Traffic jam processing method, electronic equipment, storage medium and program product

Abstract

The invention provides a traffic jam processing method, electronic equipment, a storage medium and a program product, and relates to the technical field of computers. The traffic jam processing method disclosed by the invention comprises the following steps: acquiring a video image obtained by shooting a target road; according to the video image, identification is carried out, an identification result of congestion factors related to road congestion is determined, and the congestion factors comprise road factors or personnel behavior factors; a target coping strategy is determined according to the congestion factors existing in the recognition result, and the target coping strategy comprises solutions used for eliminating the congestion factors existing on the target road; wherein a default coping strategy is determined according to congestion factors existing in the identification result; according to traffic scene information of the target road, the default coping strategy is optimized, a target coping strategy is determined, and the traffic scene information comprises time period information, weather information, holiday and festival information or traffic flow information; and executing the target coping strategy.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of computer, and particularly relates to a traffic congestion processing method, an electronic device, a storage medium and a program product. BACKGROUND

[0002] In the field of urban traffic management, with the increase of private car ownership, the contradiction between supply and demand of urban roads is increasingly intensified, and the driving environment on the road is more complex. The current traffic congestion prediction technology mainly relies on a single algorithm model, such as statistical analysis based on historical data or simple calculation of real-time traffic flow. However, such methods are often difficult to cope with complex and variable traffic scenes, resulting in low accuracy and poor timeliness of the prediction results, and cannot provide timely and effective decision support for traffic management. SUMMARY

[0003] The present disclosure provides a traffic congestion processing method, an electronic device, a storage medium and a program product.

[0004] According to one aspect of the present disclosure, a traffic congestion processing method is provided, comprising: acquiring a video image obtained by photographing a target road; identifying a congestion factor related to road congestion according to the video image to determine an identification result, the congestion factor including a road factor or a personnel behavior factor; determining a target response strategy according to the congestion factor existing in the identification result, the target response strategy including a solution measure for eliminating the existing congestion factor on the target road; wherein a default response strategy is determined according to the congestion factor existing in the identification result; the default response strategy is optimized according to traffic scene information of the target road to determine the target response strategy, the traffic scene information including time period information, weather information, holiday information or traffic flow information; and the target response strategy is executed.

[0005] According to the technical solution of one aspect, by acquiring a video image obtained by photographing a target road, the congestion factor is identified according to the video image, including a road factor or a personnel behavior factor. Then, based on the identified congestion factor, a targeted response strategy is determined and executed. In this way, without setting other sensors, traffic congestion prediction can be realized, which can effectively improve the accuracy and timeliness of traffic congestion prediction and ensure the efficiency of traffic management.

[0006] Moreover, the previously determined default response strategy is adjusted, so that the determined target response strategy can be more in line with the actual situation on site, improving the quality and efficiency of traffic management.

[0007] The method for processing traffic congestion according to at least one embodiment of the present disclosure optimizes the default response strategy according to the traffic scene information of the target road, including: in the case that the congestion factor is motor vehicle parking on the road, optimizing the disposal time and disposal range in the default response strategy according to the weather information, time period information and motor vehicle parking duration of the target road; in the case that the congestion factor is road damage, optimizing the road maintenance priority and maintenance time window in the default response strategy according to the weather information, time period information and road damage position of the target road; in the case that the congestion factor is the presence of road foreign matter, optimizing the cleaning time and cleaning method in the default response strategy according to the weather information, time period information and garbage type of the target road.

[0008] According to the technical solution of the present embodiment, the default response strategy is adjusted in combination with different traffic scene information for different congestion factors, which can ensure the pertinence of the adjusted target response strategy, improve its effectiveness, and further ensure the traffic management efficiency.

[0009] The method for processing traffic congestion according to at least one embodiment of the present disclosure further includes: in the case that the congestion factor is road damage, optimizing the temporary warning measure in the default response strategy according to the weather information and vehicle speed limit information of the target road; in the case that the congestion factor is the presence of road foreign matter, optimizing the temporary traffic control measure in the default response strategy according to the position information and traffic flow information of the road garbage on the target road.

[0010] According to the technical solution of the present embodiment, traffic confusion caused by road damage or road foreign matter can be effectively avoided, the road passing efficiency is improved, and the traffic congestion is reduced.

[0011] The method for processing traffic congestion according to at least one embodiment of the present disclosure determines the default response strategy according to the congestion factor existing in the identification result, including: matching in a strategy library according to the congestion factor existing in the identification result, the strategy library including the default response strategies corresponding to different congestion factors; determining the default response strategy corresponding to the identification result according to the matching result.

[0012] According to the technical solution of the present embodiment, the response measure corresponding to the congestion factor can be quickly determined through the pre-set strategy library, and the congestion processing time is reduced.

[0013] The method for processing traffic congestion according to at least one embodiment of the present disclosure comprises: identifying, according to the video image, determining an identification result of a congestion factor related to road congestion, which comprises: identifying, according to the video image, determining an identification result of a congestion factor, the identification result comprising a congestion factor type, a confidence level and congestion factor position information; and removing the identification result of the congestion factor with a confidence level less than a confidence threshold.

[0014] According to the technical solution of the present embodiment, by setting a confidence threshold, the identification result with high uncertainty can be effectively filtered, and the congestion factor information used for subsequent traffic management decision-making is more accurate and reliable.

[0015] The method for processing traffic congestion according to at least one embodiment of the present disclosure comprises: identifying, according to the video image, determining an identification result of a congestion factor related to road congestion, which comprises: identifying, according to the video image, determining an identification result of a congestion factor, the identification result comprising a congestion factor type, a confidence level and congestion factor position information; and removing the identification result of the congestion factor with a confidence level less than a confidence threshold.

[0016] According to the technical solution of the present embodiment, the congestion factor detected in the video image and its severity can be accurately determined.

[0017] The method for processing traffic congestion according to at least one embodiment of the present disclosure, wherein the congestion factor comprises road damage, road waterlogging, road manhole abnormality, presence of road foreign matter, road occupation or motor vehicle road parking.

[0018] According to the technical solution of the present embodiment, the congestion factors that may cause traffic congestion on the road surface can be comprehensively identified, and the effectiveness of subsequent traffic management is ensured.

[0019] The method for processing traffic congestion according to at least one embodiment of the present disclosure comprises: generating a linkage disposal instruction according to the target coping strategy; and sending the linkage disposal instruction to a target object.

[0020] According to the technical solution of the present embodiment, the strategy is converted into a specific instruction, so that the solution measures can be quickly and accurately executed, and the efficiency of traffic congestion processing is improved.

[0021] According to another aspect of the present disclosure, an electronic device is provided, comprising: a memory storing execution instructions; and a processor executing the execution instructions stored by the memory, so that the processor executes the method of any embodiment of the present disclosure.

[0022] According to still another aspect of the present disclosure, a readable storage medium is provided, in which execution instructions are stored, the execution instructions being executed by a processor to implement the method of any embodiment of the present disclosure.

[0023] According to yet another aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the method of any embodiment of the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0024] The accompanying drawings, which are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the present disclosure and together with the description serve to explain the principles of the present disclosure.

[0025] Figure 1 An application scenario schematic diagram applicable to the technical solution of the present disclosure is shown.

[0026] Figure 2 A flowchart of the traffic congestion processing method according to one embodiment of the present disclosure is shown.

[0027] FIG. 3 shows a flowchart of step S230 in the traffic congestion processing method according to one embodiment of the present disclosure.

