Track foreign matter detection method, device and equipment and storage medium
By combining ceiling-mounted cameras with target tracking and foreign object detection models, and dynamically adjusting the detection area and thread status, the problems of high false alarm rate and cumbersome process in track foreign object detection are solved, achieving efficient and accurate detection for safe railway operation.
Patent Information
- Application Number
- CN202411185194.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2025-01-10
AI Technical Summary
Existing track foreign object detection methods have a high false alarm rate, cannot automatically distinguish between train operation and maintenance status, have cumbersome processes, and cannot provide timely warnings for trains in motion, thus failing to meet the requirements for safe railway operation.
Images of the track area are captured by a ceiling-mounted camera. Combined with a target tracking model and a foreign object detection model, the detection area and thread status are dynamically updated. Foreign object detection is only activated when the train is moving, reducing the false alarm rate and server memory consumption in non-passing situations.
It significantly reduced the false alarm rate and server memory consumption in non-passage situations, improved detection efficiency and accuracy, and met the real-time detection needs of railway safety operations.
Smart Images
Figure CN119296048A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of railway safety technology, and in particular to a track foreign object detection method, device, equipment and storage medium. Background Art
[0002] The safety of the railway system is of great significance to transportation. During railway operation and maintenance, foreign objects on the track may cause serious safety hazards, especially when the train enters a maintenance site or station, the maintenance tools or debris left on the track can easily cause the train body to bump or even derail. Therefore, it is of great significance to detect foreign objects on the track.
[0003] In the early days, it was mainly based on manual inspection. Specifically, railway workers regularly patrolled the tracks and visually checked whether there were foreign objects. This was not only inefficient, but also heavily dependent on the workers' experience and attention. Long hours of work could easily lead to fatigue and missed inspections. With the continuous development of technology, sensor-based detection and image processing technology-based detection have gradually been used for track foreign object detection. However, due to the influence of sensor coverage and environmental factors, the detection effect is unstable, the false alarm rate is high, and it is difficult to meet the needs of railway safety operations. Summary of the invention
[0004] The present invention provides a track foreign object detection method, device, equipment and storage medium to solve the problem of high false alarm rate in existing track foreign object detection methods.
[0005] In a first aspect, the present disclosure provides a method for detecting foreign matter on a track, comprising:
[0006] The images of the track area along the track direction are collected by ceiling cameras;
[0007] Determine the track state according to the track area image along the track direction, the track state includes no train in the track area, a stationary train in the track area, and a moving train in the track area;
[0008] When the track state is that there is a moving train in the track area, the image collected by the side camera corresponding to the track area where the moving train exists is input into the pre-trained foreign object detection model, and the foreign object detection thread is started to perform track foreign object detection.
[0009] In some embodiments, determining the track state according to the track region image along the track direction includes:
[0010] Determine whether there is a train in the track area according to the track area image along the track direction;
[0011] When there is no train in the track area, the track state is determined as no train in the track area;
[0012] When there is a train in the track area, the train moving speed is determined based on the continuously collected images of the adjacent track area along the track direction;
[0013] When the moving speed of the train is less than a preset speed threshold, it is determined that the track state is that there is a stationary train in the track area;
[0014] When the train moving speed is greater than or equal to a preset speed threshold, it is determined that the track state is that there is a moving train in the track area.
[0015] In some embodiments, the method further comprises:
[0016] When the track status changes from a moving train in the track area to no train in the track area or a stationary train in the track area, the foreign object detection thread is closed to stop track foreign object detection.
[0017] In some embodiments, the foreign object detection model is trained based on sample images labeled with the type and location of the foreign objects. The images captured by the side camera are input into the pre-trained foreign object detection model, and the type of foreign object and its location in the image captured by the side camera are output.
[0018] In some embodiments, the method further comprises:
[0019] The spatial position information of the foreign object is determined according to the spatial position information of the side camera and the position of the foreign object in the image captured by the side camera.
[0020] In some embodiments, at least one ceiling camera is provided on both sides of each track to photograph the train entering or exiting the track; and a side camera is provided on the outside of the track to photograph the area perpendicular to the track.
