CN116882928A

CN116882928A - Commercial tenant subscription qualification auditing method and device and electronic equipment

Info

Publication number: CN116882928A
Application number: CN202310854625.3A
Authority: CN
Inventors: 曾小英; 陈新; 胡圻圻
Original assignee: AlipayCom Co ltd
Current assignee: AlipayCom Co ltd
Priority date: 2023-07-12
Filing date: 2023-07-12
Publication date: 2023-10-13

Abstract

The embodiment of the specification discloses a merchant subscription qualification auditing method, which comprises the following steps: acquiring a qualification material uploaded by a user, wherein the qualification material comprises at least one qualification picture; pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture; generating error prompt information aiming at each qualification picture if the pre-examination evaluation result is unqualified, and feeding back the error prompt information to a user; when the total times of unqualified pre-checking evaluation results of the qualification picture reach a preset threshold, submitting the qualification picture to a manual checking link; and if the pre-checking evaluation result of each qualification picture in the qualification materials is qualified, generating a qualification checking work order, and submitting the qualification checking work order to a qualification checking link. Correspondingly, the invention discloses a merchant subscription qualification auditing device.

Description

Commercial tenant subscription qualification auditing method and device and electronic equipment

Technical Field

The invention relates to the technical field of data processing, in particular to a merchant subscription qualification auditing method, a merchant subscription qualification auditing device and electronic equipment.

Background

When the merchant applies for corresponding qualification to the target organization, the target organization needs to submit the qualification material, and the target organization carries out verification on the authenticity, the integrity and the effectiveness of the qualification material and returns a verification result to the merchant. Currently, this auditing process is typically manually audited. The manual auditing mode has high cost and low efficiency, and is easy to cause risk leakage.

Disclosure of Invention

The invention aims to provide a merchant subscription qualification auditing method, which can inform auditing results in real time after a user uploads a qualification material, improve qualification auditing efficiency, and simultaneously provide a spam manual auditing mechanism to effectively improve subscription experience.

According to the above object, an embodiment of the present disclosure provides a method for checking subscription qualification of a merchant, including:

acquiring a qualification material uploaded by a user, wherein the qualification material comprises at least one qualification picture;

pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture;

generating error prompt information aiming at each qualification picture if the pre-examination evaluation result is unqualified, and feeding back the error prompt information to a user; when the total times of unqualified pre-checking evaluation results of the qualification picture reach a preset threshold, submitting the qualification picture to a manual checking link;

And if the pre-checking evaluation result of each qualification picture in the qualification materials is qualified, generating a qualification checking work order, and submitting the qualification checking work order to a qualification checking link.

The merchant subscription qualification auditing method provided in the embodiment of the specification effectively combines an algorithm and a strategy, evaluates the qualification materials from multiple sides through a pre-auditing evaluation link, feeds back the judgment result of the authenticity of the qualification materials in real time, can improve qualification auditing efficiency, and effectively detects risks in the qualification materials. When the pre-audit evaluation result is unqualified, guiding a user to upload the qualification material again through error prompt feedback, and effectively improving the conversion rate of the pre-audit page and the pre-audit passing rate; and the manual auditing link is used as a spam mechanism, so that good signing experience of the user is ensured.

Further, in some embodiments, the pre-audit process includes picture quality pre-audit; based on a preset pre-checking flow, pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

inputting the qualification picture into a pre-trained picture quality evaluation model to obtain a picture quality evaluation result of the qualification picture;

And determining a pre-checking evaluation result of the qualification picture based on the picture quality evaluation result.

Further, in some embodiments, the pre-review process includes a picture type review; based on a preset pre-checking task flow, pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

inputting the qualification picture into a pre-trained classification model to obtain the prediction type of the qualification picture;

and matching the prediction type with a preset target type, and determining a pre-checking evaluation result of the qualification picture based on a matching result.