[0028] FIG. 4 shows a flowchart of step S232 in the traffic congestion processing method according to one embodiment of the present disclosure.

[0029] FIG. 5 shows a flowchart of the optimization of temporary warning measures or temporary traffic control measures further included in the traffic congestion processing method according to one embodiment of the present disclosure.

[0030] FIG. 6 shows a flowchart of step S231 in the traffic congestion processing method according to one embodiment of the present disclosure.

[0031] FIG. 7 shows a flowchart of step S220 in the traffic congestion processing method according to one embodiment of the present disclosure.

[0032] FIG. 8 shows a flowchart of step S221 in the traffic congestion processing method according to one embodiment of the present disclosure.

[0033] Figure 9A flowchart illustrating step S240 in the method for processing traffic congestion according to one embodiment of the present disclosure is shown.

[0034] Figure 10 A flowchart illustrating the method for processing traffic congestion according to another embodiment of the present disclosure is shown.

[0035] Figure 11 A schematic framework diagram of the traffic congestion processing system according to one embodiment of the present disclosure is shown.

[0036] Figure 12 A schematic structural block diagram of an electronic device according to one embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0037] The present disclosure will be further described below in conjunction with the accompanying drawings and examples. It should be understood that the specific examples described herein are merely intended for explanation and are not limiting to the present disclosure. It should also be noted that only parts related to the present disclosure are shown in the accompanying drawings for ease of description.

[0038] It should be noted that the embodiments and features in the embodiments in the present disclosure can be combined with each other without conflict. The technical solutions of the present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0039] In small cities, although the number of vehicles is less than that in large cities, the traffic congestion problem is also prominent due to the lagging road planning and management means. The traffic congestion in small cities has its own unique characteristics and is obviously different from that in large cities.

[0040] Firstly, the road planning in small cities is relatively lagging, the road network density is low, the number of main roads is small, and some roads are narrow with limited traffic capacity. Moreover, the road infrastructure is aging and lacks facilities such as expressways and overpasses for diversion. Once the traffic volume increases, congestion is likely to occur.

[0041] Secondly, the composition of traffic participants in small cities is complex, including not only motor vehicles but also a large number of non-motor vehicles and pedestrians. During the morning and evening peak hours, electric vehicles and bicycles often mix and run, and the phenomenon of random lane changing and red light running is common, which seriously interferes with the traffic order. For example, private cars for picking up and dropping off students gather around schools, further exacerbating congestion.

[0042] Furthermore, the lack of parking space planning also easily leads to the problem of vehicle parking on the road. For example, in the surrounding areas of commercial districts and residential areas, vehicles are randomly parked and occupy the road, making the road narrower and reducing the traffic efficiency.

[0043] Finally, the traffic management technology in small cities is backward, mainly relying on manual command and simple traffic signal control, lacking intelligent traffic management system. The traffic law enforcement is insufficient, and the supervision of illegal behavior is not in place, which also affects the road traffic efficiency.

[0044] For the problems existing in traffic management, the technical scheme commonly used in large cities is to predict traffic congestion trend by combining multi-source data (such as video data, sensor data or OBD (On-Board Diagnostics, vehicle-mounted automatic diagnostic system) equipment, optimize signal control and emergency plan, which can achieve high prediction accuracy. However, the above method has significant deficiencies in small cities. Due to the lagging of road planning, aging of infrastructure and other reasons in small cities, the traditional model is difficult to adapt to the complex vehicle types, mixed non-motor vehicles and sudden road occupation and other unique road conditions, which leads to low accuracy and poor timeliness of model prediction results, and cannot guarantee the efficiency of traffic management.

[0045] Therefore, the present disclosure proposes the following technical scheme, in which, by obtaining a video image taken for a target road, the congestion factors are identified according to the video image, including road factors or personnel behavior factors. Then, based on the identified congestion factors, the corresponding countermeasures are determined and executed. In this way, without setting other sensors, traffic congestion prediction can be realized, which can effectively improve the accuracy and timeliness of traffic congestion prediction and guarantee the efficiency of traffic management.

[0046] Figure 1 An application scenario diagram applicable to the technical scheme of the present disclosure is shown. In this application scenario, it can include an image acquisition device 101, a network 102 and a server 103. The network 102 is used as a medium to provide a communication link between the image acquisition device 101 and the server 103. The network 102 can include various connection types, such as wired communication links, wireless communication links, etc.

[0047] Among them, the image acquisition device 101 can be arranged beside the target road which needs to be monitored, which can be a road camera, etc. The image acquisition device 101 can perform real-time video monitoring on the target road, and transmit the collected video image to the server 103 through the network 102.

[0048] The server 103 can be a server providing various services, which can be a physical server or a cloud server. The server 103 can identify the recognition result of the congestion factors related to road congestion according to the received video image, which can include road factors or personnel behavior factors. Then, the target countermeasures can be determined and executed according to the congestion factors existing in the recognition result.

[0049] Figure 2A flowchart of a traffic congestion processing method according to one embodiment of the present disclosure is shown. The method can be applied to a server 103 as shown, and can also be applied to any electronic device (such as a smartphone, tablet, laptop, desktop, etc.) with data receiving and data processing functions. The following description takes the method applied to a server as an example. Figure 1 The traffic congestion processing method can include steps S210 to S240 as shown.

[0050] The traffic congestion processing method can include steps S210 to S240 as shown. Figure 2

[0051] In step S210, a video image obtained by photographing a target road is acquired.

[0052] The target road can be any road that needs to be managed for traffic control.

[0053] In this embodiment, an image acquisition device (which can also be an existing monitoring camera) can be installed at a key location (such as an intersection, etc.) of the target road. The image acquisition device can be installed in a fixed manner and its position and angle can be optimized to ensure the coverage of the image acquisition device as much as possible.

[0054] The image acquisition device can capture video images of the target road in real time and transmit the video images to the server through network transmission for subsequent analysis and processing.

[0055] In step S220, the video image is identified to determine an identification result of a congestion factor related to road congestion, the congestion factor including a road factor or a personnel behavior factor.

[0056] The congestion factor can be an influencing factor that causes traffic flow to be stranded or interrupted, and can include a road factor or a personnel behavior factor. The road factor can be an influencing factor related to the road itself, such as the number of lanes, road damage, road foreign matter, road water accumulation, etc. The personnel behavior factor can be an influencing factor related to the behavior of pedestrians or drivers, such as road occupation, motor vehicle parking, etc. In one example, the congestion factor can include one or more of road damage, road water accumulation, road manhole abnormality, presence of road foreign matter, road occupation, or motor vehicle parking.

[0057] In this embodiment, the server can pre-process the received original video image to improve the clarity of the image, reduce background interference, etc. Then, the server can identify the relevant congestion factors from the pre-processed video image to obtain the identification result of the congestion factors.

[0058] ​In an example, the server can employ a deep learning algorithm (such as YOLO, etc.) to identify and locate vehicles and pedestrians in the video images, obtain their position coordinates and dynamic characteristics, and thus identify the congestion factors. For example, by analyzing the moving track of a vehicle, it can be determined whether there is a frequent lane-changing or parking behavior. By analyzing the path and behavior of a pedestrian, it can be determined whether there is a road-occupying business behavior, etc.

[0059] In step S230, a target response strategy is determined according to the congestion factors existing in the identification result, the target response strategy including a solution measure for eliminating the congestion factors existing on the target road.

[0060] Among them, the target response strategy can be a solution measure determined for the currently identified congestion factors, for eliminating the root cause of traffic congestion or interruption and improving the traffic condition.

