A smart city traffic road cleaning management method and an internet of things system

Abstract

The embodiment of the present specification provides a smart city traffic road cleaning management method and an Internet of Things system. The method is executed by a traffic management platform, and includes: acquiring a road video of a time period captured by a camera on a road; extracting a target image from the road video, and processing the target image through a prediction model to predict road cleanliness; processing the road video to predict traffic corresponding to the road; and controlling a road sweeper to clean the road based on the predicted traffic and road cleanliness. The smart city traffic road cleaning management method according to some embodiments of the present specification can realize real-time monitoring of road health cleaning, can timely process cleaning conditions that may affect traffic safety and cause traffic congestion, and can allocate road sweepers on demand, thereby saving labor costs and social resources.

Description

[0001] Cross-referencing

[0002] This application claims priority to U.S. Application No. 17664876, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This manual relates to the field of cleaning management, and in particular to a smart city traffic road cleaning management method and Internet of Things system. Background Technology

[0004] With the rapid development of society, the flow of people and vehicles on urban roads is increasing. When people and vehicles pass by, the dust they raise and the debris they leave behind affect the cleanliness of the roads, making road cleaning an issue that cannot be ignored in urban development.

[0005] Therefore, it is desirable to provide a technical solution that can better identify the cleanliness of roads and control road sweepers to perform cleaning. Summary of the Invention

[0006] One embodiment of this specification provides a smart city traffic road cleaning management method. The smart city traffic road cleaning management method includes: acquiring road video footage captured by cameras on the road over a specified time period; extracting target images from the road video footage and processing the target images using a prediction model to predict road cleanliness; processing the road video footage over the specified time period to predict traffic flow corresponding to the road; and controlling a road sweeper to clean the road based on the traffic flow and the road cleanliness.

[0007] One embodiment of this specification provides an Internet of Things (IoT) system for smart city traffic road cleaning management. The system includes a traffic management platform configured to perform the following operations: acquiring road video footage captured by cameras on the road over a specified time period; extracting target images from the road video footage and processing the target images using a prediction model to predict road cleanliness; processing the road video footage over the specified time period to predict traffic flow corresponding to the road; and controlling a road sweeper to clean the road based on the traffic flow and the road cleanliness.

[0008] One embodiment of this specification provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions from the storage medium, the computer executes the smart city traffic road cleaning management method.

[0009] Beneficial effects:

[0010] The intelligent city traffic road cleaning management method described in some embodiments of the specification can realize real-time monitoring of road health cleaning, can timely process cleaning conditions that may affect traffic safety and cause traffic congestion, and can allocate road cleaning vehicles on demand to save manpower costs and social resources. BRIEF DESCRIPTION OF DRAWINGS

[0011] The specification will be further illustrated in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same numbers represent the same structures, wherein:

[0012] Figure 1 is an application scenario diagram of the intelligent city traffic road cleaning management method according to some embodiments of the specification;

[0013] Figure 2 is a system diagram of the Internet of Things system for intelligent city traffic road cleaning management according to some embodiments of the specification;

[0014] Figure 3 is an exemplary flowchart of the intelligent city traffic road cleaning management method according to some embodiments of the specification;

[0015] Figure 4 is an exemplary diagram for predicting road cleanliness according to some embodiments of the specification;

[0016] Figure 5 is an exemplary flowchart for predicting the traffic corresponding to the road according to some embodiments of the specification. DETAILED DESCRIPTION

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the specification, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some examples or embodiments of the specification, and for those skilled in the art, the specification can be applied to other similar scenarios without creative labor on the basis of these drawings. Unless it is obvious from the language environment or otherwise stated, the same numbers in the drawings represent the same structures or operations.

[0018] It should be understood that the "system", "device", "unit" and / or "module" used herein is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0019] As shown in the specification and claims herein, unless the context clearly indicates otherwise, the words "comprise", "comprising", "consist of" and "consisting of" do not preclude the inclusion of additional steps or elements. In general, the terms "comprise" and "comprising" are used in the sense of "including" rather than "consisting of".

[0020] Flowcharts are used in the specification to illustrate the operations performed by systems in accordance with embodiments of the specification. It will be understood that the operations shown in the flowcharts are not necessarily performed in the order shown. Rather, various steps can be performed in different orders or simultaneously. Additionally, other operations can be added or removed from the flowcharts.

[0021] Embodiments of the specification relate to a smart city traffic road cleaning management method, system and storage medium. The smart city traffic road cleaning management method and the Internet of Things system can be applied to smart terminals such as mobile phones, tablet computers, notebook computers, etc. The application field can be municipal environmental sanitation cleaning, indoor cleaning, etc. In some embodiments, the smart city traffic road cleaning management method and the Internet of Things system can be applied to smart city traffic road cleaning management monitoring terminals, such as road cameras, vehicle flow monitoring systems, etc. In some embodiments, the smart city traffic road cleaning management method and the Internet of Things system can be applied to management / user terminals, such as road cleaning monitoring platforms, vehicle flow monitoring platforms, etc. In some embodiments, the smart city traffic road cleaning management method and the Internet of Things system can be applied to cleaning terminals, such as road cleaning vehicles, cleaning robots, etc. In some embodiments, the smart city traffic road cleaning management method, system and storage medium can be applied to other fields, such as vehicle flow monitoring, smart city, etc.

[0022] Figure 1 is an application scenario diagram of the smart city traffic road cleaning management method according to some embodiments of the specification.

[0023] In some embodiments, the application scenario 100 of the Internet of Things system for smart city traffic road cleaning management can include a server 110, a processing device 120, a cleaning device 130, a first terminal 140, a network 150, and a second terminal 160.

[0024] In some embodiments, the server 110 can be configured to process information and / or data related to the application scenario 100. For example, the server 110 can be configured to extract target images from road videos and predict road cleanliness by processing the target images. In some embodiments, the server 110 can be a single server or a group of servers. The group of servers can be centralized or distributed (e.g., the server 110 can be a distributed system). In some embodiments, the server 110 can be local or remote. For example, the server 110 can access information and / or data stored in the cleaning device 130, the first terminal 140, and the second terminal 160 via the network 150. For another example, the server 110 can be directly connected to the cleaning device 130, the first terminal 140, and / or the second terminal 160 to access the stored information and / or data.

