Insurance recommendation method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and provides an insurance recommendation method, which comprises the steps of preprocessing a to-be-processed image to obtain a to-be-recognized image; performing character recognition on the to-be-recognized image through a preset image recognition model to obtain a character set; performing character matching on a target character in the character set and a preset character in a target character library to obtain a character matching result; when the character matching result represents that all the target characters are correct, obtaining user information and historical data information corresponding to the to-be-recognized image; and performing risk identification on the user information and the historical data information to obtain a risk identification result, and generating an insurance recommendation result according to the risk identification result. The method is applied to the field of insurance. According to the invention, through character recognition, recognition of characters in the image is realized. Through character matching, it is ensured that the characters in the image meet requirements, so that risk identification is achieved, insurance recommendation is achieved, and the recommendation accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an insurance recommendation method, device, computer equipment and storage medium. Background Art

[0002] With the development of the economy and the improvement of people's living standards, cars have become a necessity for many families. Cars may encounter various risks during use. As an important part of the insurance industry, auto insurance plays a positive role in promoting the development of the insurance industry, increasing employment opportunities and promoting economic growth.

[0003] At present, most auto insurance products rely mainly on manual input for customer information and insured property, which is inefficient. On the other hand, the recognition rate of the recognition algorithm in the industry is low. The algorithm cannot judge the authenticity and validity of the license plate, and lacks risk assessment, resulting in inaccurate premium recommendations. Therefore, a recommendation technology solution is urgently needed to solve the above problems. Summary of the invention

[0004] The embodiments of the present invention provide an insurance recommendation method, apparatus, computer equipment and storage medium to solve the problem in the prior art that the authenticity and validity of a license plate cannot be determined and risk assessment is lacking, resulting in inaccurate premium recommendations.

[0005] An insurance recommendation method, comprising: Acquire an image to be processed, and preprocess the image to be processed to obtain an image to be recognized; Acquire a preset image recognition model, and perform boundary recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters; Acquire a target character library, and perform character matching on a target character in the character set and a preset character in the target character library to obtain a character matching result; When the character matching result indicates that all target characters are correct, obtaining user information and historical data information corresponding to the image to be recognized; Perform risk identification on the user information and the historical data information to obtain a risk identification result, and generate an insurance recommendation result based on the risk identification result.

[0006] An insurance recommendation device, comprising: An image preprocessing module is used to obtain an image to be processed and preprocess the image to be processed to obtain an image to be recognized; A character recognition module, used to obtain a preset image recognition model, and perform boundary recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters; A character matching module, used to obtain a target character library, and perform character matching on a target character in the character set and a preset character in the target character library to obtain a character matching result; An information acquisition module, used for acquiring user information and historical data information corresponding to the image to be recognized when the character matching result indicates that all target characters are correct; The risk identification module is used to perform risk identification on the user information and the historical data information, obtain risk identification results, and generate insurance recommendation results based on the risk identification results.

[0007] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned insurance recommendation method when executing the computer program.

[0008] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the insurance recommendation method is implemented.

[0009] The present invention provides an insurance recommendation method, device, equipment and medium. The method realizes the recognition of the image to be processed by preprocessing the image to be processed, ensures that the image to be processed meets the requirements, and then realizes the acquisition of the image to be recognized. Character recognition is performed on the image to be recognized by a preset image recognition model, and the recognition of the characters in the image is realized, and the construction of the character set is realized, thereby improving the recognition accuracy. By matching the target characters in the character set with the preset characters in the target character library, the detection of the target characters is realized, and it is ensured that the characters in the image meet the requirements. By performing risk identification on user information and historical data information, the identification of risks is realized, thereby realizing the recommendation of insurance and the generation of insurance recommendation results, thereby improving the recommendation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0011] Figure 1 is a schematic diagram of an application environment of an insurance recommendation method in one embodiment of the present invention; Figure 2 is a flow chart of an insurance recommendation method in one embodiment of the present invention; Figure 3 is a flow chart of step S10 of the insurance recommendation method in one embodiment of the present invention; Figure 4 is a principle block diagram of an insurance recommendation device in one embodiment of the present invention; Figure 5 is a schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] The insurance recommendation method provided by the embodiment of the present invention can be applied as follows: Figure 1 Specifically, the insurance recommendation method is applied in an insurance recommendation device, which includes: Figure 1 The client and server shown in the figure communicate with each other through the network to solve the problem in the prior art that local stylization of images cannot achieve the expected effect. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms. The client, also known as the user end, refers to a program that corresponds to the server and provides classification services to customers. The client can be installed on, but not limited to, various computers, laptops, smart phones, tablets, and portable wearable devices.