[0021] In a second aspect, the present disclosure provides a track foreign body detection device, comprising:
[0022] A collection module, used to collect images of the track area along the track direction through a ceiling camera;
[0023] A processing module, used for determining a track state according to a track area image along the track direction, wherein the track state includes no train in the track area, a stationary train in the track area, and a moving train in the track area;
[0024] The detection module is used to input the images collected by the side camera corresponding to the track area where the moving train exists into the pre-trained foreign object detection model when the track state is that there is a moving train in the track area, and start the foreign object detection thread to perform track foreign object detection.
[0025] In a third aspect, the present disclosure provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the above aspects.
[0026] In a fourth aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described in the above aspects when executed by a processor.
[0027] In a fifth aspect, the present disclosure provides a computer program product, including a computer program / instructions, which implements the steps of the method described in the above aspects when the computer program is executed by a processor.
[0028] The present disclosure provides a method, device, equipment and storage medium for detecting foreign objects on a track. The method collects images of the track area along the track direction through a ceiling camera; the track state is determined according to the images of the track area along the track direction, and the track state includes no vehicle in the track area, a stationary train in the track area and a moving train in the track area; when the track state is that there is a moving train in the track area, the image collected by the side camera corresponding to the track area where the moving train exists is input into a pre-trained foreign object detection model, and the foreign object detection thread is started to detect foreign objects on the track. When it is confirmed that there is a moving train in the track area, the foreign object detection thread is started to detect foreign objects on the track, which can not only reduce the video memory consumption of the server in the case of no passing vehicles, but also significantly reduce the false alarm rate in the case of no passing vehicles. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] The present disclosure will be described in more detail below based on embodiments and with reference to the accompanying drawings:
[0030] Figure 1 A schematic diagram of a flow chart of a track foreign body detection method provided by an embodiment of the present disclosure;
[0031] Figure 2 A schematic diagram of the on-site arrangement of a ceiling camera and a side camera provided in an embodiment of the present disclosure;
[0032] Figure 3 A schematic diagram of the viewing angles of a ceiling camera and a side camera provided in an embodiment of the present disclosure;
[0033] Figure 4 A schematic diagram of a track area image captured by a ceiling camera provided in an embodiment of the present disclosure;
[0034] Figure 5 A schematic diagram of a foreign body detection result provided by an embodiment of the present disclosure;
[0035] Figure 6 A schematic diagram of area division provided in an embodiment of the present disclosure;
[0036] Figure 7 A schematic structural diagram of a track foreign object detection device provided in an embodiment of the present disclosure.
[0037] In the drawings, the same reference numerals are used for the same components, and the drawings are not drawn to scale. DETAILED DESCRIPTION
[0038] In order to enable those skilled in the art to better understand the technical solution of the present disclosure, and to fully understand and implement how the present disclosure applies technical means to solve technical problems and achieve the corresponding technical effects, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only embodiments of a part of the present disclosure, not all of the embodiments. The embodiments of the present disclosure and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present disclosure.
[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure 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 interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.
[0040] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0041] It is of great significance to detect foreign objects on the track. Traditional methods for detecting foreign objects on the track mainly include manual inspection, sensor-based detection and simple image processing technology, based on computer vision and radar fusion, etc. However, these methods have their own limitations. For example, manual inspection is done by railway workers regularly patrolling the track to visually check whether there are foreign objects. It is inefficient and relies on the experience and attention of the workers. Long-term work can easily lead to fatigue and missed detection; sensor-based detection uses infrared sensors, ultrasonic sensors, etc. to detect foreign objects on the track. The sensor coverage is limited and is easily affected by environmental factors (such as weather and light), and the false alarm rate is high. Based on traditional image processing technology / deep learning algorithms, simple image processing algorithms (such as edge detection, morphological processing) or deep learning algorithms are used to detect foreign objects on the track. They have poor adaptability to complex backgrounds and changing lighting conditions, unstable detection effects, high false alarm rates, and cannot normally accurately locate the results to the specified area. If all tracks are to be detected at the same time, multiple deep learning algorithm models need to be continuously started, which consumes a lot of video memory. In addition, the above method is not applicable to daily railway maintenance situations. When the train enters the maintenance area and workers conduct security checks around the train, false alarms will inevitably occur due to the failure to determine the train's driving status. Frequent opening and closing of the detection status by workers is cumbersome and prone to failure.
