Data processing method and equipment for rail transit operation, and medium

Abstract

The invention discloses a data processing method and device for rail transit operation and a medium, and belongs to the technical field of rail transit, and the method comprises the following steps: collecting images and videos in a carriage and images of a waiting area, and storing the images and videos in a database; training the passenger article identification model; training the getting-off behavior recognition model; calculating the number of passengers in the compartment and the occupied space of each personal belonging; the number of passengers waiting for getting off and the occupied space of personal belongings waiting for getting off in the compartment are calculated; calculating the number of passengers waiting for getting on in the waiting area and the occupied space of personal belongings waiting for getting on; calculating the real-time congestion degree and the predicted congestion degree of the compartment; generating riding guide information according to the predicted congestion degree, and sending the riding guide information to the target station; according to the invention, passengers can be guided to select a relatively spacious compartment, important decision support is provided for rapid evacuation in an emergency, and the safety and comfort of the passengers are ensured.

Description

Technical Field

[0001] The present invention belongs to the field of rail transit technology, and specifically relates to a data processing method, device, and medium for rail transit operation. Background Art

[0002] Monitoring subway car crowding is crucial to improving passengers' travel experience and subway operation efficiency. By understanding the crowding of each car in real time, passengers can be guided to choose a more spacious car, thereby avoiding overcrowding, reducing passengers' waiting time, and improving riding comfort. In addition, crowding monitoring helps subway operators optimize car resource allocation, reasonably arrange trains and capacity, and reduce operating costs. In emergency situations, such as emergencies that lead to passenger evacuation, crowding monitoring can also provide important information for safe evacuation to ensure passenger safety; The existing method of monitoring the crowdedness of subway carriages mainly involves installing sensors at the entrance of the carriage to count the number of passengers entering and exiting, or using image recognition technology to analyze the images of passengers in the carriage to determine the number of people, thereby monitoring the crowdedness of subway carriages; However, in actual scenarios, the crowdedness of subway cars is not entirely determined by the number of passengers. The crowdedness of a car cannot be accurately judged by the number of passengers in the car alone, which affects the riding experience and operational efficiency. Therefore, people are in urgent need of a data processing method for rail transit operations to solve the above problems. Summary of the invention

[0003] The object of the present invention is to provide a data processing method for rail transit operation to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: A data processing method for rail transit operation, the method comprising the following steps: S1. Setting a camera in the carriage to collect images and videos in the carriage, pre-processing the collected images and videos, and storing them in a database; setting a camera in the waiting area to collect images of the waiting area, pre-processing the collected images, and storing them in a database; S2. Obtain historical images in the carriage from the database, annotate the historical images, and obtain annotated images; construct a passenger item recognition model, and train the passenger item recognition model based on the historical images and the annotated images; S3, obtaining historical videos in the carriage from the database, annotating the historical videos, and obtaining annotated videos; constructing a getting-off behavior recognition model, and training the getting-off behavior recognition model based on the historical videos and the annotated videos; S4, acquiring real-time images of the carriage from the database to obtain a first image set; extracting passengers and their belongings in the first image set through a passenger and item recognition model, and calculating the number of passengers in the carriage and the space occupied by each item; S5, obtaining the real-time video in the carriage from the database to obtain a first video set; extracting the passengers waiting to get off and their belongings in the first video set through the getting off behavior recognition model, and calculating the number of passengers waiting to get off in the carriage and the space occupied by each belonging to the passengers waiting to get off; S6. Acquire a real-time image of the waiting area in the target station from the database to obtain a second image set; extract the passengers waiting to board and their belongings in the second image set through the passenger and item recognition model, and calculate the number of passengers waiting to board in the waiting area and the space occupied by each of the belongings waiting to board; S7. Calculate the real-time congestion and predicted congestion of the carriage based on the number of passengers in the carriage and the space occupied by their belongings, the number of passengers waiting to get off the carriage and the space occupied by their belongings, and the number of passengers waiting to get on the carriage in the waiting area and the space occupied by their belongings; generate boarding guidance information based on the predicted congestion, and send the boarding guidance information to the target station.

[0005] According to the above technical solution, step S1 includes the following: S1-1. Cameras are installed in all carriages, and the collection range of the cameras covers the carriages; the cameras corresponding to the carriages are numbered according to the carriage numbers; the cameras collect images and videos in the corresponding carriages, and preprocess the collected images and videos, and the preprocessing includes denoising, normalization and contrast enhancement; the preprocessed images and videos are stored in a database; S1-2, setting up camera equipment in all waiting areas, the acquisition range of the camera equipment covers the waiting area; according to the waiting area number, the waiting area number is set corresponding to the carriage number, and the camera equipment corresponding to the waiting area is numbered; the camera equipment collects images in the corresponding waiting area, and pre-processes the collected images, and stores the pre-processed images and videos in the database; Multiple high-definition cameras are installed inside the carriage and the waiting area to obtain real-time images of the carriage interior. These cameras are evenly distributed in different locations in the carriage and the waiting area to ensure full coverage and capture the situation in the carriage and the waiting area without blind spots; The waiting areas correspond to the carriages one by one. There are multiple waiting areas at the target station. Passengers are located in the waiting areas. Each carriage corresponds to a waiting area, and passengers in the corresponding waiting area can enter the corresponding carriage.