[0025] In some embodiments, the server 110 can include a processing device 120. The processing device 120 can process information and / or data related to the application scenario 100 to perform one or more functions described in this application. For example, the processing device 120 can extract target images from road videos and predict road cleanliness by processing the target images using a prediction model. In some embodiments, the processing device 120 can include one or more processing engines (e.g., a single-chip processing engine or a multi-chip processing engine). For example only, the processing device 120 can include a central processing unit (CPU).

[0026] The cleaning device 130 can be configured to perform cleaning actions in response to data and / or instructions issued by the server 110 or the processing device 120. In some embodiments, the cleaning device 130 can be a road cleaning vehicle, a cleaning robot, a sweeping robot, or the like. In some embodiments, the cleaning device 130 can autonomously perform cleaning actions. In some embodiments, the cleaning device 130 can also perform other actions such as watering, wiping, spraying cleaning agents, or the like.

[0027] The first terminal 140 can be a device or other entity directly related to intelligent city traffic road cleaning management. In some embodiments, the first terminal 140 can be a terminal used by a manager of intelligent city traffic road cleaning management, such as a terminal used by a sanitation department worker. In some embodiments, the first terminal 140 can include a mobile device 140-1, a tablet computer 140-2, a notebook computer 140-3, a laptop computer 140-4, or the like, or any combination thereof. In some embodiments, the mobile terminal 140-1 can include a smart phone, a smart pager device, or the like, or other smart devices. In some embodiments, the first terminal 140 can include other smart terminals, such as wearable smart terminals, or the like. The first terminal 140 can be a smart terminal or an entity containing a smart terminal, such as a management device containing a smart computer, or the like.

[0028] The network 150 can facilitate exchange of information and / or data. In some embodiments, one or more components of the application scenario 100 (e.g., the server 110, the first terminal 140, the second terminal 160) can send information and / or data to other components of the application scenario 100 via the network 150. For example, the server 110 can obtain user instructions from the first terminal 140 via the network 150. In some embodiments, the network 150 can be a wired network or a wireless network, or any combination thereof. By way of example only, the network 150 can include a cable network. In some embodiments, the application scenario 100 can include one or more network access points. For example, base stations and / or wireless access points 150-1, 150-2,..., one or more components of the application scenario 100 can connect to the network 150 to exchange data and / or information.

[0029] The second terminal 160 is a terminal for obtaining road information, such as a camera, a microphone, a scanner, etc. In some embodiments, the second terminal 160 can be an executor of road video obtaining. In some embodiments, the second terminal 160 can obtain road videos, and other information of the road. In some embodiments, the second terminal 160 can include an image processing device, such as a device for extracting target images from road videos. In some embodiments, the functions of the first terminal 140 and the second terminal 160 can be implemented on the same device or multiple devices.

[0030] It should be noted that the application scenario 100 is provided only for illustrative purposes, and is not intended to limit the scope of the present application. Various modifications or changes can be made by those of ordinary skill in the art based on the description of the present specification. For example, the application scenario 100 can also include a storage device. For another example, the application scenario 100 can implement similar or different functions on other devices. However, these changes and modifications will not depart from the scope of the present application.

[0031] The Internet of Things system is an information processing system including part or all of the management platform, the sensing network platform, and the object platform. The management platform can realize the connection and cooperation among the various functional platforms (such as the sensing network platform and the object platform), and the management platform gathers the information of the Internet of Things running system, and can provide sensing management and control management functions for the Internet of Things running system. The sensing network platform can realize the connection between the management platform and the object platform, and plays a sensing information sensing communication and control information sensing communication function. The object platform is a functional platform for generating and executing sensing information and control information.

[0032] The processing of information in the Internet of Things system can be divided into the processing flow of sensing information and the processing flow of control information. The control information can be information generated based on the sensing information. Among them, the processing of sensing information is to obtain sensing information by the object platform and deliver it to the management platform through the sensing network platform. The control information is issued by the management platform to the object platform through the sensing network platform, and then the control of the corresponding object is realized.

[0033] In some embodiments, when the Internet of Things system is applied to urban management, it can be called a smart city Internet of Things system.

[0034] Figure 2 Figure 1 is a system diagram of an Internet of Things system for smart city traffic road cleaning management according to some embodiments of the present specification. In some embodiments, the system 200 can include a traffic management platform 210, a traffic sensing network platform 220, and an object platform 230. In some embodiments, the system 200 can be part of or implemented by the processing device 120.

[0035] In some embodiments, the system 200 can be applied to various scenarios of environmental cleaning. In some embodiments, the system 200 can obtain cleaning degree data and flow data in various scenarios respectively to obtain cleaning management strategies in each scenario. In some embodiments, the system 200 can obtain cleaning management strategies for the entire region (such as the entire city) based on the obtained cleaning degree data and flow data in each scenario.

[0036] The various scenarios of cleaning management can include, for example, road cleaning scenarios, indoor cleaning scenarios, etc. For example, cleaning equipment management, cleaning worker management, road cleaning degree evaluation, etc. It should be noted that the above scenarios are only examples and do not limit the specific application scenarios of the system 200. Those skilled in the art can apply the system 200 to any other suitable scenario based on the content disclosed in the present embodiment.

[0037] In some embodiments, the system 200 can be applied to cleaning equipment management. When applied to cleaning equipment management, the object platform 230 can be used to collect data related to cleaning equipment. For example, cleaning equipment usage years, cleaning equipment cleaning capacity, cleaning equipment model, cleaning equipment failure rate, etc. The object platform 230 can upload the collected data related to cleaning equipment to the traffic sensing network platform 220, and the traffic sensing network platform 220 can aggregate the collected data. For example, the traffic sensing network platform 220 can divide the collected data by road, classify the collected data by cleaning equipment cleaning capacity, etc. The traffic sensing network platform 220 further aggregates the data and sends the aggregated data to the traffic management platform 210, and the traffic management platform 210 makes strategies or instructions related to cleaning equipment management based on the processing of the collected data. For example, replacing cleaning equipment, repairing cleaning equipment, etc.

[0038] In some embodiments, the system 200 can be applied to cleaning worker management. When applied to cleaning worker management, the object platform 230 can be used to collect data related to cleaning worker management. For example, the number of cleaning workers, cleaning worker leave, cleaning worker labor capacity, etc. The object platform 230 can upload the collected data related to cleaning worker management to the traffic sensing network platform 220, and the traffic sensing network platform 220 can aggregate the collected data. For example, the traffic sensing network platform 220 can classify the collected data by each cleaning worker, sort the collected data by cleaning worker labor capacity, etc. The traffic sensing network platform 220 further aggregates the data and sends the aggregated data to the traffic management platform 210, and the traffic management platform 210 makes strategies or instructions related to cleaning worker management based on the processing of the collected data. For example, cleaning worker dispatch, cleaning worker recruitment plan, cleaning worker training plan, etc.

