A public transportation safety management method and system based on deep learning model

Abstract

The present invention relates to the field of intelligent transportation technology, and in particular to a public transportation safety management method and system based on a deep learning model. The method comprises: step S1, monitoring the position of a vehicle and determining the moving speed of the vehicle; step S2, analyzing whether the vehicle meets safety standards, and analyzing whether the vehicle meets safety standards based on the number of people in the vehicle, or monitoring the working status of the driver; step S3, analyzing the working status of the driver according to the current profile characteristics of the driver; step S4, reducing the speed of the vehicle and issuing an alarm command when the driver's working status is analyzed to be unqualified; in the present invention, the safety status of the vehicle is preliminarily analyzed according to the vehicle speed, and further analyzed according to the number of people in the vehicle when the vehicle speed is relatively fast, thereby improving the control precision of the vehicle status and the accuracy of analyzing the vehicle status, detecting the working status of the driver when the vehicle speed is too fast, improving the control precision of the driver's working status and improving the analysis efficiency.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a public transportation safety management method and system based on a deep learning model. Background Art

[0002] With the advancement of science and technology, video analysis, artificial intelligence, image processing, and computer vision technologies have been widely applied in various fields. In the transportation sector, these technologies are used to monitor road conditions, analyze driver behavior, and predict traffic accidents. In public transportation, they are also being used to optimize vehicle scheduling and improve passenger service quality. Driver behavior recognition, passenger density analysis, and the removal of objects left behind in carriages are key research areas in public transportation. Existing technologies primarily rely on installing cameras inside carriages to monitor in-car video in real time. These surveillance videos are then analyzed using manual or intelligent algorithms to achieve driver behavior recognition, passenger density analysis, and the removal of objects left behind in carriages. For example, deep learning algorithms can be used to identify driver behavior and determine whether the driver is fatigued or complying with traffic regulations. However, existing technologies still face some challenges in practical application. First, existing artificial intelligence algorithms may suffer from low recognition accuracy when processing videos in complex scenarios. Second, existing technologies may be affected by factors such as lighting conditions and passenger occlusion when analyzing passenger density, resulting in reduced accuracy of the analysis results.

[0003] Chinese patent application number CN202010112145.6 discloses a driving safety monitoring system for engineering vehicles in subway systems. The system comprises an onboard subsystem and a ground subsystem, which are communicatively connected to each other. The ground subsystem includes multiple ground positioning devices and a ground operation protection module for collecting automatic traffic control (ATC) information from a section of the line and remotely monitoring the engineering vehicles. The onboard subsystem includes an onboard positioning module and an onboard operation protection module for collecting onboard status information of the engineering vehicles, calculating and determining the actual location of the engineering vehicles, and generating movement authorization information and movement pattern curves. The driving safety monitoring system for engineering vehicles in subway systems according to the present invention facilitates remote monitoring and management of subway engineering vehicles and improves the safety and protection capabilities of subway engineering vehicles. It also improves the operation and maintenance management capabilities of engineering vehicles, as well as the safety and reliability of their operation. Therefore, the driving safety monitoring system for engineering vehicles in subway systems suffers from the following problems: the collected information, such as the conditions inside the vehicle and the driver's status, is easily affected by factors such as lighting and occlusion, resulting in insufficient accuracy in the analysis of vehicle safety during driving. Summary of the Invention

[0004] To this end, the present invention provides a public transportation safety management method based on a deep learning model to overcome the problem that the information collected in the prior art, such as the conditions inside the vehicle and the driver's status, is easily affected by factors such as lighting and occlusion, resulting in insufficient accuracy in the analysis of vehicle safety during driving.