[0006] According to the above technical solution, step S2 includes the following: S2-1, obtaining historical images of several different carriages from a database, marking passengers and objects in the obtained historical images of several different carriages, and marking the positions of the passengers and objects in the historical images to obtain marked images; Expert annotation of the images to identify passengers and various items such as backpacks, handbags, suitcases, etc., and mark their locations in the image (usually represented by bounding boxes); expert annotation refers to identification and annotation by professionals; S2-2, constructing a passenger item recognition model, taking the unlabeled historical image as input and the labeled image as output, training the passenger item recognition model, and obtaining a trained passenger item recognition model; The passenger-item recognition model is based on deep learning technology, especially convolutional neural network (CNN). After being trained with a large amount of image data, it can accurately recognize and distinguish between human bodies and various items.

[0007] According to the above technical solution, step S3 includes the following: S3-1. Obtaining several historical videos in different carriages from a database, extracting behavioral features of passengers in the obtained historical videos in the different carriages, wherein the behavioral features include standing behavior, sitting behavior, walking behavior, and interactive behavior; marking the behavioral features of passengers related to getting off the train, as well as the corresponding personal belongings of the passengers in the historical videos, and obtaining a marked video; The passenger's behavioral characteristics related to getting off the train, including actions of packing personal belongings, getting up from a seat and walking toward the door, or walking toward the door from other areas of the train, etc.; professional personnel shall identify and annotate the behavioral characteristics in the video; S3-2, constructing a getting-off behavior recognition model, taking the unlabeled historical video as input and the labeled video as output, training the getting-off behavior recognition model, and obtaining a trained getting-off behavior recognition model; The model is trained using labeled videos and supervised learning to learn the feature representations associated with getting off the bus, so that it can accurately identify passengers getting off the bus and their belongings.

[0008] According to the above technical solution, step S4 includes the following: S4-1. Matching real-time images in different carriages from a database according to the current time; numbering the real-time images in each carriage according to the carriage number and the current time to obtain a carriage image number; and counting the real-time images in all carriages at the current time to obtain a first image set; S4-2, according to the carriage image numbers of the images in the first image set, they are arranged from large to small, one image is extracted from the first image set each time, and the trained passenger item recognition model is input, and the labeled image of the passenger and his / her belongings in the corresponding image is output, and the number of passengers in the corresponding labeled image and the item data of each item are counted, and the item data includes size and shape, and the area ratio of the corresponding item in the corresponding labeled image in the corresponding carriage is calculated, and the actual space size of the corresponding carriage is extracted from the database, and the product of the actual size of the corresponding carriage and the area ratio of the corresponding item in the corresponding labeled image in the corresponding carriage is calculated as the occupied space of the corresponding item; According to the carriage image numbers of the images in the first image set, the carriages are matched to obtain the number of passengers in the corresponding carriages and the space occupied by each carry-on item; Provide data support for the analysis of real-time congestion and predicted congestion in subsequent carriages.

[0009] According to the above technical solution, step S5 includes the following: S5-1. According to the current time, real-time videos in different carriages are matched from the database; according to the carriage number and the current time, the real-time videos in each carriage are numbered to obtain the carriage video number; the real-time videos in all carriages at the current time are counted to obtain a first video set; S5-2, according to the carriage video numbers of the images in the first video set, they are arranged from large to small, and one video is extracted from the first video set each time, and the trained getting-off behavior recognition model is input, and a labeled video of the passengers and their belongings in the corresponding video is output; the passengers marked in the labeled video are recorded as passengers to be getting off; the belongings of the passengers marked in the labeled video are recorded as belongings to be getting off; Count the number of passengers waiting to get off the bus in the annotated video, intercept the image of the corresponding personal belongings of the passengers waiting to get off the bus in the corresponding annotated video, extract the item data of the corresponding personal belongings of the passengers waiting to get off the bus in the intercepted image, and calculate the space occupied by the corresponding personal belongings of the passengers waiting to get off the bus in the intercepted image; According to the carriage video number of the video in the first video set, the carriage is matched with the carriage to obtain the number of passengers waiting to get off in the corresponding carriage and the space occupied by the personal belongings of the passengers waiting to get off; Provide data support for the analysis of real-time congestion and predicted congestion in subsequent carriages.

