Weather Information Sharing Method, Cloud Server and System

Abstract

This application relates to a weather information sharing method, a cloud server, and a system. According to the package data uploaded by the vehicle terminal, the image in the package data is parsed to obtain an image feature map, and the environmental information is standardized to obtain a feature value. The target feature map is obtained by adding the feature value to the image feature map. After calculating the optimal probability of the target feature map, the matching weather category is obtained. By combining the location information, the statistical probability of all corresponding matching weather categories is calculated and the target weather category is output. By combining the location information, the average value of all corresponding environmental information is calculated to obtain the target environmental information. The two are integrated to obtain the target weather information, and finally, a request instruction is used to share the target weather information to the vehicle terminal. Since the package data is uploaded by the vehicle terminal in real time according to the current location information, the accuracy of the parsed target weather information is higher, and because the package data is uploaded in real time, the problem of data latency is not likely to occur.

Description

Technical Field

[0001] This application relates to the technical field of information sharing, and in particular, to a weather information sharing method, a cloud server, and a system. Background Art

[0002] Weather information is information data that needs to be obtained before traveling. Users can judge the weather conditions at the departure point, passing points, and destination point of the trip based on the received weather information, so as to decide whether to continue driving.

[0003] Currently, in the related art, there are mainly two ways to obtain weather information:

[0004] The first is the most traditional way. The weather information in the current location area is recognized by the user's eyes, and the weather conditions are judged based on life experience. This is the most direct way, but because the weather conditions change erratically, relying on the way of human eye recognition often cannot accurately judge the weather category, which will inevitably affect the user's driving travel plan;

[0005] The second is that the user uses the network to obtain weather information provided by a weather information service provider. However, there is a certain delay in the weather information published by the weather information service provider, which will also cause the weather information obtained by the user to be inaccurate, thus affecting the user's driving travel plan. Summary of the Invention

[0006] To solve or partially solve the problems existing in the related art, this application provides a weather information sharing method, a cloud server, and a system, which can provide accurate weather information.

[0007] The first aspect of this application provides a weather information sharing method, including:

[0008] Receiving package data uploaded from at least one vehicle terminal, where the package data includes pictures, environmental information, and location information;

[0009] Analyzing the picture to obtain a picture feature map, standardizing the environmental information to obtain a feature value, adding the feature value to the picture feature map to obtain a target feature map, and performing a preferred probability calculation on the target feature map to obtain a matching weather category that matches the target feature map;

[0010] Performing a statistical probability calculation on all the matching weather categories corresponding to the location information to obtain a target weather category corresponding to the location information;

[0011] Performing a mean calculation on all the environmental information corresponding to the location information to obtain a target environmental information;

[0012] Integrate the target weather category and the target environmental information into target weather information;

[0013] Receive and respond to a request instruction from at least one of the vehicle terminals, and share the target weather information to at least one of the vehicle terminals according to the relevance between the custom location information of the vehicle terminal and the location information.

[0014] Preferably, parsing the picture to obtain a picture feature map, normalizing the environmental information to obtain a feature value, adding the feature value to the picture feature map to obtain a target feature map, and performing a preference probability calculation on the target feature map to obtain a matching weather category that matches the target feature map includes:

[0015] Input into a deep learning model;

[0016] Perform multi-scale convolution calculations on the picture to obtain multiple convolution feature maps;

[0017] Perform splicing or fusion calculations on multiple convolution feature maps to obtain the picture feature map;

[0018] Use a normalization function to normalize the environmental information to obtain the feature value;

[0019] Add the feature value to the picture feature map to obtain the target feature map;

[0020] Parse several to-be-determined weather categories of the target feature map, calculate the extraction probability of each to-be-determined weather category using a category probability function, and take the to-be-determined weather category corresponding to the maximum extraction probability as the matching weather category.

[0021] Preferably, performing splicing calculations on multiple convolution feature maps to obtain a picture feature map, where the number of rows of the picture feature map is equal to the sum of the number of rows of each convolution feature map, or the number of columns of the picture feature map is equal to the sum of the number of columns of each convolution feature map.

[0022] Preferably, the fusion calculation of multiple convolution feature maps to obtain the picture feature map includes:

[0023] Obtain the weight coefficient of each scale convolution calculation;

[0024] Calculate a corrected convolution feature map according to the weight coefficient corresponding to the convolution feature map;

[0025] Perform fusion calculations on multiple corrected convolution feature maps to obtain the target feature map.

[0026] Preferably, perform at least three-scale convolutions on the environmental information.