[0061] In this embodiment, corresponding solution measures can be configured in advance for different types, different severity levels or different influence ranges of congestion factors. For example, for motor vehicle parking on the road, measures such as increasing law enforcement patrol vehicles can be adopted, and for foreign matter existing on the road, measures such as instructing a cleaning vehicle to go for cleaning can be adopted.

[0062] The server can match the congestion factors existing in the currently identified identification result, and thus determine the target response strategy corresponding to the identification result, which can include one or more solution measures for eliminating the congestion factors existing on the target road.

[0063] It should be noted that the solution measure can be configured for a single congestion factor, or it can be configured for multiple congestion factors that can exist simultaneously, so as to realize linkage and improve the processing efficiency of traffic congestion.

[0064] In step S240, the target response strategy is executed.

[0065] In this embodiment, the server can execute the solution measure in the target response strategy according to the determined target response strategy, so as to eliminate the congestion factors existing on the target road. For example, when a cleaning vehicle is needed to clean foreign matter on the road, the server can send corresponding instructions to relevant personnel or departments, indicating the position and timing of cleaning, etc.

[0066] In this way, based on the identification result of the target road, the server can determine the target response strategy corresponding to the identification result, and execute the solution measure in the target response strategy, so as to eliminate the congestion factors existing on the target road. Figure 2In the illustrated embodiment, a video image taken for a target road is acquired, and a congestion factor, including a road factor or a human behavior factor, is identified according to the video image. Then, a targeted response strategy is determined and executed based on the identified congestion factor. In this way, the traffic congestion prediction can be achieved without setting other sensors, and the accuracy and timeliness of the traffic congestion prediction can be effectively improved, and the traffic management efficiency can be ensured.

[0067] As to step S230, in some embodiments of the present disclosure, step S230 can include steps S231-S232 as shown. Figure 3

[0068] In step S231, a default response strategy is determined according to the congestion factor existing in the identification result.

[0069] The default response strategy can be a general solution preset based on the congestion factor, for example, when there is a foreign object (such as garbage or an obstacle, etc.) on the road, a cleaning vehicle is notified to clean the foreign object. When there is a motor vehicle parking on the road, a law enforcement patrol vehicle is increased, etc.

[0070] In this embodiment, the server can determine a default response strategy corresponding to the congestion factor according to the congestion factor existing in the current identification result according to a predetermined matching rule.

[0071] In step S232, the default response strategy is optimized according to the traffic scene information of the target road to determine a target response strategy, and the traffic scene information includes time period information, weather information, holiday information or traffic flow information.

[0072] The traffic scene information can be background information for reflecting the current traffic running state of the target road. It can be a dynamic or static condition such as time, weather, holiday and traffic flow, etc. describing the traffic running environment.

[0073] In this embodiment, the server can collect the current traffic scene information of the target road, for example, the server can analyze the video image corresponding to the target road by means of image recognition technology to determine the traffic flow information (such as the number of vehicles passing per minute) on the road and the weather information (whether it is raining, whether it is sunny, etc.) on the target road. Then, the time period corresponding to the current time, whether it is a holiday, etc. can also be determined based on the time information of the server itself device.

[0074] After the above traffic scene information is determined, the server can adjust the default response strategy determined in advance according to the traffic scene information according to certain rules, so that the target response strategy determined can be more in line with the actual situation on the spot, and the traffic management quality and efficiency can be improved. ​

[0075] For example, when there is a foreign matter on the road, the default coping strategy is to immediately notify a cleaning vehicle to clean the foreign matter. However, if the current time period is the morning or evening peak period, if a cleaning vehicle is dispatched for cleaning, traffic congestion will be further aggravated. Therefore, the cleaning time of the cleaning vehicle can be adjusted, for example, delayed, so as to improve the pertinence of the determined target coping strategy and ensure the traffic management efficiency.

[0076] It should be understood that different traffic scene information can be combined to optimize the default coping strategy for different congestion factors.

[0077] Regarding step S232, in some embodiments of the present disclosure, step S232 can include steps S2321 to S2323 as shown in the following table. Figure 4

[0078] In step S2321, when the congestion factor is motor vehicle parking on the road, the disposal time and range in the default coping strategy are optimized according to the weather information, time period information and motor vehicle parking duration of the target road.

[0079] In this embodiment, when the identification result indicates that the target road has the congestion factor of motor vehicle parking on the road, the disposal time and disposal range in the default coping strategy can be optimized according to the weather information, time period information and motor vehicle parking duration of the target road when optimizing the corresponding default coping strategy.

[0080] The disposal time can be the patrol time of the law enforcement patrol vehicle or the time point of executing the tow truck or the reminder. The disposal range can refer to the target road area that needs to be managed (for example, whether only the vehicle parked on the road is managed or other areas on the target road also need to be managed).

[0081] In step S2322, when the congestion factor is road damage, the road maintenance priority and maintenance time window in the default coping strategy are optimized according to the weather information, time period information and road damage position of the target road.

[0082] In this embodiment, when the identification result indicates that the target road has the congestion factor of road damage, the road maintenance priority and maintenance time window in the default coping strategy can be optimized according to the weather information, time period information and road damage position of the target road when optimizing the corresponding default coping strategy.

[0083] ​The road maintenance priority can be a priority of a road damage repair sequence, and it should be understood that the higher the priority, the greater the impact on traffic and the sooner the repair should be made. The lower the priority, the smaller the impact on traffic and the later the repair can be made. The maintenance time window can be a time period for road maintenance work.

[0084] In step S2323, in the case of the congestion factor being the presence of road debris, the cleaning time and the cleaning method in the default response strategy are optimized according to the weather information, the time period information, and the garbage type of the target road.

[0085] In this embodiment, when the result indicates that the target road has the congestion factor of road debris, the cleaning time and the cleaning method in the default response strategy corresponding to the target road are optimized in combination with the weather information, the time period information, and the garbage type (e.g., light plastic or heavy construction waste, which can be identified according to a video image) of the target road.

[0086] The cleaning time can be a time point for cleaning work, and the cleaning method can be mechanical or manual cleaning, or the use of a small and fast cleaning vehicle or a large cleaning vehicle, etc.

[0087] In this way, the default response strategy is adjusted in combination with different traffic scene information for different congestion factors, which can ensure the pertinence of the adjusted target response strategy, improve its effectiveness, and further ensure the traffic management efficiency.

[0088] Based on the foregoing embodiments, in some embodiments of the present disclosure, the traffic congestion processing method further includes steps S510 to S520 as shown in the following. Figure 5

[0089] In step S510, in the case of the congestion factor being road damage, the temporary warning measure in the default response strategy is optimized according to the weather information and the speed limit information of the target road.

[0090] The temporary warning measure can be a short-term warning means for a specific situation, such as a warning sign, a flashing light, etc.

[0091] ​In this embodiment, in the case of road damage, the temporary warning measure in the default response strategy can be adjusted in combination with the weather information and the speed limit information of the target road. For example, if the current weather is ordinary weather and the speed limit of the target road is less than or equal to 40Km / h, the temporary warning measure can be to set a warning sign in front of the damaged position. If the current weather is foggy and the speed limit of the target road is less than or equal to 80Km / h, the temporary warning measure can be adjusted to place three warning signs (with a 50m interval) at the damaged position, and then place a flashing light.

[0092] In step S520, in the case of a congestion factor of a road foreign object, the temporary traffic control measure in the default response strategy is optimized according to the position information of the road garbage on the target road and the traffic flow information.