Further, in some embodiments, the pre-verification process includes a photo text information verification; based on a preset pre-checking task flow, pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

inputting the qualification picture into a pre-trained classification model, and determining the picture type of the qualification picture;

extracting a text region image in the qualification picture, and performing optical character recognition on the text region image to obtain a text recognition result;

Determining a third party authority based on the picture type;

verifying the text recognition result based on the public data of the third party authority;

and determining a pre-verification evaluation result of the qualification picture based on the verification result.

Further, in some embodiments, the pre-audit flow includes an icon information verification, which specifically includes:

determining an icon identification template corresponding to the picture type based on the picture type of the qualification picture; the icon identification template comprises at least one target candidate frame, and the position of the target candidate frame is the position of an icon to be acquired;

determining an icon area image in the qualification picture based on at least one target candidate frame of the icon recognition template;

and matching the icon area image with a preset icon template, and determining a pre-checking evaluation result of the qualification picture based on a matching result.

Further, in some embodiments, the qualification audit worksheet includes the at least one qualification picture and at least one target field corresponding to the at least one qualification picture; the generating qualification audit worksheet specifically comprises the following steps:

Extracting a text region image of each qualification picture aiming at the qualification picture with qualified pre-examination evaluation result;

performing optical character recognition on the text region image to obtain a text recognition result;

extracting keywords from the text recognition result, and matching the extracted keywords with preset options of a target field in the qualification audit worksheet;

generating at least one selectable item of the target field based on the matching result, and recommending the selectable item to a user;

and determining the target field based on the selection result of the user on the selectable item to obtain the qualification audit work order.

The target field in the qualification auditing worksheet is filled through picture content identification and intelligent recommendation, so that the information acquisition quality based on the qualification materials can be effectively improved, the time cost for filling the materials is shortened, the steps of active operation of a user are simplified, and the subscription experience of the user is optimized.

The invention further aims to provide a merchant subscription qualification auditing device which can audit the qualification materials uploaded by the user in real time and inform auditing results, thereby improving qualification auditing efficiency, and simultaneously providing a spam manual auditing mechanism and effectively improving subscription experience.

According to the above object, an embodiment of the present disclosure provides a merchant subscription qualification auditing apparatus, including:

the data acquisition module is configured to acquire qualification materials uploaded by a user, wherein the qualification materials comprise at least one qualification picture;

the pre-checking module is configured to pre-check the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture;

the error prompt module is configured to generate error prompt information aiming at the qualification picture when the pre-checking evaluation result of the qualification picture is unqualified, and feed the error prompt information back to a user; when the total times of unqualified pre-checking evaluation results of the qualification pictures reach a preset threshold, submitting the qualification pictures to a manual checking link;

and the qualification audit work order generation module is configured to generate a qualification audit work order when the pre-audit evaluation result of each qualification picture in the qualification material is qualified, and submit the qualification audit work order to a qualification audit link.

Further, in some embodiments, the pre-audit module includes a picture quality pre-audit module; the picture quality pre-auditing module is specifically used for inputting the qualification picture into a pre-trained picture quality evaluation model to obtain a picture quality evaluation result of the qualification picture; and determining a pre-checking evaluation result of the qualification picture based on the picture quality evaluation result.

Further, in some embodiments, the pre-audit module includes a picture type audit module; the picture type auditing module is specifically used for inputting the qualification picture into a pre-trained classification model to obtain the prediction type of the qualification picture; and matching the prediction type with a preset target type, and determining a pre-checking evaluation result of the qualification picture based on a matching result.

Further, in some embodiments, the pre-verification module includes a picture text information verification module; the picture text information verification module is specifically used for inputting the qualification picture into a pre-trained classification model and determining the picture type of the qualification picture; extracting a text region image in the qualification picture, and performing optical character recognition on the text region image to obtain a text recognition result; determining a third party authority based on the picture type; and verifying the text recognition result based on the public data of the third-party authority, and determining a pre-verification evaluation result of the qualification picture based on the verification result.