[0042] In the modern train driving environment, it is required to timely and accurately detect the track area before the train enters, to ensure that there are no foreign objects in the corresponding track area, and to avoid related false alarms during the daily maintenance process (including when there are no trains or the train is stationary). While ensuring a low false alarm rate, it is also necessary to automatically complete the track foreign object detection process in the entire area of a specific track when there is a train passing by.
[0043] In order to solve at least one of the problems existing in the prior art, namely, high false alarm rate, inability to automatically distinguish between train operation and maintenance status, cumbersome procedures and inability to timely warn moving trains, and meet the needs of safe train operation, this application is based on the linkage processing of the target tracking model and the target detection model, takes the train operation status as the algorithm linkage parameter, dynamically updates the detection area and related thread status, and effectively avoids false alarms while improving detection efficiency. Compared with traditional deep learning detection algorithms, the model is opened and closed dynamically based on the track status, reducing the server video memory consumption in non-passing situations. The present application will be further explained in detail below through specific embodiments.
[0044] Embodiment 1
[0045] Figure 1 The following is a flow chart of a method for detecting foreign matter on a track provided by an embodiment of the present disclosure. Figure 1As shown, the track foreign body detection method provided by this embodiment may include:
[0046] S101. Collect images of the track area along the track direction through a ceiling camera.
[0047] In order to detect foreign objects on the track, it is necessary to arrange cameras in the track area. The cameras arranged on site can be divided into two parts: ceiling cameras and side cameras. In an optional implementation, at least one ceiling camera is arranged on both sides of each track to capture the situation of trains entering or leaving the track; the side camera is arranged on the outside of the track to capture the area perpendicular to the track.
[0048] Taking the track area with three tracks as an example, the on-site arrangement of the ceiling camera and the side camera is as follows: Figure 2 As shown. The cameras deployed on site can be divided into two parts: four cameras located in the middle and ceiling to capture the train entering or exiting the track; and side cameras to capture images of the track areas on both sides. The side cameras are distributed on the outside of the track to capture the vertical track area. There is at least one high-position ceiling camera on both sides of each track for monitoring, and the cameras on both sides are deployed on the near side of the track to ensure that the close-up coverage of each area of the track can be completed. Among them, the shooting angle of the ceiling camera is as follows Figure 3 As shown in (a), the shooting angle of the side camera is Figure 3 As shown in (b).
[0049] In this embodiment, the track area image along the track direction is collected by the ceiling camera to determine the track status. Figure 4 shown.
[0050] S102. Determine the track status according to the track area image along the track direction, where the track status includes no vehicle in the track area, a stationary train in the track area, and a moving train in the track area.
[0051] After obtaining the track area image along the track direction, the track area image can be input into the train tracking model to determine whether there is a train in the track area. If not, the track state is determined as no train in the track area, and the track area image along the track direction is continuously collected through the ceiling camera so that it can be detected in time when a train enters. If there is a train in the track area, it is further determined whether the train is moving. First, based on the image obtained by the ceiling camera, it is determined whether there is a train in the captured track area, and then the real-time position change of the train is obtained according to the track_id obtained by the tracking algorithm, and the confirmation of the train motion state is completed based on its comparison with the speed threshold. The target tracking model can assign the same track_id to the detection results of the same object in continuous time periods according to the front and rear position changes of the detection frame. For example, the real-time detection of the train position can be completed based on the tracking algorithm of YOLOV8, and the train state can be determined based on the front and rear center position changes. The center coordinates of the train with the same track_id are continuously recorded to determine the center moving speed of the train. When the center moving speed of the train is less than the preset speed threshold, the track state is determined to be a stationary train in the track area; when the center moving speed of the train is greater than or equal to the preset speed threshold, the track state is determined to be a moving train in the track area. Further determine the track number X where the train is located, and then output that there is a moving train on track X.