[0039] In some embodiments, the system 200 can be applied to road cleaning degree evaluation. When applied to road cleaning degree evaluation, the object platform 230 can be used to collect data related to road cleaning degree, such as whether there are foreign matters on the road, whether the road is flat, road cleaning level, etc. The object platform 230 can upload the collected data related to road cleaning degree to the traffic sensing network platform 220, and the traffic sensing network platform 220 can aggregate the collected data. For example, the traffic sensing network platform 220 can classify the collected data by road name, classify the collected data by area busy degree, etc. The traffic sensing network platform 220 further aggregates the data and sends the aggregated data to the traffic management platform 210, and the traffic management platform 210 makes strategies or instructions related to road cleaning degree evaluation based on the processing of the collected data. For example, scoring road cleaning degree, dispatching cleaning equipment according to road cleaning degree, etc.

[0040] In some embodiments, the system 200 can be composed of multiple intelligent city traffic road cleaning management subsystems, each of which can be applied to a scenario. In some embodiments, the system 200 can comprehensively manage and process the data obtained by each subsystem and the output data, and then obtain relevant strategies or instructions for assisting intelligent city traffic road cleaning management.

[0041] For example, the system 200 can include a subsystem applied to cleaning equipment management, a subsystem applied to cleaning worker management, a subsystem applied to road cleaning degree assessment, and the like. The system 200 is the superior system of each subsystem.

[0042] The following will be described by taking the system 200 managing each subsystem and obtaining corresponding data based on the subsystem to obtain strategies for intelligent city traffic road cleaning management as an example:

[0043] The system 200 can obtain prediction data of cleaning degree, vehicle / person flow, and the like based on the cleaning degree and flow prediction subsystem, obtain road image, road vehicle / person flow, and the like data based on the outdoor environment monitoring subsystem, and obtain road congestion degree data, road cleaning degree data, cleaning equipment scheduling data, and the like based on the cleaning equipment scheduling subsystem.

[0044] When the system 200 performs the above data acquisition, multiple object platforms can be separately set for each subsystem to collect data.

[0045] After the system 200 obtains the above data, the traffic sensing network platform 220 processes the collected data. The traffic sensing network platform 220 further uploads the data after further processing to the traffic management platform 210. The traffic management platform 210 makes prediction data related to intelligent city traffic road cleaning management based on the processing of the collected data.

[0046] For example, the traffic sensing network platform 220 can make strategies or instructions related to cleaning equipment management based on data related to cleaning equipment, such as cleaning equipment service life, cleaning equipment cleaning capacity, cleaning equipment model, cleaning equipment failure rate, and the like, such as replacing cleaning equipment, repairing cleaning equipment, and the like. The traffic sensing network platform 220 can upload the above strategies or instructions to the traffic management platform 210, and the traffic management platform 210 schedules the cleaning equipment based on the above strategies or instructions.

[0047] For another example, the traffic sensing network platform 220 can make a strategy or instruction related to the management of cleaning workers based on data related to the management of cleaning workers and data related to cleaning equipment, such as the number of cleaning workers, the leave of cleaning workers, the labor capacity of cleaning workers, and the service life of cleaning equipment, the cleaning capacity of cleaning equipment, the failure rate of cleaning equipment, etc. The traffic sensing network platform 220 can upload the above strategy or instruction to the traffic management platform 210, and the traffic management platform 210 can dispatch, schedule, etc. cleaning workers based on the above strategy or instruction.

[0048] For another example, the traffic sensing network platform 220 can determine road congestion degree data based on road cleanliness, vehicle / person flow data, etc. The traffic sensing network platform 220 can further determine cleaning equipment scheduling data based on the road congestion degree data. The traffic sensing network platform 220 can upload the above data to the traffic management platform 210, and the traffic management platform 210 can schedule cleaning equipment based on the above data, and the object platform 230 can monitor roads at different levels, for example, adjust the sampling rate of the camera, etc.

[0049] For those skilled in the art, after understanding the principle of the system, the system can be applied to any other suitable scene without departing from the principle.

[0050] The following will take the system 200 applied to the road cleaning degree evaluation scene as an example to specifically describe the system 200.

[0051] The traffic management platform can refer to an Internet of Things platform that coordinates and coordinates the contact and cooperation between various functional platforms, provides perception management and control management.

[0052] In some embodiments, the traffic management platform 210 is configured to perform the following operations: obtaining road video of a time period captured by a camera on the road; extracting a target image from the road video and processing the target image through a prediction model to predict the road cleanliness; processing the road video of the time period to predict the flow corresponding to the road; and controlling the road sweeper to clean the road based on the flow and the road cleanliness. For more details about the road video, the target image and the prediction model, please refer to the specific content of Figure 3 .

[0053] In some embodiments, the target image includes a first target image; and the prediction model includes a first prediction model.

[0054] In some embodiments, the traffic management platform 210 is configured to further perform the following operations: extracting first target images from the road videos; the first target images are images whose clarity and / or number of vehicles in the images meet preset requirements; the first prediction model processes the first target images to predict the road cleanliness. For more details about the first target images and the first prediction model, please refer to the specific content of Figure 4 .

[0055] In some embodiments, the traffic management platform 210 is configured to further perform the following operations: identifying the road videos of the time period, and identifying a plurality of target objects in the road videos; extracting features of the plurality of target objects, and filtering the plurality of target objects through clustering; determining the traffic volume based on the filtering result. For more details about the target objects and the clustering, please refer to the specific content of Figure 5 .

[0056] In some embodiments, the traffic management platform 210 is configured to further perform the following operations: determining a sampling rate corresponding to the camera for the time period; and the camera shoots the road videos based on the sampling rate for the time period.

[0057] In some embodiments, the system 200 includes an object platform 230, which refers to an Internet of Things platform for generating perception information and finally executing control information. In some embodiments, the object platform 230 can include a camera.

[0058] In some embodiments, the system 200 includes a traffic sensing network platform 220. The traffic sensing network platform refers to an Internet of Things platform for realizing sensing communication of perception information and sensing communication of control information, and for connecting the object platform and the traffic management platform. In some embodiments, the traffic sensing network platform 220 can realize mutual communication between the traffic management platform 210 and the object platform 230. For example, the camera is located in the object platform 230, the road videos are obtained from the object platform 230 by the traffic sensing network platform 220, and sent to the traffic management platform 210.