Further, in some embodiments, the pre-audit module includes an icon information audit module; the icon information verification module is specifically used for inputting the qualification picture into a pre-trained classification model and determining the picture type of the qualification picture; determining an icon identification template corresponding to the picture type based on the picture type of the qualification picture, wherein the icon identification template comprises at least one target candidate frame, and the position of the target candidate frame is the position of an icon to be acquired; determining an icon area image in the qualification picture based on at least one target candidate frame of the icon recognition template; and matching the icon area image with a preset icon template, and determining a pre-checking evaluation result of the qualification picture based on a matching result.

Further, in some embodiments, the qualification audit worksheet generating module is specifically configured to extract, for each qualified picture that is qualified as a result of pre-audit evaluation, a text area image of the qualified picture; performing optical character recognition on the text region image to obtain a text recognition result; extracting keywords from the text recognition result, and matching the extracted keywords with preset options of a target field in the qualification audit worksheet; generating at least one selectable item of the target field based on the matching result, and recommending the selectable item to a user; and determining the target field based on the selection result of the user on the selectable item to obtain the qualification audit work order.

The embodiments of the present disclosure also provide a computer readable storage medium having a computer program stored thereon, which when executed by a processor implements the merchant subscription qualification auditing method described above.

The embodiment of the specification also provides an electronic device, including:

one or more processors; and

and a memory associated with the one or more processors, the memory for storing program instructions that, when read for execution by the one or more processors, perform the steps of the merchant subscription qualification audit method described above.

The merchant subscription qualification auditing method disclosed by the embodiment of the specification has the beneficial effects that the algorithm is effectively combined with the strategy, the qualification materials are evaluated by utilizing the universal intelligent algorithm from the picture quality, the picture type, the picture text information, the icon or any suitable combination of the picture quality, the picture type, the picture text information and the icon in the pre-auditing evaluation link, the judgment result of the authenticity of the qualification materials is fed back in real time, the qualification auditing efficiency can be improved, and the risks existing in the qualification materials can be effectively detected. When the pre-audit evaluation result is unqualified, guiding a user to upload the qualification material again through error prompt feedback, and effectively improving the conversion rate of the pre-audit page and the pre-audit passing rate; and the manual auditing link is used as a spam mechanism, so that good signing experience of the user is ensured.

In addition, the target field in the qualification audit work order can be filled through picture content identification and intelligent recommendation, so that the information acquisition quality based on the qualification materials can be effectively improved, the time cost for filling the materials is shortened, and meanwhile, the step of active operation of a user is simplified.

The merchant subscription qualification auditing device disclosed by the embodiment of the specification has the beneficial effects as above.

Drawings

In order to more clearly illustrate the embodiments of the present description or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are some embodiments of the present description, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 schematically illustrates a flow chart of a merchant subscription qualification auditing method according to an embodiment of the present disclosure, in an implementation manner.

Fig. 2 schematically illustrates an application of the merchant subscription qualification auditing method according to the embodiments of the present disclosure in a specific implementation manner.

FIG. 3 is a schematic flow chart of a method for verifying subscription qualification of a merchant according to an embodiment of the present disclosure, which is executed when the method is applied to pre-verification of a business license in a specific embodiment.

Fig. 4 is a schematic flow chart of a merchant subscription qualification auditing method according to an embodiment of the present disclosure, which is executed when the method is applied to store sign pre-auditing in a specific implementation manner.

Fig. 5 is a block diagram schematically illustrating a structure of a merchant subscription qualification auditing apparatus according to an embodiment of the present disclosure.

Fig. 6 exemplarily shows a block diagram of an electronic device provided in an embodiment of the present specification.

Detailed Description

It is first noted that the terminology used in the embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. In addition, the embodiments and features in the embodiments in the present specification may be combined with each other without conflict. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.

