Rail Transit Foreign Object Intrusion Risk Identification and Assessment Method and System

Abstract

The present invention provides a method and system for identifying and evaluating the risk of foreign object intrusion in rail transit, belonging to the field of object recognition technology in image processing. Image data of the operating environment in front of the train is collected in real time; through a foreign object detection algorithm, the collected image data is subjected to foreign object detection frame by frame. If a foreign object is detected in a certain frame of image, the category and position information of the detected foreign object are obtained; the track area in this frame of image is segmented through a track segmentation algorithm to obtain the position information of the track area; based on the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, the risk level of foreign object intrusion in rail transit is comprehensively evaluated and output. The present invention can identify the risk of foreign object intrusion in the rail transit scenario in real time and accurately. At the same time, a risk level evaluation system for foreign object intrusion in rail transit is established, realizing the efficient detection, identification and evaluation of the risk of foreign object intrusion in rail transit.

Description

Technical Field

[0001] The present invention relates to the technical field of object recognition in image processing, and particularly relates to a method and system for identifying and evaluating the risk of foreign object intrusion in rail transit. Background Art

[0002] As a common means of transportation, the safety and stability during the operation of rail transit deeply affect people's lives and property safety. The intrusion of foreign objects along the track is an important factor causing rail transit safety accidents and poses a serious threat to the safe operation of trains. Therefore, it is necessary to accurately and timely detect the risk of foreign object intrusion that may occur in the rail transit scenario, so as to ensure the safety and stability during the train operation.

[0003] At present, the detection of the risk of foreign object intrusion in rail transit mainly relies on video detection. Most traditional video detection methods use cameras installed along the track to obtain video data, and then use traditional detection methods such as frame difference method and background difference method to detect the intrusion of foreign objects around the line. However, it is difficult and costly to deploy monitoring cameras along the entire railway line. And traditional detection methods can only be applied in simple scenarios, and it is difficult to balance speed and accuracy. At the same time, the existing rail transit foreign object intrusion detection technology only has the function of detecting the category of foreign objects, and cannot actively identify and evaluate the risk of foreign object intrusion, resulting in the driver needing to make a secondary risk judgment based on the category of foreign objects. This will delay the best time to handle the risk and there are huge potential safety hazards. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for identifying and evaluating the risk of foreign object intrusion in rail transit, so as to solve at least one of the technical problems existing in the above background art.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions:

[0006] In the first aspect, the present invention provides a method for identifying and evaluating the risk of foreign object intrusion in rail transit, including:

[0007] Real-time collecting image data of the running environment in front of the train;

[0008] Using a pre-trained foreign object detection model to perform foreign object detection on the collected image data frame by frame. If a foreign object is detected in a certain frame of the image, obtain the category and position information of the detected foreign object. Otherwise, continue to detect the next frame of the image until a foreign object is detected in a certain frame of the image. Among them, training the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weights; using the foreign object detection algorithm to detect each frame of the obtained image data, and output the category and position information of the foreign object when a foreign object is detected.

[0009] When a foreign object is detected in a certain frame of image, the track area in the frame of image is segmented by a pre-trained track segmentation model to obtain the position information of the track area; wherein, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories: track area and background; if the foreign object detection algorithm detects a foreign object in a certain frame of image, the track area in the frame of image is segmented by a pre-trained track segmentation algorithm; and the position information of the track area is obtained by using the output result.

[0010] Based on three key risk factors, namely, the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the risk level of rail transit foreign object intrusion and output it.

[0011] Furthermore, a comprehensive spatial risk score of the foreign object is set based on the position information of the foreign object and the relative position relationship between the foreign object and the track area; a category risk score of the foreign object is set based on the severity of the accident caused by different foreign object categories; based on the risk matrix theory, the comprehensive spatial risk score and category risk score of the foreign object are comprehensively calculated to obtain the risk score of rail transit foreign object intrusion, and finally the risk level is output according to the score.

[0012] Furthermore, the specific process of setting the comprehensive spatial risk score of the foreign object based on the position information of the foreign object and the relative position relationship between the foreign object and the track area is as follows:

[0013] Taking the on-vehicle camera as the origin, the train running direction as the Y-axis, parallel to the rail plane, and perpendicular to the train running direction as the X-axis, a plane rectangular coordinate system is established.

[0014] According to the relative position relationship between the foreign object and the track area in the horizontal direction, a spatial risk scoring standard in the horizontal direction is set.

[0015] According to the distance between the foreign object and the train in the train running direction, a spatial risk scoring standard in the train running direction is set.

[0016] According to the spatial risk scores of the foreign object in the horizontal direction and the train running direction, the comprehensive spatial risk score of the foreign object is calculated.

[0017] Furthermore, the specific process of setting the spatial risk scoring standard in the horizontal direction according to the relative position relationship between the foreign object and the track area in the horizontal direction is as follows:

[0018] Calculate the actual distance between the foreign object and the track area in the horizontal direction, and the calculation formula is as follows:

[0019]

[0020] In the formula, X Ris the abscissa of the lower right corner of the foreign object detection frame in the image, X1 and X2 are the abscissas of the intersection points of the extension lines of the lower frame line of the foreign object detection frame in the figure with the left and right rails respectively, and N is the actual distance between the foreign object and the track area;

[0021] The spatial risk scoring standard in the horizontal direction is set according to the actual distance between the foreign object and the track area in the horizontal direction as follows:

[0022] When the spatial position relationship is N > 1, the spatial risk score is 0.5;

[0023] When the spatial position relationship is N < 1, the spatial risk score is 1;

[0024] When the spatial position relationship is within the track area, the spatial risk score is 2.

[0025] Furthermore, the specific process of setting the spatial risk scoring standard in the train running direction according to the distance between the foreign object and the train in the train running direction is as follows:

[0026] Calculate the actual distance between the foreign object and the train in the train running direction. The calculation formula is as follows:

[0027]

[0028] In the formula, Y R is the ordinate of the lower right corner of the foreign object detection frame in the figure, representing the pixel distance between the foreign object and the train in the train running direction, M is the actual distance between the foreign object and the train in the train running direction, Y0 is the farthest pixel distance captured by the camera in the train running direction in the figure, and 150 is the actual farthest shooting distance of the on-vehicle camera in the train running direction;

[0029] The spatial risk scoring standard in the train running direction is set according to the actual distance between the foreign object and the train in the train running direction as follows:

[0030] When the spatial position relationship is 100 < M ≤ 150, the spatial risk score is 0.5;

[0031] When the spatial position relationship is 50 < M ≤ 100, the spatial risk score is 1;

[0032] When the spatial position relationship is M ≤ 50, the spatial risk score is 2.