[0027] Preferably, the statistical probability calculation of all the matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information includes:

[0028] Count the total number of all the matching weather categories corresponding to the location information;

[0029] Count the total number of the matching weather categories of different categories corresponding to the location information;

[0030] Calculate the screening probability of the matching weather categories of different categories according to the total number of all the matching weather categories and the total number of the matching weather categories of different categories;

[0031] Take the matching weather category corresponding to the maximum screening probability as the target weather category corresponding to the location information.

[0032] Preferably, the environmental information includes temperature and humidity.

[0033] Preferably, the picture, the environmental information and the location information are bound and packed into the package data according to the vehicle-end serial number.

[0034] A second aspect of the present application provides a cloud server, including a receiving module, a deep learning module, a statistical module, a mean module, an integration module and a sharing module;

[0035] The receiving module is configured to receive package data uploaded from at least one vehicle end, and the package data includes a picture, environmental information and location information;

[0036] The deep learning module is configured to parse the picture to obtain a picture feature map, standardize the environmental information to obtain a feature value, add the feature value to the picture feature map to obtain a target feature map, and perform a preferred probability calculation on the target feature map to obtain a matching weather category matching the target feature map;

[0037] The statistical module is configured to perform a statistical probability calculation on all the matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information;

[0038] The mean module is configured to perform a mean calculation on all the environmental information corresponding to the location information to obtain target environmental information;

[0039] The integration module is configured to integrate the target weather category and the target environmental information into target weather information;

[0040] A sharing module, configured to receive and respond to request instructions from at least one vehicle terminal, and share the target weather information to at least one of the vehicle terminals according to the relevance between the custom location information of the vehicle terminal and the location information.

[0041] A third aspect of the present application provides a weather information sharing system, including a terminal layer and a cloud server, where the terminal layer includes multiple vehicle terminals;

[0042] The vehicle terminal is configured to collect pictures and environmental information, obtain the current location information, package the pictures, the environmental information and the location information into package data and upload it to the cloud server, and / or send a request instruction to the cloud server;

[0043] The cloud server is configured to receive package data uploaded from at least one of the vehicle terminals; parse the pictures to obtain picture feature maps, standardize the environmental information to obtain feature values, add the feature values to the picture feature maps to obtain target feature maps, perform a preferred probability calculation on the target feature maps to obtain matching weather categories that match the target feature maps; perform a statistical probability calculation on all the matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information; perform an average calculation on all the environmental information corresponding to the location information to obtain target environmental information; integrate the target weather category and the target environmental information into target weather information; and share the target weather information to at least one of the vehicle terminals according to the relevance between the custom location information of the vehicle terminal and the location information.

[0044] The technical solution provided by the present application may include the following beneficial effects:

[0045] The technical solution of the present application, according to the package data uploaded by the vehicle terminal, parses the pictures in the package data to obtain picture feature maps, standardizes the environmental information to obtain feature values, adds the feature values to the picture feature maps to obtain target feature maps, performs a preferred probability calculation on the target feature maps to obtain matching weather categories that match, combines the location information to perform a statistical probability calculation on all the corresponding matching weather categories and outputs the target weather category, combines the location information to perform an average calculation on all the corresponding environmental information to obtain target environmental information, integrates the two to obtain target weather information, and finally shares the target weather information to the vehicle terminal through a request instruction. Since the package data is uploaded by the vehicle terminal in real time according to the current location information, the accuracy of the parsed target weather information is higher, and because the package data is uploaded in real time, the problem of data latency is not likely to occur.

[0046] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The above and other objects, features, and advantages of the present application will become more apparent by describing the exemplary embodiments of the present application in more detail with reference to the accompanying drawings, wherein in the exemplary embodiments of the present application, the same reference numerals generally represent the same components.

[0048] Figure 1 is a schematic flowchart of the weather information sharing method shown in the embodiments of the present application;

[0049] Figure 2 is a schematic flowchart of the target feature map obtained by the splicing method shown in the embodiments of the present application;

[0050] Figure 3 is a schematic flowchart of the target feature map obtained by the fusion method shown in the embodiments of the present application;

[0051] Figure 4 is a schematic diagram of the module structure of the cloud server shown in the embodiments of the present application;

[0052] Figure 5 is a schematic diagram of the module structure of the weather information sharing system shown in the embodiments of the present application. Detailed Embodiments

[0053] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0054] The terms used in the present application are for the purpose of describing specific embodiments only and are not intended to limit the present application. The singular forms "a", "the", and "said" used in the present application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0055] It should be understood that although the terms "first", "second", "third", etc. may be used in the present application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present application, the meaning of "a plurality" is two or more unless otherwise specifically defined.

