Intelligent control method and system of traffic lights based on deep learning
Through deep learning technology, traffic lights are monitored and adjusted in real time, and priority and adjustment order are determined based on the number and speed of vehicles passing, the problem of low efficiency of traffic light regulation in the existing technology is solved, and real-time intelligent regulation and road traffic efficiency are improved.
Patent Information
- Application Number
- CN202411256348.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-09
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-09-09
AI Technical Summary
The existing technology cannot intelligently regulate traffic lights in real time, and its operational efficiency is low, resulting in inconvenient vehicle traffic and road congestion.
Adopt a deep learning-based intelligent traffic light control system to capture vehicle images through cameras, calculate modules preprocess data, learn modules to generate vehicle condition diagrams, adjust modules to determine the priority and adjustment order of traffic lights based on the number of passes and speeds, and adjust the green light time in real time , use vehicle detection model and vehicle condition learning model for visual monitoring and adjustment.
Real-time intelligent regulation of traffic lights has been realized, operating efficiency has been improved, road congestion caused by vehicle backlogs, and driving efficiency has been improved.
Smart Images

Figure CN118762535B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent control, and in particular to a method and system for intelligent control of traffic lights based on deep learning. Background Art
[0002] In modern cities, there are many traffic lights installed at various intersections. The traffic volume at each intersection varies and changes with the time period. When using existing traffic lights, a state often occurs: during periods of low traffic volume, the green light duration of the traffic lights remains the same as during peak traffic volume, and pedestrians need to wait for a long time for the red light to pass, which is very inconvenient for pedestrians to pass. At the same time, there is another problem. During peak traffic volume, the green light duration is short when cars pass, and the number of cars passing within a single green light time is small, which easily causes traffic jams. Therefore, it is necessary to adjust the duration of the red and green lights of traffic lights in different time periods to make it more convenient for pedestrians and cars to pass.
[0003] China Patent Authorization Announcement No.: CN116363876B discloses a method and system for controlling intelligent traffic lights based on the spin glass model, which belongs to the field of computer intelligent traffic control technology. Through the road flow data provided by map surveying and mapping companies such as AutoNavi and Baidu, combined with the monitoring image data of actual road intersections, the real-time traffic flow information of the road is obtained; according to the topological connection structure of the actual road, the probability of vehicle flow turning at the current moment and other information are inferred, thereby constructing a real-time urban traffic flow model, and through mathematical equivalent operations, it can be made to correspond to the classical spin glass model. Based on the above model, combined with global optimization algorithms such as simulated annealing algorithm and simulated branching algorithm, the intelligent dynamic control of global traffic lights is realized, and road congestion is reduced as much as possible to improve the commuting efficiency of existing roads in the city. The deployment of this algorithm can effectively improve the dynamic adjustment ability of urban traffic lights to different traffic modes, thereby improving the utilization rate of urban road networks and reducing the occurrence of congestion.
[0004] China Patent Authorization Announcement No.: CN110164152B discloses a traffic light control system for a single intersection, including an image acquisition module, a vehicle flow detection module, a vehicle speed detection module, an intelligent control module and a signal light control module. The invention uses machine vision technology to obtain traffic flow information at the intersection in real time, and combines the fuzzy neural network algorithm to intelligently control the traffic lights in all directions of the intersection. According to the traffic flow conditions at the intersection, the travel time is reasonably allocated. When the traffic flow is heavy, a longer travel time is allocated, and when the traffic flow is light, a shorter travel time is allocated. The phase is switched as soon as possible, so that other phases obtain the right of way and improve the traffic efficiency.
[0005] However, the above method has the following problems: it is impossible to intelligently control the traffic lights in real time and the operation efficiency is low. Summary of the invention
[0006] To this end, the present invention provides a traffic light intelligent control method and system based on deep learning, so as to overcome the problems in the prior art of being unable to intelligently control traffic lights in real time and having low operating efficiency.