[0005] To achieve the above objectives, the present invention provides a public transportation safety management method based on a deep learning model. The method comprises:

[0006] Step S1: periodically monitoring the vehicle position based on external reference objects, and determining the vehicle speed according to the vehicle position;

[0007] Step S2: analyzing whether the vehicle meets safety standards based on the vehicle's speed, and further analyzing whether the vehicle meets safety standards based on the number of people in the vehicle if it is initially determined that the vehicle does not meet safety standards, or monitoring the working status of the driver;

[0008] Step S3: The algorithm module compares the current profile features of the driver with the profile features under normal working conditions, and analyzes the working status of the driver based on the comparison results;

[0009] Step S4: When the algorithm module analyzes that the driver's working status is unqualified, the vehicle speed is reduced and an alarm instruction is issued;

[0010] In step S2, the vehicle is analyzed to determine whether it meets safety standards based on its speed, specifically including:

[0011] The monitoring module periodically monitors the vehicle's position. The algorithm module calculates the vehicle's average speed within the time interval based on the vehicle's position at the current time node and the position at the previous time node to determine whether the vehicle meets safety standards.

[0012] If the average moving speed is less than or equal to a first preset average moving speed in the algorithm module, the algorithm module determines that the vehicle meets the safety standard and continuously monitors the vehicle position to determine the vehicle's moving speed;

[0013] If the average moving speed is greater than the first preset average moving speed and less than or equal to the second preset average moving speed in the algorithm module, the algorithm module determines whether the vehicle meets the safety standards based on the number of people in the vehicle;

[0014] A secondary determination is made as to whether the vehicle meets safety standards based on the number of people in the vehicle, specifically including:

[0015] The algorithm module determines the number of people in the vehicle based on the collected passenger profile features, and determines whether the vehicle meets safety standards based on the number of people in the vehicle, wherein:

[0016] If the number of people in the vehicle is less than or equal to the first preset number of people in the vehicle in the algorithm module, the algorithm module determines that the vehicle meets the safety standards and continuously monitors the vehicle position to determine the vehicle's moving speed;

[0017] If the number of people in the vehicle is greater than the first preset number of people in the vehicle and less than or equal to the second preset number of people in the vehicle in the algorithm module, the algorithm module determines that the vehicle does not meet the safety standards, and determines that the reason for not meeting the safety standards is a lighting problem;

[0018] If the number of people in the vehicle is greater than the second preset number of people in the vehicle, the algorithm module determines that the vehicle does not meet the safety standards and activates the in-vehicle monitoring unit to monitor the working status of the driver;

[0019] If the average moving speed is greater than the second preset average moving speed, the algorithm module determines that the vehicle does not meet the safety standards and activates the in-vehicle monitoring unit to monitor the working status of the driver;

[0020] Start the in-car monitoring unit to monitor the driver's working status, including:

[0021] The driver's status is determined to be qualified based on the similarity between the collected characteristic profile of the driver and the preset profile in the algorithm module, wherein:

[0022] If the similarity is greater than or equal to a first preset similarity preset in the algorithm module, the algorithm module determines that the driver's status is qualified and adjusts the vehicle speed based on the distance between the current vehicle and the previous vehicle;

[0023] If the similarity is less than the first preset similarity and greater than or equal to the second preset similarity in the algorithm module, the algorithm module determines whether the driver's status is qualified based on the distribution of the feature profile;

[0024] If the similarity is less than the second preset similarity, the algorithm module determines that the driver's status is unqualified and corrects the vehicle speed based on the vehicle parameters and the similarity difference;

[0025] The similarity difference is the difference between the second preset similarity and the similarity;

[0026] The algorithm module calculates the vehicle speed evaluation value Z based on the vehicle weight M and the similarity difference Q, and sets the vehicle speed evaluation value Z = m×M / M0+q×Q / Q0, where m is the vehicle weight weight coefficient, q is the similarity weight coefficient, m+q=1, M0 is the preset vehicle weight, and Q0 is the preset similarity difference. The algorithm module is provided with several correction methods for the current vehicle speed according to the vehicle speed evaluation value, and the correction amplitude of each correction method for the current vehicle speed is different.

[0027] Furthermore, when the algorithm module determines that the reason why the vehicle does not meet the safety standards is a lighting problem, it provides several adjustment methods for image brightness based on the number of people in the vehicle, and the adjustment range of the image brightness in each adjustment method is different.

[0028] Furthermore, when determining whether the driver's status is qualified, the algorithm module is provided with several adjustment methods for the current vehicle speed threshold according to the distance between the current vehicle and the previous vehicle, and the adjustment range of the speed threshold for each adjustment method is different.