With the development of social economy, more and more merchants emerge in the market, and many institutions have a more stable resource acquisition channel by establishing a cooperative relationship with merchant subscription. In order to standardize the trade market and ensure the compliance of commodity trade flow and the quality of commodity, the merchant needs to provide materials which can prove the operational compliance and the authenticity of the operation according to the admission standard of the target institution when signing up, and the materials are called as qualification materials, including but not limited to business licenses, store signboards, identity cards and in-store scenes. According to regulatory requirements and wind control requirements, the target institution needs to conduct authenticity, integrity and validity checks on the merchant qualification, including but not limited to whether the merchant qualification is counterfeit or faked to others, whether the relevant certificate, such as a business license, is within a usable period, and whether the qualification material provided by the merchant is complete.

The traditional qualification auditing is realized by adopting a manual auditing mode, and in order to further improve auditing efficiency, reduce auditing cost and enhance the capability of avoiding risks, the specification proposes an intelligent auditing scheme, namely, automatic and systematic qualification auditing is realized through a deep learning algorithm, and the algorithm is tightly combined with a strategy.

When the method is implemented, the following ideas are adopted in the specification: after the commercial tenant submits the qualification material, the interactive pre-audit is carried out, the material information extraction and the content audit are carried out in real time, then the commercial tenant is synchronously informed of the qualification pre-audit result, if the pre-audit is not passed, the error information is displayed on the audit page in real time, and the commercial tenant is reminded to re-submit. When the same qualification picture submitted by the merchant is rejected for a plurality of times, the page pops up an interactive text, prompts the merchant to supplement information of the data, and then goes to customer service for manual review.

The method and apparatus for checking subscription qualification of merchants according to embodiments of the present disclosure will be described in further detail below with reference to the accompanying drawings and specific embodiments of the present disclosure, but the detailed description is not meant to limit the embodiments of the present disclosure.

In one embodiment of the present specification, a merchant subscription qualification auditing method is provided. Fig. 1 schematically illustrates a flow chart of a merchant subscription qualification auditing method according to an embodiment of the present disclosure, in an implementation manner.

As shown in fig. 1, includes:

100: and acquiring qualification materials uploaded by the user, wherein the qualification materials comprise at least one qualification picture.

The user in this embodiment may be a merchant to sign a contract, and the qualification material may be a qualification certificate such as a business license picture, a shop sign picture, and an identity card picture of the merchant, where the merchant obtains an electronic version qualification picture through an image acquisition device such as a camera, a smart phone, and a scanner, and uploads the electronic version qualification picture to the auditing system according to the prompt of the qualification auditing page.

102: and pre-checking at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture.

The pre-audit process includes verifying the integrity, authenticity and validity of the qualification picture. On the one hand, the integrity pre-checking of the qualification picture, such as the picture quality pre-checking, can determine the pre-checking evaluation result according to whether the qualification picture is fuzzy, overexposed and whether the qualification picture has shielding and incomplete contents. On the other hand, the authenticity pre-check of the qualification picture can be carried out by identifying text content or local image content, extracting specific qualification information and comparing the specific qualification information with merchant qualification information recorded in a third-party authority, so as to judge whether the qualification is counterfeit, for example, the authenticity pre-check evaluation result can be determined based on the comparison result by identifying a official seal in a business license picture submitted by a merchant, analyzing the identification result and comparing the identification result with a standard provided by a business management department corresponding to the issued business license. In still another aspect, the validity of the qualification picture may also be evaluated by identifying text content in the picture, for example, by identifying text in a business license picture submitted by a merchant, and determining whether the business license is in a validity period, whether the business license is logged out or revoked, whether legal persons are consistent, etc. based on text information, so as to determine whether the business license is valid, and obtain a pre-review evaluation result.