[0052] by Figure 4 The track area image shown in the figure is used as an example to illustrate how to determine the track status. Figure 4 As shown, the two right tracks with stationary trains are shown in rectangular boxes; the left track with moving trains is shown in rectangular boxes, and white curves are used in the rectangular boxes to mark the moving trajectory of the center position of the object with the same track_id obtained by the YOLOV8 area tracking algorithm. Furthermore, based on the preset track area and the center position of the train, the track number where the moving train or the stationary train is located can be obtained, and the track status can be updated.
[0053] In this embodiment, the relevant track states are divided into the following categories based on whether there are trains and whether the trains are moving: no trains in the track area, stationary trains in the track area, and moving trains in the track area. Based on the above process, the original train position state can be converted into the actual track state. It should be noted that in the actual detection process, there may be multiple cameras monitoring a single track at the same time. In this state, in order to avoid missing reports, when any camera image finds that a train is moving on a specific track, the state of the track area is set to a state where a moving train exists.
[0054] S103. When the track state is that there is a moving train in the track area, the image collected by the side camera corresponding to the track area where the moving train exists is input into a pre-trained foreign object detection model, and the foreign object detection thread is started to perform track foreign object detection.
[0055] After determining the track status, it is necessary to further perform foreign object detection on the track area where there are moving trains. After obtaining the track status information, the side camera image corresponding to the track area where there are moving trains is input into the foreign object detection model, and a new thread is started to complete the start of the foreign object detection process or complete the tracking of the specified track based on the preset region of interest (ROI). The foreign object detection results are as follows: Figure 5 The dark straight line area is the track area where vehicles have passed during this inspection, and the rectangular frame is the foreign matter detected on the track (the maintenance tools left behind by workers during maintenance are shown in the figure).
[0056] The foreign body detection model in this embodiment is trained based on sample images with the type and location of foreign bodies marked. Therefore, after the image captured by the side camera is input into the pre-trained foreign body detection model, the type of foreign body and the location in the image captured by the side camera will be output.
[0057] It is understandable that in order to facilitate the transmission of specific foreign body location, type, status and other information to the processing personnel, the spatial location information of the side camera can be used to further refine the location of the foreign body. Figure 6 As shown, AB represents the detection area of the side camera B on track A, for example, 1-8 represents the detection area of the side camera 8 on track 1. Based on this division, the location of the foreign object can be refined to a specific track area and the corresponding side camera position, and the location of the foreign object in the alarm information can be further refined so that the staff can quickly and accurately locate the foreign object. That is to say, in an optional embodiment, the spatial location information of the foreign object can be determined based on the spatial location information of the side camera and the location of the foreign object in the image captured by the side camera.
[0058] The track foreign object detection method provided in this embodiment uses a ceiling camera to collect images of the track area along the track direction; determines the track state based on the images of the track area along the track direction; when the track state is that there is a moving train in the track area, the image collected by the side camera corresponding to the track area where the moving train exists is input into a pre-trained foreign object detection model, and the foreign object detection thread is started to detect foreign objects on the track. When it is confirmed that there is a moving train in the track area, the foreign object detection thread is started to detect foreign objects on the track, which can not only reduce the video memory consumption of the server in the case of no passing trains, but also significantly reduce the false alarm rate in the case of no passing trains.
[0059] Embodiment 2
[0060] On the basis of the above embodiments, the following will further explain in detail how to determine the track state according to the track area image along the track direction. In the track foreign body detection method provided by this embodiment, the track state is determined according to the track area image along the track direction, which can specifically include: determining whether there is a train in the track area according to the track area image along the track direction; when there is no train in the track area, determining the track state as no train in the track area; when there is a train in the track area, determining the train moving speed according to the continuously collected adjacent track area images along the track direction; when the train moving speed is less than a preset speed threshold, determining the track state as the presence of a stationary train in the track area; when the train moving speed is greater than or equal to the preset speed threshold, determining the track state as the presence of a moving train in the track area.