[0014] In one embodiment, if Figure 2 As shown, a method for recommending insurance is provided, and the method is applied in Figure 1 The server in the example is used as an example to illustrate the process, including the following steps: S10, obtaining an image to be processed, and preprocessing the image to be processed to obtain an image to be recognized.

[0015] It can be understood that the image to be processed refers to the image uploaded by the user, and the image to be identified refers to the regional image cut out of the uploaded image.

[0016] Specifically, the image to be processed uploaded by the target user is obtained, and then the image to be processed is preprocessed, that is, edge recognition is performed on the image to be processed to determine the target area in the image, and whether the shape of the target area is the expected rectangular shape is detected. Then, the image of the target area is sharpened to enhance the edge of the area. The color in the target area is identified to determine whether the target area includes a preset color combination. If it does, the proportion of each color in the preset color combination in the target area is calculated. When the color proportion meets the threshold, the target area is identified to segment the target area, thereby obtaining the image to be identified.

[0017] S20, obtaining a preset image recognition model, and performing character recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters.

[0018] Understandably, the target characters refer to the words and letters in the image. The character set refers to the set of all recognized characters.

[0019] Specifically, a preset image recognition model is obtained, and the image to be recognized is input into the preset image recognition model, and character recognition is performed on the image to be recognized through the preset image recognition model, that is, the boundary of each character in the image to be recognized is recognized through the boundary recognition ability learned during training, and the character boundary corresponding to each target character is obtained. Then, the characters are cut based on all the boundaries corresponding to each target character, so as to obtain all the target characters in the image to be recognized, and a character set is constructed based on all the target characters.

[0020] S30, obtaining a target character library, and performing character matching on the target characters in the character set and the preset characters in the target character library to obtain a character matching result.

[0021] Understandably, the character matching result is used to indicate whether the target characters are wrong, that is, whether all target characters are matched successfully. The target character library refers to a database storing all license plates and their characters.

[0022] Specifically, a target character library is obtained, and the encrypted target character library is decrypted by a preset decryption algorithm to obtain a target character library including preset characters. Then, character matching is performed between the target characters in the character set and the preset characters in the target character library, that is, character matching is performed between the first target character and the second target character in the order of the target characters to filter out all preset characters corresponding to the first target character and the second target character. Next, character matching is performed between the target characters in the character set except the first target character and the second target character and all preset characters corresponding to the first target character and the second target character that are filtered out to reduce the number of matching times and increase efficiency, thereby obtaining a character matching result that indicates whether the characters are incorrect.

[0023] S40, when the character matching result indicates that all target characters are correct, obtaining user information and historical data information corresponding to the image to be recognized.

[0024] Understandably, user information refers to the data of the user corresponding to the image. Historical data information refers to the historical data corresponding to the user, for example, the last insurance purchase record, or the number of accidents of the vehicle, etc.

[0025] Specifically, when the character matching result indicates that all target characters are correct, a query is performed using all target characters, that is, the string composed of all target characters and the license plate number in the user information are matched. When the match is successful, the user information corresponding to the license plate number that successfully matches the string is obtained, and the user information corresponding to the license plate number is determined as the user information corresponding to all target characters. Then, a query is performed based on the user information to obtain historical data information corresponding to the user information.

[0026] S50, performing risk identification on the user information and the historical data information to obtain a risk identification result, and generating an insurance recommendation result based on the risk identification result.

[0027] It can be understood that the risk identification result refers to the risk level predicted for the user's vehicle. The insurance recommendation result refers to the recommended insurance advice and quotation.