[0093] The temporary traffic control measure can be a short-term traffic management measure taken to ensure traffic safety and efficiency, such as adjusting the length of the red light at the upstream intersection.

[0094] In this embodiment, in the case of a road foreign object, the temporary traffic control measure in the default response strategy can be optimized in combination with the position information of the road garbage on the target road and the traffic flow information. For example, when the road foreign object is located 200m away from the intersection, the green light length of the upstream intersection can be shortened by 10 seconds during non-peak hours to reduce the traffic flow entering the road section. However, during peak hours, the upstream intersection can start the "red light flow limiting" measure 3 cycles in advance, i.e. extend the red light length to reduce the traffic flow entering the road section.

[0095] In this way, traffic chaos caused by road damage or road foreign objects can be effectively avoided, road traffic efficiency can be improved, and traffic congestion can be reduced.

[0096] In some embodiments of the present disclosure, when optimizing the default response strategy, the optimization can be implemented based on a large language model. Specifically, the determined default response strategy and the traffic scenario information of the target road can be input into the large language model, so that the large language model outputs an adjusted target response strategy according to the input. In this way, the optimization efficiency of the strategy can be improved, and the effectiveness of the adjustment result can be ensured.

[0097] Regarding step S231, in some embodiments of the present disclosure, step S231 can include steps S2311-S2312 as shown in Figure 6

[0098] In step S2311, according to the congestion factors existing in the identification result, matching is performed in a strategy library, and the strategy library includes default response strategies corresponding to different congestion factors.​

[0099] The policy library can be a database storing default coping strategies corresponding to different congestion factors, for quickly matching and providing solutions.

[0100] In this embodiment, the corresponding default coping strategies for different congestion factors can be configured in advance and added to the policy library for storage. The server can match the congestion factors in the current identification result with the data in the policy library to find the corresponding default coping strategies.

[0101] In an example, the policy library can store a correspondence table of congestion factors and default coping strategies, and the server can query the correspondence table according to the congestion factors to determine whether there is a matching default coping strategy.

[0102] In step S2312, the default coping strategy corresponding to the identification result is determined according to the matching result.

[0103] In this embodiment, according to the matching result, the server can obtain the corresponding default coping strategy from the policy library.

[0104] In this way, through the pre-configured policy library, the corresponding coping measures for the congestion factors can be quickly determined, and the congestion handling time is reduced.

[0105] Regarding step S220, in some embodiments of the present disclosure, step S220 can include steps S221 to S220 as shown in Figure 7

[0106] In step S221, the identification result of the congestion factor is determined according to the video image, including the congestion factor type, the confidence and the congestion factor position information.

[0107] The congestion factor type can be a specific reason category causing traffic congestion, such as road damage, presence of road foreign objects, motor vehicle parking on the road, etc.

[0108] The confidence can be the certainty degree of the identification algorithm for judging a certain congestion factor. It is usually represented by a value between 0 and 1, and the higher the value, the greater the algorithm's grasp of the judgment of the congestion factor.

[0109] The congestion factor position information can be the specific position of the congestion factor in the road or traffic scene, which can be represented by coordinate or distance information, etc.

[0110] In this embodiment, the traffic-related algorithm can be used to identify the video image, and the traffic-related algorithm can output the identification result of the corresponding congestion factor.

[0111] ​In an example, the traffic-related algorithms described above can include one or more of a vehicle flow identification algorithm, a road garbage identification algorithm, and a violation of road occupation identification algorithm. Each algorithm can independently analyze and process the video images and output identification results, respectively, to comprehensively capture various traffic information on the road.

[0112] For example, in the processing of the vehicle flow identification algorithm, computer vision technology can be used to achieve vehicle detection, tracking, and counting, and ultimately present the results in a structured data form, directly reflecting the quantitative results of vehicle flow. For example, the number of vehicles passing through the detection area per unit time.

[0113] In the processing of the road garbage identification algorithm, the algorithm can automatically detect and classify the types and ranges of garbage on the road. For example, the algorithm can use a target detection algorithm to identify garbage from video images and output the location, garbage range, and road area ratio, and filter out small targets (e.g., targets below 70 pixels). Thus, whether the garbage affects traffic can be analyzed.

[0114] In the processing of the road occupation identification algorithm, the algorithm can automatically detect whether there is road occupation (e.g., mobile vendors, shop operations, etc.) in the video images. The algorithm determines whether it is road occupation by overlapping the pre-marked road area with the identified target.

[0115] In the processing of the motor vehicle occupation parking identification algorithm, the algorithm can identify motor vehicles from video images and locate their positions (marked by bounding boxes). Then, the algorithm determines whether it is occupation parking by overlapping the parking position of the vehicle with the pre-marked no-parking area on the road. Meanwhile, the algorithm can also identify the duration of the motor vehicle occupation, which can be obtained by analyzing consecutive video images.

[0116] In step S222, the identification results of the congestion factors with a confidence level less than a confidence threshold are removed.

[0117] The confidence threshold can be a set value for screening the identification results of the congestion factors. Only the identification results with a confidence level higher than or equal to the confidence threshold are retained.

[0118] In this embodiment, the server can remove the identification results of the congestion factors with a confidence level less than the set confidence threshold, thereby removing invalid or interfering data and retaining valid data that is actually meaningful for traffic congestion prediction.

[0119] It should be noted that a general confidence threshold can be set for screening the identification results, or different confidence thresholds can be set for different algorithms to ensure the accuracy of the screening results.

[0120] In this way, by setting the confidence threshold, the recognition result with high uncertainty can be effectively filtered, and the congestion factor information used for subsequent traffic management decision-making is more accurate and reliable.

[0121] In some embodiments of the present disclosure, in order to ensure the efficiency, accuracy and stability of the parallel operation of multiple algorithms, an algorithm detection cooperation and fault tolerance mechanism scheme is also provided. By cooperating algorithms with different functions (such as motor vehicle lane occupation recognition algorithm, road damage detection algorithm, vehicle flow calculation algorithm, etc.), comprehensive perception and accurate recognition of congestion factors are achieved. At the same time, through the fault tolerance mechanism, problems such as algorithm mis-detection, missed detection, hardware failure, data anomaly, etc. are dealt with, ensuring that the entire traffic congestion processing process is uninterrupted and the decision is not mistaken, providing reliable data support for the formulation and execution of subsequent target response strategies, and ultimately improving the efficiency and quality of traffic management.

[0122] Specifically, when identifying and determining the recognition result of the congestion factor related to road congestion according to the video image, it can include the following steps: At the data layer, the video image is preprocessed, and the traffic scene information of the target road is integrated to form a standardized data set. At the algorithm layer, the standardized data set is input in parallel to multiple special detection algorithms, which can include at least two of the motor vehicle lane occupation parking recognition algorithm, the road damage detection algorithm, the vehicle flow calculation algorithm, and the lane occupation operation recognition algorithm. Each algorithm runs independently and outputs a preliminary detection result.

[0123] At the decision-making layer, the preliminary detection results of each algorithm are received through the interaction node, the results of different algorithms are associated and complementarily verified based on the preset cooperation strategy, and the final congestion factor recognition result is generated.

