A method, device, and electronic equipment for identifying leaning against subway car doors or platform screen doors.

Abstract

This invention discloses a method, device, and electronic equipment for identifying leaning against subway car doors or platform screen doors, used for passenger safety and to reduce deformation of subway car doors or platform screen doors due to external pressure. This invention relates to the field of subway safety technology. The method includes the following steps: S1, acquiring image frames of the door area and the vertical direction of the door; S2, detecting pedestrians in the image, determining whether the pedestrian is near the door, performing key point detection on pedestrians near the door, and calculating the included angles α and β through key point detection; S3, when α or β calculated in step S2 is greater than a set threshold Y, and simultaneously detecting that the distance d from the nearest ankle point of the person to the lower edge of the door is less than a threshold d0, then it is determined that the person is leaning against the subway door, and an alarm is issued. This invention can simply and efficiently determine the behavior of pedestrians leaning against the car door through a visual method, ensuring the safety of people and vehicles.

Description

Technical Field

[0001] A method for identifying leaning against subway car doors or platform screen doors is disclosed, for passenger safety and to reduce deformation of subway car doors or platform screen doors due to external pressure. This invention relates to the field of subway safety technology. Background Technology

[0002] With the rapid development of urban rail transit, its operational safety has become increasingly important. Due to limited seating, some passengers are forced to stand, while others, due to fatigue, habitually lean against the doors to relieve physical strain. When the train is running, the swaying of the train body causes passengers to exert greater pressure on the doors than they were designed to withstand. This pressure increases the pressure on the sliding blocks at the bottom of the doors, accelerating wear and shortening their lifespan. Simultaneously, the doors may undergo permanent deformation under prolonged stress, causing friction between the doors and the edges of the inner and outer walls of the train during opening and closing, potentially leading to jamming and preventing normal opening and closing. Regarding passenger safety, since the platform screen doors and train doors open simultaneously when the train arrives at a station, passengers who are resting or distracted may overlook the door opening procedure, potentially leading to falls or getting trapped in the gap between the platform screen doors and train doors, resulting in accidents and injuries, posing a significant safety hazard. Summary of the Invention

[0003] The purpose of this invention is to address the technical problems of passenger safety hazards and reduce the deformation of subway car doors or platform screen doors due to external pressure. This invention provides a method for identifying leaning against subway car doors or platform screen doors, which provides real-time alarms for passengers leaning against car doors or platform screen doors to remind passengers to pay attention to safety.

[0004] The technical solution adopted in this invention is as follows: A method for identifying leaning against a subway car door or platform screen door, comprising the following steps:

[0005] S1. Acquire image frames of the door area and the vertical direction of the door;

[0006] S2. Detect pedestrians in the scene and determine whether the pedestrians are near the door. Perform key point detection on pedestrians near the door. Through key point detection, obtain the key points A and B on the shoulders, key points C and D on the hips, and key points E and F on the knees of the people near the door. Connect AB, CD, and EF to obtain the midpoint G of line AB, the midpoint H of line CD, and the midpoint I of line EF. Connect GH and GI and calculate the angle α between line GH and the vertical direction and the angle β between line GI and the vertical direction.

[0007] S3. When α or β calculated in step S2 is greater than the set threshold Y, and the distance d from the nearest ankle of the person to the lower edge of the door is less than the threshold d0, it is determined that the person is leaning against the subway door and an alarm is issued.

[0008] The working principle of this invention is as follows: read the camera configuration parameters to obtain the door area and the vertical direction vector of the door; read the video stream to obtain image frames, and input the image frames into a deep neural network (such as YOLOv5s) to obtain the positions of all passengers; determine whether each person is in the door area, and perform key point detection on the passengers in the door area; obtain the passenger's body tilt vector; calculate the angle between the body tilt vector and the vertical direction of the door, and determine whether the angle is greater than a specified threshold; at the same time, detect if the distance d from the nearest ankle point of the person to the lower edge of the door is less than the threshold d0, and if the angle is greater than the specified threshold for n consecutive frames, it is determined that the person is leaning against the door; after the above steps, all people going against the flow can be detected.

[0009] If the key points A and B of the shoulders coincide, the key points C and D of the hips coincide, and the key points E and F of the knees coincide, connect EF to obtain the midpoint I, connect AC and AI, and calculate the angle α between line AC and the vertical direction, and the angle β between line AI and the vertical direction.