[0056] Currently, in the related art, there are two ways to obtain weather information. The first is to identify weather information with the human eye and determine the weather conditions based on life experience. However, this method has a too large regional limitation. The second is to obtain weather information provided by a weather information service provider. However, due to the certain delay in the release of weather information data, users often cannot obtain the most accurate weather information.

[0057] In view of the above problems, the embodiments of the present application provide a weather information sharing method, a cloud server, and a system, which can provide accurate weather information. To facilitate the understanding of the solution of the embodiments of the present application, the technical solutions of the embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0058] Please refer to Figure 1 , a weather information sharing method, including:

[0059] Step S11, receiving package data uploaded from at least one vehicle terminal, where the package data includes pictures, environmental information, and location information.

[0060] It should be noted that the package data uploaded from the vehicle terminal includes pictures, environmental information, and location information. The picture data comes from the camera video stream data installed on the vehicle terminal. The vehicle terminal can extract a certain frame of picture in the video stream as picture data at a certain time interval (such as extracting one frame of picture every 5 s). At the same time, various sensors are also installed on the vehicle terminal. The function of the sensors is to collect the environmental information of the current position of the vehicle terminal. For example, the temperature sensor collects the current temperature, and the humidity sensor collects the current humidity. The location information of the vehicle terminal is obtained by means of GPS positioning.

[0061] After extracting the picture data, collecting the environmental information (current temperature and current humidity), and obtaining the location information, it is necessary to package and bind the pictures, environmental information, and location information into package data. The binding method is based on the vehicle terminal serial number. Each vehicle terminal has a vehicle terminal serial number. For example, assume that the vehicle terminal serial number of vehicle terminal A is 0001. Then, the picture data extracted by vehicle terminal A, the collected environmental information, and the obtained location information will all be attached with the vehicle terminal serial number 001. The vehicle terminal packages and binds them into package data according to the vehicle terminal serial number. The vehicle terminal serial number corresponding to the package data uploaded from vehicle terminal A is 0001. In this way, the package data from different vehicle terminals can be distinguished.

[0062] Since multiple vehicle ends will upload package data at the same time, it is very important to distinguish them by the vehicle end serial number. For example, at the same time, there are 5 vehicle ends (vehicle end serial numbers 000A1 to 00A5) in B County, A City. The 5 vehicle ends will each upload 1 package data (vehicle end serial numbers 000A1 to 00A5). Combining the location information in each package data, the 5 package data uploaded by B County, A City can be determined.

[0063] Step S12: Analyze the picture to obtain a picture feature map, standardize the environmental information to obtain a feature value, add the feature value to the picture feature map to obtain a target feature map, and perform a preferred probability calculation on the target feature map to obtain a matching weather category that matches the target feature map.

[0064] It should be noted that after receiving 5 package data, first analyze 5 pictures to obtain 5 picture feature maps. Subsequently, standardize 5 pairs of environmental information to obtain pairs of feature values (temperature and humidity constitute environmental information, so the temperature feature value and humidity feature value obtained by standardization constitute 1 pair of feature values). Add each pair of feature values to the corresponding picture feature map to obtain a target feature map (that is, 1 picture feature map - 1 pair of feature values - 1 target feature map). Perform a preferred probability calculation on each target feature map, and the matching weather category corresponding to each target feature map can be obtained.

[0065] Furthermore, in one embodiment, step S12 includes:

[0066] Step S121: Input into a deep learning model.

[0067] The deep learning model is a deep convolutional model. The deep convolutional model is established according to a supervised learning method. By manually annotating N pictures (N is greater than or equal to 10,000), and at the same time manually determining and labeling the weather categories corresponding to the N pictures, input these N pictures into the deep convolutional model for learning, and obtain a deep learning model with recognition ability through repeated training.

[0068] Step S122: Perform multiple-scale convolution calculations on the picture to obtain multiple convolutional feature maps.

[0069] Starting from the establishment of the deep learning model, use the deep learning model to perform multiple-scale convolution calculations on the picture to obtain multiple convolutional feature maps. The number of times of scale convolution calculation is set by the user. The more times of scale convolution calculation, the more feature maps are obtained, and the richer the information data obtained. For example, the deep learning model performs 3-scale convolution calculations on 1 picture to obtain 3 convolutional feature maps, and splicing or fusing these 3 convolutional feature maps will obtain 1 picture feature map.

[0070] Step S123: Concatenate or fuse multiple convolutional feature maps to calculate an image feature map.