In some embodiments, the pre-audit flow includes picture quality pre-audit; based on a preset pre-checking flow, pre-checking at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

The picture quality pre-checking can evaluate whether the picture has quality problems such as incomplete, fuzzy or overexposure, and the corresponding picture quality evaluation result can be displayed in the form of picture quality evaluation scores. Before a user uploads a qualification picture, the qualification material itself may have defects due to improper storage, and the conditions of finger shielding, equipment shaking, too strong or too weak light and the like may also exist in the process of collecting the qualification material; after submitting the qualification picture, the picture may be distorted due to noise in the processes of compressing, transmitting, displaying, etc., which finally results in the problems of incomplete, fuzzy or overexposure of the submitted qualification picture.

In some embodiments, at least one qualification picture submitted by a user is input into a pre-trained picture quality evaluation model, in which feature extraction is performed on the qualification picture for features of interest in the picture quality evaluation, for example, whether the picture content is incomplete, occluded (slight incomplete, moderate incomplete, severe incomplete), or the blurring degree of the picture (slight blurring, moderate blurring, severe blurring), or the light environment in which the picture is located (normal light, strong light, weak light, where strong light may cause overexposure of the picture), each feature being represented by a specific numerical value. The picture quality evaluation model analyzes and classifies the extracted qualification picture features to finally obtain quality evaluation scores corresponding to the qualification pictures, and the quality evaluation scores are higher as the picture quality is better. Comparing the obtained picture quality evaluation result with a preset picture quality evaluation threshold, and if the picture quality evaluation result is higher than the threshold, qualifying the picture quality evaluation and qualifying the pre-checking evaluation; otherwise, the pre-checking is unqualified.

The picture quality evaluation model can be obtained by training the following method:

determining a sample picture, and taking a quality evaluation score corresponding to the sample picture as a classification label;

extracting features of the sample picture to obtain the features of the sample picture;

inputting the characteristics of the sample picture into a pre-built picture quality evaluation model, and training the picture quality evaluation model with the aim of minimizing the quality evaluation loss.

The quality evaluation loss may be specifically set as a difference between a quality evaluation result of the picture quality evaluation model for the sample picture and a classification label of the sample picture.

In some more specific embodiments, the picture quality evaluation model may be constructed based on a CNN network model, and of course, in other embodiments, a support vector machine may also be used to construct the picture quality evaluation model. The non-reference quality evaluation method does not need high-quality images as references, directly uses image characteristics as quality evaluation standards, and has strong flexibility and evaluation efficiency.

In the picture processing process, the quality of the picture plays a decisive role in the sufficiency and accuracy of information to be acquired, and in order to relieve distortion phenomena generated in the processes of acquisition, compression, transmission, display and the like of the picture, the quality of the picture is pre-checked and filtered through the picture quality, so that the quality of the picture does not reach the qualification picture required by further pre-check, and the qualification pre-check efficiency can be improved.

In some embodiments, the pre-review process includes a picture type review; based on a preset pre-checking task flow, pre-checking at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

and matching the prediction type with a preset target type, and determining a pre-checking evaluation result of the qualification picture based on the matching result.

Types of qualification pictures include, but are not limited to, business licenses, store signs, identity cards, and title cards. In the pre-audit page, the submitted positions of the qualification pictures of different types are also different, and the picture type audit requires that the qualification pictures of corresponding types are uploaded to the correct positions, otherwise, the pre-audit is unqualified. For example, uploading a qualification picture at a position where a business license should be submitted, wherein the preset target type is the business license, inputting the qualification picture into a pre-trained classification model to obtain a prediction type of the qualification picture, if the uploaded qualification picture is a shop sign, the prediction type is the shop sign, and if the prediction type is failed to be matched with the target type, the pre-examination evaluation result is unqualified.

In some more specific embodiments, the classification model may be built based on a CNN network model, although in other embodiments, a support vector machine may be used.

The classification model can be obtained by training the following method:

determining a sample picture, and taking the type corresponding to the sample picture as a classification label;

and inputting the characteristics of the sample picture into a pre-built classification model, and training the classification model with the aim of minimizing the classification loss.