[0028] Specifically, risk identification is performed on user information and historical data information, that is, preset risk assessment rules are obtained, and risk assessment is performed on user information and historical data information based on the risk assessment rules, that is, whether the historical data information meets the specific content of the risk assessment rules is detected, thereby obtaining risk identification results. Then, insurance recommendation results are generated based on the risk identification results, that is, based on the risk identification results, auto insurance that meets the user's needs is screened out from all insurances, and the screened insurance suggestions and quotations are fed back to the user, and the screened insurance suggestions and quotations are determined as insurance recommendation results.

[0029] In an embodiment of the present invention, an insurance recommendation method is provided. The method realizes recognition of an image to be processed by preprocessing the image to be processed, ensures that the image to be processed meets the requirements, and then realizes the acquisition of the image to be recognized. Character recognition is performed on the image to be recognized by a preset image recognition model, thereby realizing recognition of characters in the image, realizing the construction of a character set, and then improving the recognition accuracy. By matching the target characters in the character set with the preset characters in the target character library, the target characters are detected, ensuring that the characters in the image meet the requirements. By performing risk identification on user information and historical data information, risk identification is realized, thereby realizing the recommendation of insurance and the generation of insurance recommendation results, and then improving the recommendation accuracy.

[0030] In one embodiment, in step S10, that is, preprocessing the image to be processed to obtain the image to be recognized, includes: S101, performing edge recognition on the image to be processed to obtain a region image, and detecting whether the region shape of the region image is a rectangle.

[0031] S102, when the region shape of the region image is a rectangle, sharpen the region image to obtain an enhanced image, and detect whether the enhanced image includes a preset color combination; the preset color combination includes two different first preset colors and a second preset color.

[0032] S103: When the enhanced image includes a preset color combination, determine a first color proportion corresponding to the first preset color and a second color proportion corresponding to the second preset color.

[0033] S104: When the first color proportion meets the first ratio threshold range and the second color proportion meets the second ratio threshold range, perform region recognition on the enhanced image to obtain an image to be recognized.

[0034] Understandably, the regional image refers to the image corresponding to the target area. The enhanced image refers to the image after the edge of the target area is enhanced. The preset color combination refers to the color combination set in advance, for example, yellow and black, blue and white, and green and black. The preset color combination includes two different first preset colors and second preset colors. The first color proportion refers to the ratio of the first preset color to all colors in the target area. The second color proportion refers to the ratio of the second preset color to all colors in the target area. The first ratio threshold range refers to the ratio range corresponding to the first preset color. The second ratio threshold range refers to the ratio range corresponding to the second preset color. Among them, the sum of the first color proportion and the second color proportion is 100%.

[0035] Specifically, after obtaining the image to be processed, the Laplace operator is used to perform edge recognition on the image to be processed, that is, the image to be processed is first grayed to obtain a gray image, and then the Laplace template is used to multiply and sum the pixel values ​​around each pixel in the gray image with the corresponding elements in the Laplace template to obtain the Laplace value of the pixel. Then, the Laplace value of the pixel is compared with a preset threshold, and the pixels with Laplace values ​​greater than the threshold are marked as edge pixels, and the pixels with Laplace values ​​less than or equal to the threshold are marked as non-edge pixels. Finally, all edge pixels are connected to obtain a regional image. Next, it is detected whether the regional shape of the regional image is a rectangular shape, that is, the boundary of the regional shape composed of connecting all edge pixels is identified, and then it is determined whether the composed region is a rectangular shape. When the regional shape of the regional image is a rectangular shape, the regional image is sharpened, that is, the Laplace image calculated by the Laplace template and the gray image are added to enhance the edge, thereby obtaining an enhanced image. Further, it is detected whether the enhanced image includes a preset color combination, that is, the enhanced image is color recognized by a preset color recognition model, so as to obtain at least one color included in the enhanced image. The preset color combination is obtained, and the recognized color is matched with the preset color combination to determine whether the enhanced image includes the preset color combination. When the preset color combination is included in the enhanced image, the first proportion corresponding to the first preset color and the second proportion corresponding to the second preset color are determined, that is, the enhanced image is converted into a grayscale image, and according to the grayscale value range corresponding to each color, the grayscale image is traversed to calculate the first proportion corresponding to the first preset color and the second proportion corresponding to the second preset color. Next, the first proportion threshold range and the second proportion threshold range are obtained, and the first proportion is compared with the first proportion threshold range, and the second proportion is compared with the second proportion threshold range. When the first proportion is in the first proportion threshold range and the second proportion is in the second proportion threshold range, it is determined that the first proportion meets the first proportion threshold range and the second proportion meets the second proportion threshold range. Finally, the enhanced image is region-recognized, that is, the target area is region-cut according to the connected edge pixel points, and the image to be recognized can be obtained.