[0010] According to the above technical solution, step S6 includes the following: S6-1, taking the next station as the target station, matching the real-time images in different waiting areas of the target station from the database according to the current time; numbering the real-time images in the waiting area according to the waiting area number and the current time to obtain the waiting area image number; counting the real-time images in all waiting areas of the target station at the current time to obtain a second image set; S6-2, according to the waiting area image numbers of the images in the second image set, the images are arranged from large to small, one image is extracted from the second image set each time, and the trained passenger item recognition model is inputted, and annotated images of the passengers and their belongings in the corresponding images are outputted; the passengers annotated in the annotated images are recorded as passengers to be boarded; and the belongings of the passengers annotated in the annotated images are recorded as belongings to be boarded; Count the number of passengers waiting to board the bus in the annotated image, extract the item data of each item to be boarded in the corresponding annotated image, and calculate the space occupied by each item to be boarded in the corresponding annotated image; According to the waiting area picture number of the picture in the second image set, the picture is matched with the waiting area, and the number of passengers waiting to board the bus in the corresponding waiting area and the space occupied by the personal belongings of the passengers waiting to board the bus are obtained; Provide data support for the analysis of real-time congestion and predicted congestion in subsequent carriages.

[0011] According to the above technical solution, step S7 includes the following: S7-1. Calculate the average of the occupied spaces of all the carry-on items in the corresponding compartment according to the occupied spaces of the carry-on items in the compartment, as the average occupied space of the carry-on items in the corresponding compartment; The total occupied space of the carriage is calculated using the formula: M=(P×Ap×Wp)+(I×Ai×Wi); where M represents the total occupied space of the carriage, P represents the number of passengers in the carriage, Ap represents the average occupied space of passengers, Wp represents the passenger space weight, I represents the number of carry-on items in the carriage, Ai represents the average occupied space of carry-on items in the carriage, and Wi represents the item space weight; The value of Ap is obtained by dividing the actual available space in the carriage by the maximum number of passengers in the carriage to ensure that there is enough space for each passenger; Ap is determined by the carriage design parameters, and this value will be adjusted according to the actual design of the carriage to more accurately reflect the average space occupied by passengers in the carriage; The real-time crowdedness of the carriage is calculated using the formula: N=M / (S×C); where N represents the real-time crowdedness of the carriage, S represents the actual available space in the carriage, and C represents the uniformity coefficient of the carriage; The process of obtaining the uniformity of the carriage is as follows: Using image recognition technology, the real-time image of the carriage is divided into several areas of equal area, and the formula is used: ρ i =μ i / σ, calculate the passenger density of each area in the real-time image, where ρ i represents the passenger density of the ith area in the real-time image, μ i Represents the number of passengers in the i-th area in the real-time image, σ represents the area of ​​each area in the real-time image; the average passenger density of the carriage is calculated using the formula: ρ1=P / (i×σ), where ρ1 represents the average passenger density of the carriage; using the formula: , calculate the uniformity coefficient of the carriage, where C represents the uniformity coefficient of the carriage; Wp and Wi can be adjusted according to actual conditions to reflect the actual impact of passengers and items on crowding; the value of C ranges from 0 to 1, and the closer the value is to 1, the more uniform the passenger distribution is; S7-2, calculating the average of the occupied spaces of all the carry-on items to be disembarked in the corresponding carriage according to the occupied spaces of the carry-on items to be disembarked in the carriage, as the average occupied space of the carry-on items to be disembarked in the corresponding carriage; The predicted additional space in the carriage is calculated using the formula: M1=(P1×Ap×Wp)+(I1×A1×Wi); where M1 represents the predicted additional space in the carriage, P1 represents the number of passengers waiting to get off in the carriage, I1 represents the number of carry-on items in the carriage, and A1 represents the average space occupied by carry-on items in the carriage. According to the occupied space of each carry-on item to be boarded in the waiting area, the average occupied space of all carry-on items to be boarded in the corresponding waiting area is calculated as the average occupied space of the carry-on items to be boarded in the corresponding waiting area, and the corresponding carriage number is matched according to the waiting area number; The predicted reduced space of the carriage is calculated using the formula; M2=(P2×Ap×Wp)+(I2×A2×Wi); where M2 represents the predicted reduced space of the carriage, P2 represents the number of passengers waiting to board in the corresponding waiting area of ​​the carriage, I2 represents the number of carry-on items waiting to board in the corresponding waiting area of ​​the carriage, and A2 represents the average space occupied by carry-on items waiting to board in the corresponding waiting area of ​​the carriage; The predicted congestion of the carriage is calculated using the formula: N1=(M +M1-M2) / (S×C); where N1 represents the predicted congestion of the carriage; S7-3, setting a first threshold and a second threshold, if the predicted congestion of the carriage is less than the first threshold, the carriage state is comfortable; if the predicted congestion of the carriage is greater than or equal to the first threshold and the predicted congestion of the carriage is less than the second threshold, the carriage state is moderate; if the predicted congestion of the carriage is greater than or equal to the second threshold, the carriage state is crowded; Generate boarding guidance information, wherein the boarding guidance information is used to guide passengers at the target station to board a carriage with a comfortable carriage state; The boarding guidance information usually includes the specific location of the carriage, the carriage number and the direction guidance relative to the passenger's current location. The boarding guidance information is designed to be clear, concise and easy for passengers to understand.

[0012] A data processing device for rail transit operation, comprising a processor and a memory; The memory is coupled to the processor, and is used to store computer program codes, wherein the computer program codes include computer instructions.

[0013] A data processing medium for rail transit operation includes a storage medium for storing computer instructions.