[0061] Embodiment 3
[0062] On the basis of any of the above embodiments, in order to further reduce video memory consumption, in the track foreign object detection method provided in this embodiment, when the track state changes from the presence of a moving train in the track area to the presence of a stationary train in the track area, the foreign object detection thread is closed and the track foreign object detection is stopped.
[0063] After the train stops or passes through the track area, the train position tracking algorithm updates the track status to no train or stationary train in the track area based on the detection results. Based on this, the thread of the foreign object detection algorithm is terminated or the ROI of the image track area is updated to avoid false alarms of foreign objects on the track when there is no train or the train is stationary. This can not only reduce the false alarm rate in the case of no train passing, but also significantly reduce the video memory consumption of the server in the case of no train passing.
[0064] Figure 6The upper and lower cameras in the middle area monitor the status of the two tracks respectively. To improve the detection efficiency, the upper and lower cameras can be used as detection clusters to establish threads for flow acquisition and detection. For example, when a car passes by railway_1, the railway_1 detection thread can be started to collect images based on the upper camera cluster for detection to reduce video memory usage.
[0065] Compared with the traditional deep learning detection algorithm, the track foreign object detection method provided in this embodiment dynamically starts and shuts down the foreign object detection thread based on the track status, which not only reduces the server video memory consumption in the non-passing vehicle situation, but also reduces the false alarm rate in the non-passing vehicle situation.
[0066] Embodiment 4
[0067] Figure 7 The schematic diagram of the structure of a track foreign body detection device provided by the embodiment of the present disclosure is as follows. Figure 7 As shown, the track foreign object detection device 20 provided in this embodiment may include: a collection module 201, a processing module 202 and a detection module 203.
[0068] The acquisition module 201 is used to acquire images of the track area along the track direction through a ceiling camera;
[0069] A processing module 202 is used to determine the track state according to the track area image along the track direction, where the track state includes no train in the track area, a stationary train in the track area, and a moving train in the track area;
[0070] The detection module 203 is used to input the images collected by the side camera corresponding to the track area where the moving train exists into a pre-trained foreign object detection model when the track state is that there is a moving train in the track area, and start the foreign object detection thread to perform track foreign object detection.
[0071] The device of this embodiment can be used to perform Figure 1 The technical solution of the method embodiment shown has similar implementation principles and technical effects, which will not be repeated here.
[0072] In some embodiments, the processing module 202 is used to determine the track state according to the track area image along the track direction, which may specifically include:
[0073] Determine whether there is a train in the track area according to the track area image along the track direction;
[0074] When there is no train in the track area, the track state is determined as no train in the track area;
[0075] When there is a train in the track area, the train moving speed is determined based on the continuously collected images of the adjacent track area along the track direction;
[0076] When the moving speed of the train is less than a preset speed threshold, it is determined that the track state is that there is a stationary train in the track area;
[0077] When the train moving speed is greater than or equal to a preset speed threshold, it is determined that the track state is that there is a moving train in the track area.
[0078] In some embodiments, the detection module 203 is further configured to:
[0079] When the track status changes from a moving train in the track area to no train in the track area or a stationary train in the track area, the foreign object detection thread is closed to stop track foreign object detection.
[0080] In some embodiments, the foreign object detection model is trained based on sample images labeled with the type and location of the foreign objects, and the detection module 203 is also used to: input the image captured by the side camera into a pre-trained foreign object detection model, and output the type of foreign object and its location in the image captured by the side camera.
[0081] In some embodiments, the detection module 203 is further configured to determine the spatial position information of the foreign object according to the spatial position information of the side camera and the position of the foreign object in the image captured by the side camera.
[0082] In some embodiments, at least one ceiling camera is provided on both sides of each track to photograph the train entering or exiting the track; and a side camera is provided on the outside of the track to photograph the area perpendicular to the track.
[0083] Embodiment 5
[0084] Based on the above embodiments, this embodiment provides an application example.