[0071] The following introduces two different methods, concatenation and fusion. After performing three-scale convolutional calculations on one image, three convolutional feature maps (the convolutional feature maps are shown in the form of matrix diagrams) are obtained, namely the first convolutional feature map CONV1, the second convolutional feature map CONV2, and the third convolutional feature map CONV3. The first convolutional feature map CONV1, the second convolutional feature map CONV2, and the third convolutional feature map CONV3 are all 3x3 (row x column) matrix diagrams.

[0072] Concatenation method:

[0073] Please refer to Figure 2 , after concatenating the first convolutional feature map CONV1, the second convolutional feature map CONV2, and the third convolutional feature map CONV3, a 3x9 image feature map is obtained. Figure 2 The schematic concatenation method is to concatenate the first convolutional feature map CONV1, the second convolutional feature map CONV2, and the third convolutional feature map CONV3 to obtain an image feature map with the number of rows unchanged and the number of columns added. Of course, it is also possible to concatenate the first convolutional feature map CONV1, the second convolutional feature map CONV2, and the third convolutional feature map CONV3 to obtain a 9x3 image feature map by keeping the number of columns unchanged and adding the number of rows. That is, the number of rows of the image feature map is equal to the sum of the number of rows of each convolutional feature map, or the number of columns of the image feature map is equal to the sum of the number of columns of each convolutional feature map.

[0074] Fusion method:

[0075] Obtain the weight coefficients for each scale convolutional calculation. The weight coefficients are the weight coefficients generated by the deep learning model during the continuous learning process. For the number of times of scale convolutional calculation performed on the image, there will be a corresponding number of weight coefficients (in this application, three-scale convolutional calculation is selected, and there are 3 corresponding weight coefficients). Please refer to Figure 3, the weight coefficients W1, W2, and W3 represent the weight coefficients corresponding to the three-scale convolution calculations. Multiply the first convolutional feature map CONV1 by the weight coefficient W1 to obtain the first corrected convolutional feature map CONV1', multiply the second convolutional feature map CONV2 by the weight coefficient W2 to obtain the second corrected convolutional feature map CONV2', and multiply the third convolutional feature map CONV3 by the weight coefficient W3 to obtain the third corrected convolutional feature map CONV3'. Linearly add the first corrected convolutional feature map CONV1', the second corrected convolutional feature map CONV2', and the third corrected convolutional feature map CONV3' to obtain the picture feature map. The picture feature map is still a 3x3 matrix map. That is, the difference between the fusion method and the splicing method is that the number of rows and columns of the feature map in the fusion method remains unchanged, while the number of rows or columns of the feature map obtained by the splicing method will be superimposed.

[0076] Step S124: Use a normalization function to standardize the environmental information to obtain eigenvalue.

[0077] After obtaining the picture feature map, in order to establish a connection between the picture feature map and the environmental information, it is necessary to use a normalization function to standardize the environmental information to obtain eigenvalue.

[0078] The formula of the normalization function is as follows (taking the normalization of the current temperature as an example for illustration. The principle of normalizing the current humidity is the same as that of normalizing the current temperature and will not be repeated. Just refer to the principle of normalizing the current temperature):

[0079]

[0080] where, t0 represents the temperature collected at the current vehicle end, t norm represents the normalized temperature eigenvalue, t min represents the minimum value of the environmental temperature range, t max represents the maximum value of the environmental temperature range. The temperature eigenvalue can be obtained through (Equation 1).

[0081] Step S125: Add the eigenvalue to the picture feature map to obtain the target feature map.

[0082] Assume that the temperature eigenvalue obtained through (1) is 0.5. Please refer to Figure 2 , if a 3x9 picture feature map is obtained by the splicing method, expand the temperature eigenvalue into a 3x1 temperature feature matrix, where each element value in the temperature feature matrix is 0.5. After adding the temperature feature matrix to the picture feature map, a 3x10 target feature map is obtained. Please refer to Figure 3If it is a 3x3 image feature map obtained by the fusion method, the temperature feature values are still expanded into a 3x1 temperature feature matrix, where each element value in the temperature feature matrix is 0.5. After adding the temperature feature matrix to the image feature map, a 3x4 target feature map is obtained. Step S126: Parse out several to-be-determined weather categories of the target feature map, calculate the extraction probability of each to-be-determined weather category using the category probability function, and take the to-be-determined weather category corresponding to the maximum extraction probability as the matching weather category.