[0029] Furthermore, the algorithm module determines whether the standard correction standard is met based on the distribution of the contour features, wherein:

[0030] If the distribution of the contour features is dispersed, the algorithm module determines that the standard correction standard is not met;

[0031] If the distribution of the contour features is concentrated, the algorithm module determines that the standard correction standard is met, and selects the standard correction coefficient s to adjust the first preset similarity C1 to a corresponding value, and sets the adjusted first preset similarity C1'=s×C10, where C10 is the first preset similarity before adjustment.

[0032] Compared with the prior art, the beneficial effect of the present invention is that the present invention preliminarily analyzes the safety status of the vehicle based on the vehicle speed, and further analyzes based on the number of people in the vehicle when the speed is faster, thereby improving the control precision of the vehicle condition and the accuracy of analyzing the vehicle condition, detecting the working status of the driver when the speed is too fast, improving the control precision of the driver's working status and improving the analysis efficiency.

[0033] Furthermore, the present invention compares the collected profile features with the standard passenger profile features, and determines whether it is a passenger profile based on the similarity between the two, and further determines the number of people in the car, thereby improving the control accuracy of the conditions in the car and further improving the analysis accuracy.

[0034] Furthermore, the present invention compares the characteristic profile in the current image of the driver collected with the characteristic profile in the normal working state, and determines the working state of the driver based on the similarity between the two, thereby improving the control accuracy of the driver's working state and improving the analysis efficiency. The shorter the distance between the current vehicle and the previous vehicle, the greater the risk of rear-end collision. The vehicle speed is adjusted based on the distance between the two to ensure the safety of the driving process.

[0035] Furthermore, in the present invention, the working state of the driver is analyzed according to the distribution of the driver's profile features in the preset profile features. If the distribution is concentrated, it is determined that there is occlusion, thereby improving the analysis accuracy.

[0036] Furthermore, in the present invention, the heavier the vehicle, the greater the inertia and the lower the safety. The inertia of the vehicle is determined based on the vehicle's own weight and the number of people in the vehicle, and the vehicle speed is reduced to a corresponding value according to the inertia, thereby ensuring the safety of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of the public transportation safety management method based on the deep learning model of the present invention;

[0038] Figure 2 This is a structural diagram of a public transportation safety management system based on a deep learning model;

[0039] Figure 3 A flowchart for determining whether a vehicle complies with safety standards;

[0040] Figure 4 A flowchart for determining whether the driver's status is qualified. DETAILED DESCRIPTION

[0041] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0042] It should be pointed out that the data in this embodiment are obtained by comprehensive analysis and evaluation of the historical data of the six months before this judgment and the corresponding historical judgment results of the system of the present invention. Before this test, the system of the present invention comprehensively determines the values ​​of the various preset parameter standards for this judgment based on the evaluation values ​​of the 37,383 retrieval results detected cumulatively in the first three months. It can be understood by those skilled in the art that the system of the present invention can determine the single parameter mentioned above by selecting the value with the highest proportion as the preset standard parameter based on the data distribution, using weighted summation to use the obtained value as the preset standard parameter, substituting each historical data into a specific formula and using the value obtained by the formula as the preset standard parameter or other selection methods, as long as the system of the present invention can clearly define the different specific situations in the single judgment process through the obtained values.

[0043] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0044] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0045] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0046] See also Figure 1 As shown, it is a flow chart of the public transportation safety management method based on the deep learning model of the present invention.

[0047] The present invention provides a public transportation safety management method based on a deep learning model, comprising:

[0048] Step S1: periodically monitoring the vehicle position based on external reference objects, and determining the vehicle speed according to the vehicle position;

[0049] Step S2: analyzing whether the vehicle meets safety standards based on the vehicle's speed, and further analyzing whether the vehicle meets safety standards based on the number of people in the vehicle if it is initially determined that the vehicle does not meet safety standards, or monitoring the working status of the driver;

[0050] Step S3: Under the condition of determining the working status of the monitored driver, the algorithm module compares the current profile features of the driver with the profile features under the normal working state, and analyzes the working status of the driver based on the comparison results;

[0051] Step S4: When the algorithm module analyzes that the driver's working status is unqualified, the vehicle speed is reduced and an alarm instruction is issued.