Wherein the classification loss may specifically be set as a difference between the type of the sample picture obtained based on the classification model and the true type of the sample picture.

The picture type auditing is a key step of further extracting and analyzing the content of the qualification pictures, so that different auditing methods are adopted for different types of qualification pictures in the subsequent pre-auditing process, and the pre-auditing passing rate is improved.

In some embodiments, the pre-verification process includes picture text information verification; based on a preset pre-checking task flow, pre-checking at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

determining a third party authority based on the picture type;

The method for verifying the text recognition result in the picture is different according to different qualification picture types. Wherein the optical character recognition includes OCR character recognition, and text content can be recognized from the image. The third party authority may be used to represent an official entity that records and manages user information, such as an industry and commerce management, e.g., a lower business license, or a resident identification card querying system that may be used to query for identification card information.

Taking business license as an example, OCR text recognition is carried out on the text region image, and the obtained text recognition result comprises information such as merchant name, merchant type, merchant address, legal person, business deadline, operation range, registration unit and the like. A third party authority, typically an industrial and commercial administration corresponding to the merchant location, that issues and manages the business license may then be determined based on the registration entity information. And inquiring business license related information recorded in a business management department system according to the name of the merchant, and comparing the business license related information with text information such as merchant addresses, legal persons, business deadlines and the like identified in business license pictures submitted by the merchant to obtain a verification result. Wherein, whether the business license is valid is determined by checking whether the business deadline expires, whether the business license is logged off or revoked, and whether legal persons agree. If business license information recorded in the business management department system is verified to be consistent with information identified in business license pictures submitted from merchants, and the business license is valid according to the date, the verification result is qualified, and the pre-verification evaluation is also qualified; if the business license information recorded in the business management department system is inconsistent with the information identified in the business license picture submitted by the business, or the business license is verified to be invalid, the verification is failed, and the pre-verification evaluation is failed.

The authenticity and effectiveness of the qualification picture can be accurately identified by carrying out specific picture text information verification on the submitted qualification picture, the forged qualification material is identified through authoritative comparison, and the invalid qualification material is filtered.

In some embodiments, the pre-audit flow includes an icon information verification, which specifically includes:

and matching the icon area image with a preset icon template, and determining a pre-checking evaluation result of the qualification picture based on the matching result.

The icon to be verified is a certificate capable of indicating the validity of the qualification picture, such as a official seal in a business license or a shop logo in a shop sign. For different types of qualification pictures, the positions and the sizes of the icons are also different, for example, business licenses can be divided into enterprise business licenses and individual business licenses, and corresponding registration unit official stamps are positioned at different positions; the position of the official seal is also different according to whether the business license is a vertical version or a horizontal version. Therefore, corresponding icon recognition templates are preset based on different types of qualification pictures, and positions of icons to be recognized are marked by target candidate frames, so that icon information verification is performed in a targeted mode.

In some more specific embodiments, feature extraction is performed on the icon region image after the target icon is acquired, feature matching is performed on the icon region image and the icon features in the preset icon template, specifically, the feature matching threshold is preset and compared with a similarity calculation result, if the similarity calculation result is higher than the threshold, the matching is successful, otherwise, the matching is failed; and when the icon area image of the qualification picture is successfully matched with the preset icon template, the qualification picture pre-examination evaluation result is qualified.

Icon information verification can verify the authenticity and the validity of the qualification picture in a more detailed level, so that the pre-verification result is more accurate and reliable.

It should be noted that, the methods described in the pre-audit process in the foregoing embodiments may be performed alone or in combination in a suitable manner, and the corresponding methods are not necessarily performed in the order shown and described in the present specification. For example, the pre-checking of the qualification picture can be performed from three aspects of picture quality pre-checking, picture type checking and picture text information checking, preferably, the pre-checking evaluation result can also be determined based on the results of picture quality pre-checking, picture type checking, picture text information checking and icon information checking, and when all checking flows are qualified, the pre-checking evaluation is qualified. Based on a plurality of image-text multi-mode algorithms, the more comprehensive and careful the pre-examination flow, the more accurate and reliable the pre-examination evaluation result.