[0083] After obtaining the target feature map, due to different shooting angles of the image and the influence of light intensity, the deep learning model will analyze several to-be-determined weather categories of the target feature map. For example, a target feature map will parse out 4 to-be-determined weather categories, namely sunny, cloudy, rainy, and foggy. Each to-be-determined weather category will have a corresponding category value, which is the category value corresponding to different weather categories generated by the deep learning model through continuous learning. For example, the category value of sunny is 0.3, the category value of cloudy is 1.2, the category value of rainy is 0.4, and the category value of foggy is 2.1. Calculate the extraction probability of each to-be-determined weather category using the category probability function. The formula of the category probability function is as follows:

[0084]

[0085] where, e i is the category value corresponding to the i-th to-be-determined weather category, and e j is the category value corresponding to the j-th to-be-determined weather category among all weather categories. Through formula (2), the extraction probability p i of sunny can be calculated. Take the to-be-determined weather category corresponding to the maximum extraction probability as the matching weather category.

[0086] Step S13: Perform statistical probability calculation on all matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information.

[0087] Further, in one embodiment, step S13 includes:

[0088] Step S131: Count the total number of all matching weather categories corresponding to the location information;

[0089] Step S132: Count the total number of matching weather categories of different categories corresponding to the location information;

[0090] Step S133: Calculate the screening probability of matching weather categories of different categories according to the total number of all matching weather categories and the total number of matching weather categories of different categories;

[0091] Step S134: Take the matching weather category corresponding to the screening probability of the maximum value as the target weather category corresponding to the location information.

[0092] Still taking the example of 5 car ends (car end serial numbers 000A1 - 00A5) in B county of city A. After parsing and standardizing operations using a deep learning model, 5 weathers corresponding to 5 target feature maps will be obtained. For example, the matching weather category corresponding to the first target feature map is rainy, the second is rainy, the third is foggy, the fourth is rainy, and the fifth is rainy. Since these 5 target feature maps are all bound to the location information, it can be known from the location information that these 5 target feature maps all describe the weather categories in B county of city A. Therefore, the total number of all matching weather categories corresponding to the location information can be counted with the help of the location information. In the above example, the total number of all matching weather categories corresponding to the location information of B county in city A is 5. Then, count the total number of different types of matching weather categories corresponding to the location information of B county in city A. It is counted that the matching weather category of rainy days is 4, and the matching weather category of foggy days is 1. By probability calculation, the screening probability of rainy days and foggy days can be calculated, that is, the probability of rainy days in B county of city A is 80%, and the probability of foggy days is 20%. Take the matching weather category corresponding to the screening probability of the maximum value as the target weather category corresponding to B county of city A, and rainy days can be taken as the target weather category corresponding to B county of city A.

[0093] Step S14: Calculate the mean value of all the environmental information corresponding to the location information to obtain the target environmental information.

[0094] It should be noted that since 5 car ends (car end serial numbers 000A1 - 00A5) uploaded 5 pieces of environmental information (current temperature and current humidity), and according to the location information of the 5 car ends (car end serial numbers 000A1 - 00A5), they are all in B county of city A. Therefore, calculate the mean value of the 5 current temperatures and 5 current humidities to obtain the target environmental information (the mean temperature and the mean humidity).

[0095] Step S15: Integrate the target weather category and the target environmental information into the target weather information.

[0096] It should be noted that integrate the target weather category (rainy days) corresponding to B county of city A and the target environmental information (the mean temperature and the mean humidity) into the target weather information.

[0097] Step S16: Receive and respond to request instructions from at least one vehicle terminal, and share the target weather information to at least one vehicle terminal according to the relevance between the custom location information of the vehicle terminal and the location information.

[0098] Obtain the target weather information of County B in City A. At this time, two vehicle owners in City C want to travel to County B in City A and need to obtain the weather category in County B in City A at this time. The two vehicle owners set the custom location information as County B in City A and send a request instruction through the vehicle terminal. Since the custom location information is relevant to the location information, the target weather information of County B in City A can be sent to the vehicle terminals driven by the two vehicle owners. Even if the two vehicle owners are in City C, they can obtain the weather category in County B in City A in real time. Because the target weather information of County B in City A is obtained by parsing the packet data uploaded in real time by the vehicle terminal (vehicle terminal serial numbers 000A1 to 00A5), the weather information obtained by the two vehicle owners in City C is highly accurate and there will be no problem of obtaining incorrect weather information due to data update delay.

[0099] In the technical solution of this application, according to the packet data uploaded by the vehicle terminal, the picture in the packet data is parsed to obtain a picture feature map, the environmental information is standardized to obtain a feature value, the feature value is added to the picture feature map to obtain a target feature map, and the target feature map is preferably calculated for probability to obtain a matching weather category that matches the target feature map. The statistical probability of all corresponding matching weather categories is calculated in combination with the location information to output the target weather category, and the mean value of all corresponding environmental information is calculated in combination with the location information to obtain the target environmental information. The two are integrated to obtain the target weather information, and finally the request instruction shares the target weather information to the vehicle terminal. Since the packet data is uploaded in real time by the vehicle terminal according to the current location information, the accuracy of the parsed target weather information is higher, and because the packet data is uploaded in real time, the problem of data delay is not likely to occur.