[0014] Compared with the prior art, the beneficial effects achieved by the present invention are: The present invention obtains the real-time and predicted congestion of the carriages by monitoring and analyzing the historical data and real-time data of the carriages and waiting areas, and dynamically generates riding guidance information, which helps to guide passengers to choose more spacious carriages, improve riding efficiency and riding comfort, and also provides important decision-making support for rapid evacuation in emergency situations, ensuring the safety and comfort of passengers, thereby optimizing the passengers' riding experience and improving subway operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 It is a flow chart of a data processing method for rail transit operation of the present invention; Figure 2 It is a structural schematic diagram of a data processing device for rail transit operation according to the present invention.

[0016] In the figure: 401, central processing unit; 402, read-only memory; 403, random access memory; 404, bus; 405, I / O interface; 406, input part; 407, output part; 408, storage part; 409, communication part; 410, drive; 411, removable medium. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technical visitors in this field without creative work are within the scope of protection of the present invention.

[0018] See also Figure 1 , the present invention provides a technical solution: A data processing method for rail transit operation, the method comprising the following steps: S1. Setting a camera in the carriage to collect images and videos in the carriage, pre-processing the collected images and videos, and storing them in a database; setting a camera in the waiting area to collect images of the waiting area, pre-processing the collected images, and storing them in a database; S2. Obtain historical images in the carriage from the database, annotate the historical images, and obtain annotated images; construct a passenger item recognition model, and train the passenger item recognition model based on the historical images and the annotated images; S3, obtaining historical videos in the carriage from the database, annotating the historical videos, and obtaining annotated videos; constructing a getting-off behavior recognition model, and training the getting-off behavior recognition model based on the historical videos and the annotated videos; S4, acquiring real-time images of the carriage from the database to obtain a first image set; extracting passengers and their belongings in the first image set through a passenger and item recognition model, and calculating the number of passengers in the carriage and the space occupied by each item; S5, obtaining the real-time video in the carriage from the database to obtain a first video set; extracting the passengers waiting to get off and their belongings in the first video set through the getting off behavior recognition model, and calculating the number of passengers waiting to get off in the carriage and the space occupied by each belonging to the passengers waiting to get off; S6. Acquire a real-time image of the waiting area in the target station from the database to obtain a second image set; extract the passengers waiting to board and their belongings in the second image set through the passenger and item recognition model, and calculate the number of passengers waiting to board in the waiting area and the space occupied by each of the belongings waiting to board; S7. Calculate the real-time congestion and predicted congestion of the carriage based on the number of passengers in the carriage and the space occupied by their belongings, the number of passengers waiting to get off the carriage and the space occupied by their belongings, and the number of passengers waiting to get on the carriage in the waiting area and the space occupied by their belongings; generate boarding guidance information based on the predicted congestion, and send the boarding guidance information to the target station.

[0019] According to the above technical solution, step S1 includes the following: S1-1. Cameras are installed in all carriages, and the collection range of the cameras covers the carriages; the cameras corresponding to the carriages are numbered according to the carriage numbers; the cameras collect images and videos in the corresponding carriages, and preprocess the collected images and videos, and the preprocessing includes denoising, normalization and contrast enhancement; the preprocessed images and videos are stored in a database; S1-2, setting up camera equipment in all waiting areas, the acquisition range of the camera equipment covers the waiting area; according to the waiting area number, the waiting area number is set corresponding to the carriage number, and the camera equipment corresponding to the waiting area is numbered; the camera equipment collects images in the corresponding waiting area, and pre-processes the collected images, and stores the pre-processed images and videos in the database; Multiple high-definition cameras are installed inside the carriage and the waiting area to obtain real-time images of the carriage interior. These cameras are evenly distributed in different locations in the carriage and the waiting area to ensure full coverage and capture the situation in the carriage and the waiting area without blind spots; The waiting areas correspond to the carriages one by one. There are multiple waiting areas at the target station. Passengers are located in the waiting areas. Each carriage corresponds to a waiting area, and passengers in the corresponding waiting area can enter the corresponding carriage.

[0020] According to the above technical solution, step S2 includes the following: S2-1, obtaining historical images of several different carriages from a database, marking passengers and objects in the obtained historical images of several different carriages, and marking the positions of the passengers and objects in the historical images to obtain marked images; Expert annotation of the images to identify passengers and various items such as backpacks, handbags, suitcases, etc., and mark their locations in the image (usually represented by bounding boxes); expert annotation refers to identification and annotation by professionals; S2-2, constructing a passenger item recognition model, taking the unlabeled historical image as input and the labeled image as output, training the passenger item recognition model, and obtaining a trained passenger item recognition model; The passenger-item recognition model is based on deep learning technology, especially convolutional neural network (CNN). After being trained with a large amount of image data, it can accurately recognize and distinguish between human bodies and various items.

