Method, system and equipment for detecting invading personnel in track supervision area and medium

Abstract

The invention relates to the technical field of rail traffic safety monitoring, and discloses a method, system and device for detecting invading personnel in a rail supervision area and a medium, and the method comprises the steps: carrying out the intrusion detection of a first supervision area and a second supervision area of a rail traffic station, the first supervision area comprises an area allowing train attendants to arrive at a station for inspection around the train track, and the second supervision area comprises a lingering forbidding area except the first supervision area around the train track; and closing intrusion detection on the first supervision area in response to the existence of the arrival train at the rail transit station and the opening state of the target door of the arrival train. According to the invention, whether to close or start monitoring of the first supervision area is determined by identifying the opening and closing state of the target vehicle door, so that false alarms caused by misjudgment of personnel invasion when a trainman gets off and inspects are avoided, flexible subarea monitoring is realized, false alarms can be avoided, the false alarm rate is reduced, and missing detection can be effectively prevented.

Description

Technical Field

[0001] The present application relates to the technical field of rail transit safety monitoring, and in particular to a method, system, device and medium for detecting persons intruding into a rail transit supervision area. Background Art

[0002] With the rapid development of urban rail transit, subways, high-speed railways, light rails, etc., as convenient public transportation tools, have been widely used in urban travel. However, the safety issues of these rail transit tools are becoming increasingly prominent, especially in the detection of illegal intrusions in the supervision area. For example, when the subway arrives at the station, the behavior of the conductor or train conductor getting off the train to patrol increases the complexity of security detection. If only relying on a single detection method in the existing technology, the conductor or train conductor may be misjudged as a dangerous intruder, resulting in unnecessary security alarms and patrol interference. Summary of the invention

[0003] The main purpose of the present application is to provide a method, system, device and medium for detecting intruders in a track supervision area, aiming to solve the technical problem in the prior art of misjudging train conductors as intruders, resulting in false alarms.

[0004] The first aspect of the present application provides a method for detecting an intruder in a rail transit supervision area, the method for detecting an intruder in a rail transit supervision area comprising: Perform intrusion detection on a first supervision area and a second supervision area of ​​a rail transit station, wherein the first supervision area includes an area around the train track where train crews are allowed to inspect at the station, and the second supervision area includes a prohibited stay area around the train track except the first supervision area; In response to the presence of an arriving train at the rail transit station and the target door of the arriving train being in an open state, intrusion detection on the first supervision area is closed.

[0005] The present application also provides a system for detecting intruders in a rail transit supervision area, the system for detecting intruders in a rail transit supervision area comprising: A detection control module, used to perform intrusion detection on a first supervision area and a second supervision area of ​​a rail transit station, wherein the first supervision area includes an area around the train track where train crews are allowed to inspect at the station, and the second supervision area includes a prohibited staying area around the train track except the first supervision area; The detection control module is also used to close the intrusion detection of the first supervision area in response to the presence of an arriving train at the rail transit station and the target door of the arriving train is in an open state.

[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 method for detecting intruders in a rail transit regulatory area.

[0007] A fourth aspect of the present application provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the above-mentioned method for detecting intruders in a rail transit supervision area.

[0008] This embodiment determines whether to temporarily close the monitoring of the first supervision area by identifying and monitoring the switch status of the target door, thereby avoiding the conductor's misjudgment of human intrusion when getting off the train for inspection, resulting in false alarms; during this period, the second supervision area is always monitored to achieve flexible zoning monitoring, which can avoid false alarms as much as possible and reduce the false alarm rate, and effectively prevent missed detections, thereby improving the intelligence level of rail transit safety monitoring and management; this embodiment can accurately distinguish the stay behavior of staff and non-staff under various conditions of different lighting conditions and changes in the position of the car head, ensure that no false alarms are generated when the staff is patrolling, avoid unnecessary interference, and provide reliable protection for rail transit operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a flow chart of a first embodiment of a method for detecting an intruder in a rail transit supervision area in an embodiment of the present application; Figure 2 A schematic diagram of the positions of the track, the first supervision area and the second supervision area in an embodiment of the present application; Figure 3 A scene diagram of the track, the first supervision area and the second supervision area in an embodiment of the present application; Figure 4 This is a schematic diagram of a normal patrol state of a train conductor in an embodiment of the present application; Figure 5 This is a schematic diagram of simulating an intruder entering the second supervision area in an embodiment of the present application; Figure 6 This is a flow chart of a second embodiment of a method for detecting an intruder in a rail transit supervision area in an embodiment of the present application; Figure 7 This is a functional module diagram of an embodiment of a detection system for intruders in a rail transit supervision area 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, rail transit such as subways has been widely used in urban travel as a convenient public transportation tool. However, the safety issues in the front area of ​​subways are becoming increasingly prominent, especially in the timely detection of people climbing over the screen door. When the subway arrives at the station, the driver's behavior of getting off the train to patrol increases the complexity of security detection. If only a single detection method is relied upon, the driver may be misjudged as a dangerous intruder, resulting in unnecessary security alarms and patrol interference.

[0012] Based on this, the present application provides a solution for detecting intruders in rail transit supervision areas.

[0013] refer to Figure 1 In one embodiment, the present application provides a method for detecting an intruder in a rail transit supervision area, and the method for detecting an intruder in a rail transit supervision area includes: S100: Perform intrusion detection on a first supervision area and a second supervision area of ​​a rail transit station, wherein the first supervision area includes an area around the train track where train crew are allowed to inspect at the station, and the second supervision area includes a prohibited staying area around the train track except the first supervision area.

[0014] Specifically, the rail transit in this embodiment includes public transportation such as subways, high-speed railways, light rails, trains, and motor vehicles, but is not limited to these.

[0015] The supervision area is the key supervision area of ​​the rail transit station. For example, there are the first supervision area and the second supervision area in the front area. The first supervision area is an area where train conductors (or conductors) are allowed to get off the train to patrol, inspect, and direct, and it is strictly prohibited for passengers and other illegal persons to stay. The second supervision area is an area where any person, especially passengers, is generally prohibited from staying.

[0016] The first supervision area and the second supervision area are both adjacent to the track and are used by station staff. Normally, passengers and other illegal persons are not allowed to stay in the first supervision area and the second supervision area.