[0036] In this embodiment, edge recognition is used to obtain the regional image, thereby realizing the detection of the shape of the regional image. Sharpening processing is used to obtain the enhanced image, thereby realizing the detection of the color in the image. By calculating the first proportion and the second proportion, it is realized to determine whether the proportion meets the threshold, thereby realizing the cutting of the region and the acquisition of the image to be identified.

[0037] In one embodiment, in step S20, character recognition is performed on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters, including: S201, performing character recognition on the image to be recognized by using the preset image recognition model to obtain character boundaries corresponding to each target character.

[0038] S202: Perform character segmentation on the target characters based on all the character boundaries to obtain a character set including all the target characters.

[0039] It can be understood that the character boundary refers to the top, bottom, left and right boundaries of each character in the image.

[0040] Specifically, after obtaining the preset image recognition model, the image to be recognized is input into the preset image recognition model, and character recognition is performed on the image to be recognized by the preset image recognition model, that is, the image to be recognized is convolved by the convolution layer, that is, the convolution is performed from top to bottom and from left to right in sequence by the convolution kernel, so as to obtain the convolution feature. Then, the convolution feature is pooled by the pooling layer, that is, the convolution feature is pooled by the pooling kernel to obtain the pooled feature, and then the pooled feature is calculated by the linear function to obtain the character boundary corresponding to each target character. Furthermore, the target characters are cut based on all character boundaries, that is, each target character is cut according to all character boundaries corresponding to each target character, and sorted according to the position of each target character, and a character set is constructed based on the character sorting result.

[0041] In this embodiment, the boundary of the target character is recognized by the preset image recognition model, thereby obtaining the boundary of the character. The target character is cut through the boundary of the character, thereby constructing the character set.

[0042] In one embodiment, after step S202, that is, after character segmentation is performed on the target characters based on all the character boundaries to obtain a character set including all the target characters, the following further includes: S203, obtaining a preset number of characters corresponding to the preset color combination, and detecting whether the number of all target characters in the character set meets the preset number of characters.

[0043] S204: If the number of characters in the character set does not meet the preset number of characters, it is determined that the image to be recognized is not a license plate image.

[0044] S205: If the number of characters in the character set meets the preset number of characters, it is determined that the image to be recognized is a license plate image.

[0045] Understandably, the license plate image refers to a license plate picture. The preset number of characters refers to the number of characters in the license plate set in advance.

[0046] Specifically, after obtaining a character set including all target characters, obtain the preset number of characters corresponding to the preset color combination, that is, query and obtain through the association between the preset color combination and the preset number of characters. Then, count the number of characters of all target characters in the character set, and detect whether the number of characters of all target characters in the character set meets the preset number of characters, that is, determine whether the two are equal. If the number of characters in the character set does not meet the preset number of characters, it is determined that the image to be recognized is not a license plate image. If the number of characters in the character set meets the preset number of characters, it is determined that the image to be recognized is a license plate image. For example, a blue license plate has white characters on a blue background, and the license plate number has seven characters, and the color of the recognized processed image is also white characters on a blue background, but it has six characters, so the image is not a license plate image.

[0047] In this embodiment, by detecting whether the number of all target characters in the character set meets the preset number of characters, further determination of the image is achieved, ensuring that the image uploaded by the user is a license plate image, thereby improving the accuracy of subsequent risk identification and insurance recommendations.

[0048] In one embodiment, in step S30, character matching is performed on the target character in the character set and the preset character in the target character library to obtain a character matching result, including: S301, determining all screening characters corresponding to the character set from the target character library according to a first target character and a second target character in the character set.

[0049] S302: Perform character matching on target characters other than the first target character and the second target character in the character set and the screening characters.