[0021] According to the above technical solution, step S3 includes the following: S3-1. Obtaining several historical videos in different carriages from a database, extracting behavioral features of passengers in the obtained historical videos in the different carriages, wherein the behavioral features include standing behavior, sitting behavior, walking behavior, and interactive behavior; marking the behavioral features of passengers related to getting off the train, as well as the corresponding personal belongings of the passengers in the historical videos, and obtaining a marked video; The passenger's behavioral characteristics related to getting off the train, including actions of packing personal belongings, getting up from a seat and walking toward the door, or walking toward the door from other areas of the train, etc.; professional personnel shall identify and annotate the behavioral characteristics in the video; S3-2, constructing a getting-off behavior recognition model, taking the unlabeled historical video as input and the labeled video as output, training the getting-off behavior recognition model, and obtaining a trained getting-off behavior recognition model; The model is trained using labeled videos and supervised learning to learn the feature representations associated with getting off the bus, so that it can accurately identify passengers getting off the bus and their belongings.

[0022] According to the above technical solution, step S4 includes the following: S4-1. Matching real-time images in different carriages from a database according to the current time; numbering the real-time images in each carriage according to the carriage number and the current time to obtain a carriage image number; and counting the real-time images in all carriages at the current time to obtain a first image set; S4-2, according to the carriage image numbers of the images in the first image set, they are arranged from large to small, one image is extracted from the first image set each time, and the trained passenger item recognition model is input, and the labeled image of the passenger and his / her belongings in the corresponding image is output, and the number of passengers in the corresponding labeled image and the item data of each item are counted, and the item data includes size and shape, and the area ratio of the corresponding item in the corresponding labeled image in the corresponding carriage is calculated, and the actual space size of the corresponding carriage is extracted from the database, and the product of the actual size of the corresponding carriage and the area ratio of the corresponding item in the corresponding labeled image in the corresponding carriage is calculated as the occupied space of the corresponding item; According to the carriage image numbers of the images in the first image set, the carriages are matched to obtain the number of passengers in the corresponding carriages and the space occupied by each carry-on item; Provide data support for the analysis of real-time congestion and predicted congestion in subsequent carriages.

[0023] According to the above technical solution, step S5 includes the following: S5-1. According to the current time, real-time videos in different carriages are matched from the database; according to the carriage number and the current time, the real-time videos in each carriage are numbered to obtain the carriage video number; the real-time videos in all carriages at the current time are counted to obtain a first video set; S5-2, according to the carriage video numbers of the images in the first video set, they are arranged from large to small, and one video is extracted from the first video set each time, and the trained getting-off behavior recognition model is input, and a labeled video of the passengers and their belongings in the corresponding video is output; the passengers marked in the labeled video are recorded as passengers to be getting off; the belongings of the passengers marked in the labeled video are recorded as belongings to be getting off; Count the number of passengers waiting to get off the bus in the annotated video, intercept the image of the corresponding personal belongings of the passengers waiting to get off the bus in the corresponding annotated video, extract the item data of the corresponding personal belongings of the passengers waiting to get off the bus in the intercepted image, and calculate the space occupied by the corresponding personal belongings of the passengers waiting to get off the bus in the intercepted image; According to the carriage video number of the video in the first video set, the carriage is matched with the carriage to obtain the number of passengers waiting to get off in the corresponding carriage and the space occupied by the personal belongings of the passengers waiting to get off; Provide data support for the analysis of real-time congestion and predicted congestion in subsequent carriages.

[0024] According to the above technical solution, step S6 includes the following: S6-1, taking the next station as the target station, matching the real-time images in different waiting areas of the target station from the database according to the current time; numbering the real-time images in the waiting area according to the waiting area number and the current time to obtain the waiting area image number; counting the real-time images in all waiting areas of the target station at the current time to obtain a second image set; S6-2, according to the waiting area image numbers of the images in the second image set, the images are arranged from large to small, one image is extracted from the second image set each time, and the trained passenger item recognition model is inputted, and annotated images of the passengers and their belongings in the corresponding images are outputted; the passengers annotated in the annotated images are recorded as passengers to be boarded; and the belongings of the passengers annotated in the annotated images are recorded as belongings to be boarded; Count the number of passengers waiting to board the bus in the annotated image, extract the item data of each item to be boarded in the corresponding annotated image, and calculate the space occupied by each item to be boarded in the corresponding annotated image; According to the waiting area picture number of the picture in the second image set, the picture is matched with the waiting area, and the number of passengers waiting to board the bus in the corresponding waiting area and the space occupied by the personal belongings of the passengers waiting to board the bus are obtained; Provide data support for the analysis of real-time congestion and predicted congestion in subsequent carriages.