[0124] In this embodiment, a "hierarchical cooperation + module linkage" architecture is adopted to build an algorithm detection cooperation system, which is specifically divided into three levels of data layer, algorithm layer and decision-making layer. The data and instructions are efficiently transferred between each level through standardized interfaces. The processing flow of this architecture is as follows: Data layer cooperation: responsible for uniformly receiving video data transmitted by front-end image acquisition devices (such as road cameras), and pre-processing the data (denoising, enhancement, key frame extraction, etc.) to form a standardized image data set. At the same time, the data layer needs to realize the fusion of multiple sources of data. In addition to video image data, traffic scene information (such as time period, weather, holiday, traffic flow information, etc.) can also be integrated, and the processed data is broadcasted and distributed to each algorithm module in synchronization to ensure the consistency and timeliness of the data obtained by each algorithm.

[0125] Algorithm layer coordination: This layer can include various special detection algorithms, such as motor vehicle parking on the road recognition algorithm, road damage detection algorithm, road foreign object recognition algorithm, traffic volume calculation algorithm, and parking on the road recognition algorithm. The algorithm layer realizes coordination through "parallel running + result interaction": (1) Parallel running: each algorithm module independently processes the standardized data distributed by the data layer, such as the motor vehicle parking on the road recognition algorithm focusing on detecting whether the vehicle is illegally parked, and the road damage detection algorithm identifying whether the road surface is damaged, ensuring that each algorithm completes the special detection task in the same time dimension and improves the overall detection efficiency. (2) Result interaction: set algorithm interaction nodes, each algorithm uploads the preliminary detection results (such as congestion factor type, confidence, location information, and retention time) to the interaction node in real time, and the interaction node performs preliminary integration and correlation analysis on the results. For example, when the traffic volume calculation algorithm detects a sudden drop in traffic volume on a road section, it will trigger the interaction node to retrieve the motor vehicle parking on the road recognition results and road foreign object recognition results for that road section to determine whether the traffic drop is caused by vehicle parking or road foreign objects, achieving complementary verification of different algorithm detection results.

[0126] Decision layer coordination: Based on the detection results integrated by the algorithm layer interaction node, the decision layer realizes coordinated decision-making through the rule engine and large language model. The rule engine first matches the default response strategy corresponding to the congestion factor according to the pre-set strategy library. The large language model optimizes the default response strategy based on traffic scene information to generate the target response strategy. At the same time, the decision layer will return the feedback information generated during the decision-making process (such as strategy optimization basis) to the algorithm layer to provide reference for parameter adjustment, forming a closed-loop coordination of "detection - decision - feedback".

[0127] In an example, the specific coordination strategy can be as follows: 1) When the motor vehicle parking on the road recognition algorithm detects the presence of vehicle parking, it needs to coordinate with the traffic volume calculation algorithm and the time period recognition algorithm. The traffic volume calculation algorithm provides real-time traffic data for the road section, and if the traffic volume is large (such as peak period traffic density exceeding the pre-set threshold), the retention time of the parked vehicle (provided by the motor vehicle parking on the road recognition algorithm) needs to be confirmed to determine its impact on traffic congestion. The time period recognition algorithm determines whether it is a special period such as morning or evening peak, holiday, etc., providing a basis for subsequent optimization of response strategies (such as adjustment of disposal timing). For example, if a vehicle is detected to be parked on the road during the morning peak period and the retention time exceeds 10 minutes, and the traffic volume calculation algorithm shows that traffic congestion has formed, the disposal instruction needs to be triggered first to avoid exacerbating congestion.

[0128] 2) Road damage related coordination strategy: After the road damage detection algorithm detects road damage, it needs to coordinate with the weather interface data, road speed limit configuration, and vehicle flow calculation algorithm. The weather interface data provides the current weather conditions (such as whether it is raining or snowing), and if it is raining, it needs to determine whether the rain will exacerbate the spread of road damage. The road speed limit configuration provides the speed limit standard of the road section, combined with the location of the road damage (such as whether it is on the main road or the bend), to evaluate the safety risk of the damage to vehicle traffic. The vehicle flow calculation algorithm provides real-time traffic data to determine the best time window for maintenance (such as during the low traffic period). For example, if there is road damage on the main road bend and the weather is rainy and the traffic is heavy, the coordination result of the road damage detection algorithm with the weather interface and the speed algorithm will trigger the optimization of temporary warning measures (such as increasing the number of warning signs and flashing lights), and combined with the traffic data to determine the specific time for night maintenance.

[0129] 3) Road foreign matter related coordination strategy: After the road foreign matter recognition algorithm detects that there is foreign matter on the road, it needs to coordinate with the garbage type recognition algorithm, weather interface, and vehicle flow calculation algorithm. The garbage type recognition algorithm determines the material of the foreign matter (such as lightweight plastic, heavy construction waste), and judges the difficulty of cleaning and the required equipment. The weather interface provides information such as wind power and rainfall to evaluate whether the foreign matter has a risk of moving (such as strong winds may cause lightweight plastic foreign matter to blow to the opposite lane). The vehicle flow calculation algorithm provides real-time traffic data and traffic trend prediction to determine the cleaning time and temporary traffic control measures. For example, if lightweight plastic foreign matter is detected on the main road and the wind power reaches level 5 and it is currently a flat peak period, the coordination result will trigger an immediate cleaning instruction, and combined with the traffic data to shorten the green light duration at the upstream intersection to reduce the traffic entering the road section.

[0130] In an embodiment, the fault-tolerant mechanism can include: Setting a confidence filtering and threshold dynamic adjustment mechanism to filter the confidence of the recognition results output by each algorithm, and only retaining results with a confidence greater than or equal to a preset reference threshold; When it is detected that the false detection rate or the missed detection rate of a certain algorithm in a specific scenario exceeds a preset threshold, the confidence threshold in that scenario is automatically adjusted, and the results of other algorithms are combined for complementary verification; When it is detected that the algorithm false detection rate, missed detection rate, or hardware failure duration exceeds a preset duration, a multi-level warning is triggered, and the third level warning starts the manual intervention mode, suspends automatic decision-making, and remotely adjusts the parameters or arranges on-site troubleshooting by the operation and maintenance personnel.

[0131] In this implementation, each detection algorithm needs to attach a confidence parameter (to reflect the reliability of the detection result) when outputting the detection result. A unified confidence threshold benchmark (such as an initial threshold of 0.7) is set, and detection results with a confidence lower than the threshold are filtered to avoid low reliability results affecting subsequent decisions. At the same time, based on historical detection data, a threshold dynamic adjustment mechanism is established: if the false detection rate of a certain algorithm increases in a specific scenario (such as rain), such as misjudging accumulated water as road damage, the confidence threshold in that scenario is automatically increased (such as from 0.7 to 0.8); if the false detection rate increases (such as difficulty in identifying road foreign objects in fog), the threshold is appropriately reduced (such as from 0.7 to 0.6), and complementary verification is performed in combination with other algorithm results to ensure the accuracy and completeness of the detection results.

[0132] Single algorithm avoids relying on a single feature for detection and needs to integrate multiple features to improve fault tolerance. For example, in addition to judging the overlap between vehicle position and no-parking area, the motor vehicle parking on the road recognition algorithm also needs to combine features such as vehicle residence time and surrounding traffic speed: if the overlap between vehicle position and no-parking area is low, but the residence time exceeds 15 minutes and the surrounding traffic speed is significantly lower than the normal level, it is still determined as suspected parking on the road, which needs to be further confirmed by manual review or other algorithm results. The road damage detection algorithm needs to integrate road surface texture changes, flatness differences, edge contour features, etc., to avoid false positives or false negatives caused by changes in light (such as strong light, shadows).