[0010] The image frames are acquired from a camera installed at an angle along the direction of the door.

[0011] Preferably, the specific steps for detecting pedestrians in the image in step S2 are as follows:

[0012] S2.1 Read the camera configuration parameters to obtain the door area and the vertical vector of the door;

[0013] S2.2 Read the video stream, obtain image frames, input the image frames into a deep neural network to obtain the positions of all passengers, and use visual methods to simply and efficiently identify pedestrians, thereby improving computational efficiency.

[0014] Preferably, if α or β of n consecutive frames is greater than the set angle value Y, it is determined that the target is leaning against the car door and an alarm is generated. An alarm is only issued when α or β of n consecutive frames is greater than the set angle value Y, thereby improving the accuracy of the warning.

[0015] Preferably, the camera is installed on the outside of the platform screen door and the inside of the subway car door, with the shooting direction being the direction of the subway's travel. Whether waiting for the subway or riding the subway, one can receive a warning when leaning against the door.

[0016] Preferably, the set angle value is 3-8°, which is outside the normal standing posture range of the human body, but can still be used to issue a warning within the detection area.

[0017] A recognition device for leaning against a subway car door or platform screen door includes:

[0018] Acquisition unit: used to acquire image frames of the door area and the vertical direction of the door;

[0019] Detection and calculation unit: Used to detect pedestrians in the image, determine whether the pedestrian is near the door, perform key point detection on pedestrians near the door, obtain key points A and B on the shoulders, key points C and D on the hips, and key points E and F on the knees of the person near the door through key point detection, connect lines AB, CD, and EF to obtain the midpoint G of line AB, the midpoint H of line CD, and the midpoint I of line EF, connect lines GH and GI, and calculate the angle α between line GH and the vertical direction, and the angle β between line GI and the vertical direction;

[0020] Alarm Judgment Unit: Used to determine if a person is leaning against a subway door when the calculated α or β is greater than the set threshold Y, and at the same time the distance d from the nearest ankle point to the bottom edge of the door is less than the threshold d0.

[0021] An electronic device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method described above.

[0022] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0023] This invention can easily and efficiently determine pedestrians leaning against train doors using visual methods, ensuring the safety of both people and vehicles. It prevents door malfunctions caused by passengers leaning against subway doors, helps reduce wear on the bottom slider and friction plate of the door, and also avoids door deformation caused by passengers leaning against the door, which could lead to friction or even jamming when the door is opened, preventing it from opening normally. This reduces the failure rate of the doors, improves their reliability and safety, reduces the impact of human factors, and improves operational efficiency. Attached Figure Description

[0024] The present invention will be described by way of example and with reference to the accompanying drawings, wherein:

[0025] Figure 1 This is a flowchart of the workflow of this invention;

[0026] Figure 2 This is a schematic diagram of a subway scene captured by a camera according to engineering constraints, as described in Embodiment 1 of the present invention.

[0027] Figure 3 This is a schematic diagram of a subway scene captured by a camera, which conforms to engineering constraints according to Embodiment 2 of the present invention.

[0028] Figure 4 This is a schematic diagram of a subway scene camera shooting scenario that does not meet engineering constraints according to the present invention;

[0029] Figure 5This is a schematic diagram of the pedestrian detection result scenario in Embodiment 1 of the present invention;

[0030] Figure 6 This is a result image after filtering passengers who are not in the suspicious area according to Embodiment 1 of the present invention;

[0031] Figure 7 This is a schematic diagram of the key point detection results in Embodiment 1 of the present invention;

[0032] Figure 8 This is a schematic diagram of the result of the judgment vector in Embodiment 1 of the present invention;

[0033] Figure 9 This is a schematic diagram of the result of the judgment vector in Embodiment 2 of the present invention;

[0034] Figure 10 This is a schematic diagram of the identification device for leaning against a subway car door or platform screen door according to the present invention;

[0035] Figure 11 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0036] To further illustrate the technical means and effects of the present invention, the following detailed description is provided in conjunction with the preferred embodiments of the present invention and their accompanying drawings.

[0037] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art.

[0038] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0039] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0040] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily need to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the actual situation.

[0041] Example 1

[0042] like Figure 1 As shown in Figures 2, 5-8, a method for identifying leaning against a subway car door or platform screen door includes the following steps:

[0043] S1. Read the image captured by the camera along the direction of the door, obtain the camera configuration parameters, obtain the door area and the vertical direction vector of the door, and obtain the image frames of the door area and the vertical direction of the door.