[0059] It should be understood that Figure 2 The system and its platforms shown can be implemented in various ways. For example, in some embodiments, the traffic management platform 210 can be arranged in the server 110 in Figure 1 .

[0060] It should be noted that the above description of the system 200 and its platforms is for the convenience of description, and cannot limit the scope of the embodiments. It can be understood that, for those skilled in the art, after understanding the principle of the system, any combination of the platforms or connection of the platforms with other platforms can be made without departing from the principle. In some embodiments, Figure 2The traffic management platform 210, object platform 230, and traffic sensing network platform 220 disclosed in the present specification can be different platforms in a system, or can be a platform that implements the functions of two or more of the above platforms. For example, the platforms can share an object platform, or each platform can have its own object platform. Variations such as these are within the scope of protection of the present specification.

[0061] Figure 3 is an exemplary flowchart of a smart city traffic road cleaning management method according to some embodiments of the present specification. As shown in Figure 3 , the flow 300 includes the following steps. In some embodiments, the flow 300 can be performed by the traffic management platform 210.

[0062] The traffic management platform can be an Internet of Things platform that coordinates the connection and cooperation between various functional platforms, and provides perception management and control management. The traffic management platform can include Figure 1 processing devices and other components in the present specification. In some embodiments, the traffic management platform can obtain road videos of a road from the traffic sensing network platform, and determine a cleaning plan for a cleaning device such as a road cleaning vehicle based on its processing of the road videos. In some embodiments, the traffic management platform can be a remote platform operated by a manager, artificial intelligence, or by preset rules. For more information about the traffic management platform, see Figure 2 .

[0063] Step 310: Obtain road videos of a time period taken by a camera on the road.

[0064] In some embodiments, the road can be a highway, road, or path on which motor vehicles, non-motor vehicles, people, or other living beings travel. In some embodiments, the individuals and groups traveling on the road will generate traffic, which can be used to determine the number of individuals passing through the road per unit time.

[0065] The camera can be a device for obtaining road videos. For example, the camera can include a camera, a camera, a smartphone, a smart computer, a smart bracelet, a wearable device, etc. In some embodiments, the camera can obtain videos, images of a road in a certain time period. In some embodiments, the camera can further process the road videos, such as extracting target images in the road videos.

[0066] The time period can be the length of time during which the camera takes pictures. For example, 5 minutes, 1.5 hours, etc. In some embodiments, the time period can be determined by the user or by the system preset.

[0067] The road video can be a video related to the road. In some embodiments, the road video can reflect the situation of the road in the shooting time period, such as the cleaning situation, whether there are foreign matters, the traffic / passenger flow situation, the accident / violation situation, etc.

[0068] In some embodiments, the road video can be obtained based on a camera, such as a monitoring camera, or obtained through a network monitoring platform, user uploading, etc.

[0069] In some embodiments, the smart city traffic road cleaning management method can include: determining a sampling rate corresponding to the camera in the time period, and the camera shooting the road video based on the sampling rate in the time period.

[0070] The sampling rate can be the sampling frequency of the camera. The camera shoots the road video based on the sampling rate; for example, the sampling rate can be 30Hz to 10kHz. In some embodiments, the size of the sampling rate can reflect the definition and smoothness of the road video, and the larger the sampling rate, the more clear and smooth the road video.

[0071] In some embodiments, the sampling rate can also be the frame rate or frame frequency of the camera, for example, 60 frames per second, that is, sampling 60 times per second. The larger the sampling rate, the more road videos the camera shoots in a unit of time, the more image materials are provided, and the more accurate the road cleanliness data obtained based on more image materials; but the number of samplings increases, and the operation and storage load of the camera and the processing device also increase, so the sampling rate needs to be reasonably controlled.

[0072] In some embodiments, the sampling rate can vary with other factors. For example, the sampling rate of the camera can be determined by the traffic predicted in the previous time period, such as the larger the traffic predicted in the previous time period, the higher the sampling rate thereafter. When the traffic is large, the road is more likely to be dirty, and a clearer road video is obtained through a high sampling rate to facilitate the cleaning of the road sweeper. Figure 5 See the specific description of traffic prediction and related description.

[0073] In some embodiments, the sampling rate of the camera is determined by the distance between the road sweeper and the camera shooting position, such as the closer the distance, the higher the sampling rate. When the road sweeper is close to the camera shooting position, a clearer road video is obtained through a high sampling rate, which can facilitate the cleaning of the road sweeper to the camera shooting position.

[0074] In some embodiments, the sampling rate can be determined based on the attributes of the road, for example, a higher sampling rate is used when the camera shoots a highway, and a lower sampling rate is used when the camera shoots an ordinary road or a sidewalk. In some embodiments, the sampling rate can also be determined in other ways, such as user settings, etc.

[0075] By the method for determining the sampling rate of the embodiments of the present specification, the sampling rate can be adjusted as needed for the vehicle / person flow and the position of the road sweeper, reducing the wear and tear of the camera caused by high sampling rate and the occupation of a large amount of invalid road video on the memory space. In addition, when the road sweeper is close to the shooting position, it cleans the road near the shooting position, saving the road sweeper's journey and improving the cleaning efficiency.

[0076] In some embodiments, the camera is located on the object platform 230, and the road video is obtained by the traffic sensing network platform 220 from the object platform 230 and sent to the traffic management platform 210.

[0077] The object platform 230 can be a functional platform for generating perception information and executing control information. In some embodiments, the object platform 230 can be a management platform of the second terminal 160. In some embodiments, the object platform can obtain road video, for example, by a camera. In some embodiments, the object platform can also include other terminals in addition to the second terminal 160, such as smart devices, etc. In some embodiments, the object platform can be a remote platform operated by a manager, artificial intelligence, or by a preset rule.

[0078] The traffic sensing network platform can be a platform for sensing communication of perception information and sensing communication of control information, bridging between the object platform and the traffic management platform. In some embodiments, the traffic sensing network platform can be a decision-making platform of the user. In some embodiments, the traffic platform can preprocess the information obtained from the object platform. For example, adjusting the road video of the object platform to a format readable by the traffic management platform, deleting invalid content in the road video of the object platform, changing the clarity, file size of the road video of the object platform, etc.

[0079] In some embodiments, the traffic sensing network platform obtains the original road video preliminarily shot from the object platform, preprocesses the road video, and sends the preprocessed road video to the traffic management platform.