[0050] S303, when there is a matching failure among all the target characters, obtaining a character matching result indicating the incorrect character among all the target characters.

[0051] S304, when the character matching result indicates that all the target characters are matched to preset characters, a correct character matching result is obtained for all the target characters.

[0052] Understandably, the screening characters refer to the preset characters corresponding to the first target character and the second target character. The character matching result is used to indicate whether all target characters are successfully matched. The first target character refers to the first target character. The second target character refers to the second target character, that is, the character positions are different.

[0053] Specifically, after obtaining the target character library, according to the character positions in the image to be recognized, the first target character and the second target character in the character set corresponding to the character positions in the image to be recognized are determined. Then, according to the first target character and the second target character in the character set, all the screened characters corresponding to the character set are determined from the target character library, that is, the first target character is matched with all the preset characters in the target character library. When the match is successful, all the preset characters corresponding to the first target character are obtained, and the second target character is matched with all the preset characters corresponding to the first target character, so as to obtain all the screened characters corresponding to the character set. Further, character matching is performed on the target characters and the screened characters in the character set except the first target character and the second target character, that is, the remaining target characters and the screened characters in the character set are matched one by one. That is, when the target character at the previous position is successfully matched, the target character and the screened character at the next position are matched until all the remaining target characters are successfully matched and the matching ends, or the target character at a certain position fails to match. When the match fails, the matching ends. When there is a match failure among all the target characters, a character matching result indicating that there is an error among all the target characters is obtained. When the character matching result indicates that all the target characters are matched to the preset characters, a character matching result indicating that there is no error among all the target characters is obtained. For example, for a license plate of Shanghai A, all the preset characters corresponding to Shanghai can be determined first through Shanghai, and then A is matched with all the preset characters corresponding to Shanghai, so as to obtain the screened characters corresponding to Shanghai A. Then, the remaining characters of the license plate are matched to obtain the character matching result.

[0054] In this embodiment, by determining all the screened characters corresponding to the character set from the target character library according to the first target character and the second target character in the character set, the screening of characters is realized, thereby improving the character matching efficiency and obtaining the character matching result.

[0055] In one embodiment, in step S50, that is, risk identification is performed on the user information and the historical data information to obtain a risk identification result, including: S501, obtain a risk identification model, and input the user information and the historical data information into the risk identification model.

[0056] S502, perform feature extraction on the user information and the historical data information through the feature extraction layer in the risk identification model to obtain information features corresponding to the user information and data features corresponding to the historical data information.

[0057] S503, perform risk identification on the information features and the data features through the risk prediction layer in the risk identification model to obtain a risk identification result.

[0058] It can be understood that information features are features of extracted user information, and data features are features of extracted historical data information.

[0059] Specifically, after obtaining the user information and historical data information, a risk identification model is obtained, and the user information and the historical data information are input into the risk identification model. Then, the feature extraction layer in the risk identification model is used to extract features from the user information and the historical data information, that is, the capabilities learned by the risk identification model during training are used to extract features from the user information and the historical data information, respectively, thereby obtaining information features corresponding to the user information and data features corresponding to the historical data information. Next, the risk identification is performed on the information features and the data features through the risk prediction layer in the risk identification model, that is, the risk level is predicted based on the information features and the data features through the capabilities learned during training, thereby obtaining a risk identification result. For example, if the number of accidents is 5, it is determined to be a high risk; if the number of accidents is 0, it is determined to be a low risk.

[0060] In this embodiment, feature extraction of user information and historical data information is performed through the feature extraction layer in the risk identification model, thereby realizing the extraction of information features and data features, thereby realizing the identification of risk levels and the acquisition of risk identification results, thereby improving the efficiency and accuracy of identification.

[0061] In one embodiment, before step S20, that is, before obtaining the preset image recognition model, the process further includes: S106, obtaining a sample data set, where the sample data set includes at least one sample training data and a sample label corresponding to each sample training data.