[0052] See also Figure 2 As shown in the figure, it is a structural block diagram of a public transportation safety management system based on a deep learning model.

[0053] The present invention provides a public transportation safety management system based on a deep learning model, comprising:

[0054] A monitoring module, which includes an in-vehicle monitoring unit and an out-vehicle monitoring unit;

[0055] an algorithm module, connected to the monitoring module, configured to analyze the safety status of the vehicle and the driver's status based on the monitoring data from the monitoring module, and adjust corresponding parameters to corresponding values ​​when the vehicle's safety status is determined to be unqualified, wherein the parameters include: a vehicle speed threshold, a first preset similarity, and a current vehicle speed;

[0056] An application module, connected to the algorithm module, for executing corresponding instructions according to the analysis results of the algorithm module;

[0057] The host computer is connected to the monitoring module and is used to receive the transmission data of the monitoring module.

[0058] The present invention preliminarily analyzes the safety status of the vehicle based on the vehicle speed, and further analyzes based on the number of people in the vehicle when the vehicle speed is relatively fast, thereby improving the control precision of the vehicle condition and the accuracy of analyzing the vehicle condition. When the vehicle speed is too fast, the working status of the driver is detected, thereby improving the control precision of the driver's working status and improving the analysis efficiency.

[0059] See also Figure 3 As shown, it is a flow chart for determining whether a vehicle meets safety standards.

[0060] Specifically, the monitoring module periodically monitors the vehicle's position, and the algorithm module calculates the vehicle's average speed within the time interval based on the vehicle's position at the current time node and the position at the previous time node to determine whether the vehicle meets the safety standards.

[0061] The first preliminary determination method is that the algorithm module determines that the vehicle meets the safety standards and continuously monitors the vehicle; the first preliminary determination method satisfies that the average moving speed is less than or equal to a first preset average moving speed in the algorithm module;

[0062] The second preliminary determination method is that the algorithm module determines whether the vehicle meets the safety standards based on the number of people in the vehicle; the second preliminary determination method satisfies that the average moving speed is greater than the first preset average moving speed and less than or equal to the second preset average moving speed in the algorithm module;

[0063] The third preliminary determination method is that the algorithm module determines that the vehicle does not meet the safety standards and activates the in-vehicle monitoring unit to monitor the driver's working status; the third preliminary determination method satisfies that the average moving speed is greater than the second preset average moving speed.

[0064] In the embodiment of the present invention, the average speed of the vehicle is calculated based on the distance between reference objects outside the vehicle. The first preset average speed is 40 km / h, and the second preset average speed is 55 km / h.

[0065] Specifically, the algorithm module determines the number of people in the vehicle based on the collected passenger profile features under the second preliminary determination method, and determines whether the vehicle meets the safety standards based on the number of people in the vehicle, wherein:

[0066] The first determination method is that the algorithm module determines that the vehicle meets the safety standards; the first determination method satisfies that the number of people in the vehicle is less than or equal to the first preset number of people in the vehicle in the algorithm module;

[0067] The second determination method is that the algorithm module determines that the vehicle does not meet the safety standards and determines that the reason for not meeting the safety standards is a lighting problem; the second determination method satisfies that the number of people in the vehicle is greater than the first preset number of people in the vehicle and is less than or equal to the second preset number of people in the vehicle in the algorithm module;

[0068] The third determination method is that the algorithm module determines that the vehicle does not meet the safety standards and activates the in-vehicle monitoring unit to monitor the driver's working status; the third determination method satisfies that the number of people in the vehicle is greater than the second preset number of people in the vehicle.

[0069] In the embodiment of the present invention, the first preset number of people in the vehicle is 10, and the second preset number of people in the vehicle is 25.

[0070] The present invention compares the collected profile features with the standard passenger profile features, and determines whether it is a passenger profile based on the similarity between the two, and further determines the number of people in the car, thereby improving the control accuracy of the conditions in the car and further improving the analysis accuracy.