[0085] The track foreign body detection method provided in this embodiment can be divided into the following steps:
[0086] Step 1: The YOLOV8-based tracking algorithm is used to complete real-time detection of the train position, and the train status is determined based on the changes in the front and rear center positions.
[0087] like Figure 2 As shown in the figure, the on-site layout under the reference situation consists of three track areas and several cameras. The cameras arranged on site can be divided into two parts: four cameras located in the middle and ceiling to capture the train entering or leaving the track; and side cameras to capture images of the track areas on both sides. The side cameras are distributed on the outside of the track to capture the vertical track area.
[0088] There is at least one high-position ceiling camera on both sides of each track for monitoring, and the cameras on both sides are deployed on the near side of the track to ensure that the close-up view of each area of the track can be fully covered. Figure 3 As shown in (a), the side camera shooting angle is Figure 3 As shown in (b).
[0089] The specific process of the train tracking algorithm is to first determine whether there is a train in the track area captured based on the image obtained by the ceiling camera, then obtain the real-time position change of the train based on the track_id obtained by the tracking algorithm, and complete the confirmation of the train's motion state based on its comparison with the speed threshold. The target tracking model can assign the same track_id to the detection results of the same object in consecutive time periods based on the front and back position changes of the detection frame.
[0090] Step 2: Analyze the position of the train in motion and pass the corresponding track information to the foreign object detection model.
[0091] After obtaining the track area train status from step 1, we can get Figure 4 The detection results. The two right tracks with stationary trains are marked with light-colored square boxes, and the left track with moving trains is marked with dark-colored boxes. The curve in the left track box is the moving trajectory of the center position of the same object obtained by the YOLOV8 area tracking algorithm. Based on the preset track area and train center position, the track number where the moving train or stationary train is located can be obtained, and the track status can be updated.
[0092] Based on whether there is a train and whether the train is moving, the relevant track status is divided into: no train in the track area, stationary train in the track area, and train in a moving state on the track; based on the above process, the original train position state can be converted into the actual track state. In the actual detection process, there may be multiple cameras monitoring a single track at the same time. In this state, in order to avoid missing reports, when any camera image finds that a train is moving on a specific track, the track area state is set to a moving train state.
[0093] Step 3: The foreign object detection model dynamically adjusts the ROI based on the track information and completes the detection, outputs the type and location of the relevant foreign objects, and inputs the alarm information to the front end.
[0094] After obtaining the track status information output in step 2, the side camera image corresponding to the track area where the moving train exists is input into the foreign object detection model, and a new thread is started to complete the start of the foreign object detection process or complete the tracking of the specified track based on the preset ROI area. The detection results are as follows: Figure 5The dark straight line area is the track area where vehicles have passed during this inspection, and the box is the foreign matter detected on the track (the maintenance tools left behind by workers during the inspection are shown in the figure).
[0095] Based on the camera position, the position of the foreign object can be further refined to facilitate the transmission of the specific foreign object position and status to the processing personnel. The area divisions obtained by the track area and camera division are as follows: Figure 6 As shown in the figure, AB represents the detection area of track A and camera B. In this way, the alarm information can be refined to the specified area and camera position.
[0096] After the train stops or passes through the area, the train position tracking algorithm updates the track status to no vehicle or stationary train based on the detection results. Based on this, the thread of the foreign object detection algorithm is terminated or the ROI of the image track area is updated to avoid false alarms of foreign objects on the track when there is no vehicle or the train is stationary.
[0097] In summary, the track foreign body detection method provided by this embodiment solves the problems of high false alarm rate, inability to automatically distinguish the train operation and maintenance status, cumbersome process and inability to timely warn the moving trains in the traditional railway foreign body detection algorithm. Based on the linkage processing of the target tracking model and the target detection model, the train operation status is used as the algorithm linkage parameter, and the detection area and related thread status are dynamically updated, which effectively avoids false alarms while improving the detection efficiency. Compared with the traditional deep learning detection algorithm, the model is opened and closed dynamically based on the track status, which reduces the server video memory consumption in the non-passing situation.
[0098] Embodiment 6
[0099] On the basis of the above embodiments, this embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the above embodiments.