[0080] Step 320, extracting a target image from the road video, and processing the target image through a prediction model to predict the road cleanliness.

[0081] The target image can be an image in the road video that is convenient for the prediction model to predict. In some embodiments, the target image can be an image whose image content or image parameters meet the preset standard. For example, an image with higher clarity, smaller vehicle / person flow, and reflecting the road cleaning condition.

[0082] In some embodiments, the target images can be extracted from the road video by the second terminal or the processing device. The extraction manner can include identifying and extracting images meeting preset standards based on an image recognition algorithm. For example, the sharpness of each frame of image of the road video is identified by a Laplacian operator, and the image with greater sharpness is taken as the target image.

[0083] The prediction model can be a model for predicting the road cleanliness. In some embodiments, the prediction model can be a machine learning model, such as a Convolutional Neural Network (CNN), a Deep Neural Network (DNN), or the like, or a combination thereof. For specific description of the prediction model, see Figure 4 and the related description.

[0084] The road cleanliness can be a measurement for judging the cleanliness of the road. In some embodiments, the road cleanliness can be represented by a numerical value, for example, the greater the numerical value, the higher the cleanliness of the road. In some embodiments, the road cleanliness can be represented by a cleanliness level, such as clean, relatively clean, relatively dirty, and seriously dirty, and the like.

[0085] In some embodiments, the road cleanliness can be determined based on the processing of the target images by the prediction model. For more description of predicting the road cleanliness, see Figure 4 for specific description.

[0086] Step 330, processing the road video of the time period to predict the traffic corresponding to the road.

[0087] The traffic can be the number of vehicles and pedestrians passing a certain position of the road per unit time. The traffic can reflect the congestion and popularity of the road. In some embodiments, the road cleanliness is related to the traffic on the road, for example, the greater the traffic, the worse the cleanliness of the road.

[0088] In some embodiments, processing the road video of the time period includes manual processing, intelligent processing, and the like. The manual processing can include manually identifying the traffic size corresponding to the road in the road video to predict the traffic; the intelligent processing can include identifying the road video based on a target object recognition model, and extracting features, clustering based on the identification result to predict the traffic. For specific description of processing the road video of the time period to predict the traffic, see Figure 5 and the related description.

[0089] In some embodiments, step 320 and step 330 can be performed simultaneously, or step 330 can be performed first and then step 320. The numbering and sequence of the above steps are not intended to limit the order of the above steps.

[0090] Step 340, based on the traffic flow and road cleanliness, controlling the road sweeper to clean the road.

[0091] The road sweeper can be a device for cleaning the road. In some embodiments, the road sweeper can be the cleaning device 130. For specific description of the road sweeper, see Figure 1 the description of the cleaning device 130 in

[0092] In some embodiments, the traffic management platform can control the road sweeper to clean the road with poor road cleanliness and low traffic flow.

[0093] In some embodiments, the traffic management platform can send control instructions to the road sweeper through the network. The road sweeper can execute the corresponding operation based on the instructions, or indirectly realize the operation corresponding to the instructions through the road sweeper driver.

[0094] The intelligent city traffic road cleaning management method described in some embodiments of the present specification can realize real-time monitoring of road health cleaning; the cleaning condition that may affect traffic safety and cause traffic congestion can be handled in time; the road sweeper is allocated on demand, saving manpower cost and social resources.

[0095] It should be noted that the above description of the process 300 is only for example and illustration, and does not limit the scope of application of the present specification. Those skilled in the art can make various modifications and changes to the process 300 under the guidance of the present specification. However, these modifications and changes are still within the scope of the present specification. For example, the process 300 can also include an information storage step.

[0096] Figure 4 is an exemplary schematic diagram of predicting road cleanliness according to some embodiments of the present specification. In some embodiments, this step 320 can be performed by the processing device 120 of the traffic management platform 210.

[0097] Step 321, extracting a target image from the road video.

[0098] In some embodiments, the target image can include a first target image.

[0099] The first target image can be an image whose sharpness and / or number of vehicles in the image meets the preset requirement. For example, the first target image can be an image with a sampling rate greater than 120Hz or greater than 20 frames per second, or an image with less than 5 vehicles.

[0100] In some embodiments, the first target image can be determined by algorithm recognition. For example, the sharpness of the image is identified by Laplacian operator. The target image with a sampling rate greater than 120 Hz is taken as the first target image. In some embodiments, the first target image can be determined automatically by the camera. For example, the camera automatically outputs the image with a sampling rate meeting the preset requirement during acquisition as the first target image.

[0101] In some embodiments, the first target image can be determined by a first target image recognition model. For example, the vehicles in the image are identified by a Yolo model, and the target image with less than 5 vehicles is taken as the first target image. In some embodiments, the input of the Yolo model can be a single frame image of the road video, and the output can be a segmentation result of the single frame image, a recognition result of the single frame image, etc.

[0102] In some embodiments, the target image can include a second target image.

[0103] The second target image can be an image containing complete information of foreign matter. The complete information of foreign matter refers to image information that can reflect the type of foreign matter and the positional relationship between the foreign matter and the road or ground. For example, an image that can reflect the existence of a plastic bottle garbage on the ground, an image that can reflect the existence of a pile of sand in the middle of the road, etc. In some embodiments, the second target image can also be an image containing partial information of foreign matter. For example, an incomplete stone image due to occlusion by pedestrians or vehicles, etc.

[0104] In some embodiments, the images in the road video can be scored based on a scoring model, and the second target image can be determined based on the score. The score represents the completeness of the foreign matter in the image, and the score of the second target image meets the preset requirement.

[0105] The scoring model can be a model for scoring images based on the completeness of foreign matter to determine the second target image. Through the scoring model, each frame of image can be combined. By excluding the moving parts (such as vehicles, pedestrians, etc.) in each frame of image, the foreign matter in the road is determined, and the completeness of the foreign matter is further determined before scoring. In some embodiments, the scoring model can be a sequence-to-sequence model. For example, the scoring model can be a recurrent neural network (RNN), a long short-term memory model (LSTM), an encoder-decoder model, etc.

[0106] The score can be a reference value for evaluating the completeness of the foreign object information in the image. For example, when the image can fully reflect a certain side of the foreign object, such as a side, the scoring model can give a higher score, such as 95 points; when the image has no foreign object, or the foreign object is partially blocked by a vehicle or a pedestrian, the scoring model can give a lower score, such as 15 points. When the image score exceeds the score threshold (for example, the score is greater than 60 points), the image is taken as the second target image. In some embodiments, when the image contains at least two foreign objects, each foreign object is scored separately, and the highest score is compared with the score threshold to determine the second target image.