[0017] When there is no train arriving at the rail transit station, both the first supervision area and the second supervision area need to be monitored. Therefore, it is necessary to enable the intrusion detection function of the first supervision area and the second supervision area of ​​the rail transit station at the same time.

[0018] Figure 2 Schematic diagram of the positions of the track, the first supervision area and the second supervision area in the embodiment of the present application; Figure 2 The first supervision area and the second supervision area are the areas beside the track. Under normal circumstances, the conductor can stand in the first supervision area when getting off the train for patrol.

[0019] Among them, the intrusion detection function of the first supervision area and the second supervision area of ​​the rail transit station can be realized through computer vision technology.

[0020] If a person intrusion is detected in at least one of the first supervision area and the second supervision area when no train arrives at the station, an alarm will be issued.

[0021] Taking the subway as an example, the first supervision area and the second supervision area can be the area between the track and the screen door at the front of the train. If someone climbs over the screen door to enter this area, it may be dangerous. Therefore, it is necessary to monitor whether there is illegal intrusion in this area.

[0022] S200: In response to the presence of an arriving train at the rail transit station and the target door of the arriving train being in an open state, closing the intrusion detection on the first supervision area.

[0023] Specifically, the target door is the door of the carriage where the conductor or the driver gets on and off the train. If it is detected that there is an arriving train at the rail transit station, and the target door of the arriving train is in the open state, it means that the conductor or the driver may get off the train for inspection, patrol or command, etc. The conductor or the driver may stay in the first supervision area. Therefore, in order to prevent misjudgment and false alarm, in this case, the intrusion detection function of the first supervision area will be turned off, while the intrusion detection function of the second supervision area will be maintained.

[0024] This embodiment determines whether to temporarily close the monitoring of the first supervision area by identifying and monitoring the switch status of the target door, thereby avoiding the conductor's misjudgment of human intrusion when getting off the train for inspection, resulting in false alarms; during this period, the second supervision area is always monitored to achieve flexible zoning monitoring, which can avoid false alarms as much as possible and reduce the false alarm rate, and effectively prevent missed detections, thereby improving the intelligence level of rail transit safety monitoring and management; this embodiment can accurately distinguish the stay behavior of staff and non-staff under various conditions of different lighting conditions and changes in the position of the car head, ensure that no false alarms are generated when the staff is patrolling, avoid unnecessary interference, and provide reliable protection for rail transit operations.

[0025] In one embodiment, in response to the presence of an arriving train at the rail transit station and the target door of the arriving train being in an open state in step S200, after closing the intrusion detection of the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: In response to the target door changing from an open door state to a closed door state, the intrusion detection on the first supervision area is restarted.

[0026] Specifically, if the target door is monitored to be in a closed state, it means that the conductor or the conductor has boarded the train. Under normal circumstances, the conductor or the conductor will not continue to stay in the first supervision area. Therefore, it is necessary to reopen the intrusion detection function for the first supervision area. In this case, the intrusion detection functions for both the first supervision area and the second supervision area are already turned on.

[0027] After the intrusion detection function for the first supervision area is turned on, if a human intrusion is detected in the first supervision area again, it means that the first supervision area has determined that there is an illegal intrusion by a person other than the train attendant or the conductor, and an alarm needs to be issued, and no misjudgment will occur.

[0028] This embodiment determines whether to temporarily close the monitoring of the first supervision area by identifying and monitoring the switch status of the target door, thereby avoiding the conductor from misjudging the intrusion of people when getting off the train for inspection, resulting in false alarms; this embodiment can promptly restart the monitoring of the first supervision area after the target door is closed again. During this period, the second supervision area is always monitored, realizing flexible zoning monitoring, which can avoid false alarms as much as possible and reduce the false alarm rate, and effectively prevent missed detections, thereby improving the intelligent level of rail transit safety monitoring and management; this embodiment can accurately distinguish the stay behavior of staff and non-staff under various conditions of different lighting conditions and changes in the position of the car head, ensuring that no false alarms are generated when the staff is patrolling, avoiding unnecessary interference, and providing reliable protection for rail transit operations.

[0029] In one embodiment, the method for detecting an intruder in a rail transit supervision area further includes: In response to the target door being in an open state and a person intruding into the second supervision area, a first warning is issued; and / or, In response to the presence of an intruder in at least one of the first supervision area and the second supervision area, a second early warning is issued.

[0030] Specifically, when the target door is in the door open state, if human intrusion into the second supervision area is detected, a first warning will be issued.

[0031] Computer vision technology can be used to detect intrusion in the second supervision area. More specifically, a camera is used to take pictures of the area including the supervision area to obtain pictures, or a video is taken of the area including the supervision area to obtain a video.

[0032] The target detection model is used to perform target detection and classification on the captured image, locate the second supervision area in the captured image, and determine whether there is human intrusion or stay in the second supervision area.

[0033] or, Video frames (frame images) are captured from the camera video, and the target detection model is used to perform target detection and classification on the video frames, locate the second supervision area in the video frames, and determine whether there is human intrusion or stay in the second supervision area.

[0034] The video frames captured from the camera video may be captured frame by frame or at intervals according to a preset time interval, and this application does not limit this. Interval capture at preset time intervals can ensure the diversity of pictures, reduce duplicate pictures, and avoid pictures that are too repetitive and dense.

[0035] If human intrusion into the second supervision area is detected, a first warning is issued.

[0036] If no human intrusion is detected in the second supervision area, there is no need to issue a first warning.

[0037] Among them, the first warning may include but is not limited to at least one of the following: turning on the alarm system of the rail transit station to sound an alarm to instruct the staff to deal with the human intrusion in a timely manner and clear out the intruders; displaying the target image on the monitoring system, selecting the intruder and / or flashing red and orange alternately on the target image as a warning, wherein the target image is a photograph or video frame in which a human intrusion is detected in the second supervision area; displaying the camera video to the staff and flashing red and orange alternately in the camera video as a warning, etc.

[0038] After the target door is in the door-closed state, the monitoring system will perform intrusion detection on the first supervision area and the second supervision area. If there is a human intrusion in at least one of the two supervision areas, a second warning will be issued.