[0033] Furthermore, the comprehensive spatial risk score of the foreign object is:

[0034] L = L h + L v

[0035] In the formula, L is the comprehensive spatial risk score of the foreign object, L h is the spatial risk score in the horizontal direction, Lv is the spatial risk score in the train running direction;

[0036] The risk score of foreign object intrusion in rail transit is:

[0037] R = L × P

[0038] In the formula, R is the risk score of foreign object intrusion in rail transit, L is the comprehensive spatial risk score of the foreign object, and P is the category risk score of the foreign object.

[0039] In a second aspect, the present invention provides a rail transit foreign object intrusion risk identification and assessment system, including:

[0040] An acquisition module, configured to collect image data of the running environment in front of the train in real time;

[0041] A detection module, configured to use a pre-trained foreign object detection model to detect foreign objects frame by frame in the collected image data. If a foreign object is detected in a certain frame of image, obtain the category and position information of the detected foreign object, otherwise continue to detect the next frame of image until a foreign object is detected in a certain frame of image; wherein, the training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain optimal weights; using the foreign object detection algorithm to detect each frame of image data obtained, and output the category and position information of the foreign object when a foreign object is detected;

[0042] A segmentation module, configured to, when a foreign object is detected in a certain frame of image, segment the track area in the frame of image through a pre-trained track segmentation model to obtain the position information of the track area; wherein, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories: track area and background; if the foreign object detection algorithm detects a foreign object in a certain frame of image, segment the track area in the frame of image through a pre-trained track segmentation algorithm; use the output result to obtain the position information of the track area;

[0043] An evaluation module, configured to comprehensively evaluate the risk level of rail transit foreign object intrusion based on three key risk factors: the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, and output it.

[0044] In a third aspect, the present invention provides a non-transitory computer-readable storage medium, which is used to store computer instructions. When the computer instructions are executed by a processor, the rail transit foreign object intrusion risk identification and assessment method as described in the first aspect is implemented.

[0045] Fourth aspect, the present invention provides a computer device, including a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the rail transit foreign object intrusion risk identification and assessment method as described in the first aspect.

[0046] Fifth aspect, the present invention provides an electronic device, including: a processor, a memory, and a computer program; wherein, the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device runs, the processor executes the computer program stored in the memory so that the electronic device executes the instructions for implementing the rail transit foreign object intrusion risk identification and assessment method as described in the first aspect.

[0047] Advantages of the present invention: By first detecting foreign objects and then performing track segmentation only when there are foreign objects in the image, real-time detection of foreign object intrusion is achieved, effectively improving the speed of the entire rail transit foreign object intrusion risk identification and assessment process. Based on three key risk factors, namely the positional relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, the rail transit foreign object intrusion risk level is comprehensively evaluated. Compared with other risk assessment methods, this method considers more comprehensive factors and has a more reasonable evaluation system, and can accurately judge the intrusion risks of various foreign objects in real time in complex and diverse rail transit scenarios, ensuring the safe operation of the rail transit scenario.

[0048] The advantages of the additional aspects of the present invention will be more clearly given in the following description part, or can be understood through the practice of the present invention. Description of the Drawings

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0050] Figure 1 It is a flowchart of the rail transit foreign object intrusion risk identification and assessment method described in the embodiments of the present invention.

[0051] Figure 2 It is a detection effect diagram of the foreign object detection algorithm described in the embodiments of the present invention.

[0052] Figure 3 It is a segmentation effect diagram of the track segmentation algorithm described in the embodiments of the present invention. Detailed Embodiments

[0053] The embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary only for explaining the present invention and should not be construed as limiting the present invention.

[0054] Those skilled in the art of the present technology can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those of ordinary skill in the art to which the present invention pertains.

[0055] It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art and will not be interpreted with an idealized or overly formal meaning unless defined as herein.

[0056] Those skilled in the art of the present technology can understand that, unless specifically stated otherwise, the singular forms "a", "an", "the", and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the description of the present invention means the presence of the described features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, and / or groups thereof.

[0057] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples.

[0058] To facilitate the understanding of the present invention, the present invention will be further explained below with specific embodiments in conjunction with the accompanying drawings, and the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0059] Those skilled in the art should understand that the drawings are only schematic diagrams of the embodiments, and the components in the drawings are not necessarily essential for implementing the present invention.

[0060] Embodiment 1

[0061] In Embodiment 1, first, a risk identification and assessment system for foreign object intrusion in rail transit is provided, including: an acquisition module for real-time collecting image data of the operating environment in front of the train; a detection module for using a pre-trained foreign object detection model to perform foreign object detection on the collected image data frame by frame. If a foreign object is detected in a certain frame of the image, the category and position information of the detected foreign object are obtained; otherwise, the next frame of the image is continuously detected until a foreign object is detected in a certain frame of the image; wherein, the training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weights; using the foreign object detection algorithm to detect each frame of the obtained image data, and outputting the category and position information of the foreign object when a foreign object is detected; a segmentation module for, when a foreign object is detected in a certain frame of the image, segmenting the track area in the frame of the image through a pre-trained track segmentation model to obtain the position information of the track area; wherein, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories: track area and background; if the foreign object detection algorithm detects a foreign object in a certain frame of the image, segmenting the track area in the frame of the image through a pre-trained track segmentation algorithm; using the output result to obtain the position information of the track area; an evaluation module for comprehensively evaluating the risk level of foreign object intrusion in rail transit and outputting based on three key risk factors: the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object.