[0071] Specifically, the algorithm module records the difference between the number of people in the vehicle and the first preset number of people in the vehicle as a first-level difference under the second determination method, and determines an adjustment method for the image brightness based on the first-level difference, wherein:

[0072] The first adjustment method is that the algorithm module selects a first adjustment coefficient α1 to adjust the image brightness L to a corresponding value, and sets the adjusted image brightness L'=α1×L0, where L0 is the initial image brightness before adjustment; the first adjustment method satisfies that the first-level difference is less than or equal to a preset first-level difference in the algorithm module;

[0073] The second adjustment method is that the algorithm module selects the second adjustment coefficient α2 to adjust the image brightness L to the corresponding value, and sets the adjusted image brightness L'=α2×L0; the second adjustment method satisfies that the first-level difference is greater than the preset first-level difference.

[0074] In the embodiment of the present invention, the preset first-level difference is 2, the first adjustment coefficient α1 is 0.85, and the second adjustment coefficient α2 is 0.75.

[0075] See also Figure 4 As shown, it is a flowchart for determining whether the driver's status is qualified.

[0076] Specifically, the algorithm module determines whether the driver's status is qualified based on the similarity between the collected characteristic profile of the driver and the preset profile in the algorithm module under the first preset condition, wherein:

[0077] The first determination method is that the algorithm module determines that the driver's status is qualified and adjusts the vehicle speed based on the distance between the current vehicle and the previous vehicle; the first determination method satisfies that the similarity is greater than or equal to a first preset similarity preset in the algorithm module;

[0078] The second determination method is that the algorithm module determines whether the driver's status is qualified based on the distribution of the characteristic profile; the second determination method satisfies that the similarity is less than the first preset similarity and greater than or equal to the second preset similarity in the algorithm module;

[0079] The third determination method is that the algorithm module determines that the driver's state is unqualified and corrects the vehicle speed based on the difference between the vehicle parameters and the similarity; the third determination method satisfies that the similarity is less than the second preset similarity;

[0080] The first preset condition is that the algorithm module determines to start the in-vehicle monitoring unit to detect the working status of the driver.

[0081] In the embodiment of the present invention, the preset profile is a characteristic profile of the driver in a normal working state, the first preset similarity is 0.9, the second preset similarity is 0.75, and the similarity difference is the difference between the second preset similarity and the similarity.

[0082] In the present invention, the characteristic profile in the current image of the driver is compared with the characteristic profile in the normal working state, and the working state of the driver is determined based on the similarity between the two, thereby improving the control accuracy of the driver's working state and improving the analysis efficiency. The shorter the distance between the current vehicle and the previous vehicle, the greater the risk of rear-end collision. The vehicle speed is adjusted based on the distance between the two to ensure the safety of the driving process.

[0083] Specifically, the algorithm module determines the vehicle speed adjustment method for the current vehicle speed threshold according to the distance between the current vehicle and the previous vehicle in the first determination method, wherein:

[0084] The first vehicle speed adjustment method is that the algorithm module selects a first vehicle speed adjustment coefficient β1 to adjust the vehicle speed threshold D to a corresponding value, and sets the adjusted vehicle speed threshold D'=β1×D0, where D0 is the initial vehicle speed threshold before adjustment; the first vehicle speed adjustment method satisfies that the distance is less than or equal to the preset distance in the algorithm module;

[0085] The second speed adjustment method is that the algorithm module selects the second speed adjustment coefficient β2 to adjust the speed threshold D to a corresponding value, and sets the adjusted speed threshold D'=β2×D0; the second speed adjustment method satisfies that the distance is greater than the preset distance.

[0086] In the embodiment of the present invention, the vehicle speed threshold is the maximum vehicle speed. If it is detected that the current vehicle speed exceeds the vehicle speed threshold, an overspeed signal is issued.

[0087] In the embodiment of the present invention, the first vehicle speed adjustment coefficient is 0.9, the second vehicle speed adjustment coefficient is 0.8, and the preset distance is 1000m.

[0088] Specifically, the algorithm module determines whether the analysis method of the standard correction standard is met according to the distribution of the contour features under the second determination method, wherein:

[0089] The first analysis method is that the algorithm module determines that the standard correction standard is not met; the first analysis method satisfies the distribution dispersion of the contour features;

[0090] The second analysis method is that the algorithm module determines that the standard correction standard is met, and selects the standard correction coefficient s to adjust the first preset similarity C1 to the corresponding value, and sets the adjusted first preset similarity C1'=s×C10, where C10 is the first preset similarity before adjustment; the second analysis method satisfies the distribution concentration of the contour features.