[0100] Please refer to Figure 4 , a cloud server 900 includes a receiving module 910, a deep learning module 920, a statistical module 930, a mean value module 940, an integration module 950, and a sharing module 960. Among them:

[0101] The receiving module 910 is used to receive packet data uploaded from at least one vehicle terminal, and the packet data includes pictures, environmental information, and location information;

[0102] The deep learning module 920 is used to parse pictures to obtain a picture feature map, standardize environmental information to obtain a feature value, add the feature value to the picture feature map to obtain a target feature map, and perform a preferred probability calculation on the target feature map to obtain a matching weather category that matches the target feature map;

[0103] The statistical module 930 is used to calculate the statistical probability for all matching weather categories corresponding to the location information, and obtain the target weather category corresponding to the location information;

[0104] The mean module 940 is used to calculate the mean of all environmental information corresponding to the location information to obtain the target environmental information;

[0105] The integration module 950 is used to integrate the target weather category and the target environmental information into the target weather information;

[0106] The sharing module 960 is used to receive and respond to request instructions from at least one vehicle terminal, and share the target weather information to at least one vehicle terminal according to the relevance between the custom location information of the vehicle terminal and the location information.

[0107] The cloud server 900 of the present application receives the package data uploaded by the vehicle terminal through the receiving module 910. The deep learning module 920 analyzes and processes the pictures and environmental information in the package data to obtain the matching weather categories that match the target feature map. The statistical module 930 calculates the statistical probability according to all the matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information. The mean module 940 calculates the mean of all the environmental information corresponding to the location information to obtain the target environmental information. The integration module 950 integrates the two to obtain the target weather information. The sharing module 960 shares the target weather information to the requesting vehicle terminal according to the request instruction of the vehicle terminal and the relevance between the custom location information of the vehicle terminal and the location information. Since the package data is uploaded by the vehicle terminal in real time according to the current location information, the accuracy of the target weather information is higher, and because the package data is uploaded in real time, the problem of data delay is not likely to occur.

[0108] Please refer to the figure Figure 5 , a weather information sharing system 10 includes a terminal layer 800 and a cloud server 900. The terminal layer 800 includes multiple vehicle terminals 810. Among them:

[0109] The vehicle terminal 810 is used to collect pictures and environmental information, obtain the current location information, package the pictures, environmental information and location information into package data and upload it to the cloud server, and / or send a request instruction to the cloud server;

[0110] The cloud server 900 is used to receive the package data uploaded from at least one vehicle terminal 810; parse the picture to obtain the picture feature map, standardize the environmental information to obtain the feature value, add the feature value to the picture feature map to obtain the target feature map, perform the optimal probability calculation on the target feature map to obtain the matching weather category that matches the target feature map; calculate the average value of all environmental information corresponding to the location information to obtain the target environmental information; integrate the target weather category and the target environmental information into the target weather information; perform the statistical probability calculation on all matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information; according to the relevance between the custom location information of the vehicle terminal and the location information, share the target weather category to at least one vehicle terminal 810.

[0111] The vehicle terminal 810 (vehicle terminal serial number is 0002) extracts a certain frame of photo from the video stream captured by the camera as the picture data, and uses the temperature sensor to collect the current temperature and the humidity sensor to collect the current humidity, and obtains the current location information (D City, F County) of the vehicle terminal 810 (vehicle terminal serial number is 0002) by means of GPS positioning. The picture data, the current temperature, the current humidity and the location information are packaged and bound into the package data and sent into the cloud server 900 through the vehicle terminal serial number.

[0112] Further, in one of the embodiments, please refer to Figure 5 , the cloud server 900 includes a receiving module 910, a deep learning module 920, a statistical module 930, an average value module 940, an integration module 950 and a sharing module 960. Among them:

[0113] The receiving module 910 is used to receive the package data uploaded from at least one vehicle terminal 810, and the package data includes pictures, environmental information and location information.

[0114] The deep learning module 920 is used to parse the picture to obtain the picture feature map, standardize the environmental information to obtain the feature value, add the feature value to the picture feature map to obtain the target feature map, and perform the optimal probability calculation on the target feature map to obtain the matching weather category that matches the target feature map.