[0107] In some embodiments, the input of the scoring model can be a plurality of images sorted in time sequence. The output can be whether each frame of the group of images is a second target image, or the specific score of each frame of the group of images. When the specific score of each frame of the group of images is output, the scoring model further includes a score judgment layer. The score judgment layer is used to judge whether the specific score exceeds the score threshold, and the image exceeding the score threshold can be considered as the second target image.

[0108] In some embodiments, the scoring model can be trained by a plurality of labeled training samples. For example, a plurality of labeled training samples can be input into an initial scoring model. A loss function is constructed by the label and the result of the initial scoring model, and the parameters of the initial scoring model are iteratively updated based on the loss function. When the loss function of the initial scoring model meets a preset condition, the model training is completed, and a trained scoring model is obtained. The preset condition can be that the loss function converges, the number of iterations reaches a threshold, etc. The loss function can be an index for measuring the performance of the scoring model. In some embodiments, the loss function of the training sample includes at least one weight coefficient, which is determined based on the sampling rate. For example, n training samples 、 、… corresponding label 、 、… Suppose when the regression loss function is the mean square error loss function, a weight coefficient is added to the loss function of each training sample, and the formula is as follows:

[0109]

[0110] wherein, is the mean square error loss function, is the weight coefficient corresponding to the sample is the label corresponding to the sample is the weight coefficient corresponding to the sample is the label corresponding to the sample is the weight coefficient corresponding to the sample Corresponding supervised data. In some embodiments, the training samples can be samples with higher definition, higher sampling rate, and the ability to feed back key information Assign higher weight coefficients to increase their impact on the loss function.

[0111] In some embodiments, the training samples can be historical scoring images, and the training samples can be obtained by calling historical scoring cases.

[0112] By adding weight coefficients in the loss function, the training proportion of clear images can be increased, and the adverse effects of invalid training content can be reduced. Ultimately, a more accurate training model can be obtained.

[0113] In some embodiments, the label of an image in which a foreign object appears can be set to 1, and the labels of other images in which the same foreign object appears can be determined based on the similarity between the image with the label of 1 and the other images. In some embodiments, the similarity can be determined based on the image features of the image with the label of 1 and the other images by calculating the Euclidean distance between the image features. The image features of the image can be determined by the convolutional neural network of the first prediction model or the second prediction model. For specific descriptions of the determination of image features, see the related descriptions of the first prediction model and the second prediction model. In some embodiments, the similarity can also be determined by manual annotation.

[0114] Through the scoring model in some embodiments of the present specification, the determination of foreign objects in the road can be realized. By adding weight coefficients in the loss function, the training proportion of clear images can be increased, and a more accurate training model can be obtained. In addition, based on the similarity, the samples can be labeled, and each frame of the road video can be determined, the number of samples can be increased, and the training process can be closer to the actual situation.

[0115] Step 323, processing the target image through the prediction model to predict the road cleanliness.

[0116] In some embodiments, the prediction model includes a first prediction model. Processing the target image through the prediction model to predict the road cleanliness includes processing the first target image through the first prediction model to predict the road cleanliness.

[0117] The first prediction model can be a model for predicting the road cleanliness. For example, the first prediction model can be a combined model of a convolutional neural network (CNN) and a deep neural network (DNN).

[0118] In some embodiments, the input of the first prediction model can be a single frame image of the first target image, and the output can be the cleanliness of the road. In some embodiments, the input of the convolutional neural network of the first prediction model can be a single frame image of the first target image, and the output can be image features corresponding to the single frame image; the input of the deep neural network of the first prediction model can be the output of the convolutional neural network, i.e., image features corresponding to the single frame image, and the output can be the cleanliness of the road. Wherein, the image features can be color features, texture features, shape features, spatial relationship features, etc. in the image.

[0119] In some embodiments, when the first prediction model is a combination model of convolutional neural network + deep neural network, the first prediction model can be obtained by joint training. The processing device 120 can train the initial convolutional neural network and the initial deep neural network based on a large number of labeled training samples. Specifically, the labeled training samples are input into the initial convolutional neural network, and the parameters of the initial convolutional neural network and the initial deep neural network are updated by training until the trained convolutional neural network and deep neural network meet the preset condition, to obtain the trained convolutional neural network and deep neural network. Wherein, the preset condition can be that the loss function is less than a threshold, converges, or the training period reaches a threshold.

[0120] In some embodiments, the training sample can be a historical road video, which can be obtained by calling historical monitoring.

[0121] In some embodiments, the label of the first prediction model training can be the cleanliness of the first target image in the historical road video, and the label can be obtained by manual annotation.

[0122] Through the first prediction model of some embodiments of the present specification, intelligent judgment of the cleanliness of the road can be realized; in addition, the combination of road sanitation work and urban road monitoring increases the diversity of the use of road monitoring images.

[0123] In some embodiments, the first target image can be segmented to determine a target region; the target region is processed based on the first prediction model to determine the cleanliness of the road.

[0124] In some embodiments, the input of the first prediction model is a target region of the first target image. The target region can be a region where vehicles travel and / or pedestrians walk in the first target image. For example, the target region can be a part of the road, sidewalk, etc. in the first target image.

[0125] In some embodiments, the target region can be determined by manual division or determined by a machine learning model, for example, determined by a Yolo model.

[0126] In some embodiments, the input of the Yolo model can be the first target image, and the output can be a target region partition result of the first target image. The result includes a plurality of identified regions and a category corresponding to each identified region. For example, an identified region with a category of "road" can be taken as a target region.

[0127] In some embodiments, the prediction model can include a second prediction model. By processing the target image through the prediction model, predicting the road cleanliness further includes: processing the second target image through the second prediction model to determine whether there is foreign matter on the road; and correcting the road cleanliness based on the foreign matter condition of the road.

[0128] The second prediction model can be a model for determining whether there is foreign matter on the road. For example, the second prediction model can be a combined model of a convolutional neural network (CNN) and a deep neural network (DNN).

[0129] The foreign matter can be other objects in the image that are different from the road, vehicles, and pedestrians. For example, the foreign matter can be an empty plastic bottle, a pile of garbage, etc.