[0062] In this embodiment, using the above system, a risk identification and assessment method for foreign object intrusion in rail transit is implemented, including: using an on-vehicle front-view camera installed at the head of the train to real-time collect image data of the operating environment in front of the train; using a pre-trained foreign object detection model to perform foreign object detection on the collected image data frame by frame. If a foreign object is detected in a certain frame of the image, the category and position information of the detected foreign object are obtained; otherwise, the next frame of the image is continuously detected until a foreign object is detected in a certain frame of the image; wherein, the training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weights; using the foreign object detection algorithm to detect each frame of the obtained image data, and outputting the category and position information of the foreign object when a foreign object is detected;

[0063] When a foreign object is detected in a certain frame of image, the track area in the frame of image is segmented by a pre-trained track segmentation model to obtain the position information of the track area. Among them, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories: track area and background. If the foreign object detection algorithm detects a foreign object in a certain frame of image, the track area in the frame of image is segmented by a pre-trained track segmentation algorithm, and the position information of the track area is obtained by using the output result.

[0064] Based on three key risk factors: the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the risk level of rail transit foreign object intrusion and output it.

[0065] Set the comprehensive spatial risk score of the foreign object based on the position information of the foreign object and the relative position relationship between the foreign object and the track area; set the category risk score of the foreign object based on the severity of the accident caused by different foreign object categories; based on the risk matrix theory, comprehensively calculate the comprehensive spatial risk score and category risk score of the foreign object, calculate the risk score of rail transit foreign object intrusion, and finally output the risk level according to the score.

[0066] The specific process of setting the comprehensive spatial risk score of the foreign object based on the position information of the foreign object and the relative position relationship between the foreign object and the track area is as follows:

[0067] Taking the on-vehicle camera as the origin, the train running direction as the Y-axis, parallel to the rail plane, and perpendicular to the train running direction as the X-axis, establish a plane rectangular coordinate system.

[0068] Set the spatial risk scoring standard in the horizontal direction according to the relative position relationship between the foreign object and the track area in the horizontal direction.

[0069] Set the spatial risk scoring standard in the train running direction according to the distance between the foreign object and the train in the train running direction.

[0070] Calculate the comprehensive spatial risk score of the foreign object according to the spatial risk scores of the foreign object in the horizontal direction and the train running direction.

[0071] The specific process of setting the spatial risk scoring standard in the horizontal direction according to the relative position relationship between the foreign object and the track area in the horizontal direction is as follows:

[0072] Calculate the actual distance between the foreign object and the track area in the horizontal direction. The calculation formula is as follows:

[0073]

[0074] In the formula, X Ris the abscissa of the lower right corner of the foreign object detection frame in the image, X1 and X2 are the abscissas of the intersection points of the extension lines of the lower frame line of the foreign object detection frame in the figure with the left and right rails respectively, and N is the actual distance between the foreign object and the track area;

[0075] Set the spatial risk scoring standard in the horizontal direction according to the size of the actual distance between the foreign object and the track area in the horizontal direction as:

[0076] When the spatial position relationship is N > 1, the spatial risk score is 0.5;

[0077] When the spatial position relationship is N < 1, the spatial risk score is 1;

[0078] When the spatial position relationship is within the track area, the spatial risk score is 2.

[0079] The specific process of setting the spatial risk scoring standard in the train running direction according to the distance between the foreign object and the train in the train running direction is as follows:

[0080] Calculate the actual distance between the foreign object and the train in the train running direction, and the calculation formula is as follows:

[0081]

[0082] In the formula, Y R is the ordinate of the lower right corner of the foreign object detection frame in the figure, representing the pixel distance between the foreign object and the train in the train running direction, M is the actual distance between the foreign object and the train in the train running direction, Y0 is the farthest pixel distance photographed by the camera in the figure in the train running direction, and 150 is the actual farthest shooting distance of the on-vehicle camera in the train running direction;

[0083] Set the spatial risk scoring standard in the train running direction according to the actual distance between the foreign object and the train in the train running direction as:

[0084] When the spatial position relationship is 100 < M ≤ 150, the spatial risk score is 0.5;

[0085] When the spatial position relationship is 50 < M ≤ 100, the spatial risk score is 1;

[0086] When the spatial position relationship is M ≤ 50, the spatial risk score is 2.

[0087] The comprehensive spatial risk score of the foreign object is:

[0088] L = L h + L v

[0089] In the formula, L is the comprehensive spatial risk score of the foreign object, L h is the spatial risk score in the horizontal direction, L vis the spatial risk score in the train running direction;

[0090] The risk score of foreign object intrusion in rail transit is:

[0091] R = L × P

[0092] In the formula, R is the risk score of foreign object intrusion in rail transit, L is the comprehensive spatial risk score of the foreign object, and P is the category risk score of the foreign object.

[0093] Embodiment 2

[0094] In this embodiment, in view of the problems of poor real-time performance, low accuracy, and non-refined evaluation system existing in the existing risk identification and evaluation of foreign object intrusion in rail transit, a method for risk identification and evaluation of foreign object intrusion in rail transit is provided. This method can actively detect the intrusion of foreign objects on the front line in real time during the train operation based on the video images captured by the on-vehicle front camera, and comprehensively evaluate the risk level of foreign object intrusion in rail transit according to three key risk factors: the position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, so as to assist the train driver to avoid risks according to the risk level, effectively ensuring the safety and stability during the operation of the rail transit scenario.

[0095] As Figure 1 shown, the method for risk identification and evaluation of foreign object intrusion in rail transit described in this embodiment includes the following steps:

[0096] Step 1: Use the on-vehicle front camera installed at the train head to collect the image data of the running environment in front of the train in real time;

[0097] Step 2: Through the foreign object detection algorithm, perform foreign object detection on the image data in Step 1 frame by frame. If a foreign object is detected in a certain frame of image, obtain the category and position information of the detected foreign object, otherwise continue to detect the next frame of image until a foreign object is detected in a certain frame of image;

[0098] The specific steps of Step 2 include the following steps:

[0099] Step 2.1, construct a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects;

[0100] Step 2.2, use the foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weight;

[0101] Step 2.3, use the foreign object detection algorithm to detect each frame of image data obtained in Step 1, and output the category and position information of the foreign object when a foreign object is detected.

[0102] Step 3: When a foreign object is detected in a certain frame of image, the track area in this frame of image is segmented by the track segmentation algorithm to obtain the position information of the track area;

[0103] The specific steps of Step 3 are as follows:

[0104] Step 3.1, construct a track segmentation dataset, including two categories: track area and background;

[0105] Step 3.2, if the foreign object detection algorithm detects a foreign object in a certain frame of image, the track area in this frame of image is segmented by the pre-trained track segmentation algorithm;

[0106] Step 3.3, use the output result to obtain the position information of the track area.

