etc., aiming to significantly improve the monitoring and identification capabilities of non-standard gate-passing behaviors. Through advanced technical means, this system can achieve efficient and accurate fare evasion detection in a complex subway environment, providing strong support for the operation and management of public places. Specifically, this application has the following advantages: High-precision motion recognition: This application uses YOLO-Pose technology for key point detection, which can accurately identify 17 key points of an individual, including the head, shoulders, legs, etc. This detailed recognition enables the system to accurately judge non-standard gate entry and exit behaviors such as fare evasion, thereby significantly reducing misjudgments and missed detections.

[0131] Adapt to complex environment: Even during peak hours when there is a lot of traffic, the system can effectively extract key information and maintain efficient identification of non-standard gate-crossing behaviors. Figure 2 As shown, even in the case of a large flow of people, the system can still accurately identify the key points of each person, and its powerful real-time monitoring capability ensures accuracy and reliability in complex environments.

[0132] Diversified testing capabilities: This application has the ability to identify different types of non-standard gate-passing behaviors, including jumping over, stepping over with one foot from above, and crawling from below, etc. Through multi-dimensional detection methods such as height judgment, angle analysis, and hip position confirmation, the accuracy of identifying fare evasion behaviors is enhanced.

[0133] In summary, the non-standard gate-passing behavior monitoring technology based on key point analysis in this application not only improves the ability to identify non-standard gate-passing behaviors, but also provides an innovative solution for the safe operation and management of public places, which has important practical application value and broad market prospects. The implementation of this system will effectively ensure the safety and management efficiency of the subway, while improving the overall travel experience and safety assurance.

[0134] The monitoring technology of non-standard gate-passing behavior based on key point analysis in this application can be applied to various public places such as urban public transportation systems, office buildings, shopping malls, especially places with high passenger density such as subways and light rails. This technology can not only effectively identify fare evasion and improve the safety of operation management, but also provide data support for transportation operation companies to help analyze passenger flow trends and the occurrence patterns of fare evasion. In addition, this technology is also suitable for large-scale events and gathering places, such as sports events, concerts and exhibitions, and can monitor passenger flow management in real time to ensure the smooth progress of activities.

[0135] In addition, with the advancement of smart city construction, this technology can be combined with other monitoring systems and intelligent traffic management platforms to achieve comprehensive traffic management and improve urban safety and management efficiency. In the future, this technology has the potential to expand into areas such as smart security, retail monitoring and behavior analysis, providing more accurate data support and decision-making basis for all walks of life.

[0136] refer to Figure 7 In one embodiment of the present application, a monitoring device for non-standard gate-passing behavior is provided, and the monitoring device for non-standard gate-passing behavior includes: The target detection module 100 is used to detect the image frame and obtain the human key point data of the person passing through the gate, wherein the image frame is extracted from the video stream of the gate area of ​​the target site, and the human key point data includes the coordinates of each human key point; The posture data acquisition module 200 is used to acquire human posture data according to human key point data; The behavior judgment module 300 is used to judge the non-standard gate-crossing behavior of the person passing through the gate based on the height information and body angle information in the human posture data.

[0137] In one embodiment, the key points of the human body include ankles and knees. The behavior judgment module 300 includes an upper non-standard gate-crossing behavior judgment module, which is used to determine that the person passing the gate has non-standard gate-crossing behavior in response to the leg key points of the human posture data being higher than the top horizontal height of the gate; wherein the leg key points include at least one of the ankle joint and the knee.

[0138] In one embodiment, the upper non-standard gate-passing behavior judgment module includes: The first upper jump-through gate behavior judgment module is used to judge that the gate-passing person has jumped through the gate from above in response to the fact that the left and right sets of leg key points are both higher than the top horizontal height of the gate; The module for judging the behavior of crossing the gate diagonally from above is used to judge that the person passing the gate has crossed the gate diagonally from above in response to one group of leg key points being higher than the top horizontal height of the gate and the other group of leg key points being lower than the top horizontal height of the gate.

[0139] In one embodiment, the human body key points include ankles, knees, hips, and shoulders; The posture data acquisition module 200 is specifically used to calculate the first angle between the trunk and thigh of the person passing through the gate and the second angle between the thigh and calf according to the coordinates of the key points of the human body; The behavior judgment module 300 includes: The second upper jumping over the gate behavior judgment module is used to determine that the gate-passing person has jumped over the gate in response to the first angle and the second angle being smaller than the corresponding angle thresholds.

[0140] In one embodiment, the human body key points also include elbows and wrists; The posture data acquisition module 200 is specifically used to calculate the third angle between the arm and torso of the person passing through the gate according to the coordinates of the key points of the human body; The first module for judging the behavior of jumping over the gate is used to respond to the situation that the left and right groups of leg key points are both higher than the top horizontal height of the gate, and the third angle is greater than the corresponding angle threshold, and then determine that the person passing the gate has jumped over the gate.

[0141] In one embodiment, the key points of the human body further include the crotch; The module for judging the behavior of crossing the gate diagonally from above is used to judge that the person passing the gate has crossed the gate diagonally from above in response to one group of leg key points being higher than the top horizontal height of the gate, and the other group of leg key points being lower than the top horizontal height of the gate, and the hips being in the top area of ​​the gate.

[0142] In one embodiment, the behavior determination module 300 includes: A gate-passing behavior judgment module is used to judge that the gate-passing person has passed through the gate from below the gate in response to the horizontal height of all key points of the body of the gate-passing person being lower than the bottom horizontal height of the gate; or, The behavior determination module 300 includes: The module for judging the behavior of passing through the gate from below is used to judge that the person passing through the gate has passed through the gate from below in response to the fact that the horizontal heights of all key points of the person's body are lower than the bottom horizontal height of the gate and the head is lower than the shoulders.