[0130] In some embodiments, the input of the second prediction model can be a single frame image, and the output can be whether there is foreign matter on the road. In some embodiments, the input of the convolutional neural network of the second prediction model can be a single frame image, and the output can be image features corresponding to the single frame image; the input of the deep neural network of the second prediction model can be the output of the convolutional neural network, i.e., the image features corresponding to the single frame image, and the output can be whether there is foreign matter on the road. In some embodiments, the convolutional neural network of the second prediction model can be obtained by migrating the convolutional neural network of the first prediction model. For the description of the image features, refer to the description of the first prediction model.

[0131] In some embodiments, the trained first prediction model or the convolutional neural network of the trained second prediction model can also be used in the aforementioned scoring model. Specifically, it can be used in the calculation of the similarity between images, for extracting image features of the input image. For example, a single frame image can be input to the convolutional neural network of the trained first prediction model or the second prediction model, and the image features corresponding to the image can be output. Based on the image features, the Euclidean distance between the image features is calculated to obtain the similarity.

[0132] In some embodiments, when the second prediction model is a combined model of a convolutional neural network and a deep neural network, the second prediction model can be obtained by joint training. The training module can train the initial convolutional neural network and the initial deep neural network based on a large number of labeled training samples. Specifically, the labeled training samples are input into the initial convolutional neural network, and the parameters of the initial convolutional neural network and the initial deep neural network are updated by training until the trained convolutional neural network and the deep neural network meet a preset condition, to obtain the trained convolutional neural network and the deep neural network. The preset condition can be that a loss function is less than a threshold, convergence, or a training period reaches a threshold.

[0133] In some embodiments, the training sample can be a historical road video, which can be obtained by calling a historical monitoring.

[0134] In some embodiments, the label for training the second prediction model can be whether a single frame image of the historical road video has a foreign object, and the label can be obtained by manual labeling.

[0135] In some embodiments, the road cleanliness is corrected based on the foreign object condition of the road, to obtain a corrected road cleanliness. In some embodiments, the correction can be that when the second prediction model outputs the presence of a foreign object, the road cleanliness output by the first prediction model is reduced in score and / or grade. For example, when the first prediction model judges that the road cleanliness is high, such as 90 points, if the second prediction model judges that there is a foreign object on the road, the corrected road cleanliness is 20 points. In some embodiments, the correction rule can also be that the road cleanliness output by the first prediction model is corrected to half, 0 points, a more dirty grade or a serious dirty grade, etc.

[0136] In some embodiments, the correction can also be that when the second prediction model outputs the absence of a foreign object, the road cleanliness output by the first prediction model is scored and / or upgraded. For example, when the first prediction model judges that the road cleanliness is 60 points, if the second prediction model judges that there is no foreign object on the road, the corrected road cleanliness is 90 points. In some embodiments, the correction rule can also be that the road cleanliness output by the first prediction model is corrected to 100 points, a clean grade or a more clean grade, etc.

[0137] Through the correction process described in some embodiments of the present specification, the model can be prevented from ignoring a certain foreign object to cause a misjudgment of the road cleanliness; large foreign objects such as a brick and an oil drum are prevented from being ignored by the model to cause an impact on the road cleanliness and even the safety of driving and pedestrians, so that the judgment of the cleanliness is more in line with the actual situation.

[0138] Figure 5is an exemplary flowchart of predicting traffic corresponding to a road according to some embodiments of the present specification. In some embodiments, the step 330 can be performed by the processing device 120 of the traffic management platform 210.

[0139] At step 510, a road video of a time period is identified, and a plurality of target objects in the road video are identified.

[0140] The target object can be a biological or non-biological object related to the traffic to be counted, such as a motor vehicle, a non-motor vehicle, a pedestrian, etc.

[0141] In some embodiments, the identification process can be implemented by a target object identification model, for example, a Yolo model to identify target objects in the road video. The input of the target object identification model can be a single frame image of the road video, and the output can include a bounding box of the target object.

[0142] At step 520, features of the plurality of target objects are extracted, and the plurality of target objects are filtered by clustering.

[0143] The features of the target object can be image features related to the target object, such as a vehicle license plate, a vehicle tire, a pedestrian head, etc. In some embodiments, the features of the target object further include movement features of the target object, such as vehicle driving, pedestrian walking, etc.

[0144] In some embodiments, the features of the target object are extracted by a feature extraction model. The feature extraction model can be a convolutional neural network model, a feature detection algorithm (Histogram of Oriented Gridients, HOG), etc.

[0145] In some embodiments, the input of the feature extraction model can be a bounding box of the target object, and the output can be a feature of the target object.

[0146] In some embodiments, by a clustering method, target objects with the same feature are clustered in the same cluster center to achieve filtering of the target objects. The cluster center can be a set of target objects with the same or similar features. The clustering method can include K-means clustering, mean shift clustering, etc. For example, in a road video, a target object in the 10th frame image and a target object in the 20th frame image are clustered in the same cluster center. This indicates that the two target objects are the same person or the same vehicle. Different target objects are clustered in different cluster centers, and the same target objects are clustered in the same cluster center to achieve filtering of the target objects. Prevent the same object from being counted repeatedly when determining the traffic.

[0147] At step 530, the traffic is determined based on the filtered results.

[0148] In some embodiments, the result of the filtering can include at least one cluster center. In some embodiments, by counting the number of cluster centers in a time period, the traffic condition in the time period is determined. For example, when the number of cluster centers is 2000, it can be approximately considered that the number of people / vehicles passing through the road in the time period is 2000.

[0149] By the traffic determination method of some embodiments of the present specification, repeated counting of the same pedestrian / vehicle can be avoided based on clustering. Moreover, the counting process is based on a model, reducing labor cost and improving counting efficiency.

[0150] Some embodiments of the present specification also disclose a computer readable storage medium storing computer instructions, when the computer reads the computer instructions in the storage medium, the computer executes the smart city traffic road cleaning management method described above.

[0151] The above has described the basic concept, and it is obvious that the above detailed disclosure is only used as an example for the person skilled in the art, and does not constitute a limitation on the present specification. Although it is not explicitly stated here, the person skilled in the art can make various modifications, improvements and corrections to the present specification. Such modifications, improvements and corrections are suggested in the present specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of the present specification.

[0152] Meanwhile, specific words are used in the present specification to describe the embodiments of the present specification. As "one embodiment", "an embodiment", and / or "some embodiments" means a certain feature, structure or characteristic related to at least one embodiment of the present specification. Therefore, it should be emphasized and noted that the "an embodiment" or "one embodiment" or "one alternative embodiment" mentioned in different places in the present specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present specification can be properly combined.