[0107] Step 4: Based on three key risk factors, namely the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the risk level of foreign object intrusion into rail transit and output it.

[0108] The specific steps of Step 4 are as follows:

[0109] Step 4.1, set a comprehensive spatial risk score for the foreign object based on the position information of the foreign object and the relative position relationship between the foreign object and the track area.

[0110] The specific steps of Step 4.1 are as follows:

[0111] Step 4.1.1, take the on-vehicle camera as the origin, the train running direction as the Y-axis, parallel to the rail plane, and perpendicular to the train running direction as the X-axis to establish a plane rectangular coordinate system;

[0112] Step 4.1.2, set the spatial risk scoring standard in the horizontal direction according to the relative position relationship between the foreign object and the track area in the horizontal direction;

[0113] The distance calculation formula between the foreign object and the track area in the horizontal direction in Step 4.1.2 is:

[0114]

[0115] In the formula, X R is the abscissa of the lower right corner of the foreign object detection box in the image, X1 and X2 are the abscissas of the intersections of the extension lines of the lower frame line of the foreign object detection box in the figure with the left and right rails respectively, N is the actual distance between the foreign object and the track area, and 1.435 is the actual rail gauge (unit: meter).

[0116] The spatial risk scoring standard in the horizontal direction in Step 4.1.2 is:

[0117] When the actual distance between the foreign object and the track area is greater than 1 meter, there is a low risk of foreign object intrusion; when the actual distance between the foreign object and the track area is less than 1 meter, there is a high risk of foreign object intrusion; when the foreign object is located within the track area, foreign object intrusion has occurred, and the risk is the greatest at this time. Based on this, the spatial risk score of the foreign object in the horizontal direction is set as shown in Table 1.

[0118] Table 1 Spatial risk scoring criteria in the horizontal direction

[0119] Spatial position relationship Spatial risk score N>1 0.5 N<1 1 Located within the orbital area 2

[0120] Step 4.1.3, set the spatial risk scoring criteria in the train running direction according to the distance between the foreign object and the train in the train running direction;

[0121] The formula for calculating the distance between the foreign object and the train in the train running direction in Step 4.1.3 is:

[0122]

[0123] In the formula, Y R is the ordinate of the lower right corner of the foreign object detection frame in the figure, representing the pixel distance between the foreign object and the train in the train running direction, M is the actual distance between the foreign object and the train in the train running direction (unit: meter), Y0 is the farthest pixel distance captured by the camera in the train running direction in the figure, and 150 is the actual farthest shooting distance of the on-vehicle camera in the train running direction;

[0124] The spatial risk scoring criteria in the train running direction in Step 4.1.3 are:

[0125] When 100 < M ≤ 150, at this time the foreign object is far from the train, and there is a low risk of foreign object intrusion; when 50 < M ≤ 100, since the foreign object is relatively close to the train, there is a high risk of foreign object intrusion; when M ≤ 50, since the foreign object is very close to the train, there is a very high risk of foreign object intrusion. Based on this, the spatial risk score of the foreign object in the train running direction is set as shown in Table 2.

[0126] Table 2 Spatial risk scoring criteria in the train running direction

[0127] Spatial position relationship Spatial risk score 100＜M≤150 0.5 50＜M≤100 1 M≤50 2

[0128] Step 4.1.4, calculate the comprehensive spatial risk score of the foreign object according to the spatial risk scores of the foreign object in the horizontal direction and the train running direction;

[0129] The formula for calculating the comprehensive spatial risk score of the foreign object in Step 4.1.4 is:

[0130] L = L h + Lv

[0131] Wherein, L is the comprehensive spatial risk score of the foreign object, and L h is the spatial risk score in the horizontal direction, and L v is the spatial risk score in the train running direction.

[0132] The comprehensive spatial risk score of the foreign object in step 4.1.4 is shown in Table 3 as follows:

[0133] Table 3 Comprehensive Spatial Risk Score of Foreign Objects

[0134]

[0135] Step 4.2: Set the category risk score of the foreign object according to the severity of the accidents caused by different foreign object categories;

[0136] The category risk score standard of the foreign object in step 4.2 is shown in Table 4 as follows:

[0137] Table 4 Category Risk Score Standard of Foreign Objects

[0138] Foreign object type Risk score Pedestrian 2 Large foreign object 1.5 Floating object 1 Small foreign object 0.5

[0139] Step 4.3: Based on the risk matrix theory, comprehensively consider the comprehensive spatial risk score of the foreign object and the category risk score of the foreign object, calculate the risk score of foreign object intrusion in rail transit, and finally output the risk level according to the score.

[0140] The risk score of foreign object intrusion in rail transit in step 4.3 is:

[0141] R = L × P

[0142] Wherein, R is the risk score of foreign object intrusion in rail transit, L is the comprehensive spatial risk score of the foreign object, and P is the category risk score of the foreign object.

[0143] The risk level of foreign object intrusion in rail transit in step 4.3 is:

[0144] When R ≤ 1, it means that a small foreign object appears outside the track area, and the risk level is the lowest at this time, which is risk level I; when 1 < R ≤ 2.5, it means that there is a small foreign object intrusion in the track area or other foreign objects outside the track, and there is a certain risk at this time, and the risk level is risk level II; when 2.5 < R ≤ 4, it means that one of the three situations occurs: there is a floating object intrusion in the nearby track area, a pedestrian approaches outside the track area, or there is a large foreign object intrusion in the far track; at this time, the risk is relatively high, which is risk level III; when 4 < R ≤ 8, it means that there is a large foreign object intrusion in the nearby track area or a pedestrian intrusion in the track area; at this time, the risk is extremely high, and the risk level is the highest, which is risk level IV. The risk level of foreign object intrusion in rail transit is shown in Table 5.