[0025] According to the above technical solution, step S7 includes the following: 7-1. Calculate the average of the space occupied by all the carry-on items in the corresponding compartment based on the space occupied by each carry-on item in the compartment, and use it as the average space occupied by the carry-on items in the corresponding compartment; The total occupied space of the carriage is calculated using the formula: M=(P×Ap×Wp)+(I×Ai×Wi); where M represents the total occupied space of the carriage, P represents the number of passengers in the carriage, Ap represents the average occupied space of passengers, Wp represents the passenger space weight, I represents the number of carry-on items in the carriage, Ai represents the average occupied space of carry-on items in the carriage, and Wi represents the item space weight; The value of Ap is obtained by dividing the actual available space in the carriage by the maximum number of passengers in the carriage to ensure that there is enough space for each passenger; Ap is determined by the carriage design parameters, and this value will be adjusted according to the actual design of the carriage to more accurately reflect the average space occupied by passengers in the carriage; The real-time crowdedness of the carriage is calculated using the formula: N=M / (S×C); where N represents the real-time crowdedness of the carriage, S represents the actual available space in the carriage, and C represents the uniformity coefficient of the carriage; The process of obtaining the uniformity of the carriage is as follows: Using image recognition technology, the real-time image of the carriage is divided into several areas of equal area, and the formula is used: ρ i =μ i / σ, calculate the passenger density of each area in the real-time image, where ρ i represents the passenger density of the ith area in the real-time image, μ i Represents the number of passengers in the i-th area in the real-time image, σ represents the area of ​​each area in the real-time image; the average passenger density of the carriage is calculated using the formula: ρ1=P / (i×σ), where ρ1 represents the average passenger density of the carriage; using the formula: , calculate the uniformity coefficient of the carriage, where C represents the uniformity coefficient of the carriage; Wp and Wi can be adjusted according to actual conditions to reflect the actual impact of passengers and items on crowding; the value of C ranges from 0 to 1, and the closer the value is to 1, the more uniform the passenger distribution is; S7-2, calculating the average of the occupied spaces of all the carry-on items to be disembarked in the corresponding carriage according to the occupied spaces of the carry-on items to be disembarked in the carriage, as the average occupied space of the carry-on items to be disembarked in the corresponding carriage; The predicted additional space in the carriage is calculated using the formula: M1=(P1×Ap×Wp)+(I1×A1×Wi); where M1 represents the predicted additional space in the carriage, P1 represents the number of passengers waiting to get off in the carriage, I1 represents the number of carry-on items in the carriage, and A1 represents the average space occupied by carry-on items in the carriage. According to the occupied space of each carry-on item to be boarded in the waiting area, the average occupied space of all carry-on items to be boarded in the corresponding waiting area is calculated as the average occupied space of the carry-on items to be boarded in the corresponding waiting area, and the corresponding carriage number is matched according to the waiting area number; The predicted reduced space of the carriage is calculated using the formula; M2=(P2×Ap×Wp)+(I2×A2×Wi); where M2 represents the predicted reduced space of the carriage, P2 represents the number of passengers waiting to board in the corresponding waiting area of ​​the carriage, I2 represents the number of carry-on items waiting to board in the corresponding waiting area of ​​the carriage, and A2 represents the average space occupied by carry-on items waiting to board in the corresponding waiting area of ​​the carriage; The predicted congestion of the carriage is calculated using the formula: N1=(M +M1-M2) / (S×C); where N1 represents the predicted congestion of the carriage; S7-3, setting a first threshold and a second threshold, if the predicted congestion of the carriage is less than the first threshold, the carriage state is comfortable; if the predicted congestion of the carriage is greater than or equal to the first threshold and the predicted congestion of the carriage is less than the second threshold, the carriage state is moderate; if the predicted congestion of the carriage is greater than or equal to the second threshold, the carriage state is crowded; For example: The first threshold is 0.5, the second threshold is 0.8; there are 3 carriages, numbered 1, 2 and 3, the predicted crowdedness of carriage 1 is 0.3, so the carriage state of carriage 1 is comfortable, the predicted crowdedness of carriage 2 is 0.6, so the carriage state of carriage 2 is moderate, the predicted crowdedness of carriage 3 is 0.9, so the carriage state of carriage 3 is crowded; Generate boarding guidance information, wherein the boarding guidance information is used to guide passengers at the target station to board a carriage with a comfortable carriage state; The boarding guidance information usually includes the specific location of the carriage, the carriage number and the direction guidance relative to the passenger's current location. The boarding guidance information is designed to be clear, concise and easy for passengers to understand; For example, the boarding guidance information states: "Go to carriage No. 3, to the right of your current position, and enjoy a more comfortable riding environment." In addition, the system can dynamically adjust the guidance information based on changes in passenger flow and carriage congestion to ensure the real-time and accuracy of the information.

[0026] See also Figure 2 , a data processing device for rail transit operation, comprising a processor and a memory; The memory is coupled to the processor, the memory is used to store computer program code, the computer program code includes computer instructions; For example: A data processing device for rail transit operation includes a central processing unit 401, a read-only memory 402, a random access memory 403, a bus 404, an I / O interface 405, an input part 406, an output part 407, a storage part 408, a communication part 409, a drive 410 and a removable medium 411.

[0027] A data processing medium for rail transit operation includes a storage medium for storing computer instructions.

[0028] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants 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.