[0153] In addition, unless the claim explicitly states, the order of the processing elements and sequences described in the present specification, the use of numerals and letters, or the use of other names, is not intended to limit the order of the processes and methods of the present specification. Although some currently considered useful embodiments of the invention are discussed in the above disclosure through various examples, it should be understood that such details are only for the purpose of illustration, and the additional claims are not limited to the disclosed embodiments, on the contrary, the claims are intended to cover all modifications and equivalent combinations that meet the spirit and scope of the embodiments of the present specification. For example, although the system components described above can be realized by hardware devices, they can also be realized by only software solutions, such as installing the described system on existing servers or mobile devices.

[0154] For simplicity and to facilitate understanding of one or more embodiments, a description of an embodiment sometimes refers to a plurality of features in a single embodiment, drawing, or description of an embodiment. However, this method of disclosure is not to be interpreted as meaning that the claimed embodiment requires more features than are explicitly recited in the claims. In fact, claims that do not specifically claim a combination of features are intended to cover the various possible combinations of features as would be understood by a person of ordinary skill in the art.

[0155] Some embodiments use numerical values to describe components, quantities of attributes. It should be understood that such numerical values used in the description of embodiments are, in some examples, modified by the adjectives "about," "approximately," or "substantially." Unless otherwise stated, "about," "approximately," or "substantially" indicate that the described numerical value allows for a variation of ±20%. Accordingly, numerical values used in the specification and claims of some embodiments are approximations that can vary depending on the desired characteristics of the individual embodiments. In some embodiments, numerical values used in the specification and claims are approximations that can vary depending on the desired characteristics of the individual embodiments. In some embodiments, numerical values should be considered in the context of the number of significant digits used in the number and the accepted bits of precision of the number. Although the numerical ranges and parameters setting forth the broadest scope of some embodiments of the specification are approximations, the numerical values set forth in the specific examples are reported as precisely as reasonably possible. The application is not limited to the specific numerical values set forth in the examples.

[0156] Each patent, patent application, patent publication, and other material cited in this specification is hereby incorporated by reference in its entirety. In the event of inconsistencies between the disclosure of this specification and the materials incorporated by reference, the disclosure of this specification shall prevail. In the event of inconsistencies between the disclosure of this specification and the claims, the claims shall prevail. In the event of inconsistencies between the disclosure of this specification and the materials incorporated by reference (whether attached hereto or subsequently added), the disclosure of this specification shall prevail. In the event of inconsistencies between the disclosure of this specification and the description, definitions, and / or terminology used in the materials incorporated by reference, the description, definitions, and / or terminology used in this specification shall prevail.

[0157] Finally, it should be understood that the embodiments described herein are merely exemplary of the principles of the embodiments described herein. Other variations having essentially the same structure and function but different values for components, and / or different arrangements of the components can also be utilized. Accordingly, the embodiments described herein are not to be considered as limited to the embodiments specifically set forth and any embodiments that are functionally equivalent are within the scope of the embodiments described herein. Accordingly, the embodiments described herein are not to be considered as being limited to the particular embodiments described herein, but rather only as being limited by the scope of the appended claims.

Claims

1. A smart city traffic road cleaning management method, the method being performed by a traffic management platform, comprising: obtaining road videos of a time period captured by a camera on a road; extracting target images from the road videos, and processing the target images by a prediction model to predict road cleanliness, the target images including first target images and second target images, the first target images being images whose definition and / or number of vehicles in the images meet preset requirements, and the second target images being images containing complete information of foreign matters or images containing partial information of foreign matters; the prediction model including a first prediction model and a second prediction model, both of which being combined models of convolutional neural networks and deep neural networks; wherein comprising: extracting the first target images from the road videos; segmenting the first target images to determine target regions, the target regions being regions in which vehicles travel and / or pedestrians walk in the first target images; processing the target regions by the first prediction model to determine road cleanliness; processing the second target images by the second prediction model to determine road foreign matter conditions; correcting the road cleanliness based on the road foreign matter conditions; processing the road videos of the time period to predict traffic corresponding to the road; and controlling a road sweeper to clean the road based on the traffic and the road cleanliness. 2.The method of claim 1, wherein processing the road videos of the time period to predict traffic corresponding to the road comprises: identifying the road videos of the time period to identify a plurality of target objects in the road videos; extracting features of the plurality of target objects, and filtering the plurality of target objects by clustering; determining traffic based on a result of the filtering. 3.The method of claim 1, wherein the camera is located on an object platform, the road videos are obtained by a traffic sensing network platform from the object platform, and sent to the traffic management platform. 4.An Internet of Things system for smart city traffic road cleaning management, the system comprising a traffic management platform configured to perform the following operations: obtaining road videos of a time period captured by a camera on a road; extracting target images from the road videos, and processing the target images by a prediction model to predict road cleanliness, the target images including first target images and second target images, the first target images being images whose definition and / or number of vehicles in the images meet preset requirements, and the second target images being images containing complete information of foreign matters or images containing partial information of foreign matters; The prediction model comprises a first prediction model and a second prediction model, and the first prediction model and the second prediction model are both combined models of a convolutional neural network and a deep neural network. wherein comprising: extracting the first target images from the road videos; segmenting the first target images to determine target regions, the target regions being regions in which vehicles travel and / or pedestrians walk in the first target images; processing the target regions by the first prediction model to determine road cleanliness; processing the second target image through the second prediction model to determine a road foreign matter condition; correcting the road cleanliness based on the road foreign matter condition; processing the road video of the time period to predict a traffic volume corresponding to the road; and controlling a road sweeper to clean the road based on the traffic volume and the road cleanliness. 5.The system of claim 4, wherein the traffic management platform is configured to further perform the following operations: identifying the road video of the time period to identify a plurality of target objects in the road video; extracting features of the plurality of target objects, and filtering the plurality of target objects through clustering; determining a traffic volume based on a result of the filtering. 6.The system of claim 4, further comprising an object platform and a traffic sensing network platform. The camera device is located on the object platform, and the road video is obtained from the object platform by the traffic sensing network platform and sent to the traffic management platform.

7. A smart city traffic road cleaning management device, characterized in that, The device comprises at least one processor and at least one memory. The at least one memory is configured to store computer instructions. The at least one processor is configured to execute at least part of the computer instructions to implement the smart city traffic road cleaning management method of any one of claims 1-3. 8.A computer readable storage medium, the storage medium storing computer instructions, when a computer reads the computer instructions in the storage medium, the computer executes the smart city traffic road cleaning management method of any one of claims 1-3.

Images