[0145] Table 5 Risk Levels of Foreign Object Intrusion in Rail Transit

[0146] Risk score R≤1 1＜R≤2.5 2.5＜R≤4 4＜R≤8 Risk level Level I risk Level II risk Level III risk Level IV risk

[0147] In summary, the common detection process for foreign object intrusion in rail transit generally involves segmenting the track first and then detecting foreign objects. Since the time-consuming of the track segmentation step is much higher than that of the foreign object detection step, when there are no foreign objects in the image, the track segmentation step will waste a lot of time. In this embodiment, a detection process is adopted where foreign objects are detected first, and track segmentation is only performed when there are foreign objects in the image, achieving real-time detection of foreign object intrusion and effectively improving the speed of the entire risk identification and assessment process for foreign object intrusion in rail transit. Based on three key risk factors, namely the positional relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, this embodiment comprehensively evaluates the risk level of foreign object intrusion in rail transit. Compared with other risk assessment methods, this method takes more comprehensive factors into account, and the evaluation system is more reasonable. It can accurately judge the intrusion risks of various foreign objects in real time in complex and diverse rail transit scenarios, ensuring the safe operation of the rail transit scenario.

[0148] Embodiment 3

[0149] In this Embodiment 3, a method for detecting intrusion into a rail transit tunnel is provided, and this method includes the following steps:

[0150] Step 1: Use an on-vehicle forward-looking camera installed at the head of the train to collect image data of the operating environment in front of the train in real time;

[0151] Step 2: Through a foreign object detection algorithm, perform foreign object detection on the image data in Step 1 frame by frame. If a foreign object is detected in a certain frame of the image, obtain the category and position information of the detected foreign object; otherwise, continue to detect the next frame of the image until a foreign object is detected in a certain frame of the image.

[0152] The specific steps of Step 2 include the following steps:

[0153] Step 2.1, construct a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects;

[0154] Step 2.2, use the foreign object detection algorithm to train and validate the foreign object data set to obtain the optimal weights;

[0155] Step 2.3, use this foreign object detection algorithm to detect each frame of image data obtained in Step 1, and output the category and position information of the foreign object when a foreign object is detected.

[0156] Step 3: When a foreign object is detected in a certain frame of image, segment the track area in this frame of image through the track segmentation algorithm to obtain the position information of the track area;

[0157] The specific steps of Step 3 are as follows:

[0158] Step 3.1, construct a track segmentation data set, including two categories: track area and background;

[0159] Step 3.2, if the foreign object detection algorithm detects a foreign object in a certain frame of image, segment the track area in this frame of image through the pre-trained track segmentation algorithm;

[0160] Step 3.3, use the output result to obtain the position information of the track area.

[0161] Step 4: Based on the three key risk factors of the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the risk level of foreign object intrusion into rail transit and output it.

[0162] The specific steps of Step 4 are as follows:

[0163] Step 4.1, set a comprehensive spatial risk score for the foreign object based on the position information of the foreign object and the relative position relationship between the foreign object and the track area.

[0164] The specific steps of Step 4.1 are as follows:

[0165] Step 4.1.1, take the on-vehicle camera as the origin, the train running direction as the Y-axis, parallel to the rail plane, and perpendicular to the train running direction as the X-axis to establish a plane rectangular coordinate system;

[0166] Step 4.1.2, set the spatial risk scoring standard in the horizontal direction according to the relative position relationship between the foreign object and the track area in the horizontal direction;

[0167] The distance calculation formula between the foreign object and the track area in the horizontal direction in Step 4.1.2 is:

[0168]

[0169] In the formula, X R is the abscissa of the lower right corner of the foreign object detection box in the image, X1 and X2 are the abscissas of the intersection points of the extended lines of the lower frame line of the foreign object detection box in the figure and the left and right rails respectively, N is the actual distance between the foreign object and the track area, and 1.435 is the actual rail gauge (unit: meter);

[0170] The spatial risk scoring standard in the horizontal direction in Step 4.1.2 is:

[0171] When the actual distance between the foreign object and the track area is greater than 1 meter, there is a low risk of foreign object intrusion; when the actual distance between the foreign object and the track area is less than 1 meter, there is a high risk of foreign object intrusion; when the foreign object is located within the track area, foreign object intrusion has occurred, and the risk is the greatest at this time. Based on this, the spatial risk score of the foreign object in the horizontal direction is set as shown in Table 1.

[0172] Table 1 Spatial risk scoring criteria in the horizontal direction

[0173] Spatial position relationship Spatial risk score N>1 0.5 N<1 1 Located within the orbital area 2

[0174] Step 4.1.3, set the spatial risk scoring criteria in the train running direction according to the distance between the foreign object and the train in the train running direction;

[0175] The calculation formula for the distance between the foreign object and the train in the train running direction in Step 4.1.3 is:

[0176]

[0177] In the formula, Y R is the vertical coordinate of the lower right corner of the foreign object detection frame in the figure, representing the pixel distance between the foreign object and the train in the train running direction, M is the actual distance between the foreign object and the train in the train running direction (unit: meter), Y0 is the farthest pixel distance captured by the camera in the train running direction in the figure, and 150 is the actual farthest shooting distance of the on-vehicle camera in the train running direction.

[0178] The spatial risk scoring criteria in the train running direction in Step 4.1.3 are:

[0179] When 100 < M ≤ 150, at this time the foreign object is far from the train, and there is a low risk of foreign object intrusion; when 50 < M ≤ 100, since the foreign object is relatively close to the train, there is a high risk of foreign object intrusion; when M ≤ 50, since the foreign object is very close to the train, there is a very high risk of foreign object intrusion. Based on this, the spatial risk score of the foreign object in the train running direction is set as shown in Table 2.

[0180] Table 2 Spatial risk scoring criteria in the train running direction

[0181] Spatial position relationship Spatial risk score 100＜M≤150 0.5 50＜M≤100 1 M≤50 2

[0182] Step 4.1.4, calculate the comprehensive spatial risk score of the foreign object according to the spatial risk scores in the horizontal direction and the train running direction;

[0183] The calculation formula for the comprehensive spatial risk score of the foreign object in Step 4.1.4 is:

[0184] L = L h + Lv

[0185] In the formula, L is the comprehensive spatial risk score of the foreign object, and L h is the spatial risk score in the horizontal direction, and L v is the spatial risk score in the train running direction.