[0115] It should be noted that the deep learning module 920 inputs a deep learning model. The deep learning module 920 obtains multiple convolutional feature maps by performing convolutional calculations on the picture at multiple scales, and obtains a picture feature map by splicing or fusing the multiple convolutional feature maps. Then, the current temperature and current humidity are standardized to obtain a temperature feature value and a humidity feature value respectively. The temperature feature value and the humidity feature value are added to the picture feature map to obtain a target feature map, and a preferred probability calculation is performed on the target feature map, and then a matching weather category that matches the target feature can be obtained. For example, by parsing and calculating the package data uploaded from the vehicle terminal 810 (vehicle terminal serial number is 0002) as described above, the matching weather category that matches the target feature map is sunny.

[0116] The statistical module 930 is used to perform statistical probability calculation on all matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information.

[0117] It should be noted that the statistical module 930 statistically counts all matching weather categories corresponding to the location information (County F, City D) of the vehicle terminal 810 (vehicle terminal serial number 0002). The statistical result shows that the matching weather category is rainy, and the determined rainy probability is 100%. Therefore, rainy is the target weather category for County F, City D.

[0118] The mean module 940 is used to perform mean calculation on all environmental information corresponding to the location information to obtain the target environmental information;

[0119] It should be noted that the mean module 940 statistically counts all temperature values corresponding to the location information (County F, City D) of the vehicle terminal 810 (vehicle terminal serial number 0002), and performs mean calculation on all temperature values to obtain the target environmental information (the mean temperature value).

[0120] The integration module 950 is used to integrate the target weather category and the target environmental information into the target weather information. The sharing module 960 is used to receive and respond to request instructions from at least one vehicle terminal 810, and share the target environmental information to at least one vehicle terminal 810 according to the relevance between the custom location information of the vehicle terminal 810 and the location information.

[0121] It should be noted that the sharing module 960 receives a request instruction from the vehicle terminal 810 (vehicle terminal serial number 0003) in City K. The custom location information of the vehicle terminal 810 (vehicle terminal serial number 0003) is County F, City D. When the sharing module 960 determines that the custom location information (County F, City D) of the vehicle terminal 810 (vehicle terminal serial number 0003) is associated with the location information (County F, City D), it shares the target environment information corresponding to County F, City D to the vehicle terminal 810 (vehicle terminal serial number 0003). Since the package data is uploaded by the vehicle terminal 810 in real time according to the current location information, the accuracy of the target environment information is higher. And because the package data is uploaded in real time, the problem of data latency is not likely to occur.

[0122] The embodiments of the present application have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technologies in the market, or to enable other ordinary skill in the art in the technical field to understand the embodiments disclosed herein.

Claims

1. A weather information sharing method, characterized in that, Applied to a cloud server communicatively connected to a vehicle end, including: Receiving package data uploaded from multiple vehicle ends, where the package data includes pictures, environmental information, and location information; the pictures are obtained by extracting video data collected by a camera installed on the vehicle end; Analyzing the pictures to obtain picture feature maps, normalizing the environmental information to obtain feature values, adding the feature values to the picture feature maps to obtain target feature maps, and performing a preferred probability calculation on the target feature maps to obtain a matching weather category that matches the target feature maps; wherein, each piece of the package data uploaded by each vehicle end corresponds to one such matching weather category; Performing a statistical probability calculation on all the matching weather categories corresponding to the location information to obtain a target weather category corresponding to the location information; wherein, it includes counting the total number of all the matching weather categories corresponding to the location information; counting the total number of the matching weather categories of different categories corresponding to the location information; calculating the screening probability of the matching weather categories of different categories based on the total number of all the matching weather categories and the total number of the matching weather categories of different categories; taking the matching weather category corresponding to the maximum screening probability as the target weather category corresponding to the location information; Performing a mean value calculation on all the environmental information corresponding to the location information to obtain target environmental information; Integrating the target weather category and the target environmental information into target weather information; Receiving and responding to a request instruction from at least one vehicle end that uploads package data or does not upload package data, and sharing the target weather information to at least one vehicle end that sends the request instruction according to the relevance between the custom location information of the vehicle end that sends the request instruction and the location information.

2. The weather information sharing method according to claim 1, wherein The analyzing the pictures to obtain picture feature maps, normalizing the environmental information to obtain feature values, adding the feature values to the picture feature maps to obtain target feature maps, and performing a preferred probability calculation on the target feature maps to obtain a matching weather category that matches the target feature maps includes: Inputting into a deep learning model; Performing a multi-scale convolution calculation on the pictures to obtain multiple convolution feature maps; Performing a splicing or fusion calculation on the multiple convolution feature maps to obtain the picture feature maps; Normalizing the environmental information by using a normalization function to obtain the feature values; Adding the feature values into the picture feature maps to obtain the target feature maps; Analyzing several to-be-determined weather categories of the target feature maps, calculating the extraction probability of each to-be-determined weather category by using a category probability function, and taking the to-be-determined weather category corresponding to the maximum extraction probability as the matching weather category.