[0133] Then, by setting multiple levels of warning thresholds, when the system detects that the algorithm false detection rate, false detection rate exceeds the preset threshold (such as false detection rate exceeds 15%), or hardware failure, data anomaly duration exceeds the set time (such as device failure exceeds 5 minutes), triggers different levels of warning: 1) First level warning (minor anomaly): the system automatically starts fault tolerance measures (such as threshold adjustment, data completion), and records the abnormal information to the log, without human intervention; 2) Second level warning (moderate anomaly): In addition to starting automatic fault tolerance measures, the system also needs to notify traffic management and operation personnel through SMS, platform message, etc. to inform the type, location and impact range of the anomaly, and the operation personnel need to monitor the anomaly handling progress in real time; Third level warning (serious anomaly): When the automatic fault tolerance measures cannot solve the problem (such as multiple collection devices fail at the same time, algorithm false detection in a large area), trigger the third level warning, the system suspends automatic decision-making and switches to manual intervention mode, the operation personnel can adjust the algorithm parameters, switch to standby equipment, or arrange on-site personnel to troubleshoot the fault through remote control to ensure that the system resumes normal operation as soon as possible.

[0134] In an embodiment, for fault recovery and data rollback, missing or abnormal data during the fault can be supplemented by front-end equipment (such as camera recovery to supplement cached data during the fault period), peripheral equipment data integration, and the like, to ensure the continuity of the data. On the other hand, the algorithm parameters can be restored to the optimal configuration before the fault, and the algorithm model can be fine-tuned in combination with the manual decision results during the fault (such as training the algorithm by using historical correct decision data to reduce the subsequent mis-detection rate). After the automatic decision function is restored, the first batch of generated target response strategies can be manually reviewed to confirm the rationality of the strategies, avoid decision errors caused by unstable algorithms in the early stage of fault recovery, and then completely hand over to the automatic decision mechanism for processing when the system is stable.

[0135] Regarding step S221, in some embodiments of the present disclosure, step S221 can include steps S2211 to S2215 as shown in the following table. Figure 8

[0136] In step S2211, the video image is preprocessed to determine the image key frame.

[0137] In this embodiment, the server can preprocess the received video image, including but not limited to denoising, enhancement, key frame extraction, and the like. In an example, the acquired video image can be first preprocessed by denoising, enhancement, and the like to improve the clarity and quality of the image. Then, by calculating the difference between adjacent frames or using a motion detection algorithm, the area with significant changes or motion in the video is detected. According to these change or motion information, the image key frame that can represent the change of the video content is selected according to certain rules and thresholds.

[0138] In step S2212, target recognition is performed according to the image key frame to determine the target object existing in the image key frame.

[0139] Among them, the target recognition can be a recognition process of identifying a specific target from an image by using computer vision technology and algorithm. The specific target can be a vehicle, a pedestrian, or a static obstacle (such as garbage), and the like.

[0140] In this embodiment, the extracted image key frame can be input into a pre-trained target recognition model, such as a convolutional neural network (CNN) model based on deep learning. The model extracts and analyzes the features of the key frame image, matches with known target categories, and thus identifies the target object existing in the image and gives the position, category, and the like of the target object.

[0141] In step S2213, the target object detected in the image key frame is calculated for overlap with the sensitive detection area to determine the overlap area ratio. ​

[0142] The sensitive detection area can be an area that has special detection significance for events such as traffic congestion, for example, a lane, an intersection, a curb, etc. The sensitive detection area can be obtained by manual annotation for each road camera and different algorithms.

[0143] The overlap area ratio can be the ratio of the overlap area of the identified target and the sensitive detection area in the image plane to the area of the target or the area of the sensitive detection area.

[0144] In this embodiment, the position and range of the sensitive detection area can be set in the video image in advance according to the actual traffic scene and monitoring requirements. For each target identified in the key frame, the area of the overlap between the boundary box of the target and the boundary of the sensitive detection area is calculated. Then, the overlap area is divided by the area of the target and the area of the sensitive detection area, respectively, to obtain two overlap area ratio values.

[0145] It should be noted that for different targets, it can be selected whether to calculate the overlap area ratio. For example, when the target is a vehicle and the detection target is whether the vehicle is parked on the road, the area of the vehicle in the image can be calculated with the no-parking area on the target road. If the target is a road manhole cover and the detection target is whether the road manhole cover is missing, no overlap calculation is needed.

[0146] In step S2214, the residence time of the target detected in the image key frame is determined according to the time sequence information of the video image.

[0147] The time sequence information can refer to the time sequence and time interval information between the frames in the video image.

[0148] The residence time can refer to the length of time that the target continuously exists in the sensitive detection area.

[0149] In this embodiment, the key frame sequence is time-labeled using the time sequence information of the video image. For each target that enters the sensitive detection area, the time stamp corresponding to the first appearance of the key frame is recorded as the entering time, and the time stamp corresponding to the last appearance in the key frame is recorded as the leaving time. The time difference between the entering time and the leaving time (or the current time if the target has not left) is calculated, which is the residence time of the target in the sensitive detection area.

[0150] It should be understood that the determination of the residence time can reflect the staying situation of the target in the sensitive detection area, which is helpful for analyzing the influence degree of the target on the traffic flow and the persistence of the congestion. Long residence time can indicate that the target is one of the important factors causing the congestion, providing time dimension information for subsequent traffic management and decision-making.

[0151] Similarly, for different target objects, it can be selected whether to determine the residence time. For example, when the target object is a road manhole cover, it only needs to be detected whether it exists, and it is not necessary to determine its residence time. When the target object is a vehicle parked on the road, it is necessary to analyze its residence time in order to analyze its influence on traffic congestion in the subsequent.

[0152] In step S2215, according to the target object, the overlapping area ratio and / or the residence time, the identification result of the congestion factor is determined.

[0153] In this embodiment, the categories, positions, overlapping area ratios and residence times of the target objects and other factors can be considered comprehensively to establish a congestion factor evaluation model or rule. According to the preset threshold and weight, the related parameters of each target object are analyzed and evaluated. For example, if the target object is a vehicle and its overlapping area ratio in the sensitive detection area is large and its residence time is long, it can be determined that the vehicle is an important factor causing congestion. In this way, the congestion factors and their severity detected in the video image can be accurately determined.

[0154] In an embodiment, for the picture that is blocked and cannot be identified, a warning prompt can be given to the relevant personnel. And for different algorithms, non-target interference objects (such as pedestrians, static obstacles, vehicles in motion, etc.) can be excluded.

[0155] In an embodiment, the identification rules can be preset according to the characteristics and application scenarios of different algorithms, and the corresponding pre-plan rules can be configured in advance. Specifically, the identification data after the above filtering (such as confidence threshold screening) can be input into the rule engine. The rule engine can analyze and integrate the data according to the preset identification rules, associate different algorithm results with the corresponding pre-plan (i.e. default coping strategy), and generate related data containing event type, severity, impact range and other information.

[0156] For example, for the motor vehicle parking / flowing vendor identification algorithm, and the residence time reaches a certain time, for example, 10 minutes, at this time, another algorithm (such as vehicle flow detection algorithm) identifies that the passing vehicle data decreases from 10 vehicles per minute to only 2 vehicles per minute, the decrease = [(10-2) / 10] x 100% = (8 / 10) x 100% = 80%, and the duration exceeds 5 minutes, then based on this rule, according to the scheme generated by the rule engine configuration, the surrounding personnel can be notified through various ways such as SMS, APP message, etc. for manual on-site disposal operation.

[0157] For example, if the detection range of the garbage on the road occupies a large part of the road, and the type of garbage identified is an obstacle that can easily cause accidents, the APP message or SMS can be used to link the nearby road workers for disposal.