[0007] To achieve the above objectives, the present invention provides a traffic light intelligent control system based on deep learning, comprising:
[0008] A plurality of acquisition modules, which are used to capture images of vehicles entering the entrances of a plurality of traffic lights in various directions;
[0009] A computing module, connected to the plurality of acquisition modules, for preprocessing the vehicle image to generate corresponding preprocessing data;
[0010] A learning module, connected to the calculation module, for selecting several index features of the preprocessed data for learning and generating a corresponding vehicle condition map;
[0011] an adjustment module connected to the learning module, for determining the priorities of the traffic lights according to the learning results of the learning module, determining the adjustment order of the traffic lights according to the priorities, and adjusting a plurality of the traffic lights according to the adjustment order;
[0012] Wherein, the pre-processed data includes the number of vehicles in the vehicle image and the passing time;
[0013] The learning module is provided with a vehicle condition learning model for fitting according to the preprocessed data and generating a number of data with a sampling rate of a standard sampling rate, and generating a corresponding vehicle condition map according to corresponding indicator features.
[0014] Furthermore, the acquisition module includes:
[0015] A camera, which is used to capture the vehicle image.
[0016] The camera is provided with an interval time for controlling the camera to take pictures of the vehicle image at intervals of the interval time;
[0017] A memory, connected to the camera, for storing a plurality of vehicle images;
[0018] A transmitter is connected to the memory and is used to transmit a plurality of vehicle images to the computing module.
[0019] Furthermore, the calculation module includes:
[0020] A quantity monitor, which is used to monitor the waiting quantity and the stranded quantity of vehicles according to the vehicle image;
[0021] A time monitor, which is used to monitor the passing time between adjacent green lights and red lights according to the vehicle image;
[0022] A preprocessor, connected to the quantity monitor and the time monitor, for filtering the waiting quantity, the retention quantity and the passing time to obtain the preprocessing data;
[0023] Wherein, the pre-processed data includes the number of traffic and the traffic speed;
[0024] The passing number is the difference between the waiting number and the detained number, and the passing rate is the quotient of the passing number and the passing time;
[0025] The waiting number is the length of the queue of waiting vehicles accumulated during the red light period of the traffic light when the green light starts;
[0026] The number of stranded vehicles is the length of the queue of waiting vehicles that is stranded after the traffic light turns green when the red light starts.
[0027] Furthermore, the learning module includes:
[0028] A quantity processor, which is used to learn the passing quantity and generate corresponding quantity data;
[0029] A rate processor, which is used to learn the passing rate and generate corresponding rate data;
[0030] The quantity processor is provided with a quantity threshold, the rate processor is provided with a rate threshold, the quantity threshold is the maximum standard of the difference between the waiting quantity and the stranded quantity, and the rate threshold is the minimum standard of the quotient of the passing quantity and the passing time.
[0031] Furthermore, a plurality of the cameras take pictures of the vehicle images, the memory stores the plurality of the vehicle images, and transmits the images to the calculation module through the transmitter, the calculation module monitors and records the waiting number, the detention number and the passing time, and generates corresponding pre-processing data.
[0032] Wherein, the quantity monitor and the time monitor are provided with a vehicle monitoring model for monitoring and calculating the waiting quantity, the detained quantity and the passing time in the vehicle image.
[0033] Furthermore, the vehicle condition learning model in the quantity processor learns the passing quantity, and if the quantity data is higher than the quantity threshold, the adjustment module does not adjust the traffic light.
[0034] Furthermore, if the quantity data is lower than the quantity threshold, the adjustment module adjusts the time of the traffic lights, determines the priorities of several of the traffic lights according to the passing quantity, determines the adjustment order of the traffic lights according to the priorities, and adjusts the green light time of several of the traffic lights according to the adjustment order.
[0035] Furthermore, after the time adjustment, the acquisition module takes pictures of the vehicle image again and transmits the corresponding pre-processed data to the rate processor, and the vehicle condition learning model learns the pre-processed data.
[0036] Furthermore, when the rate data is higher than the rate threshold, the adjustment module does not adjust the traffic light; when the rate data is lower than the rate threshold, the adjustment module makes a secondary time adjustment to the traffic light, determines the priorities of several traffic lights again according to the traffic rate, determines the adjustment order of the traffic lights according to the new priorities, and adjusts the green light time of several traffic lights again according to the adjustment order.