[0039] At this time, the second warning will not misjudge the conductor or train attendant as an intruder, thus avoiding misjudgment.

[0040] Computer vision technology can be used to detect intrusions in the first supervision area and the second supervision area. More specifically, a camera is used to take pictures of the area including the supervision area to obtain pictures, or a video is taken of the area including the supervision area to obtain a video.

[0041] The target detection model is used to perform target detection and classification on the captured images, locate the first supervision area and the second supervision area in the captured images, and determine whether there is human intrusion or stay in the first supervision area and the second supervision area.

[0042] or, Video frames (frame images) are captured from the camera video, and the target detection model is used to perform target detection and classification on the video frames, locate the first supervision area and the second supervision area in the video frames, and determine whether there is human intrusion or stay in the first supervision area and the second supervision area.

[0043] The video frames captured from the camera video may be captured frame by frame or at intervals according to a preset time interval, and this application does not impose any restrictions on this.

[0044] If human intrusion is detected in the first supervision area and / or the second supervision area, a second warning is issued.

[0045] If no human intrusion is detected in the first supervision area and the second supervision area, there is no need to issue a second warning.

[0046] Among them, the second warning may include but is not limited to at least one of the following: turning on the alarm system of the rail transit station to sound an alarm to instruct the staff to deal with the human intrusion in a timely manner and clear out the intruders; displaying the target image on the monitoring system, selecting the intruder and / or flashing red and orange alternately on the target image as a warning, wherein the target image is a photographed image or video frame in which a human intrusion is detected in the first supervision area or the second supervision area; displaying the camera video to the staff and flashing red and orange alternately in the camera video as a warning, etc.

[0047] This embodiment determines whether to temporarily close the monitoring of the first supervision area by identifying and monitoring the switch status of the target door, thereby avoiding the conductor from misjudging the intrusion of personnel when getting off the train for inspection, resulting in false alarms; during this period, the second supervision area is always monitored, and flexible zoning monitoring is achieved, which can avoid false alarms as much as possible, reduce the false alarm rate, and effectively prevent missed detections, thereby improving the intelligent level of rail transit safety monitoring and management; this embodiment can accurately distinguish the stay behavior of staff and non-staff under various conditions of different lighting conditions and changes in the position of the head of the car, ensuring that no false alarms are generated when the staff patrols, avoiding unnecessary interference, and providing reliable protection for rail transit operations. This embodiment can provide timely warnings for the second supervision area where there is human intrusion, or timely warnings for the first supervision area and the second supervision area where there is human intrusion during different supervision periods.

[0048] Taking the subway as an example of rail transit, intelligent detection is performed to see whether there are non-captains or non-staff members climbing over the screen door in the front area of ​​the subway. To achieve this goal, it can be judged by monitoring whether the human body detection center point is located in the set intrusion area. In order to effectively reduce the area where the conductor gets off the train for patrol, the intrusion area is divided into two parts: area 1 and area 2. Among them, area 1 is the area that the conductor is allowed to enter. The system is designed so that the conductor is allowed to enter area 1 without triggering an alarm only when the subway arrives at the station and the door is open, which can shorten the time for closing the detection of area 1. The specific detection opening and closing logic of area 1 and area 2 is described in the above method. Through this division, the system can more accurately monitor the intrusion behavior of non-captains, thereby ensuring that no false alarms are generated during the normal patrol of the conductor and avoiding unnecessary interference. In addition, the subway arrival detection and the front door opening and closing detection can be simplified to a comprehensive judgment of the door status of the conductor. Through this solution, the safety monitoring capability of the front area of ​​the subway can be significantly improved, the false alarm rate can be reduced, and the stability and safety of operation can be ensured.

[0049] Figure 3 A scene diagram of the track, the first supervision area and the second supervision area in an embodiment of the present application; Figure 4 This is a schematic diagram of a normal patrol state of a train conductor in an embodiment of the present application; Figure 5 This is a schematic diagram of simulating an intruder entering the second supervision area in an embodiment of the present application; Figure 3-Figure 5 It can be seen that under normal circumstances, the conductor door is open, and the conductor or conductor can temporarily stay in the first supervision area, and the monitoring of the first supervision area is turned off. In the case of no train arriving, both the first supervision area and the second supervision area should be monitored, and no one should stay in the second supervision area or the first supervision area.

[0050] The method for detecting intruders in the rail transit supervision area of ​​the present embodiment can be applied to, for example, detecting people climbing over the screen door of a subway train head. It can accurately distinguish the intrusion behaviors of train conductors or drivers from those of non-staff members under different lighting conditions and changes in the position of the train head, thereby improving the intelligent level of subway safety management and providing more reliable protection for operations.

[0051] The method for detecting intruders in the rail transit supervision area of ​​the present application can be applied to urban rail transit systems, especially in the field of security monitoring of subway operations. Its technology can be widely used for intrusion detection in the front area of ​​subways to ensure that unauthorized persons cannot easily climb over the screen door and protect the safety of passengers and staff. In addition, this method is also applicable to the security management of other public transportation vehicles, such as light rail and high-speed rail, and the monitoring and management of large public places, and has a wide range of applications.

[0052] In one embodiment, the method for detecting intruders in the rail transit supervision area also includes: if the arrival of a train is monitored, monitoring the door switch status of the target door of the arriving train, wherein the door switch status is the door open state or the door closed state.

[0053] Specifically, if the arriving train includes a conductor door and a passenger door, the target door is the conductor door.

[0054] If all doors of the arriving train are of the same type, or the conductor (carriage conductor) and passengers can be in the same carriage, the conductor or carriage conductor may get off from any door, and the target door can be any door.

[0055] When a train arrives at a station, the arriving train can automatically open the carriage doors. Therefore, the arriving train can send the door switch status to the monitoring system by communication when or after opening the doors.

[0056] or, The monitoring system can monitor the opening and closing status of the target doors of the arriving train through computer vision technology.

[0057] In one embodiment, before shutting down the intrusion detection on the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: The video frames captured from the camera video data are input into the trained target detection model, and the target door is detected by the trained target detection model; The door switch state of the target door is determined according to the obtained target detection result, wherein the door switch state includes a door open state and a door closed state.