[0062] Understandably, sample data refers to different license plate images, for example, images at different angles, images at different distances, images of different license plates, etc. One sample training data is associated with a sample label, and the sample label is used to characterize the content of the sample training data, that is, the specific character information in the license plate image. The sample training data and sample labels can be collected from different databases, or they can be prepared in advance and sent from the client to the database. Then, a sample training data set is constructed based on all the sample training data obtained and the sample labels corresponding to each sample training data.

[0063] S107, identifying and predicting all the sample training data using a preset training model to obtain a prediction result corresponding to each of the sample training data.

[0064] It can be understood that the prediction result refers to the result obtained by the preset training model identifying the sample training data. In this embodiment, the preset training model adopts a convolutional neural network.

[0065] Specifically, a preset training model is obtained, and all sample training data are input into the preset training model, and all sample training data are identified and predicted through the preset training model, that is, each sample training data is convoluted through the convolution layer in the preset training model to obtain convolution features, and then the convolution features are activated by the linear activation function, and then the convolution features are pooled through the maximum pooling layer to reduce the dimension size, thereby obtaining pooling features, and finally, the pooling features are identified and predicted through the fully connected layer to obtain the prediction results corresponding to the sample training data. Among them, the number of convolution layers and pooling layers can be increased to perform deep convolution processing on the sample training data to extract more complex features.

[0066] S108, determining a prediction loss value of a preset training model according to sample labels and prediction results corresponding to the same sample training data.

[0067] Understandably, the prediction loss value is generated during the prediction process of the sample training data.

[0068] Specifically, after obtaining the prediction result, all prediction results corresponding to the sample data are arranged according to the order of the sample data in the sample data set, and then the prediction results associated with the sample data are compared with the sample labels of the sample data with the same sequence; that is, according to the sorting of the sample data, the sample label corresponding to the first sample data is compared with the prediction result corresponding to the first sample data, and the loss value between the sample label and the prediction result is determined by the loss function; the sample label corresponding to the second sample data is compared with the prediction result corresponding to the second sample data, and the loss value between the sample label and the prediction result is determined by the loss function, until all sample labels are compared with the prediction results, the prediction loss value of the preset training model can be obtained.

[0069] S109, when the predicted loss value reaches a preset convergence condition, determining the preset training model after convergence as a preset image recognition model.

[0070] Understandably, the convergence condition may be that the predicted loss value is less than a set threshold, or that the predicted loss value is very small and will not decrease after 50,000 calculations, and the training is stopped.

[0071] Specifically, after obtaining the predicted loss value, when the predicted loss value does not reach the preset convergence condition, the initial parameters of the preset training model are adjusted according to the predicted loss value, and all sample training data are re-input into the preset training model with the adjusted initial parameters, and the preset training model with the adjusted initial parameters is iteratively trained to obtain the predicted loss value corresponding to the preset training model with the adjusted initial parameters. Furthermore, when the predicted loss value does not reach the preset convergence condition, the initial parameters of the preset training model are adjusted again according to the predicted loss value, so that the predicted loss value of the preset training model with the adjusted initial parameters reaches the preset convergence condition. In this way, the accuracy of the preset training model becomes higher and higher, and the quality inspection results obtained are constantly approaching the correct results, until the predicted loss value of the preset training model reaches the preset convergence condition, and the preset training model after convergence is determined as the preset image recognition model.

[0072] In the embodiment of the present invention, the preset training model is iteratively trained through a large amount of sample training data, and the overall loss value of the preset training model is calculated by comparing the loss function, thereby realizing the determination of the predicted loss value of the preset training model. The initial parameters of the preset training model are adjusted according to the predicted loss value until the model converges, thereby realizing the training of the preset image recognition model, thereby ensuring that the preset image recognition model has a higher expected effect.

[0073] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0074] In one embodiment, an insurance recommendation device is provided, which corresponds one-to-one to the insurance recommendation method in the above embodiment. Figure 4 As shown, the insurance recommendation device includes an image preprocessing module 10, a character recognition module 20, a character matching module 30, an information acquisition module 40 and a risk identification module 50. The functional modules are described in detail as follows: The image preprocessing module 10 is used to obtain an image to be processed and preprocess the image to be processed to obtain an image to be recognized; The character recognition module 20 is used to obtain a preset image recognition model, and perform boundary recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters; The character matching module 30 is used to obtain a target character library, and perform character matching on the target character in the character set and the preset character in the target character library to obtain a character matching result; An information acquisition module 40 is used to acquire user information and historical data information corresponding to the image to be recognized when the character matching result indicates that all target characters are correct; The risk identification module 50 is used to perform risk identification on the user information and the historical data information, obtain risk identification results, and generate insurance recommendation results based on the risk identification results.