3. The weather information sharing method according to claim 2, wherein For the splicing calculation of the multiple convolution feature maps to obtain picture feature maps, the number of rows of the picture feature maps is equal to the sum of the number of rows of each convolution feature map, or the number of columns of the picture feature maps is equal to the sum of the number of columns of each convolution feature map.

4. The weather information sharing method according to claim 2, wherein The fusion calculation of the multiple convolution feature maps to obtain the picture feature maps includes: Obtain the weight coefficients for each of the scale convolution calculations; Calculate a corrected convolution feature map based on the weight coefficients corresponding to the convolution feature map; Fusion calculate multiple corrected convolution feature maps to obtain the picture feature map.

5. The weather information sharing method according to claim 2, wherein Perform at least three-scale convolutions on the picture.

6. The weather information sharing method according to claim 1, wherein The environmental information includes temperature and humidity.

7. The weather information sharing method according to claim 1, wherein The picture, the environmental information, and the location information are bound and packaged into the package data according to the vehicle-end serial number.

8. A cloud server, characterized in that, It includes a receiving module, a deep learning module, a statistical module, a mean module, and a sharing module; The receiving module is used to receive the package data uploaded from multiple vehicle ends. The package data includes pictures, environmental information, and location information; the pictures are extracted from the video data collected by the cameras installed on the vehicle ends; The deep learning module is used to parse the picture to obtain a picture feature map, standardize the environmental information to obtain a feature value, add the feature value to the picture feature map to obtain a target feature map, perform a preferred probability calculation on the target feature map, and obtain a matching weather category that matches the target feature map; wherein, each package data uploaded by each vehicle end corresponds to a matching weather category; The statistical module is used to perform a statistical probability calculation on all the matching weather categories corresponding to the location information to obtain the target weather category corresponding to the location information; wherein, it includes counting the total number of all the matching weather categories corresponding to the location information; counting the total number of the matching weather categories of different categories corresponding to the location information; calculating the screening probability of the matching weather categories of different categories according to the total number of all the matching weather categories and the total number of the matching weather categories of different categories; taking the matching weather category corresponding to the screening probability with the maximum value as the target weather category corresponding to the location information; The mean module is used to perform a mean calculation on all the environmental information corresponding to the location information to obtain the target environmental information; The integration module is used to integrate the target weather category and the target environmental information into the target weather information; The sharing module is used to receive and respond to the request instructions from at least one vehicle end that uploads package data or does not upload package data. According to the relevance between the custom location information of the vehicle end that sends the request instruction and the location information, share the target weather information to at least one vehicle end that sends the request instruction.

9. A weather information sharing system, characterized in that, It includes a terminal layer and a cloud server, and the terminal layer includes multiple vehicle ends; The vehicle end is used to collect pictures and environmental information, obtain the current location information, package the pictures, the environmental information, and the location information into package data and upload it to the cloud server, and / or send a request instruction to the cloud server; The pictures are extracted from the video data collected by the cameras installed on the vehicle ends; The cloud server is used to receive the package data uploaded from multiple vehicle ends; Parse the picture to obtain a picture feature map, standardize the environmental information to obtain a feature value, add the feature value to the picture feature map to obtain a target feature map, perform a preferred probability calculation on the target feature map to obtain a matching weather category that matches the target feature map; wherein, each piece of the package data uploaded by each vehicle end corresponds to a kind of the matching weather category; perform a statistical probability calculation on all the matching weather categories corresponding to the location information to obtain a target weather category corresponding to the location information; wherein, it includes counting the total number of all the matching weather categories corresponding to the location information; counting the total number of the matching weather categories of different categories corresponding to the location information; calculating the screening probability of the matching weather categories of different categories according to the total number of all the matching weather categories and the total number of the matching weather categories of different categories; taking the matching weather category corresponding to the screening probability with the maximum value as the target weather category corresponding to the location information; perform a mean calculation on all the environmental information corresponding to the location information to obtain target environmental information; integrate the target weather category and the target environmental information into target weather information; receive and respond to a request instruction from at least one vehicle end that uploads or does not upload package data, and share the target weather information to at least one vehicle end that sends the request instruction according to the relevance between the custom location information of the vehicle end that sends the request instruction and the location information.

Images