[0029] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, a technical visitor in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A data processing method for rail transit operation, characterized in that: The method comprises the following steps: S1. Setting a camera in the carriage to collect images and videos in the carriage, pre-processing the collected images and videos, and storing them in a database; setting a camera in the waiting area to collect images of the waiting area, pre-processing the collected images, and storing them in a database; S2. Obtain historical images in the carriage from the database, annotate the historical images, and obtain annotated images; construct a passenger item recognition model, and train the passenger item recognition model based on the historical images and the annotated images; S3, obtaining historical videos in the carriage from the database, annotating the historical videos, and obtaining annotated videos; constructing a getting-off behavior recognition model, and training the getting-off behavior recognition model based on the historical videos and the annotated videos; S4, acquiring real-time images of the carriage from the database to obtain a first image set; extracting passengers and their belongings in the first image set through a passenger and item recognition model, and calculating the number of passengers in the carriage and the space occupied by each item; S5, obtaining the real-time video in the carriage from the database to obtain a first video set; extracting the passengers waiting to get off and their belongings in the first video set through the getting off behavior recognition model, and calculating the number of passengers waiting to get off in the carriage and the space occupied by each belonging to the passengers waiting to get off; S6. Acquire a real-time image of the waiting area in the target station from the database to obtain a second image set; extract the passengers waiting to board and their belongings in the second image set through the passenger and item recognition model, and calculate the number of passengers waiting to board in the waiting area and the space occupied by each of the belongings waiting to board; S7. Calculate the real-time congestion and predicted congestion of the carriage based on the number of passengers in the carriage and the space occupied by their belongings, the number of passengers waiting to get off the carriage and the space occupied by their belongings, and the number of passengers waiting to get on the carriage in the waiting area and the space occupied by their belongings; generate boarding guidance information based on the predicted congestion, and send the boarding guidance information to the target station.

2. A data processing method for rail transit operation according to claim 1, characterized in that: The step S1 comprises the following: S1-1. Cameras are installed in all carriages, and the collection range of the cameras covers the carriages; the cameras corresponding to the carriages are numbered according to the carriage numbers; the cameras collect images and videos in the corresponding carriages, and preprocess the collected images and videos, and the preprocessing includes denoising, normalization and contrast enhancement; the preprocessed images and videos are stored in a database; S1-2. Cameras are installed in all waiting areas, and the acquisition range of the cameras covers the waiting areas. The cameras corresponding to the waiting areas are numbered according to the waiting area numbers, which are set corresponding to the carriage numbers. The cameras acquire images in the corresponding waiting areas, and pre-process the acquired images, and store the pre-processed images and videos in a database.

3. A data processing method for rail transit operation according to claim 2, characterized in that: The step S2 comprises the following: S2-1, obtaining historical images of several different carriages from a database, marking passengers and objects in the obtained historical images of several different carriages, and marking the positions of the passengers and objects in the historical images to obtain marked images; S2-2. Construct a passenger item recognition model, take the unlabeled historical image as input, take the labeled image as output, train the passenger item recognition model, and obtain a trained passenger item recognition model.

4. A data processing method for rail transit operation according to claim 3, characterized in that: The step S3 includes the following: S3-1. Obtaining several historical videos in different carriages from a database, extracting behavioral features of passengers in the obtained historical videos in the different carriages, wherein the behavioral features include standing behavior, sitting behavior, walking behavior, and interactive behavior; marking the behavioral features of passengers related to getting off the train, as well as the corresponding personal belongings of the passengers in the historical videos, and obtaining a marked video; S3-2. Construct a getting-off behavior recognition model, take the unlabeled historical video as input, take the labeled video as output, train the getting-off behavior recognition model, and obtain a trained getting-off behavior recognition model.

5. A data processing method for rail transit operation according to claim 4, characterized in that: The step S4 comprises the following: S4-1. Matching real-time images in different carriages from a database according to the current time; numbering the real-time images in each carriage according to the carriage number and the current time to obtain a carriage image number; and counting the real-time images in all carriages at the current time to obtain a first image set; S4-2, according to the carriage image numbers of the images in the first image set, they are arranged from large to small, one image is extracted from the first image set each time, and the trained passenger item recognition model is input, and the labeled image of the passenger and his / her belongings in the corresponding image is output, and the number of passengers in the corresponding labeled image and the item data of each item are counted, and the item data includes size and shape, and the area ratio of the corresponding item in the corresponding labeled image in the corresponding carriage is calculated, and the actual space size of the corresponding carriage is extracted from the database, and the product of the actual size of the corresponding carriage and the area ratio of the corresponding item in the corresponding labeled image in the corresponding carriage is calculated as the occupied space of the corresponding item; The carriage image numbers of the images in the first image set are matched with the carriages to obtain the number of passengers in the corresponding carriage and the space occupied by each carry-on item.