[0075] In one embodiment, the image preprocessing module 10 includes: a shape detection unit, configured to perform edge recognition on the image to be processed to obtain a region image, and detect whether the region shape of the region image is a rectangular shape, A color detection unit is used to perform a sharpening process on the regional image when the regional shape of the regional image is a rectangular shape to obtain an enhanced image, and detect whether the enhanced image includes a preset color combination; the preset color combination includes two different first preset colors and a second preset color; a ratio determination unit, configured to determine a first color ratio corresponding to the first preset color and a second color ratio corresponding to the second preset color when the enhanced image includes a preset color combination; The region identification unit is used to perform region identification on the enhanced image to obtain an image to be identified when the first color proportion meets the first ratio threshold range and the second color proportion meets the second ratio threshold range.

[0076] In one embodiment, the character recognition module 20 includes: A character boundary unit, used to perform boundary recognition on the image to be recognized by using the preset image recognition model to obtain character boundaries corresponding to each target character; The character segmentation unit is used to segment the target characters based on all the character boundaries to obtain a character set including all the target characters.

[0077] In one embodiment, the character recognition module 20 further includes: A character quantity detection unit, used to obtain a preset character quantity corresponding to the preset color combination, and detect whether the character quantity of all target characters in the character set meets the preset character quantity; A quantity non-conforming unit, used for determining that the image to be recognized is not a license plate image if the number of characters in the character set does not match a preset number of characters; The quantity matching unit is used to determine that the image to be recognized is a license plate image when the number of characters in the character set matches a preset number of characters.

[0078] In one embodiment, the character matching module 30 includes: a character screening unit, configured to determine all screening characters corresponding to the character set from the target character library according to the first target character and the second target character in the character set; A character matching unit, used for performing character matching on target characters other than the first target character and the second target character in the character set and the screening character; A character error unit, used for obtaining a character matching result indicating an error among all the target characters when a matching failure occurs among all the target characters; The character correctness unit is used to obtain a character matching result indicating that all the target characters are correct when the character matching result indicates that all the target characters are matched to preset characters.

[0079] In one embodiment, the risk identification module 50 includes: An information input unit, used to obtain a risk identification model, and input the user information and the historical data information into the risk identification model; A feature extraction unit, configured to extract features from the user information and the historical data information through a feature extraction layer in the risk identification model to obtain information features corresponding to the user information and data features corresponding to the historical data information; The risk identification unit is used to perform risk identification on the information features and the data features through the risk prediction layer in the risk identification model to obtain a risk identification result.

[0080] In one embodiment, the device further comprises: A sample acquisition unit, used to acquire a sample data set, wherein the sample data set includes at least one sample training data and a sample label corresponding to each sample training data; An identification and prediction unit, used to identify and predict all the sample training data through a preset training model to obtain a prediction result corresponding to each of the sample training data; A loss prediction unit, used to determine a predicted loss value of a preset training model according to a sample label and a prediction result corresponding to the same sample training data; The model convergence unit is used to determine the preset training model after convergence as the preset image recognition model when the predicted loss value reaches the preset convergence condition.

[0081] The specific definition of the insurance recommendation device can be found in the definition of the insurance recommendation method above, which will not be repeated here. Each module in the above insurance recommendation device can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0082] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data used in the insurance recommendation method in the above-mentioned embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an insurance recommendation method is implemented.

[0083] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned insurance recommendation method when executing the computer program.

[0084] In one embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above-mentioned insurance recommendation method when executed by a processor.

[0085] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0086] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0087] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. An insurance recommendation method, characterized in that: include: Acquire an image to be processed, and preprocess the image to be processed to obtain an image to be recognized; Acquire a preset image recognition model, and perform boundary recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters; Acquire a target character library, and perform character matching on a target character in the character set and a preset character in the target character library to obtain a character matching result; When the character matching result indicates that all target characters are correct, obtaining user information and historical data information corresponding to the image to be recognized; Perform risk identification on the user information and the historical data information to obtain a risk identification result, and generate an insurance recommendation result based on the risk identification result.