[0058] Specifically, the method for detecting intruders in a rail transit supervision area of ​​the present embodiment is an intelligent method for detecting intruders in a rail transit supervision area based on computer vision.

[0059] The first supervision area and the second supervision area are both located beside the train track of the rail transit station. Therefore, a camera can be arranged near the track. If a train arrives at the station, the camera can capture a first captured image or a first camera video including a target door of the arriving train. In addition, the first camera video data can also be a camera video captured after the train arrives at the station.

[0060] Of course, in a specific embodiment, the camera shooting angle can also be set so that the camera can capture the train track (which can capture the target door of the arriving train), the first supervision area and the second supervision area in the same video. That is, the first camera video can also include the first supervision area and the second supervision area, etc., and this application does not limit this. This can reduce the use and waste of cameras, while ensuring that the detection of the arriving train, the target door, and the supervision area has time consistency.

[0061] In order to prevent missing shots, the present embodiment preferably captures the camera video and transmits the camera video as camera video data to the computer device of the monitoring system.

[0062] The computer device may capture video frames from the camera video data frame by frame, or may capture video frames from the camera video data at intervals according to a predetermined time interval, which is not limited in this embodiment.

[0063] The video frames captured from the camera video data are input into the trained object detection model in the order of shooting. The trained object detection model can realize the recognition and detection of target doors in different opening and closing states of the train. For example, it can recognize the target door in the door open state and the target door in the door closed state, both of which are targets to be detected.

[0064] In addition, if the target door is a captain's door, which is different from a passenger door, the target detection model also has the function of identifying and distinguishing the target door from the passenger door.

[0065] The trained target detection model detects or captures the target door in the door-open state and the target door in the door-closed state in the video frame, frames the target area or the candidate area containing the target, and classifies the target area or the candidate area to obtain the target detection result, wherein the target detection result includes the target type and calibration coordinates of the detected target. The target detection result indicates the target type of the target detected in the video frame, and the target type is the target door in the door-open state or the target door in the door-closed state. Therefore, the door switch state of the target door can be determined according to the target detection result.

[0066] In a specific embodiment, the target detection model may be constructed based on a YOLO (You Only Look Once) series of detection algorithms, such as YOLO V1, YOLO V5, YOLO V7, or YOLO V8.

[0067] Taking the subway as an example, when the subway arrives at the station, the system will analyze the video captured by the on-site camera frame by frame to detect the status of the conductor door. By analyzing the detection results, the system can accurately determine the arrival of the subway and the open and close status of the doors. When the conductor door is detected, it means that the subway has arrived; otherwise, it means that the subway has not arrived. When the subway arrives at the station, the system will continue to monitor the opening status of the doors. If the conductor door is open, the system will allow station staff to patrol and suspend the intrusion detection function for area 1 (the first supervision area) according to the process design. The system will then continue to monitor the status of the conductor door until the door is closed again, at which time the intrusion detection of area 1 will be restarted.

[0068] This embodiment can accurately detect target doors in different switch states through the target detection model, thereby improving the efficiency and reliability of overall safety management.

[0069] In one embodiment, in step S200, in response to the presence of an arriving train at the rail transit station and the target door of the arriving train is in an open state, the intrusion detection on the first supervision area is turned off, including: in response to the presence of an arriving train at the rail transit station and the target doors are monitored in an open state from a first number of consecutive first video frames, the intrusion detection on the first supervision area is turned off.

[0070] Specifically, the camera video contains multiple video frames. In order to prevent the false detection of the door switch state of the target door, several video frames can be sequentially input into the target detection model according to the sequence of the timestamps of the video frames. If the target detection results of the first video frames indicate that the target door is in the door open state after the first number of first video frames are continuously input into the target detection model, it is determined to close the intrusion detection function of the first supervision area. This ensures the accurate determination of the door switch state of the target door and prevents false judgments, and also prevents the false closing of the intrusion detection function of the first supervision area, resulting in missed detections.

[0071] Through the above measures, this application can ensure accurate monitoring of the status of subway car doors and improve the efficiency and reliability of overall safety management.

[0072] This embodiment uses continuous detection to confirm that the target door is open or closed when a preset number is reached, thereby effectively reducing the false alarm rate.

[0073] In one embodiment, before shutting down the intrusion detection on the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: The video frames captured from the camera video data are input into the trained target detection model, and the target door is detected by the trained target detection model; Determine the door switch state of the target door according to the obtained target detection result; In response to the target door changing from an open state to a closed state, the intrusion detection for the first supervision area is reopened, including: if it is monitored from a second number of consecutive second video frames that the target doors are all in a closed state, the intrusion detection function for the first supervision area is reopened, wherein the time point of the second video frame is later than that of the first video frame.

[0074] Specifically, the video frames are input into the target detection model in the order of timestamps. Therefore, under normal circumstances, the target door will switch from the door closed state (for example, when just entering the station) to the door open state (midway) and finally to the door closed state (for example, when leaving the station).

[0075] Input a number of video frames into the target detection model in sequence. If, after continuously inputting a second number of second video frames into the first target detection model, the target detection results of these second video frames all indicate that the target door is in the door closed state, and the door is switched from the door open state to the door closed state, it is determined to reopen the intrusion detection function for the first supervision area. This not only ensures accurate determination of the door switch state of the target door and prevents misjudgment, but also prevents the intrusion detection function for the first supervision area from being mistakenly turned on when the conductor or the driver gets off the train for inspection or before getting on the train, resulting in false alarms.

[0076] The first quantity and the second quantity are configured according to actual application scenarios, and this embodiment does not limit this.

[0077] Through the above measures, this application can ensure accurate monitoring of the status of subway car doors and improve the efficiency and reliability of overall safety management.

[0078] This embodiment uses continuous detection to confirm that the target door is open or closed when a preset number is reached, thereby effectively reducing the false alarm rate.

[0079] In one embodiment, the method for detecting intruders in a rail transit supervision area further includes: monitoring whether there are arriving trains at the rail transit station.