[0158] In an embodiment, the data after the integration of the rule engine can be input into a large language model. The large language model can combine current traffic scene information such as time period, weather condition, holiday, etc., and use deep learning and natural language processing technologies to optimize and adjust the associated plans, generating optimized plans (i.e., target coping strategies) that are more suitable for actual situations.

[0159] For example, for the default response strategy of motor vehicle parking on the road, the original default response strategy is triggered only according to "motor vehicle parking on the road + current traffic flow", without considering the influence of weather, time period, etc. on traffic efficiency (e.g., long braking distance of vehicles in rainy weather, and early diversion needs to be started).

[0160] The optimization direction and parameter adjustment of the large language model can include: (1) Dynamic adjustment of disposal time and disposal range: ① Associated factors: weather (rain / snow / sunny), time period (peak / non-peak), and occupancy time (<10 minutes / 10 minutes-2 hours / >2 hours). Based on these associated elements, the large language model can achieve more accurate and dynamic management by analyzing the current visual analysis algorithm recognition results and historical data. The specific logic can be: by learning the traffic flow decay curve of the "rain / snow + occupancy" scene in historical data, the congestion diffusion speed is predicted, and the severity, influence range, etc. are automatically adjusted.

[0161] Alternatively, for the default response strategy of road damage, the original default response strategy is triggered only according to "damage area (>1㎡ triggers maintenance) + traffic flow (>500 vehicles / hour starts emergency disposal)", without considering factors such as weather (e.g., heavy rain may exacerbate damage spread) and time period (night maintenance has less impact on traffic).

[0162] The optimization direction and parameter adjustment of the large language model can include: (1) Dynamic determination of road maintenance priority and maintenance time window ① Associated factors: weather, time period, and damage location (main road / secondary road).

[0163] ② Optimization parameters include: 1) Response time: 1㎡ of damage on the main road, response time is 2 hours in sunny daytime; in bad weather, the response time is compressed to 1 hour.

[0164] 2) Maintenance period: according to whether it is a holiday, traffic flow, etc., the maintenance time window is determined.

[0165] 3) Maintenance priority: Through historical data training, according to the "rain + high temperature alternation" weather, the diffusion speed of road damage under the traffic flow, the maintenance priority is promoted.

[0166] (2) Strengthening of temporary warning measures ① Associated factors: visibility, road speed limit.

[0167] ② Optimization parameters: 1) Warning intensity: ordinary weather + low-speed road section (<40 km / h), set 1 warning sign; fog + high-speed road section (>80 km / h), increase to 3 warning signs (interval 50 meters) + flashing light (default response strategy only 1 warning sign, easy to ignore).

[0168] Large language model processing logic can be: combined with NLP analysis of historical story data, extract the accident causes in the "fog + road damage" scene (such as insufficient warning leading to rear-end collision), and specifically strengthen the temporary measures.

[0169] Alternatively, for the default response strategy caused by road garbage, it originally only starts the plan according to "garbage size (>50cm triggers cleaning) + lane (main lane / auxiliary lane)", without considering weather (such as strong wind may move garbage), time (morning and evening peak cleaning has greater impact on traffic).

[0170] Then the optimization direction and parameter adjustment of the large language model can include: (1) Cleaning time and method selection ① Associated factors: wind force (>5 levels easy to move), time (peak / flat peak), garbage type (lightweight plastic / heavy construction waste).

[0171] ② Optimization parameters: 1) Cleaning time: there is lightweight plastic garbage on the main lane, and it is a 5-level windy weather, even if it is in the morning peak (7:30-9:00), it still arranges immediate fast cleaning (because the garbage may be blown to the opposite lane to cause accidents); no wind weather + flat peak period, delay to 10:00 for cleaning.

[0172] 2) Cleaning equipment: small fast cleaning vehicles (occupy 1 lane) instead of large cleaning vehicles (occupy 2 lanes) on the roads around scenic spots during holidays (traffic flow is 2 times that of weekdays), to reduce traffic interference (traditional rules always use large vehicles, which easily exacerbate congestion).

[0173] ③ Large model logic: through visual algorithm to identify garbage material + wind force prediction data, calculate the garbage movement trajectory probability (such as the probability of lightweight garbage moving to the opposite lane within 10 minutes under 5-level wind is 65%), and preferentially handle high-risk scenarios.

[0174] (2) Temporary traffic control cooperation ① Associated elements: distance from garbage location to intersection (<300 meters can be linked to signal), current speed (<20 km / h can be considered as congestion state).

[0175] ② Optimization parameters: signal linkage: garbage is located 200 meters away from the intersection, during the flat peak period, the green light duration of the upstream intersection is shortened by 10 seconds when cleaning, reducing the vehicle flow entering the road section; during the peak period + congestion state, the upstream intersection starts "red light flow limiting" 3 periods in advance (traditional rules without signal linkage, cleaning is easy to form a queue).

[0176] The processing logic of the large language model can be: combined with the traffic flow prediction model (such as XGBoost), calculate the traffic delay during cleaning, and automatically trigger upstream signal control to reduce the impact.

[0177] Regarding step S240, in some embodiments of the present disclosure, step S240 can include steps S241-S242 as shown in Figure 9 .

[0178] In step S241, a linkage disposal instruction is generated according to the target response strategy.

[0179] The linkage disposal instruction can be an instruction information for guiding the operation of related equipment or personnel.

[0180] In this embodiment, each solution measure in the target response strategy can be converted into a specific linkage disposal instruction according to the target response strategy.

[0181] In step S242, the linkage disposal instruction is sent to the target object.

[0182] The target object can be a device, system or personnel that needs to receive and execute the linkage disposal instruction.

[0183] In this embodiment, the server can send the generated linkage disposal instruction to the target object, such as a traffic signal control system, a traffic guidance screen, a law enforcement personnel terminal, etc., thereby realizing timely intervention and dredging of traffic congestion.

[0184] In this way, the strategy is converted into a specific instruction, so that the solution measure can be quickly and accurately executed, improving the efficiency of traffic congestion processing.

[0185] Based on the technical solutions of the above embodiments, a specific application scenario of the embodiments of the present application is introduced as follows: Figure 10 A flowchart of a traffic congestion processing method according to another embodiment of the present disclosure is shown. As Figure 10As shown, it can include steps S1010 to S1060.

[0186] In step S1010, the access front-end camera collects video data. The front-end camera can be deployed at key locations of the target road to collect real-time video data of the road.

[0187] In step S1020, multi-algorithm parallel analysis is performed. The collected video data is input into multiple parallel running analysis algorithms. These algorithms may include target detection, target tracking, behavior recognition, etc., for analyzing key information in the video data from different angles and levels.

[0188] In step S1030, algorithm threshold is set to filter data. According to the analysis requirements and algorithm characteristics, the corresponding threshold is set to screen out key information. For example, a confidence threshold is set for the road foreign object recognition algorithm, only retaining foreign object detection results with a confidence higher than the threshold, and removing low-confidence detection data to improve the accuracy and efficiency of subsequent processing.

[0189] In step S1040, the rule engine integrates data. The filtered data is integrated and analyzed by the rule engine. The rule engine can integrate the analysis results of different algorithms according to the pre-set rules and logic, and associate the corresponding default response strategy.

[0190] In step S1050, the large language model combines the scene to optimize the plan. The integrated data is input into the large language model, combined with the current traffic scene and historical data, to generate an optimized target response strategy.