2. The insurance recommendation method according to claim 1, characterized in that: The preprocessing of the image to be processed to obtain the image to be recognized includes: Performing edge recognition on the image to be processed to obtain a regional image, and detecting whether the regional shape of the regional image is a rectangular shape; When the region shape of the region image is a rectangular shape, the region image is sharpened to obtain an enhanced image, and it is detected whether the enhanced image includes a preset color combination; the preset color combination includes two different first preset colors and a second preset color; When a preset color combination is included in the enhanced image, determining a first color proportion corresponding to the first preset color and a second color proportion corresponding to the second preset color; When the first color proportion meets the first ratio threshold range, and the second color proportion meets the second ratio threshold range, region recognition is performed on the enhanced image to obtain an image to be recognized.

3. The insurance recommendation method according to claim 1, characterized in that: The step of performing boundary recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters includes: Performing boundary recognition on the image to be recognized by using the preset image recognition model to obtain character boundaries corresponding to each target character; The target characters are segmented based on all the character boundaries to obtain a character set including all the target characters.

4. The insurance recommendation method according to claim 1, characterized in that: After performing character segmentation on the target characters based on all the character boundaries to obtain a character set including all the target characters and before character matching, the method further includes: Obtaining a preset number of characters corresponding to the preset color combination, and detecting whether the number of all target characters in the character set meets the preset number of characters; If the number of characters in the character set does not meet the preset number of characters, determining that the image to be recognized is not a license plate image; If the number of characters in the character set meets the preset number of characters, it is determined that the image to be recognized is a license plate image.

5. The insurance recommendation method according to claim 1, characterized in that: The performing character matching on the target character in the character set and the preset character in the target character library to obtain a character matching result includes: Determining all screening characters corresponding to the character set from the target character library according to the first target character and the second target character in the character set; Performing character matching on target characters other than the first target character and the second target character in the character set and the screening characters; When there is a matching failure among all the target characters, obtaining a character matching result representing the incorrect character among all the target characters; When the character matching result indicates that all the target characters are matched to the preset characters, a correct character matching result is obtained for all the target characters.

6. The insurance recommendation method according to claim 1, characterized in that: The performing risk identification on the user information and the historical data information to obtain a risk identification result includes: Acquire a risk identification model, and input the user information and the historical data information into the risk identification model; Performing feature extraction on the user information and the historical data information through the feature extraction layer in the risk identification model to obtain information features corresponding to the user information and data features corresponding to the historical data information; The risk prediction layer in the risk identification model performs risk identification on the information features and the data features to obtain a risk identification result.

7. The insurance recommendation method according to claim 1, characterized in that: Before obtaining the preset image recognition model, the method further includes: Acquire a sample data set, wherein the sample data set includes at least one sample training data and a sample label corresponding to each sample training data; Identify and predict all the sample training data using a preset training model to obtain a prediction result corresponding to each of the sample training data; Determining a prediction loss value of a preset training model according to sample labels and prediction results corresponding to the same sample training data; When the predicted loss value reaches a preset convergence condition, the preset training model after convergence is determined as the preset image recognition model.

8. An insurance recommendation device, characterized in that: include: An image preprocessing module is used to obtain an image to be processed and preprocess the image to be processed to obtain an image to be recognized; A character recognition module, used to obtain a preset image recognition model, and perform boundary recognition on the image to be recognized by using the preset image recognition model to obtain a character set including all target characters; A character matching module, used to obtain a target character library, and perform character matching on a target character in the character set and a preset character in the target character library to obtain a character matching result; An information acquisition module, configured to acquire user information and historical data information corresponding to the image to be recognized when the character matching result indicates that all target characters are correct; The risk identification module is used to perform risk identification on the user information and the historical data information, obtain risk identification results, and generate insurance recommendation results based on the risk identification results.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the insurance recommendation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the insurance recommendation method according to any one of claims 1 to 7 is implemented.

Images