[0091] In the embodiment of the present invention, the correction coefficient s is 0.85.

[0092] In the present invention, the working state of the driver is analyzed according to the distribution of the driver's profile features in the preset profile features. If the distribution is concentrated, it is determined that there is occlusion, thereby improving the analysis accuracy.

[0093] Specifically, the algorithm module calculates the vehicle speed evaluation value Z based on the vehicle weight M and the similarity difference Q under the third determination method, and sets the vehicle speed evaluation value Z=m×M / M0+q×Q / Q0, where m is the vehicle weight weight coefficient, q is the similarity weight coefficient, m+q=1, M0 is the preset vehicle weight, and Q0 is the preset similarity difference. The algorithm module determines a correction method for the current vehicle speed V based on the vehicle speed evaluation value, where:

[0094] The first correction method is that the algorithm module selects a first correction coefficient γ1 to correct the current vehicle speed V to a corresponding value, and sets the corrected current vehicle speed V'=γ1×V0, where V0 is the initial current vehicle speed before correction; the first correction method satisfies that the vehicle speed evaluation value is less than or equal to the preset vehicle speed evaluation value in the algorithm module;

[0095] The second correction method is that the algorithm module selects the second correction coefficient γ2 to correct the current vehicle speed V to the corresponding value, and sets the corrected current vehicle speed V'=γ2×V0; the second correction method satisfies that the vehicle speed evaluation value is greater than the preset vehicle speed evaluation value.

[0096] In the embodiment of the present invention, the first correction coefficient γ1 is 0.85, and the second correction coefficient is 0.75.

[0097] In the present invention, the heavier the vehicle, the greater the inertia and the lower the safety. The inertia of the vehicle is determined based on the vehicle's own weight and the number of people in the vehicle. The vehicle speed is reduced to a corresponding value according to the inertia, thereby ensuring the safety of the vehicle.

[0098] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0099] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A public transportation safety management method based on a deep learning model, characterized in that: include: Step S1: periodically monitoring the vehicle position based on external reference objects, and determining the vehicle speed according to the vehicle position; Step S2: analyzing whether the vehicle meets safety standards based on the vehicle's speed, and further analyzing whether the vehicle meets safety standards based on the number of people in the vehicle if it is initially determined that the vehicle does not meet safety standards, or monitoring the working status of the driver; Step S3: The algorithm module compares the current profile features of the driver with the profile features under normal working conditions, and analyzes the working status of the driver based on the comparison results; Step S4: When the algorithm module analyzes that the driver's working status is unqualified, the vehicle speed is reduced and an alarm instruction is issued; In step S2, the vehicle is analyzed to determine whether it meets safety standards based on its speed, specifically including: The monitoring module periodically monitors the vehicle's position. The algorithm module calculates the vehicle's average speed within the time interval based on the vehicle's position at the current time node and the position at the previous time node to determine whether the vehicle meets safety standards. If the average moving speed is less than or equal to a first preset average moving speed in the algorithm module, the algorithm module determines that the vehicle meets the safety standard and continuously monitors the vehicle position to determine the vehicle's moving speed; If the average moving speed is greater than the first preset average moving speed and less than or equal to the second preset average moving speed in the algorithm module, the algorithm module determines whether the vehicle meets the safety standards based on the number of people in the vehicle; A secondary determination is made as to whether the vehicle meets safety standards based on the number of people in the vehicle, specifically including: The algorithm module determines the number of people in the vehicle based on the collected passenger profile features, and determines whether the vehicle meets safety standards based on the number of people in the vehicle, wherein: If the number of people in the vehicle is less than or equal to the first preset number of people in the vehicle in the algorithm module, the algorithm module determines that the vehicle meets the safety standards and continuously monitors the vehicle position to determine the vehicle's moving speed; If the number of people in the vehicle is greater than the first preset number of people in the vehicle and less than or equal to the second preset number of people in the vehicle in the algorithm module, the algorithm module determines that the vehicle does not meet the safety standards, and determines that the reason for not meeting the safety standards is a lighting problem; If the number of people in the vehicle is greater than the second preset number of people in the vehicle, the algorithm module determines that the vehicle does not meet the safety standards and activates the in-vehicle monitoring unit to monitor the working status of the driver; If the average moving speed is greater than the second preset average moving speed, the algorithm module determines that the vehicle does not meet the safety standards and activates the in-vehicle monitoring unit to monitor the working status of the driver; Start the in-car monitoring unit to monitor the driver's working status, including: The driver's status is determined to be qualified based on the similarity between the collected characteristic profile of the driver and the preset profile in the algorithm module, wherein: If the similarity is greater than or equal to a first preset similarity preset in the algorithm module, the algorithm module determines that the driver's status is qualified and adjusts the vehicle speed based on the distance between the current vehicle and the previous vehicle; If the similarity is less than the first preset similarity and greater than or equal to the second preset similarity in the algorithm module, the algorithm module determines whether the driver's status is qualified based on the distribution of the feature profile; If the similarity is less than the second preset similarity, the algorithm module determines that the driver's status is unqualified and corrects the vehicle speed based on the vehicle parameters and the similarity difference; The similarity difference is the difference between the second preset similarity and the similarity; The algorithm module calculates the vehicle speed evaluation value Z based on the vehicle weight M and the similarity difference Q, and sets the vehicle speed evaluation value Z = m×M / M0+q×Q / Q0, where m is the vehicle weight weight coefficient, q is the similarity weight coefficient, m+q=1, M0 is the preset vehicle weight, and Q0 is the preset similarity difference. The algorithm module is provided with several correction methods for the current vehicle speed according to the vehicle speed evaluation value, and the correction amplitude of each correction method for the current vehicle speed is different.