[0037] On the other hand, the present invention provides a method for intelligently controlling a traffic light based on deep learning, comprising:
[0038] Determining the priorities of a plurality of traffic lights according to the number of traffic lights and the speed of traffic;
[0039] Determining the adjustment order of the traffic lights according to the priorities;
[0040] The green light times of the traffic lights are adjusted according to the learning results of the learning module.
[0041] Compared with the prior art, the present invention utilizes several acquisition modules to capture vehicle images at entrances in various directions of several traffic lights, a calculation module to preprocess vehicle images and generate corresponding preprocessed data, a learning module to select several indicator features of the preprocessed data for learning and generate corresponding vehicle condition maps, an adjustment module to determine the priority of the traffic lights according to the learning results of the learning modules, and determine the adjustment order of the traffic lights according to the priority, adjust several traffic lights according to the adjustment order, and visualize the number and rate of vehicle passage using a vehicle detection model and a vehicle condition learning model, so as to monitor the traffic conditions at road intersections in real time, and adjust the traffic lights in real time according to the current traffic conditions, thereby effectively improving the operational efficiency of the intelligent control of traffic lights, avoiding road congestion caused by vehicle backlogs, and further improving driving efficiency.
[0042] Furthermore, by setting the interval duration in the acquisition module, real-time control of the traffic lights can be ensured, thereby improving the control efficiency of the traffic lights.
[0043] Furthermore, by filtering the waiting number, detention number and passing time in the calculation module, preprocessed data is obtained, which eliminates the errors in the learning results in deep learning caused by errors in the preprocessing data processing, making the learning results more accurate.
[0044] Furthermore, by setting a quantity threshold and a rate threshold in the learning module, the operation status of traffic lights can be screened, making it easier to adjust traffic lights in real time according to current traffic conditions.
[0045] Furthermore, by setting up a vehicle monitoring model, the number of waiting vehicles, the number of stranded vehicles and the length of time required for passing through can be quickly calculated, further improving driving efficiency.
[0046] Furthermore, by setting up a vehicle condition learning model, the number and speed of vehicle traffic were visualized, which effectively improved the operational efficiency of the intelligent control of traffic lights and avoided road congestion caused by vehicle backlogs.
[0047] Furthermore, by performing time adjustment and secondary time adjustment on the traffic light, the signal time of the traffic light is effectively regulated, thereby reducing the complexity of regulating the traffic light. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 This is a schematic diagram of the structure of the intelligent control system of traffic lights based on deep learning of the present invention;
[0049] Figure 2 This is a schematic diagram of the structure of the acquisition module of the present invention;
[0050] Figure 3 This is a schematic diagram of the structure of the computing module of the present invention;
[0051] Figure 4 This is a schematic diagram of the structure of the learning module of the present invention;
[0052] Figure 5 This is a flow chart of the intelligent control method of traffic lights based on deep learning in the present invention. DETAILED DESCRIPTION
[0053] 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 only used to explain the present invention and are not used to limit the present invention.
[0054] 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 protection scope of the present invention.
[0055] 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 drawings. This is merely 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.
[0056] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0057] See also Figure 1 As shown, it is a schematic diagram of the structure of the intelligent control system of traffic lights based on deep learning of the present invention, including:
[0058] A plurality of acquisition modules, which are used to capture images of vehicles entering the entrances of a plurality of traffic lights in various directions;
[0059] A computing module, which is connected to a plurality of acquisition modules and is used to pre-process vehicle images and generate corresponding pre-processed data;
[0060] A learning module, which is connected to the calculation module, is used to select several indicator features of the preprocessed data for learning and generate a corresponding vehicle condition map;
[0061] An adjustment module, which is connected to the learning module, and is used to determine the priority of the traffic lights according to the learning result of the learning module, and determine the adjustment order of the traffic lights according to the priority, and adjust the plurality of traffic lights according to the adjustment order;
[0062] The preprocessed data includes the number of vehicles and the passing time in the vehicle image;
[0063] The learning module is provided with a vehicle condition learning model, which is used to fit the preprocessed data and generate a number of data with a standard sampling rate, and generate a corresponding vehicle condition map according to the corresponding indicator characteristics.