[0186] The comprehensive spatial risk score of the foreign object in step 4.1.4 is shown in Table 3 as follows:

[0187] Table 3 Comprehensive Spatial Risk Score of Foreign Objects

[0188]

[0189] Step 4.2: Set the category risk score of the foreign object according to the severity of the accidents caused by different foreign object categories;

[0190] The category risk score standard of the foreign object in step 4.2 is shown in Table 4 as follows:

[0191] Table 4 Category Risk Score Standard of Foreign Objects

[0192] Foreign object type Risk score Pedestrian 2 Large foreign object 1.5 Floating object 1 Small foreign object 0.5

[0193] Step 4.3: Based on the risk matrix theory, comprehensively consider the comprehensive spatial risk score of the foreign object and the category risk score of the foreign object, calculate the risk score of foreign object intrusion in rail transit, and finally output the risk level according to the score.

[0194] The risk score of foreign object intrusion in rail transit in step 4.3 is as follows:

[0195] R = L × P

[0196] In the formula, R is the risk score of foreign object intrusion in rail transit, L is the comprehensive spatial risk score of the foreign object, and P is the category risk score of the foreign object.

[0197] The risk level of foreign object intrusion in rail transit in step 4.3 is as follows:

[0198] When R ≤ 1, it represents that a small foreign object appears outside the track area, and the risk level is the lowest at this time, which is level I risk; when 1 < R ≤ 2.5, it represents that there is a small foreign object intrusion in the track area or there are other foreign objects outside the track, and there is a certain risk at this time, and the risk level is level II risk; when 2.5 < R ≤ 4, it represents one of the three situations: there is a floating object intrusion in the nearby track area, a pedestrian approaches outside the track area, or there is a large foreign object intrusion in the far track. At this time, the risk is relatively high, which is level III risk; when 4 < R ≤ 8, it represents that there is a large foreign object intrusion in the nearby track area or a pedestrian intrusion in the track area. At this time, the risk is extremely high, and the risk level is the highest, which is level IV risk. The risk level of foreign object intrusion in rail transit is shown in Table 5.

[0199] Table 5 Risk Levels of Foreign Object Invasion in Rail Transit

[0200] Risk score R≤1 1＜R≤2.5 2.5＜R≤4 4＜R≤8 Risk level Level I risk Level II risk Level III risk Level IV risk

[0201] In this embodiment, the YOLOv5 algorithm is used as the foreign object detection algorithm, and the STDC algorithm is used as the track segmentation algorithm. The risk identification and assessment of foreign object invasion in rail transit are carried out according to the above technical solutions. First, two datasets constructed by 1788 images of four types of foreign objects, namely pedestrians, rocks, plastic sheets, and cardboard boxes, invading or about to invade different tracks from the perspective of on-vehicle cameras are used to train and test these two algorithms respectively. The test results show that both algorithms have high speed and accuracy. The test results of the two are shown in Tables 6-8, Figures 2 - 3 as shown below.

[0202] Table 6 Test Results of Foreign Object Detection Accuracy

[0203] Category Number of images Number of samples Accuracy rate Recall rate F1 Total 179 526 0.875 0.806 0.839 Pedestrian 179 262 0.840 0.821 0.830 Rock 179 154 0.896 0.897 0.896 Plastic sheet 179 24 0.900 0.749 0.818 Cardboard box 179 86 0.865 0.756 0.807

[0204] Table 7 Test Results of Foreign Object Detection Speed

[0205] Pre - process time Inference time Non - maximum suppression (NMS) time 0.6 milliseconds 7.8 milliseconds 0.9 milliseconds

[0206] Table 8 Test Results of Track Segmentation Algorithm

[0207] <![CDATA[IoU background > <![CDATA[IoU trac k]]> MIoU <![CDATA[PA background > <![CDATA[PA trac k]]> MPA 0.9983 0.9706 0.9845 0.9991 0.9877 0.9934

[0208] Subsequently, based on the position information of the foreign object in the image and the relative position relationship between the foreign object and the track area, the comprehensive spatial risk score of the foreign object is calculated. At the same time, according to the severity of the accidents caused by different foreign object categories, the category risk score of the foreign object is set. In this embodiment, rocks belong to large foreign objects, plastic sheets belong to floating objects, and cardboard boxes belong to small foreign objects. The risk scores of pedestrians, rocks, plastic sheets, and cardboard boxes are set as shown in Table 9 according to the foreign object category risk score standard in the above technical solutions:

[0209] Table 9 Foreign Object Category Risk Scores

[0210] Foreign object type Risk score Pedestrian 2 Rock 1.5 Plastic sheet 1 Cardboard box 0.5

[0211] Finally, based on the risk matrix theory, combining the comprehensive spatial risk score of the foreign object and the category risk score of the foreign object, the risk score of foreign object invasion in rail transit is calculated, and finally the risk level is output according to the score. The risk score of foreign object invasion in rail transit in this embodiment is shown in Table 10.

[0212] Table 10 Risk Score Table of Foreign Object Invasion in Rail Transit

[0213]

[0214] According to the calculation results in Table 10, in this embodiment, a level-I risk represents that there is a cardboard box foreign object outside the track area. Relevant staff can effectively avoid this risk by cleaning it in time; a level-II risk represents that there is a plastic sheet, a rock or a pedestrian outside the track area, or there is a cardboard box intrusion inside the track area. At this time, the driver and relevant personnel need to pay concentrated attention to prevent dangerous situations from occurring; a level-III risk represents one of the three situations where there is a plastic sheet intrusion in the track area close to the train, a rock intrusion in the track far from the train, or a pedestrian is close to the train but does not intrude into the track area. Relevant personnel need to handle it in time to avoid dangerous situations; a level-IV risk represents that there is a rock intrusion in the track area close to the train or a pedestrian intrusion in the track area. At this time, the risk is extremely high, and serious accidents may occur if not handled properly, and the risk level is the highest.