[0080] Specifically, the method for detecting intruders in the rail transit supervision area of ​​this embodiment can be applied to the monitoring system of rail transit stations. Whether a train has arrived at the station can be determined by communicating with the arriving train. More specifically, the arriving train will send a station arrival message to the monitoring system of the rail transit station when entering the station or stopping at the station, so that the monitoring system can determine the arrival of the train in time.

[0081] Of course, the monitoring system can also monitor whether there are arriving trains or trains entering the station on the track through computer vision technology, thereby determining whether there are arriving trains on the track of the rail transit station.

[0082] It should be noted that in different application scenarios, the arriving train may be a subway, high-speed rail, EMU, light rail, etc. This application does not impose any restrictions on this.

[0083] In one embodiment, before shutting down the intrusion detection on the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: The video frames captured from the camera video data are input into the trained target detection model, and the target door is detected by the trained target detection model. If the target door is detected from a third number of consecutive third video frames by the trained target detection model, it is determined that there is an arriving train at the rail transit station; or, Before closing the intrusion detection on the first supervision area, the method for detecting the intruder in the rail transit supervision area further includes: The video frames captured from the camera video data are input into the trained object detection model, and the door detection is performed through the trained object detection model. If the carriage door is detected from a fourth number of fourth consecutive video frames by the trained target detection model, it is determined that there is an arriving train at the rail transit station.

[0084] Specifically, the camera can record the rail transit station in real time to obtain a video stream. Therefore, the first camera video data can include videos before and after the train enters the station. Based on this, a third video frame is intercepted from the first camera video data, wherein the third video frame is earlier than the first video frame. Multiple third video frames are input into the target detection model for target detection. If the target door (specifically the target door in the closed door state) is detected based on the third number of consecutive third video frames, it can be accurately determined that there is an arriving train at the rail transit station.

[0085] This embodiment can detect the target door and the switch state of the target door by using the same target detection model, which reduces the complexity of model training and improves the target detection efficiency. In addition, this embodiment determines that the train has arrived at the station after the target door is detected according to the third number of consecutive third video frames, which reduces the misjudgment rate and improves the accuracy of arrival detection.

[0086] In another embodiment, the vehicle compartment doors include target doors and other non-target doors.

[0087] If the train includes multiple compartment doors such as the conductor door and the passenger door, other passenger doors may also be monitored to determine whether the train has arrived at the station, without being limited to the target door detection.

[0088] In addition to detecting the target door in the door-open state and the target door in the door-closed state, the target detection model of this embodiment can also detect other compartment doors, such as passenger doors, etc. Based on this, a fourth video frame is intercepted from the camera video data, wherein the fourth video frame is earlier than the first video frame. Multiple fourth video frames are input into the target detection model for target detection. If the compartment door is detected according to the fourth number of consecutive fourth video frames, it can be accurately determined that there is an arriving train at the rail transit station.

[0089] This embodiment can more accurately determine whether a train has arrived at a station by detecting the carriage doors, thereby further improving the accuracy of train arrival detection.

[0090] Taking the subway as an example, in this process, the door switch status may be misdetected. To solve this problem, this application introduces a continuous detection mechanism: only when ten frames of the train door are detected continuously, it is determined that the subway has arrived at the station; similarly, only when ten frames of the train door are detected continuously, it is determined that the train door is open, thereby effectively reducing the false alarm rate. In addition, in order to avoid misjudging the structure of the passenger door with a similar shape between carriages as the train door during the train running, the discrimination logic of the carriage category is added to the target detection algorithm to improve the accuracy and robustness of the detection.

[0091] Through the above measures, this application can ensure accurate monitoring of the status of subway car doors and improve the efficiency and reliability of overall safety management.

[0092] In one embodiment, in response to the target door being in an open state and a person intruding into the second supervision area, a first warning is issued, including: Inputting the video frames captured from the camera video data into the trained target detection model, and performing intruder target detection on the second supervision area through the trained target detection model; If it is determined according to the obtained target detection result that an intruder is detected in the second supervision area, a first warning is issued.

[0093] Specifically, the first supervision area and the second supervision area are both beside the train track of the rail transit station, so a camera can be deployed near the track, and pictures or videos including the first supervision area and the second supervision area can be taken by the camera.

[0094] Of course, in a specific embodiment, the camera shooting angle can also be set so that the camera can capture the train track (which can capture the target door of the arriving train), the first supervision area and the second supervision area in the same video. That is, the camera video can also include the arriving train, etc., and this application does not limit this. This can reduce the use and waste of cameras, while ensuring that the detection of the arriving train, the target door, and the supervision area has time consistency.

[0095] In order to prevent missing shots, the present embodiment preferably captures the camera video and transmits the camera video as camera video data to the computer device of the monitoring system.

[0096] The computer device may capture video frames from the camera video data frame by frame, or may capture video frames from the camera video data at intervals according to a predetermined time interval, which is not limited in this embodiment.

[0097] The video frames captured from the camera video data are input into the trained target detection model in the order of shooting. The trained target detection model can realize the recognition and detection of the second supervision area without human intrusion and the second supervision area with human intrusion. That is, the second supervision areas in the above two states are both targets to be detected.

[0098] The trained target detection model detects or captures targets in the second supervision area without human intrusion and the second supervision area with human intrusion in the video frame, frames the target area or the candidate area containing the target, and classifies the target area or the candidate area to obtain the target detection result, wherein the target detection result includes the target type and calibration coordinates of the detected target. The target detection result indicates the target type of the target detected in the video frame, and the target type is the second supervision area without human intrusion or the second supervision area with human intrusion. Therefore, according to the target detection result, it can be determined whether there is human intrusion in the second supervision area.

[0099] In a specific embodiment, the target detection model may be constructed based on a YOLO (You Only Look Once) series of detection algorithms, such as YOLO V1, YOLO V5, YOLO V7, or YOLO V8.

[0100] In a specific embodiment, the target detection model has the function of detecting target doors with different door switch states, and also has the function of detecting whether there is a person intruding in the second supervision area.

[0101] In a specific embodiment, the trained target detection model can realize target detection in the first supervision area without human intrusion and the first supervision area with human intrusion.