[0064] In a specific implementation, the indicator features of the preprocessed data can be selected as quantity, shooting time, duration, etc. It can be understood that the quantity corresponds to the number of vehicles, and the shooting time and duration correspond to the elapsed time.
[0065] Preferably, setting the sampling rate to 10 per minute has a better learning effect on the preprocessed data, and for the system described in the present application, the corresponding analysis effect is optimal.
[0066] By setting up several acquisition modules to capture vehicle images at entrances in various directions of several traffic lights, a calculation module to preprocess vehicle images and generate corresponding preprocessed data, a learning module to select several indicator features of the preprocessed data for learning and generate a corresponding vehicle condition map, an adjustment module to determine the priority of the traffic lights according to the learning results of the learning module, and determine the adjustment order of the traffic lights according to the priority, and adjust several traffic lights according to the adjustment order, the vehicle detection model and the vehicle condition learning model are used to visualize the number and rate of vehicle passage, which can monitor the traffic conditions at road intersections in real time, and adjust the traffic lights in real time according to the current traffic conditions, effectively improving the operating efficiency of the intelligent control of traffic lights, avoiding road congestion caused by vehicle backlogs, and further improving driving efficiency.
[0067] See also Figure 2 As shown, it is a schematic diagram of the structure of the acquisition module of the present invention, including:
[0068] A camera is used to capture images of the vehicle.
[0069] The camera is provided with an interval time for controlling the camera to take pictures of the vehicle image at intervals;
[0070] A memory, connected to the camera, for storing a plurality of vehicle images;
[0071] The transmitter is connected to the memory and is used for transmitting a plurality of vehicle images to the computing module.
[0072] By setting the interval time in the acquisition module, real-time control of traffic lights can be guaranteed, thereby improving the control efficiency of traffic lights.
[0073] See also Figure 3 As shown, it is a schematic diagram of the structure of the computing module of the present invention, including:
[0074] A number monitor, which is used to monitor the number of waiting vehicles and the number of stranded vehicles based on vehicle images;
[0075] A time monitor, which is used to monitor the passing time between adjacent green lights and red lights based on the vehicle image;
[0076] A preprocessor, which is connected to the quantity monitor and the time monitor, is used to filter the waiting quantity, the detention quantity and the passing time to obtain preprocessing data;
[0077] Among them, the preprocessed data includes the number of passes and the passing speed;
[0078] The number of passes is the difference between the waiting number and the detained number, and the passing rate is the quotient of the number of passes and the passing time;
[0079] The waiting number is the length of the queue of waiting vehicles accumulated during the red light period of the traffic light when the green light starts;
[0080] The number of stranded vehicles is the length of the queue of waiting vehicles after the traffic light turns green when the red light starts.
[0081] By filtering the waiting number, detention number and passing time in the calculation module, preprocessed data is obtained, which eliminates the errors in the learning results in deep learning caused by errors in the preprocessing data processing, making the learning results more accurate.
[0082] See also Figure 4 As shown, it is a schematic diagram of the structure of the learning module of the present invention, including:
[0083] A quantity processor, which is used to learn the number of passes and generate corresponding quantity data;
[0084] A rate processor, which is used to learn the passing rate and generate corresponding rate data;
[0085] Among them, the quantity processor is provided with a quantity threshold, and the rate processor is provided with a rate threshold. The quantity threshold is the maximum standard of the difference between the waiting quantity and the stranded quantity, and the rate threshold is the minimum standard of the quotient of the quantity data and the passing time.
[0086] By setting the quantity threshold and rate threshold in the learning module, the operation status of traffic lights can be screened, making it easier to adjust the traffic lights in real time according to the current traffic conditions.
[0087] Specifically, several cameras take vehicle images, the memory stores several vehicle images, and transmits them to the calculation module through the transmitter. The calculation module monitors and records the number of waiting vehicles, the number of stranded vehicles, and the length of time for passing through, and generates corresponding pre-processed data.