6. A data processing method for rail transit operation according to claim 5, characterized in that: The step S5 comprises the following: S5-1. According to the current time, real-time videos in different carriages are matched from the database; according to the carriage number and the current time, the real-time videos in each carriage are numbered to obtain the carriage video number; the real-time videos in all carriages at the current time are counted to obtain a first video set; S5-2, according to the carriage video numbers of the images in the first video set, arrange them from large to small, extract one video from the first video set each time, input the trained getting-off behavior recognition model, and output the labeled video of the passenger and his / her belongings in the corresponding video; The passengers marked in the marked video are recorded as passengers waiting to get off; The personal belongings of the passengers marked in the marked video are recorded as the personal belongings to be taken off the bus; Count the number of passengers waiting to get off the bus in the annotated video, intercept the image of the corresponding personal belongings of the passengers waiting to get off the bus in the corresponding annotated video, extract the item data of the corresponding personal belongings of the passengers waiting to get off the bus in the intercepted image, and calculate the space occupied by the corresponding personal belongings of the passengers waiting to get off the bus in the intercepted image; According to the carriage video number of the video in the first video set and the carriage, the number of passengers waiting to get off in the corresponding carriage and the space occupied by the personal belongings of each passenger waiting to get off are obtained.

7. A data processing method for rail transit operation according to claim 6, characterized in that: The step S6 comprises the following: S6-1, taking the next station as the target station, matching the real-time images in different waiting areas of the target station from the database according to the current time; numbering the real-time images in the waiting area according to the waiting area number and the current time to obtain the waiting area image number; counting the real-time images in all waiting areas of the target station at the current time to obtain a second image set; S6-2, according to the waiting area image numbers of the images in the second image set, the images are arranged from large to small, one image is extracted from the second image set each time, the trained passenger and item recognition model is input, and the labeled image of the passenger and his / her belongings in the corresponding image is output; The passengers marked in the marked image are recorded as passengers waiting to board the bus; the belongings of the passengers marked in the marked image are recorded as belongings waiting to board the bus; Count the number of passengers waiting to board the bus in the annotated image, extract the item data of each item to be boarded in the corresponding annotated image, and calculate the space occupied by each item to be boarded in the corresponding annotated image; According to the waiting area picture number of the picture in the second image set and the waiting area, the number of passengers waiting to get on the bus in the corresponding waiting area and the space occupied by each personal belongings waiting to get on the bus are obtained.

8. A data processing method for rail transit operation according to claim 7, characterized in that: The step S7 includes the following: S7-1. Calculate the average of the occupied spaces of all the carry-on items in the corresponding compartment according to the occupied spaces of the carry-on items in the compartment, as the average occupied space of the carry-on items in the corresponding compartment; The total occupied space of the carriage is calculated using the formula: M=(P×Ap×Wp)+(I×Ai×Wi); where M represents the total occupied space of the carriage, P represents the number of passengers in the carriage, Ap represents the average occupied space of passengers, Wp represents the passenger space weight, I represents the number of carry-on items in the carriage, Ai represents the average occupied space of carry-on items in the carriage, and Wi represents the item space weight; The real-time crowdedness of the carriage is calculated using the formula: N=M / (S×C), where N represents the real-time crowdedness of the carriage, S represents the actual available space in the carriage, and C represents the uniformity coefficient. S7-2, calculating the average of the occupied spaces of all the carry-on items to be disembarked in the corresponding carriage according to the occupied spaces of the carry-on items to be disembarked in the carriage, as the average occupied space of the carry-on items to be disembarked in the corresponding carriage; The predicted additional space in the carriage is calculated using the formula: M1=(P1×Ap×Wp)+(I1×A1×Wi); where M1 represents the predicted additional space in the carriage, P1 represents the number of passengers waiting to get off in the carriage, I1 represents the number of carry-on items in the carriage, and A1 represents the average space occupied by carry-on items in the carriage. According to the occupied space of each carry-on item to be boarded in the waiting area, the average occupied space of all carry-on items to be boarded in the corresponding waiting area is calculated as the average occupied space of the carry-on items to be boarded in the corresponding waiting area, and the corresponding carriage number is matched according to the waiting area number; The predicted reduced space of the carriage is calculated using the formula; M2=(P2×Ap×Wp)+(I2×A2×Wi); where M2 represents the predicted reduced space of the carriage, P2 represents the number of passengers waiting to board in the corresponding waiting area of ​​the carriage, I2 represents the number of carry-on items waiting to board in the corresponding waiting area of ​​the carriage, and A2 represents the average space occupied by carry-on items waiting to board in the corresponding waiting area of ​​the carriage; The predicted congestion of the carriage is calculated using the formula: N1=(M +M1-M2) / (S×C); where N1 represents the predicted congestion of the carriage; S7-3, setting a first threshold and a second threshold, if the predicted congestion of the carriage is less than the first threshold, the carriage state is comfortable; if the predicted congestion of the carriage is greater than or equal to the first threshold and the predicted congestion of the carriage is less than the second threshold, the carriage state is moderate; if the predicted congestion of the carriage is greater than or equal to the second threshold, the carriage state is crowded; Generate boarding guidance information, which is used to guide passengers at the target station to board a car with a comfortable car state.

9. A data processing device for rail transit operation, the device being used to implement a data processing method for rail transit operation according to any one of claims 1 to 8, characterized in that: including a processor and a memory; The memory is coupled to the processor, and is used to store computer program codes, wherein the computer program codes include computer instructions.

10. A data processing medium for rail transit operation, the medium being used to implement a data processing method for rail transit operation according to any one of claims 1 to 8, characterized in that: Includes a storage medium for storing computer instructions.

Images