[0100] In some implementations of this embodiment, a computer-readable storage medium is provided, on which a computer program is stored, characterized in that when the computer program is executed by a processor, the steps of the method described in the above embodiment are implemented.
[0101] In some implementations of this embodiment, a computer program product is provided, including a computer program / instructions, characterized in that when the computer program is executed by a processor, the steps of the method described in the above embodiment are implemented.
[0102] The processor may include, but is not limited to, one or more processors or microprocessors. Each processor may be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components to execute the method in the above embodiments.
[0103] The computer-readable storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, and the computer-readable storage medium may include but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (e.g., hard disk, floppy disk, solid-state drive, removable disk, CD-ROM, DVD-ROM, Blu-ray disc, etc.).
[0104] The computer-readable storage medium may also store at least one computer executable program / instruction, which may be, for example, a computer-readable instruction. The computer-readable storage medium includes, but is not limited to, for example, a volatile memory and / or a non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The computer-readable storage medium may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device runs the computer-readable instructions stored on the computer-readable storage medium, the various methods described above may be performed.
[0105] In addition, the computer device may also include (but not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.), etc.
[0106] The processor may communicate with external devices via an I / O bus via a wired or wireless network.
[0107] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product / computer program product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.
[0108] In the embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.
[0109] It should be noted that in the present disclosure, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element limited by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.
[0110] Although the embodiments disclosed in the present disclosure are as above, the above contents are only embodiments adopted for facilitating the understanding of the present disclosure and are not intended to limit the present disclosure. Any technician in the technical field to which the present disclosure belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present disclosure, but the scope of patent protection of the present disclosure shall still be subject to the scope defined in the attached claims.
Claims
1. A method for detecting foreign matter on a track, characterized in that: include: The images of the track area along the track direction are collected by ceiling cameras; Determining a track state according to the track area image along the track direction, the track state including no train in the track area, a stationary train in the track area, and a moving train in the track area; When the track state is that there is a moving train in the track area, the image collected by the side camera corresponding to the track area where the moving train exists is input into the pre-trained foreign object detection model, and the foreign object detection thread is started to perform track foreign object detection.
2. The method according to claim 1, characterized in that The determining the track state according to the track area image along the track direction includes: Determining whether there is a train in the track area according to the track area image along the track direction; When there is no train in the track area, the track state is determined as no train in the track area; When there is a train in the track area, the train moving speed is determined based on the continuously collected images of the adjacent track area along the track direction; When the moving speed of the train is less than a preset speed threshold, determining that the track state is that there is a stationary train in the track area; When the moving speed of the train is greater than or equal to a preset speed threshold, it is determined that the track state is that there is a moving train in the track area.
3. The method according to claim 1, characterized in that The method further comprises: When the track status changes from a moving train in the track area to no train in the track area or a stationary train in the track area, the foreign object detection thread is closed to stop track foreign object detection.
4. The method according to claim 1, characterized in that: The foreign object detection model is trained based on sample images marked with the type and location of the foreign objects. The image captured by the side camera is input into the pre-trained foreign object detection model, and the type of foreign object and the location in the image captured by the side camera are output.
5. The method according to claim 4, characterized in that The method further comprises: The spatial position information of the foreign object is determined according to the spatial position information of the side camera and the position of the foreign object in the image captured by the side camera.
6. The method according to any one of claims 1 to 5, characterized in that: At least one ceiling camera is arranged on both sides of each track for photographing the train entering or exiting along the track; the side camera is arranged on the outside of the track for photographing the area perpendicular to the track.
7. A track foreign body detection device, characterized in that: include: A collection module, used to collect images of the track area along the track direction through a ceiling camera; A processing module, configured to determine a track state according to the track area image along the track direction, wherein the track state includes no vehicle in the track area, a stationary train in the track area, and a moving train in the track area; The detection module is used to input the images collected by the side camera corresponding to the track area where the moving train exists into the pre-trained foreign object detection model when the track state is that there is a moving train in the track area, and start the foreign object detection thread to perform track foreign object detection.
8. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product comprising a computer program / instructions, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.