[0044] S2. Detect pedestrians in the image and determine if they are near the door. Perform key point detection on pedestrians near the door. When the detected person is leaning to the side, obtain the key points A and B on the shoulders, C and D on the hips, and E and F on the knees of the person near the door through key point detection. Connect lines AB, CD, and EF to obtain the midpoint G of line AB, the midpoint H of line CD, and the midpoint I of line EF. Connect lines GH and GI, and calculate the angle α between line GH and the vertical direction, and the angle β between line GI and the vertical direction. The specific steps for detecting pedestrians in the image are as follows:

[0045] S2.1 Read the camera configuration parameters to obtain the door area and the vertical vector of the door;

[0046] S2.2 Read the video stream, obtain image frames, input the image frames into a deep neural network to obtain the positions of all passengers, and use visual methods to simply and efficiently identify pedestrians, thereby improving computational efficiency;

[0047] The specific steps for calculating the included angles α and β are as follows: First, determine whether all passenger positions in step S2.2 are within the door area, and then perform key point detection on passengers in the door area; that is, let the vertical direction vector be V, and the key point detection locations on the human body be the shoulder points A and B, the hip key points C and D, and the knee key points E and F; where the midpoint of AB is G, the midpoint of CD is H, and the midpoint of EF is I, then the human body offset vector GH is recorded as g, GI is recorded as h, and the included angle α between g and v, and the included angle β between h and v are calculated; people outside the area are filtered out, and key point detection is only performed on people within the area to improve the effective work efficiency;

[0048] S3. When α or β calculated in step S2 is greater than the set threshold Y, and the distance d from the nearest ankle of the person to the lower edge of the door is less than the threshold d0, it is determined that the person is leaning against the subway door and an alarm is issued. Specifically, if α or β of n consecutive frames is greater than the set angle value Y, it is determined that the target is leaning against the door and an alarm is generated. An alarm is only issued when α or β of n consecutive frames is greater than the set angle value Y, thereby improving the accuracy of the warning.

[0049] The working principle of this invention is as follows: read the camera configuration parameters to obtain the door area and the vertical direction vector of the door; read the video stream to obtain image frames, and input the image frames into a deep neural network (such as YOLOv5s) to obtain the positions of all passengers; determine whether each person is in the door area, and perform key point detection on the passengers in the door area; obtain the passenger's body tilt vector; calculate the angle between the body tilt vector and the vertical direction of the door, and determine whether the angle is greater than a specified threshold; if the angle is greater than the specified threshold for n consecutive frames, it is determined that the person is leaning against the door; after the above steps, all people going against the flow can be detected.

[0050] Example 2

[0051] Based on Example 1, such as Figure 3 As shown in Figure 9, when the person being tested is leaning against the door with their back to it, the key points A and B on the shoulders coincide, the key points C and D on the hips coincide, and the key points E and F on the knees coincide. Connect EF to get the midpoint I, connect AC and AI, and calculate the angle α between line AC and the vertical direction, and the angle β between line AI and the vertical direction.

[0052] Example 3

[0053] Based on Example 1, the camera is installed on the outside of the shielding door, and the captured images are as follows: Figure 2 As shown, the camera is installed on the inside of the subway car door, and the images it captures are as follows. Figure 3 As shown, the camera is positioned in the direction of the subway's movement, so whether you are waiting for the subway or riding it, you will be alerted when you lean against the door.

[0054] Example 4

[0055] Based on Example 1, the set included angle value is 3-8°, which exceeds the normal standing posture range of the human body, but can still be warned within the detection area.

[0056] Please see Figure 10 A recognition device for leaning against a subway car door or platform screen door, comprising:

[0057] Acquisition unit 10: used to acquire image frames of the door area and the vertical direction of the door;

[0058] Detection and calculation unit 20: used to detect pedestrians in the picture, determine whether the pedestrian is near the door, perform key point detection on the pedestrian near the door, obtain the key points A and B on the shoulders, key points C and D on the hips, and key points E and F on the knees of the person near the door through key point detection, connect AB, CD, EF to obtain the midpoint G of line AB, the midpoint H of line CD, and the midpoint I of line EF, connect GH and GI, and calculate the angle α between line GH and the vertical direction and the angle β between line GI and the vertical direction;

[0059] Alarm Judgment Unit 30: It is used to determine whether a person is leaning against the subway door when the calculated α or β is greater than the set threshold Y, and at the same time the distance d from the nearest ankle point of the person to the lower edge of the door is less than the threshold d0. Then it is determined that the person is leaning against the subway door and an alarm is issued.