[0088] Among them, the number monitor and time monitor are equipped with a vehicle monitoring model to monitor and calculate the waiting number, detention number and passing time in the vehicle image.
[0089] By setting up a vehicle monitoring model, we can quickly calculate the number of waiting vehicles, the number of stranded vehicles and the passing time, further improving driving efficiency.
[0090] Specifically, the vehicle condition learning model in the quantity processor learns the number of traffic, and if the quantity data is higher than the quantity threshold, the adjustment module does not adjust the traffic light.
[0091] Embodiment 1:
[0092] Set the quantity threshold to 10 / minute and the quantity data to 12 / minute. If the quantity data is higher than the quantity threshold, the adjustment module will not adjust the traffic light.
[0093] By setting up a vehicle condition learning model, the number and speed of vehicle traffic are visualized, which effectively improves the operational efficiency of intelligent control of traffic lights and avoids road congestion caused by vehicle backlogs.
[0094] Specifically, if the quantity data is lower than the quantity threshold, the adjustment module will adjust the time of the traffic lights, determine the priority of several traffic lights according to the number of passes, determine the adjustment order of the traffic lights according to the priority, and adjust the green light time of several traffic lights according to the adjustment order.
[0095] Embodiment 2:
[0096] If the number of vehicles passing through Road A is 5 / minute and the green light time is 20s, the number of vehicles passing through Road B is 3 / minute and the green light time is 15s, and Roads A and B intersect at intersection C, the number threshold is set to 10 / minute, the number data at intersection C is 8 / minute, and the number data is lower than the number threshold. According to the number of vehicles passing through, Road A has a higher priority than Road B. The adjustment module first adjusts the green light time of Road A to 22s, and then adjusts the green light time of Road B to 18s.
[0097] Specifically, after the time adjustment, the acquisition module takes pictures of the vehicle image again and transmits the corresponding pre-processed data to the rate processor, and the vehicle condition learning model learns the pre-processed data.
[0098] Specifically, when the rate data is higher than the rate threshold, the adjustment module does not adjust the traffic light. When the rate data is lower than the rate threshold, the adjustment module makes a secondary time adjustment to the traffic light, determines the priority of several traffic lights again according to the traffic rate, and determines the adjustment order of the traffic lights according to the new priority, and adjusts the green light time of several traffic lights again according to the adjustment order.
[0099] Embodiment 3:
[0100] Set the rate threshold to 2m / s and the rate data to 3m / s. If the rate data is higher than the rate threshold, the adjustment module will not adjust the traffic light.
[0101] Embodiment 4:
[0102] If the traffic speed of Road D is 1m / s and the green light time is 20s, the traffic speed of Road E is 1.5m / s and the green light time is 15s, and Roads D and E intersect at intersection F, the speed threshold is set to 2m / s, the speed data at intersection F is 1.25m / s, and the speed data is lower than the speed threshold. According to the traffic speed, it is determined that Road D has a higher priority than Road E. The adjustment module first adjusts the green light time of Road D to 25s, and then adjusts the green light time of Road E to 20s.
[0103] By performing time adjustment and secondary time adjustment on the traffic lights, the signal time of the traffic lights is effectively regulated, thereby reducing the complexity of regulating the traffic lights.
[0104] See also Figure 5 As shown, it is a flow chart of the intelligent control method of traffic lights based on deep learning of the present invention, including:
[0105] Step S1, determining the priorities of a plurality of traffic lights according to the number of traffic lights and the traffic speed;
[0106] Step S2, determining the adjustment order of the traffic lights according to the priority;
[0107] Step S3, adjusting the green light time of several traffic lights according to the learning result of the learning module.