2. The public transportation safety management method based on deep learning model according to claim 1 is characterized in that: When the algorithm module determines that the reason why the vehicle does not meet the safety standards is a lighting problem, it sets several adjustment methods for the image brightness based on the number of people in the vehicle, and each adjustment method has a different adjustment range for the image brightness. The algorithm module records the difference between the number of people in the car and the first preset number of people in the car as a first-level difference, and adjusts the image brightness according to the first-level difference. If the first-level difference is less than or equal to the preset first-level difference in the algorithm module, the algorithm module selects the first adjustment coefficient α1 to adjust the image brightness L to the corresponding value, and sets the adjusted image brightness L'=α1×L0, where L0 is the initial image brightness before adjustment; If the first-level difference is greater than the preset first-level difference, the algorithm module selects a second adjustment coefficient α2 to adjust the image brightness L to a corresponding value, and sets the adjusted image brightness L'=α2×L0.

3. The public transportation safety management method based on deep learning model according to claim 1 is characterized in that: When the algorithm module determines that the driver's status is qualified, it has several adjustment methods for the current vehicle speed threshold according to the distance between the current vehicle and the previous vehicle, and the adjustment range of the speed threshold in each adjustment method is different. The algorithm module adjusts the current vehicle speed threshold according to the distance between the current vehicle and the previous vehicle. If the distance is less than or equal to the preset distance in the algorithm module, the algorithm module selects the first vehicle speed adjustment coefficient β1 to adjust the vehicle speed threshold D to the corresponding value, and sets the adjusted vehicle speed threshold D'=β1×D0, where D0 is the initial vehicle speed threshold before adjustment; If the distance is greater than the preset distance, the algorithm module selects the second vehicle speed adjustment coefficient β2 to adjust the vehicle speed threshold D to a corresponding value, and sets the adjusted vehicle speed threshold D'=β2×D0.

4. The public transportation safety management method based on deep learning model according to claim 1 is characterized in that: The algorithm module determines whether the distribution of the contour features meets the standard correction standard, wherein: If the distribution of the contour features is dispersed, the algorithm module determines that the standard correction standard is not met; If the distribution of the contour features is concentrated, the algorithm module determines that the standard correction standard is met, and selects the standard correction coefficient s to adjust the first preset similarity C1 to a corresponding value, and sets the adjusted first preset similarity C1'=s×C10, where C10 is the first preset similarity before adjustment.

Images