[0060] Please see Figure 11 The present invention also provides an electronic device, including: a memory 100 and a processor 200, wherein the memory 100 stores a computer program, and when the computer program is executed by the processor 200, the processor performs the steps of the above method.

[0061] In summary, this invention determines whether each person is in the door area, performs key point detection on passengers in the door area, obtains the passenger's body tilt vector, calculates the angle between the body tilt vector and the vertical direction of the door, and determines whether the angle is greater than a specified threshold. At the same time, if the distance d from the person's nearest ankle point to the lower edge of the door is less than the threshold d0, it is determined that the target is leaning against the door, and an alarm is generated. This simple and efficient method of determining pedestrians leaning against the door ensures the safety of people and vehicles.

[0062] As described above, those skilled in the art can make various other corresponding changes and modifications based on the technical solutions and concepts of this invention, and all such changes and modifications should fall within the protection scope of the claims of this invention.

Claims

1. A method for identifying leaning against a subway car door or platform screen door, characterized in that, Includes the following steps: S1. Acquire image frames of the door area and the vertical direction of the door; the image frames are acquired from a camera whose installation angle is along the direction of the door; the camera is installed on the outside of the platform screen door and the inside of the subway car door, and the shooting direction is the direction of the subway's travel. S2. Detect pedestrians in the image. This involves reading camera configuration parameters to obtain the door area and the vertical vector of the door; reading the video stream to acquire image frames, and inputting these frames into a deep neural network to obtain the positions of all passengers; determining if a pedestrian is near the door, and performing keypoint detection on pedestrians near the door. Keypoint detection yields the shoulder keypoints A and B, hip keypoints C and D, and knee keypoints E and F. Connecting lines AB, CD, and EF, we obtain the midpoint G of line AB, the midpoint H of line CD, and the midpoint I of line EF. Connecting lines GH and GI, we calculate the angle α between line GH and the vertical direction, and the angle β between line GI and the vertical direction. If shoulder keypoints A and B coincide, hip keypoints C and D coincide, and knee keypoints E and F coincide, connecting EF yields the midpoint I. Connecting lines AC and AI, we calculate the angle α between line AC and the vertical direction, and the angle β between line AI and the vertical direction. S3. When α or β calculated in step S2 is greater than the set threshold Y, and the distance d from the nearest ankle of the person to the lower edge of the door is less than the threshold d0, it is determined that the person is leaning against the subway door and an alarm is issued.

2. A recognition device for leaning against subway car doors or platform screen doors, characterized in that, include: Acquisition Unit: Used to acquire image frames of the door area and the vertical direction of the door; the image frames are acquired from a camera whose installation angle is along the direction of the door; the camera is installed on the outside of the platform screen door and the inside of the subway car door, and the shooting direction is the direction of the subway's travel; The detection and calculation unit is used to detect pedestrians in the image. It reads camera configuration parameters to obtain the door area and the vertical vector of the door; reads the video stream to acquire image frames, and inputs these frames into a deep neural network to obtain the positions of all passengers; it determines whether a pedestrian is near the door, performs keypoint detection on pedestrians near the door, and obtains the shoulder keypoints A and B, hip keypoints C and D, and knee keypoints E and F for people near the door. It connects lines AB, CD, and EF to obtain the midpoint G of line AB, the midpoint H of line CD, and the midpoint I of line EF. It then connects lines GH and GI, and calculates the angle α between line GH and the vertical direction, and the angle β between line GI and the vertical direction. If the shoulder keypoints A and B coincide, the hip keypoints C and D coincide, and the knee keypoints E and F coincide, it connects EF to obtain the midpoint I, connects AC and AI, and calculates the angle α between line AC and the vertical direction, and the angle β between line AI and the vertical direction. Alarm Judgment Unit: Used to determine if a person is leaning against a subway door when the calculated α or β is greater than the set threshold Y, and at the same time the distance d from the nearest ankle point to the bottom edge of the door is less than the threshold d0.

3. An electronic device, characterized in that, include: A memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in claim 1.

Images