[0108] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. 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. The intelligent traffic light control system based on deep learning is characterized by: include: A plurality of acquisition modules, which are used to capture images of vehicles entering the entrances of a plurality of traffic lights in various directions; A computing module, connected to the plurality of acquisition modules, for preprocessing the vehicle image to generate corresponding preprocessing data; A learning module, connected to the calculation module, for selecting several index features of the preprocessed data for learning and generating a corresponding vehicle condition map; an adjustment module connected to the learning module, for determining the priorities of the traffic lights according to the learning results of the learning module, determining the adjustment order of the traffic lights according to the priorities, and adjusting a plurality of the traffic lights according to the adjustment order; The vehicle condition learning model in the quantity processor learns the number of passing vehicles, and if the number data is higher than the number threshold, the adjustment module does not adjust the traffic light; After the time adjustment, the acquisition module takes the vehicle image again and transmits the corresponding pre-processed data to the rate processor, and the vehicle condition learning model learns the pre-processed data; When the speed data is higher than the speed threshold, the adjustment module does not adjust the traffic light; when the speed data is lower than the speed threshold, the adjustment module makes a secondary time adjustment to the traffic light, determines the priorities of the traffic lights again according to the traffic speed, determines the adjustment order of the traffic lights according to the new priorities, and adjusts the green light time of the traffic lights again according to the adjustment order; Wherein, the pre-processed data includes the number of vehicles in the vehicle image and the passing time; The learning module is provided with the vehicle condition learning model, which is used to perform fitting according to the preprocessed data and generate a number of data with a sampling rate of a standard sampling rate, and generate a corresponding vehicle condition map according to corresponding indicator features.
2. The traffic light intelligent control system based on deep learning according to claim 1 is characterized in that: The acquisition module comprises: A camera, which is used to capture the vehicle image. The camera is provided with an interval time for controlling the camera to take pictures of the vehicle image at intervals of the interval time; A memory, connected to the camera, for storing a plurality of vehicle images; A transmitter is connected to the memory and is used to transmit a plurality of vehicle images to the computing module.
3. The traffic light intelligent control system based on deep learning according to claim 2 is characterized in that: The computing module comprises: A quantity monitor, which is used to monitor the waiting quantity and the stranded quantity of vehicles according to the vehicle image; A time monitor, which is used to monitor the passing time between adjacent green lights and red lights according to the vehicle image; A preprocessor, connected to the quantity monitor and the time monitor, for filtering the waiting quantity, the retention quantity and the passing time to obtain the preprocessing data; Wherein, the pre-processed data includes the number of traffic and the traffic speed; The passing number is the difference between the waiting number and the detained number, and the passing rate is the quotient of the passing number and the passing time; The waiting number is the length of the queue of waiting vehicles accumulated during the red light period of the traffic light when the green light starts; The number of stranded vehicles is the length of the queue of waiting vehicles that is stranded after the traffic light turns green when the red light starts.
4. The traffic light intelligent control system based on deep learning according to claim 3 is characterized in that: The learning module includes: The quantity processor is used to learn the passing quantity and generate corresponding quantity data; A rate processor, which is used to learn the passing rate and generate corresponding rate data; The quantity processor is provided with the quantity threshold, the rate processor is provided with the rate threshold, the quantity threshold is the maximum standard of the difference between the waiting quantity and the stranded quantity, and the rate threshold is the minimum standard of the quotient of the passing quantity and the passing time.
5. The traffic light intelligent control system based on deep learning according to claim 4 is characterized in that: The plurality of cameras take pictures of the vehicle images, the memory stores the plurality of vehicle images, and transmits the images to the calculation module through the transmitter, the calculation module monitors and records the waiting number, the detention number and the passing time, and generates corresponding pre-processing data, Wherein, the quantity monitor and the time monitor are provided with a vehicle monitoring model for monitoring and calculating the waiting quantity, the detained quantity and the passing time in the vehicle image.
6. The traffic light intelligent control system based on deep learning according to claim 5 is characterized in that: If the quantity data is lower than the quantity threshold, the adjustment module makes the time adjustment to the traffic lights, determines the priorities of several traffic lights according to the passing quantity, determines the adjustment order of the traffic lights according to the priorities, and adjusts the green light time of several traffic lights according to the adjustment order.
7. A method for intelligently controlling traffic lights according to the system of any one of claims 1 to 6, characterized in that: include: Determining the priorities of a plurality of traffic lights according to the number of traffic lights and the speed of traffic; Determining the adjustment order of the traffic lights according to the priorities; The green light times of the traffic lights are adjusted according to the learning results of the learning module.