[0143] Figure 87 is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. The computer device 700 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 710 (for example, one or more processors) and a memory 720, and one or more storage media 730 (for example, one or more mass storage devices) storing application programs 733 or data 732. Among them, the memory 720 and the storage medium 730 can be short-term storage or permanent storage. The program stored in the storage medium 730 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the computer device 700. Furthermore, the processor 710 may be configured to communicate with the storage medium 730 to execute a series of instruction operations in the storage medium 730 on the computer device 700.

[0144] The computer device 700 may also include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input and output interfaces 760, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 8 The illustrated computer device structure does not constitute a limitation on the computer device, and may include more or fewer components than illustrated, or combine certain components, or arrange the components differently.

[0145] The present application also provides a computer device, the computer device includes a memory and a processor, the memory stores computer-readable instructions, when the computer-readable instructions are executed by the processor, the processor executes the steps of the monitoring method for non-standard gate-passing behavior in the above-mentioned embodiments. The present application also provides a computer-readable storage medium, the computer-readable storage medium can be a non-volatile computer-readable storage medium, the computer-readable storage medium can also be a volatile computer-readable storage medium, the computer-readable storage medium stores instructions, when the instructions are executed on the computer, the computer executes the steps of the monitoring method for non-standard gate-passing behavior.

[0146] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0147] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., and other media that can store program codes.

[0148] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A monitoring method for non-standard gate-passing behavior, characterized in that: The monitoring method of the non-standard gate-passing behavior includes: Detecting the image frame to obtain the key point data of the human body of the person passing through the gate, wherein the image frame is extracted from the video stream of the gate area of ​​the target site, and the key point data of the human body includes the coordinates of each key point of the human body; Acquire human body posture data according to the human body key point data; The non-standard gate-crossing behavior of the person passing through the gate is judged based on the height information and limb angle information in the human posture data.

2. The monitoring method for non-standard gate-crossing behavior according to claim 1 is characterized in that: The key points of the human body include ankle joints and knees, The judging of the non-standard gate-passing behavior of the gate-passing person according to the height information and the body angle information in the human posture data includes: In response to the leg key points of the human posture data being higher than the top horizontal height of the gate, it is determined that the person passing through the gate has non-standard gate-passing behavior; wherein the leg key points include at least one of the ankle joint and the knee.

3. The monitoring method for non-standard gate-crossing behavior according to claim 2 is characterized in that: In response to the leg key point of the human body posture data being higher than the top horizontal height of the gate, it is determined that the gate-passing person has non-standard gate-passing behavior, including: In response to the two groups of left and right leg key points being higher than the top level of the gate at the same time, it is determined that the gate-passing person has jumped over the gate; and / or, In response to one group of the leg key points being higher than the top horizontal height of the gate, and another group of the leg key points being lower than the top horizontal height of the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above.

4. The monitoring method for non-standard gate-crossing behavior according to claim 1 is characterized in that: The key points of the human body include ankles, knees, hips and shoulders; The step of obtaining human body posture data according to the human body key point data includes: Calculating a first angle between the trunk and thigh of the gate passer and a second angle between the thigh and calf according to the coordinates of the key points of the human body; The judging of the non-standard gate-passing behavior of the gate-passing person according to the height information and the body angle information in the human posture data includes: In response to the first angle and the second angle being both smaller than corresponding angle thresholds, it is determined that the person passing through the gate has jumped over the gate.

5. The monitoring method for non-standard gate-crossing behavior according to claim 3 is characterized in that: The key points of the human body also include elbows and wrists; The step of acquiring human body posture data according to the human body key point data further comprises: Calculating a third angle between the arm and torso of the gate passer according to the coordinates of the key points of the human body; In response to the two groups of leg key points being higher than the top level of the gate at the same time, the method further includes: In response to the third angle being greater than the corresponding angle threshold, it is determined that the person passing through the gate has jumped over the gate.

6. The monitoring method for non-standard gate-crossing behavior according to claim 3 is characterized in that: The key points of the human body also include the crotch; In response to one group of the leg key points being higher than the top level of the gate and another group of the leg key points being lower than the top level of the gate, the method further includes: In response to the crotch being in the top area of ​​the gate, it is determined that the person passing through the gate has crossed the gate diagonally from above.

7. The monitoring method for non-standard gate-crossing behavior according to claim 1 is characterized in that: The judging of the non-standard gate-passing behavior of the gate-passing person according to the height information and the body angle information in the human posture data includes: In response to the fact that the levels of all key points of the body of the person passing through the gate are lower than the bottom level of the gate, it is determined that the person passing through the gate has passed through the gate from below; or, The key points of the human body include the head and shoulders, and the non-standard gate-passing behavior of the gate-passing person is judged based on the height information and limb angle information in the human body posture data, including: In response to the fact that the levels of all key points of the body of the person passing through the gate are lower than the bottom level of the gate, and the head is lower than the shoulders, it is determined that the person passing through the gate has passed through the gate from below.

8. A monitoring device for non-standard gate-passing behavior, characterized in that: The monitoring device for the non-standard gate-passing behavior comprises: A target detection module is used to detect image frames and obtain human key point data of people passing through the gate, wherein the image frames are extracted from the video stream of the gate area of ​​the target site, and the human key point data includes the coordinates of each human key point; A posture data acquisition module, used to acquire human posture data according to the human key point data; The behavior judgment module is used to judge the non-standard gate-crossing behavior of the person passing through the gate based on the height information and limb angle information in the human posture data.

9. A computer device, characterized in that: The computer device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the computer device to execute the monitoring method for non-standard gate-passing behavior according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, a monitoring method for non-standard gate-crossing behavior as described in any one of claims 1 to 7 is implemented.

Images