[0215] Embodiment 4

[0216] Embodiment 4 of the present invention provides a non-transitory computer-readable storage medium, which is used to store computer instructions. When the computer instructions are executed by a processor, the above-mentioned risk identification and assessment method for foreign object intrusion in rail transit is realized. The method includes:

[0217] Real-time collect image data of the operating environment in front of the train;

[0218] Use a pre-trained foreign object detection model to detect foreign objects frame by frame for the collected image data. If a foreign object is detected in a certain frame of the image, obtain the category and position information of the detected foreign object. Otherwise, continue to detect the next frame of the image until a foreign object is detected in a certain frame of the image; among them, the training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories of pedestrians, large foreign objects, small foreign objects and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weights; using the foreign object detection algorithm to detect each frame of the obtained image data, and output the category and position information of the foreign object when a foreign object is detected;

[0219] When a foreign object is detected in a certain frame of the image, use a pre-trained track segmentation model to segment the track area in this frame of the image to obtain the position information of the track area; among them, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories of track area and background; if the foreign object detection algorithm detects a foreign object in a certain frame of the image, then use a pre-trained track segmentation algorithm to segment the track area in this frame of the image; use the output result to obtain the position information of the track area;

[0220] Based on three key risk factors, namely the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the risk level of foreign object intrusion in rail transit and output it.

[0221] Example 5

[0222] Example 5 provides a computer device, including a memory and a processor, where the processor and the memory communicate with each other. The memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the rail transit foreign object intrusion risk identification and assessment method as described above. The method includes:

[0223] Real-time collect image data of the operating environment in front of the train;

[0224] Use a pre-trained foreign object detection model to perform foreign object detection on the collected image data frame by frame. If a foreign object is detected in a certain frame of the image, obtain the category and location information of the detected foreign object; otherwise, continue to detect the next frame of the image until a foreign object is detected in a certain frame of the image. Among them, the training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weights; using the foreign object detection algorithm to detect each frame of the obtained image data, and output the category and location information of the foreign object when a foreign object is detected;

[0225] When a foreign object is detected in a certain frame of the image, use a pre-trained track segmentation model to segment the track area in this frame of the image to obtain the location information of the track area. Among them, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories: track area and background; if the foreign object detection algorithm detects a foreign object in a certain frame of the image, then use a pre-trained track segmentation algorithm to segment the track area in this frame of the image; use the output result to obtain the location information of the track area;

[0226] Based on three key risk factors, namely the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the rail transit foreign object intrusion risk level and output it.

[0227] Example 6

[0228] Example 6 provides an electronic device, including: a processor, a memory, and a computer program. Among them, the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device runs, the processor executes the computer program stored in the memory so that the electronic device executes the instructions for implementing the rail transit foreign object intrusion risk identification and assessment method as described above. The method includes:

[0229] Real-time collect image data of the operating environment in front of the train;

[0230] Using a pre-trained foreign object detection model, perform foreign object detection on the collected image data frame by frame. If a foreign object is detected in a certain frame of the image, obtain the category and location information of the detected foreign object; otherwise, continue to detect the next frame of the image until a foreign object is detected in a certain frame of the image; among them, the training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects, and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weights; using the foreign object detection algorithm to detect each frame of the obtained image data, and output the category and location information of the foreign object when a foreign object is detected;

[0231] When a foreign object is detected in a certain frame of the image, segment the track area in this frame of the image through a pre-trained track segmentation model to obtain the location information of the track area; among them, the training of the track segmentation model includes: constructing a track segmentation data set, including two categories: track area and background; if the foreign object detection algorithm detects a foreign object in a certain frame of the image, then segment the track area in this frame of the image through a pre-trained track segmentation algorithm; use the output result to obtain the location information of the track area;

[0232] Based on three key risk factors: the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the category of the foreign object, comprehensively evaluate the risk level of rail transit foreign object intrusion and output it.

[0233] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0234] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate for implementing in the process Figure 1 one process or multiple processes and / or blocks Figure 1 a device for the functions specified in one block or multiple blocks.

[0235] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one or more of the procedures Figure 1 one or more procedures and / or blocks Figure 1 specified in a block or blocks.

[0236] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to perform a series of operational steps on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the procedures Figure 1 one or more procedures and / or blocks Figure 1 specified in a block or blocks.

[0237] Although the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, they are not limitations on the scope of the present invention. Those skilled in the art should understand that, based on the technical solutions disclosed in the present invention, various modifications or variations that can be made by those skilled in the art without creative efforts should be covered within the scope of the present invention.

Claims

1. A method for identifying and assessing the risk of foreign body intrusion in rail transit, characterized in that: include: Collect image data of the running environment ahead of the train in real time; Using a pre-trained foreign object detection model, the collected image data is detected for foreign objects frame by frame. If a foreign object is detected in a certain frame of image, the category and location information of the detected foreign object is obtained. Otherwise, the next frame of image is detected until a foreign object is detected in a certain frame of image. The training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weight; using the foreign object detection algorithm to detect each frame of image data obtained, and outputting the category and location information of the foreign object when a foreign object is detected; When a foreign object is detected in a certain frame image, the track area in the frame image is segmented by a pre-trained track segmentation model to obtain the position information of the track area; wherein the training of the track segmentation model includes: constructing a track segmentation data set, including two categories of track area and background; if the foreign object detection algorithm detects that a foreign object exists in a certain frame image, the track area in the frame image is segmented by the pre-trained track segmentation algorithm; and the position information of the track area is obtained by using the output result; Based on the three key risk factors of the relative position between the foreign object and the track area, the distance between the foreign object and the train, and the type of the foreign object, the risk level of foreign object intrusion in rail transit is comprehensively evaluated and output; including: The spatial risk comprehensive score of foreign objects is set based on the location information of foreign objects and the relative position relationship between foreign objects and the track area; the category risk score of foreign objects is set based on the severity of accidents caused by different categories of foreign objects; based on the risk matrix theory, the spatial risk comprehensive score and category risk score of foreign objects are combined to calculate the foreign object intrusion risk score of rail transit, and finally the risk level is output according to the score; The specific process of setting the spatial risk comprehensive score of foreign objects based on the location information of foreign objects and the relative position relationship between foreign objects and the track area is as follows: A plane rectangular coordinate system is established with the onboard camera as the origin, the train running direction as the Y axis, parallel to the track plane, and perpendicular to the train running direction as the X axis; The spatial risk scoring standard in the horizontal direction is set according to the relative position relationship between the foreign matter and the track area in the horizontal direction; The spatial risk scoring standard in the train running direction is set according to the distance between the foreign object and the train in the train running direction; the specific process of setting the spatial risk scoring standard in the train running direction according to the distance between the foreign object and the train in the train running direction is as follows: Calculate the actual distance between the foreign object and the train in the direction of train operation. The calculation formula is as follows: Where Y R is the ordinate of the lower right corner of the foreign object detection frame in the figure, indicating the pixel distance between the foreign object and the train in the running direction of the train, M is the actual distance between the foreign object and the train in the running direction of the train, Y0 is the farthest pixel distance captured by the camera in the running direction of the train in the figure, and 150 is the actual farthest shooting distance of the on-board camera in the running direction of the train; According to the actual distance between the foreign object and the train in the direction of train operation, the spatial risk scoring standard in the direction of train operation is set as follows: When the spatial position relationship is 100<M≤150, the spatial risk score is 0.5; When the spatial position relationship is 50<M≤100, the spatial risk score is 1; When the spatial position relationship is M≤50, the spatial risk score is 2; The comprehensive spatial risk score of foreign objects is calculated based on the spatial risk scores of foreign objects in the horizontal direction and the train running direction.