[0102] Turning off the intrusion detection function for the first supervision area may be turning off the output function of the trained target detection model for the second target detection result of the first supervision area, or turning off the intrusion detection function for the first supervision area may be shielding or discarding the target detection result output by the target detection model for the first supervision area, or suspending the execution of the code for intrusion detection or alarm for the first supervision area. In this way, the monitoring system will not obtain the target detection result corresponding to the first supervision area. Therefore, when the target door is in the door open state, the train conductor or the conductor will not trigger the alarm function even if he stays in the first supervision area.

[0103] Through the target detection model, this embodiment can detect intruders only in the second supervision area during the opening of the target door, and issue timely warnings to prevent the driver or conductor from being misjudged as an intruder, thereby improving the accuracy and reliability of illegal intrusion detection in rail transit and the efficiency and reliability of overall safety management.

[0104] In one embodiment, in response to the presence of an intruder in at least one of the first supervision area and the second supervision area, a second warning is issued, including: Inputting the video frames captured from the camera video data into the trained target detection model, and performing intruder target detection on the first supervision area and the second supervision area through the trained target detection model; If it is determined according to the obtained target detection result that there is an intruder entering from the non-supervised area into at least one of the first supervised area and the second supervised area, a second warning is issued.

[0105] Specifically, the first supervision area and the second supervision area are both beside the train track of the rail transit station, so a camera can be deployed near the track, and pictures or videos including the first supervision area and the second supervision area can be taken by the camera.

[0106] Of course, in a specific embodiment, the camera shooting angle can also be set so that the camera can capture the train track (which can capture the target door of the arriving train), the first supervision area and the second supervision area in the same video. That is, the camera video can also include the arriving train, etc., and this application does not limit this. This can reduce the use and waste of cameras, while ensuring that the detection of the arriving train, the target door, and the supervision area has time consistency.

[0107] In order to prevent missing shots, the present embodiment preferably captures the camera video and transmits the camera video as the third camera video data to the computer device of the monitoring system.

[0108] The computer device may capture video frames from the camera video data frame by frame, or may capture video frames from the camera video data at intervals according to a predetermined time interval, which is not limited in this embodiment.

[0109] The video frames captured from the camera video data are input into the trained target detection model in the order of shooting. The trained target detection model can realize the recognition and detection of the first supervision area without human intrusion, the first supervision area with human intrusion, the second supervision area without human intrusion, and the second supervision area with human intrusion. That is, the supervision areas in the above four states are all targets to be detected.

[0110] The trained target detection model detects or captures targets in the first supervision area without human intrusion, the first supervision area with human intrusion, the second supervision area without human intrusion, and the second supervision area with human intrusion in the video frame, frames the target area or the candidate area containing the target, and classifies the target area or the candidate area to obtain the target detection result, wherein the target detection result includes the target type and calibration coordinates of the detected target. The target detection result indicates the target type of the target detected in the video frame, and the target type is the second supervision area without human intrusion or the second supervision area with human intrusion or the first supervision area without human intrusion or the first supervision area with human intrusion. Therefore, according to the target detection result, it can be determined whether there is human intrusion in the first supervision area and the second supervision area.

[0111] In a specific embodiment, within the supervision area, the system will perform intelligent personnel detection on each frame to ensure timely detection of intrusion by unauthorized personnel. When the system detects that someone has entered the supervision area, an alarm will be immediately sounded (for example, the area in the video will flash alternately in red and orange). In order to avoid the possibility of misjudging the rails as human bodies, resulting in false intrusion warnings. This embodiment adds a judgment mechanism based on detection: the system will only trigger an intrusion warning when it detects that a human body enters the supervision area from outside the non-supervision area (for example, climbing over the screen door to enter the supervision area or exiting the train to enter the supervision area, etc.). This measure effectively reduces the probability of false alarms and ensures the safety and efficiency of the system.

[0112] In a specific embodiment, the target detection model may be constructed based on a YOLO (You Only Look Once) series of detection algorithms, such as YOLO V1, YOLO V5, YOLO V7, or YOLO V8.

[0113] In a specific embodiment, the target detection model has the function of simultaneously detecting whether there is human intrusion in the first supervision area and the second supervision area.

[0114] In a specific embodiment, the target detection model has the function of detecting target doors with different door switch states, and also has the function of simultaneously detecting whether there is human intrusion in the first supervision area and the second supervision area.

[0115] Turning off the intrusion detection function for the first supervision area may be turning off the output function of the trained target detection model for the target detection results of the first supervision area, or turning off the intrusion detection function for the first supervision area may be shielding or discarding the target detection results output by the target detection model for the first supervision area, or suspending the execution of the code for intrusion detection or alarm for the first supervision area, etc. This application does not impose any restrictions on this. In this way, the monitoring system will not obtain the target detection results corresponding to the first supervision area. Therefore, when the target door is in the door open state, the conductor or the driver will not trigger the alarm function even if he stays in the first supervision area.

[0116] Similarly, enabling the intrusion detection function for the first supervision area may be enabling the output function of the trained target detection model for the target detection result of the first supervision area, or enabling the intrusion detection function for the first supervision area may be transmitting the target detection result output by the target detection model for the target detection of the first supervision area to the computer device or the relevant module of the monitoring system. In this way, the monitoring system will obtain the target detection result corresponding to the first supervision area, so after the target door is in the door closed state, if there is a person staying in the first supervision area, the alarm function will be triggered.

[0117] In a specific embodiment, the same camera can be used to record videos of different time periods. The camera can simultaneously record the track or the train on the track, the target door of the train, the first supervision area and the second supervision area, so that the target door, the first supervision area and the second supervision area can exist simultaneously in the captured video frame.

[0118] This embodiment uses the target detection model to detect intruders in the first supervision area and the second supervision area at the same time after the target door is closed, and issue a timely warning.

[0119] In one embodiment, the detection of intruders in the rail transit supervision area further includes: Acquire first sample data, wherein the first sample data includes a frame image of an arriving train captured from a sample video, wherein a target door in an open state and a target door in a closed state are framed in the frame image of the arriving train; The first sample data is used to train the first target detection model to be trained to obtain a trained first target detection model.

[0120] Specifically, the first sample data includes multiple frames of arriving trains, and each frame of the arriving train contains frames and category labels for target doors in the door-open state and the door-closed state. Therefore, the first sample data can be used to train the first target detection model to detect target doors in different states.