[0191] In step S1060, the output linkage disposal instruction is output. According to the optimized target response strategy, specific linkage disposal instructions are generated. These instructions are sent to relevant traffic management equipment and personnel to guide them to take corresponding measures to alleviate traffic congestion.

[0192] Figure 11 A schematic diagram of the framework of a traffic congestion processing system according to one embodiment of the present disclosure is shown. As shown, the system can include a video image acquisition module 1110, a visual algorithm recognition module 1120, and an intelligent analysis module 1130. Figure 11

[0193] ​The video image acquisition module 1110 is configured to acquire real-time video images of a target road, to provide original data support for subsequent congestion factor identification and analysis. The visual algorithm identification module 1120 is configured to analyze and process the acquired video images by using various identification algorithms, to identify various factors related to traffic congestion in the video images, to provide a basis for subsequent congestion evaluation and decision-making. The intelligent analysis module 1130 is configured to comprehensively analyze the traffic congestion situation by using the congestion factor information output by the visual algorithm identification module 1120 and other auxiliary data (such as vehicle flow, pedestrian flow, whether a main road, weather, etc.), to evaluate the congestion degree, to predict the congestion development trend, and to determine whether intervention is needed.

[0194] The present disclosure also provides an electronic device. Figure 12 A schematic diagram of a hardware implementation of a processing system is shown.

[0195] As shown in Figure 12 The hardware structure of the electronic device 1000 can be implemented by using a bus architecture. The bus architecture can include any number of interconnected buses and bridges, depending on the particular application of the hardware and overall design constraints. The bus 1100 connects various circuits including one or more processors 1200, memories 1300, and / or hardware modules. The bus 1100 can also connect various other circuits 1400 such as peripheral devices, voltage regulators, power management circuits, external antennas, etc. The bus 1100 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one connection line is shown in the figure, but it does not mean that there is only one bus or only one type of bus.

[0196] The present disclosure also provides a readable storage medium having a computer program stored therein, the computer program being executed by a processor to implement the above method. The "readable storage medium" can be any device that can contain a program for use by or in connection with an instruction execution system, apparatus, or device, or in conjunction with such an instruction execution system, apparatus, or device. More specific examples of the readable storage medium include the following: an electrical connection having one or more wires (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM), etc.

[0197] The present disclosure also provides a computer program product. The methods of the present disclosure can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, the computer program or instructions can be implemented in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed by a computer, the processes or functions of the present disclosure are performed in whole or in part.

[0198] The computer program or instructions can be stored in a readable storage medium or transmitted from one readable storage medium to another readable storage medium, for example, the computer program or instructions can be transmitted from one website site, computer, server, or data center to another website site, computer, server, or data center through wired or wireless means. The readable storage medium can be any available medium that can be accessed or a data storage device such as a server, data center, etc. that integrates one or more available media. The available media can be a magnetic medium, such as a floppy disk, a hard disk, a magnetic tape; an optical medium, such as a digital video disc; and a semiconductor medium, such as a solid state disk. The computer readable storage medium can be a volatile or non-volatile storage medium, or can include both volatile and non-volatile storage media.

[0199] Those skilled in the art will appreciate that embodiments of the present disclosure can be provided as methods, systems, or computer program products. Therefore, the present disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present disclosure can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0200] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, apparatuses, electronic devices, and computer program products according to the present disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the computer or other programmable data processing apparatus produce a device that implements the processes described in the flowcharts and / or block diagrams. Figure 1 The apparatus for performing the functions specified in one or more flows and / or blocks Figure 1 The apparatus for performing the functions specified in one or more flows and / or blocks

[0201] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 The flow or flows and / or blocks Figure 1 The flow or flows and / or blocks

[0202] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 The flow or flows and / or blocks Figure 1 The flow or flows and / or blocks

[0203] In the description of the specification, the description of the terms "one embodiment / way", "some embodiments / ways", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, or characteristics described in connection with the embodiment / way or example are included in at least one embodiment / way or example of the present disclosure. In the specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / way or example. Also, the specific features, structures, or characteristics described can be combined in any appropriate manner in one or more embodiments / ways or examples. In addition, the person skilled in the art can combine and combine the different embodiments / ways or examples described in the specification and the features of the different embodiments / ways or examples, without contradiction.

[0204] The person skilled in the art should understand that the above-mentioned embodiments are only for clearly illustrating the present disclosure, and are not intended to limit the scope of the present disclosure. Other changes or modifications can be made on the basis of the above disclosure, and these changes or modifications are still within the scope of the present disclosure.

Claims

1. A method for handling traffic congestion, characterized in that, include: Acquire video images captured on the target road; Based on the video images, the identification results of congestion factors related to road congestion are determined, including road factors or human behavior factors; Based on the congestion factors identified in the identification results, a target response strategy is determined. The target response strategy includes measures to eliminate congestion factors on the target road. A default response strategy is determined based on the congestion factors identified in the identification results. The default response strategy is then optimized based on traffic scene information of the target road to determine the target response strategy. The traffic scene information includes time period information, weather information, holiday information, or traffic flow information. Implement the target response strategy.

2. The method as described in claim 1, characterized in that, Based on the traffic scenario information of the target road, the default response strategy is optimized, including: When the congestion factor is parked vehicles occupying the road, the timing and scope of the response in the default response strategy are optimized based on the weather information, time period information, and duration of vehicle occupation on the target road. When the congestion factor is road damage, the road maintenance priority and maintenance time window in the default response strategy are optimized based on the weather information, time period information and road damage location of the target road. When the congestion factor is the presence of foreign objects on the road, the cleaning timing and method in the default response strategy are optimized based on the weather information, time period information, and garbage type of the target road.

3. The method as described in claim 2, characterized in that, Also includes: When the congestion factor is road damage, the temporary warning measures in the default response strategy are optimized based on the weather information and speed limit information of the target road. When the congestion factor is the presence of road debris, the temporary traffic control measures in the default response strategy are optimized based on the location information of road debris and traffic flow information on the target road.

4. The method as described in claim 1, characterized in that, Based on the congestion factors identified in the results, a default response strategy is determined, including: Based on the congestion factors identified in the results, a matching process is performed in a strategy library, which includes default response strategies corresponding to different congestion factors. Based on the matching results, determine the default response strategy corresponding to the identification results.

5. The method as described in claim 1, characterized in that, Based on the video images, the identification results of congestion factors related to road congestion are determined, including: Based on the video images, the identification results of congestion factors are determined, including the type of congestion factor, confidence level, and location information of the congestion factor; The identification results of congestion factors with confidence levels below the confidence threshold are removed.

6. The method as described in claim 5, characterized in that, The identification results of congestion factors based on the video images include: The video images are preprocessed to determine key frames; Target recognition is performed based on the image keyframes to determine the target objects present in the image keyframes; The overlap between the target object detected in the keyframe of the image and the sensitive detection area is calculated to determine the overlap area ratio. Based on the timing information of the video image, determine the dwell time of the target object detected in the key frame of the image; The identification results of congestion factors are determined based on the target object, the overlapping area ratio, and / or the dwell time.

7. The method according to any one of claims 1-6, characterized in that, Implementing the target response strategy includes: Based on the target response strategy, generate coordinated response instructions; The coordinated action command is sent to the target object.

8. An electronic device, characterized in that, include: The memory stores execution instructions; as well as A processor that executes execution instructions stored in the memory, causing the processor to perform the method of any one of claims 1 to 7.

9. A readable storage medium, characterized in that, The readable storage medium stores execution instructions, which, when executed by a processor, are used to implement the method of any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images