2. The method for identifying and assessing the risk of foreign body intrusion in rail transit according to claim 1, characterized in that: The specific process of setting the horizontal spatial risk scoring standard according to the relative position relationship between foreign objects and the track area in the horizontal direction is as follows: Calculate the actual distance between the foreign object and the track area in the horizontal direction. The calculation formula is as follows: Where, X R is the horizontal coordinate of the lower right corner of the foreign object detection frame in the image, X1 and X2 are the horizontal coordinates of the intersection of the extension line of the lower frame line of the foreign object detection frame in the figure and the left and right rails, and N is the actual distance between the foreign object and the track area; According to the actual distance between the foreign object and the track area in the horizontal direction, the spatial risk scoring standard in the horizontal direction is set as follows: When the spatial position relationship is N>1, the spatial risk score is 0.5; When the spatial position relationship is N<1, the spatial risk score is 1; When the spatial position relationship is located within the track area, the spatial risk score is 2.

3. The method for identifying and assessing the risk of foreign body intrusion in rail transit according to claim 1, characterized in that: The comprehensive spatial risk score of foreign matter is: L=L h +L v Where, L is the comprehensive score of spatial risk of foreign matter, L h is the spatial risk score in the horizontal direction, L v Score the spatial risk in the direction of train movement; The risk score of foreign body intrusion in rail transit is: R=L×P Where R is the foreign body intrusion risk score of rail transit, L is the comprehensive spatial risk score of foreign bodies, and P is the category risk score of foreign bodies.

4. A rail transit foreign body intrusion risk identification and assessment system, characterized in that: include: An acquisition module is used to collect image data of the running environment in front of the train in real time; The detection module is used to use a pre-trained foreign object detection model to perform foreign object detection on the collected image data frame by frame. If a foreign object is detected in a certain frame of image, the category and location information of the detected foreign object is obtained. Otherwise, the next frame of image is detected until a foreign object is detected in a certain frame of image. The training of the foreign object detection model includes: constructing a foreign object detection data set, including four categories: pedestrians, large foreign objects, small foreign objects and floating objects; using a foreign object detection algorithm to train and verify the foreign object data set to obtain the optimal weight; using the foreign object detection algorithm to detect each frame of image data obtained, and outputting the category and location information of the foreign object when a foreign object is detected; The segmentation module is used to segment the track area in a frame image by using a pre-trained track segmentation model when a foreign object is detected in the frame image, and obtain the position information of the track area; wherein the training of the track segmentation model includes: constructing a track segmentation data set, including two categories of track area and background; if the foreign object detection algorithm detects that a foreign object exists in a frame image, segmenting the track area in the frame image by using the pre-trained track segmentation algorithm; and obtaining the position information of the track area by using the output result; The assessment module is used to comprehensively assess and output the risk level of foreign object intrusion in rail transit based on the three key risk factors: the relative position relationship between the foreign object and the track area, the distance between the foreign object and the train, and the type of the foreign object; including: The spatial risk comprehensive score of foreign objects is set based on the location information of foreign objects and the relative position relationship between foreign objects and the track area; the category risk score of foreign objects is set based on the severity of accidents caused by different categories of foreign objects; based on the risk matrix theory, the spatial risk comprehensive score and category risk score of foreign objects are combined to calculate the foreign object intrusion risk score of rail transit, and finally the risk level is output according to the score; The specific process of setting the spatial risk comprehensive score of foreign objects based on the location information of foreign objects and the relative position relationship between foreign objects and the track area is as follows: A plane rectangular coordinate system is established with the onboard camera as the origin, the train running direction as the Y axis, parallel to the track plane, and perpendicular to the train running direction as the X axis; The spatial risk scoring standard in the horizontal direction is set according to the relative position relationship between the foreign matter and the track area in the horizontal direction; The spatial risk scoring standard in the train running direction is set according to the distance between the foreign object and the train in the train running direction; the specific process of setting the spatial risk scoring standard in the train running direction according to the distance between the foreign object and the train in the train running direction is as follows: Calculate the actual distance between the foreign object and the train in the direction of train operation. The calculation formula is as follows: Where Y R is the ordinate of the lower right corner of the foreign object detection frame in the figure, indicating the pixel distance between the foreign object and the train in the running direction of the train, M is the actual distance between the foreign object and the train in the running direction of the train, Y0 is the farthest pixel distance captured by the camera in the running direction of the train in the figure, and 150 is the actual farthest shooting distance of the on-board camera in the running direction of the train; According to the actual distance between the foreign object and the train in the direction of train operation, the spatial risk scoring standard in the direction of train operation is set as follows: When the spatial position relationship is 100<M≤150, the spatial risk score is 0.5; When the spatial position relationship is 50<M≤100, the spatial risk score is 1; When the spatial position relationship is M≤50, the spatial risk score is 2; The comprehensive spatial risk score of foreign objects is calculated based on the spatial risk scores of foreign objects in the horizontal direction and the train running direction.

5. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the rail transit foreign object intrusion risk identification and assessment method as described in any one of claims 1-3 is implemented.

6. A computer device, characterized in that: It includes a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the rail transit foreign object intrusion risk identification and assessment method as described in any one of claims 1-3.

7. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the rail transit foreign object intrusion risk identification and assessment method as described in any one of claims 1-3.

Images