[0121] In addition, the performance of the trained first target detection model can also be tested through the test set.

[0122] In one embodiment, second sample data is obtained, wherein the second sample data includes a frame image of the second supervision area captured from the sample video, and the second supervision area with human invasion and the second supervision area without human invasion are framed in the frame image of the second supervision area; The second sample data is used to train the second target detection model to be trained to obtain a trained second target detection model.

[0123] In one embodiment, third sample data is obtained, wherein the third sample data includes a first supervision area frame image and a second supervision area frame image captured from a sample video, wherein the first supervision area with human intrusion and the first supervision area without human intrusion are framed and labeled by category in the first supervision area frame image, and the second supervision area with human intrusion and the second supervision area without human intrusion are framed and labeled by category in the second supervision area frame image; The third sample data is used to train the third object detection model to obtain a trained third object detection model.

[0124] It should be noted that the above-mentioned first target detection model, second target detection model and third target detection model can be the same target detection model, which also has the functions of arrival detection, target door detection, target door switch status detection, whether there is human intrusion detection in the first supervision area, and whether there is human intrusion detection in the second supervision area.

[0125] Figure 6 This is a flow chart of a second embodiment of the method for detecting intruders in a rail transit supervision area in the embodiment of the present application; Figure 6 In this embodiment, taking the subway as an example, the camera video data is taken, and the camera video data is intercepted frame by frame. It is detected whether there is a conductor door (target door) according to the intercepted frame image. If the conductor door is detected, it means that there is an arriving train on the track, that is, the subway has arrived. In the case of the subway arriving, the detection of area 1 (the first supervision area) and the detection of area 2 (the second supervision area) are turned on. Continue to detect whether the conductor door is open according to the intercepted frame image. If the conductor door is open, it is determined that the subway door is open, and the detection of area 1 is closed, and the detection of area 2 is maintained. If it is detected that the conductor door is not open or closed, it is determined that the subway door is closed, and the detection of area 1 is reopened and the detection of area 2 is maintained.

[0126] If the train door is not detected, it means that there is no arriving train on the track, that is, the subway has not arrived. In the case of no subway arriving, this embodiment also starts detection in area 1 (the first supervision area) and detection in area 2 (the second supervision area).

[0127] In this embodiment, regardless of whether the subway arrives at the station or not, the intruder detection in area 1 and area 2 is enabled. When the subway arrives at the station and the conductor door is open, the intruder detection in area 1 is disabled and the intruder detection in area 2 is retained. When the conductor door is closed, the detection in (the first supervision area) and the detection in area 2 (the second supervision area) are enabled.

[0128] This application significantly improves the safety monitoring effect of rail transit, including the front area of ​​subway cars, by introducing advanced computer vision technology and intelligent detection mechanisms. Under different lighting conditions, the system can accurately identify the intrusion behavior of unauthorized personnel, effectively reduce the false alarm rate, and ensure the accurate triggering of alarms. The flexible area division design enables monitoring to dynamically adapt to the patrol of the driver and environmental changes, further enhancing the robustness of the system. In addition, the discrimination logic of the car category in the target detection algorithm ensures the accurate identification of similar structures and improves the stability of monitoring. Overall, this application effectively improves the safety and management efficiency of rail transit operations, and provides a solid guarantee for the safety of passengers and vehicles.

[0129] In one embodiment, the present application further provides a system for detecting intruders in a rail transit regulatory area, the system for detecting intruders in a rail transit regulatory area comprising: A detection control module, used to perform intrusion detection on a first supervision area and a second supervision area of ​​a rail transit station, wherein the first supervision area includes an area around the train track where train crews are allowed to inspect at the station, and the second supervision area includes a prohibited staying area around the train track except the first supervision area; The detection control module is also used to close the intrusion detection of the first supervision area in response to the presence of an arriving train at the rail transit station and the target door of the arriving train is in an open state.

[0130] In one embodiment, the detection control module is further configured to restart the intrusion detection on the first supervision area in response to the target door changing from an open door state to a closed door state.

[0131] In one embodiment, the rail transit supervision area intruder detection system further includes: A first intrusion detection module, configured to issue a first warning in response to the target door being in an open state and a person intruding into the second supervision area; and / or, The second intrusion detection module is used to issue a second warning in response to the presence of an intruder in at least one of the first supervision area and the second supervision area.

[0132] In one embodiment, the rail transit supervision area intruder detection system further includes: a door status detection module, which is used to: The video frames captured from the camera video data are input into the trained target detection model, and the target door is detected by the trained target detection model; The door switch state of the target door is determined according to the obtained target detection result, wherein the door switch state includes a door open state and a door closed state.

[0133] In one embodiment, the detection control module is specifically used to: in response to the presence of an arriving train at a rail transit station and the detection of target doors being in an open state from a first number of consecutive first video frames, shut down intrusion detection on the first supervision area.

[0134] In one embodiment, the rail transit supervision area intruder detection system further includes: an arrival detection module; Among them, the arrival detection module is used to: The video frames captured from the camera video data are input into the trained target detection model, and the target door is detected by the trained target detection model. If the target door is detected from a third number of consecutive third video frames by the trained target detection model, it is determined that there is an arriving train at the rail transit station; or, Arrival detection module, used for: The video frames captured from the camera video data are input into the trained object detection model, and the door detection is performed through the trained object detection model. If the carriage door is detected from a fourth number of fourth consecutive video frames by the trained target detection model, it is determined that there is an arriving train at the rail transit station.

[0135] In one embodiment, the second intrusion detection module is specifically configured to: Inputting the video frames captured from the camera video data into the trained target detection model, and performing intruder target detection on the first supervision area and the second supervision area through the trained target detection model; If it is determined according to the obtained target detection result that there is an intruder entering from the non-supervised area into at least one of the first supervised area and the second supervised area, a second warning is issued.

[0136] refer to Figure 7In a specific embodiment, the present application also provides a system for detecting intruders in a rail transit control area, and the system for detecting intruders in a rail transit control area includes: a detection control module 100, an arrival detection module 200, a door status detection module 300, a first intrusion detection module 400 and a second intrusion detection module 500.

[0137] The functions of the detection control module 100, the arrival detection module 200, the door status detection module 300, the first intrusion detection module 400 and the second intrusion detection module 500 are as described above and will not be repeated here.

[0138] The working principle of the system for detecting intruders in the rail transit supervision area of ​​the present application is described in the above-mentioned method for detecting intruders in the rail transit supervision area, which will not be repeated here.

[0139] This application can effectively improve the intelligence and accuracy of security monitoring in rail transit, such as the front area of ​​subways, with a special focus on the detection of non-train captains climbing over screen doors. By combining advanced computer vision technology, the system can monitor the position of the human detection center point under different lighting conditions and changes in the front position of the train, thereby accurately judging the intrusion behavior of unauthorized personnel. This technology reasonably divides the intrusion detection area into area 1 and area 2, and sets an effective captain patrol detection logic, which greatly reduces false alarms and interference and significantly enhances the overall monitoring effect. Compared with existing methods, this application has the following advantages: Accurate intrusion detection and intelligent monitoring: This application significantly reduces the false alarm rate by introducing a continuous detection mechanism, ensuring that the intrusion alarm is triggered only when confirmed, thereby improving the accuracy of detection. In addition, the system automatically identifies the arrival of trains and the status of doors, realizing automated management. This level of intelligent security monitoring reduces the need for manual intervention, thereby improving operational efficiency and safety.

[0140] Flexible area division and strong robustness: The system divides the intrusion detection area into Area 1 and Area 2 to flexibly adapt to different monitoring conditions and optimize the relationship between the captain's patrol and intrusion detection. At the same time, the target detection algorithm introduces the discrimination logic of the carriage category to effectively avoid misjudging other structures with similar shapes as the captain's door. This design significantly enhances the stability and reliability of monitoring and ensures effective operation in complex environments.

[0141] Improve safety management efficiency: This application not only enhances the ability to detect intrusions by unauthorized personnel, but also effectively reduces false alarms during normal patrols by train drivers, ensuring the safety and smoothness of subway operations. Through these innovations, the system can significantly improve security management efficiency in actual operations and effectively ensure the safety of passengers and vehicles.

[0142] Figure 8 7 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.

[0143] 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.

[0144] 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 method for detecting intruders in the rail transit regulatory area in the above 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 method for detecting intruders in the rail transit regulatory area.

[0145] 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.

[0146] 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.

[0147] As described above, 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 make equivalent replacements for some of the technical features therein. 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 method for detecting intruders in a rail transit supervision area, characterized in that: The method for detecting an intruder in a rail transit supervision area includes: Performing intrusion detection on a first supervision area and a second supervision area of ​​a rail transit station, wherein the first supervision area includes an area around the train track where train crews are allowed to inspect at the station, and the second supervision area includes a prohibited staying area around the train track except the first supervision area; In response to the presence of an arriving train at the rail transit station and the target door of the arriving train being in an open state, intrusion detection on the first supervision area is closed.

2. The method for detecting intruders in a rail transit supervision area according to claim 1 is characterized in that: After the intrusion detection on the first supervision area is turned off in response to the presence of an arriving train at the rail transit station and the target door of the arriving train is in an open state, the method further includes: In response to the target door changing from an open door state to a closed door state, intrusion detection on the first supervision area is restarted.

3. The method for detecting intruders in a rail transit supervision area according to claim 1 or 2, characterized in that: The method for detecting intruders in the rail transit supervision area also includes: In response to the target door being in an open state and a person intruding into the second supervision area, a first warning is issued; and / or, In response to the presence of an intruder in at least one of the first supervision area and the second supervision area, a second early warning is issued.

4. The method for detecting intruders in a rail transit supervision area according to claim 1 or 2, characterized in that: Before shutting down the intrusion detection on the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: Inputting the video frames captured from the camera video data into the trained target detection model, and performing target door detection by the trained target detection model; The door switch state of the target door is determined according to the obtained target detection result, wherein the door switch state includes a door open state and a door closed state.

5. The method for detecting intruders in a rail transit supervision area according to claim 4 is characterized in that: In response to the presence of an arriving train at the rail transit station and the target door of the arriving train being in an open state, closing the intrusion detection on the first supervision area includes: In response to the presence of an arriving train at the rail transit station and the detection of the target doors being in an open state from a first number of consecutive first video frames, intrusion detection on the first supervision area is closed.

6. The method for detecting intruders in a rail transit supervision area according to claim 1 or 2, characterized in that: Before shutting down the intrusion detection on the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: The video frames captured from the camera video data are input into the trained target detection model, and the target door detection is performed by the trained target detection model. If the target door is detected from a third number of consecutive third video frames by the trained target detection model, it is determined that there is an arriving train at the rail transit station; or, Before shutting down the intrusion detection on the first supervision area, the method for detecting an intruder in the rail transit supervision area further includes: The video frames captured from the camera video data are input into the trained object detection model, and the door detection is performed by the trained object detection model. If the carriage door is detected from a fourth number of fourth consecutive video frames by the trained target detection model, it is determined that there is an arriving train at the rail transit station.

7. The method for detecting intruders in a rail transit supervision area according to claim 3, characterized in that: In response to the presence of an intruder in at least one of the first supervision area and the second supervision area, issuing a second warning includes: Inputting the video frames captured from the camera video data into the trained target detection model, and performing intruder target detection on the first supervision area and the second supervision area through the trained target detection model; If it is determined according to the obtained target detection result that there is an intruder entering from the non-supervised area into at least one of the first and second supervised areas, a second warning is issued.

8. A detection system for intruders in a rail transit supervision area, characterized in that: The rail transit supervision area intruder detection system includes: A detection control module, configured to perform intrusion detection on a first supervision area and a second supervision area of ​​a rail transit station, wherein the first supervision area includes an area around the train track where train crews are allowed to inspect at the station, and the second supervision area includes a prohibited staying area around the train track except the first supervision area; The detection control module is further configured to shut down intrusion detection on the first supervision area in response to the presence of an arriving train at the rail transit station and the target door of the arriving train being in an open state.

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 so that the computer device executes the method for detecting an intruder in a rail transit supervision area as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by the processor, the method for detecting a person intruding into a rail transit supervision area as described in any one of claims 1 to 7 is implemented.

Images