104: for each qualification picture, if the pre-checking evaluation result is unqualified, generating error prompt information, and feeding back the error prompt information to a user; and submitting the qualification picture to a manual auditing link when the total times of unqualified pre-auditing evaluation results of the qualification picture reach a preset threshold.

The pre-checking evaluation of each qualification picture needs to go through one or more aspects of pre-checking flow, such as picture quality pre-checking, picture type checking, picture text information checking or icon information checking, when one aspect of checking fails, clear and comprehensive error prompt information is displayed in real time in a page to describe the reason of rejection, a user is guided to submit the qualification materials again after correcting errors, interactive pre-checking is effectively realized, and the conversion rate of the checking page and the passing rate of the qualification materials after submitting are improved.

For the same qualification picture, when the accumulated reject times reach a preset threshold, in order to avoid the influence of algorithm misjudgment on user experience, a manual auditing entrance is provided for secondary verification, and user subscription experience is effectively improved.

In some embodiments, the preset threshold may be set to three times, when the pre-checking evaluation result of the same qualification picture is that the cumulative number of times of failure reaches three or more, whether the data is correct or not is confirmed to the user through the popup window, if the user does not recognize the checking result, the supplementary explanation in the popup window may be filled in through text and/or picture form to further prove the authenticity and effectiveness of the qualification material, and then the qualification material is submitted to the manual customer service department for secondary checking.

Taking a shop signboard as an example, when a user wants to upload a shop signboard picture shot by the user to an audit page, the user submits a business license picture to the position of the shop signboard by mistake during the first uploading, and when the picture type audit is carried out, the picture type (business license) predicted by a classification model cannot be successfully matched with the target type (shop signboard), if the picture type audit is not passed, the page can pop a window to prompt the user to upload the correct qualification picture type; the user uploads the correct shop sign picture at the correct position for the second time, but because the picture is overexposed and fails to pass the picture quality pre-checking, the page prompts the user to upload a clearer qualification picture through a popup window; and when the icon information verification is carried out, the user re-shoots the clear shop signboards, and effective shop logo is not identified to cause the verification failure due to the lack of a proper icon identification template, at the moment, the accumulated reject times reach a preset threshold value, the verification page prompts the popup window to the user, and if the user confirms that the submitted qualification pictures are correct, the user can carry out supplementary explanation through texts or pictures and then carry out manual customer service verification.

106: if the pre-checking evaluation result of each qualification picture in the qualification material is qualified, generating a qualification checking work order, and submitting the qualification checking work order to a qualification checking link.

The qualification audit worksheet reflects text information extracted from the qualification material that contains basic information about the user's qualification, such as the business' business category or the name of the store.

In some embodiments, the qualification audit worksheet includes at least one qualification picture and at least one target field corresponding to the at least one qualification picture; generating a qualification audit work order, which specifically comprises the following steps:

extracting keywords from the text recognition result, and matching the extracted keywords with preset options of target fields in the qualification audit worksheets;

generating at least one selectable item of the target field based on the matching result, and recommending the selectable item to the user;

and determining a target field based on a selection result of the user on the selectable items to obtain the qualification audit work order.

Basic information related to the qualification is extracted to be matched with target fields by carrying out real-time text recognition on the content of the qualification picture, wherein the target fields comprise, but are not limited to, business categories, store names or contact names. Recommending the matching result to the user according to the confidence score, and automatically filling in a target field in the qualification audit work order based on the matching result. Taking business category filling as an example, aiming at business license pictures with qualified pre-auditing evaluation results, text information in the pictures is extracted through OCR recognition, keywords related to the business categories, such as business names and business ranges, are filtered out from the text recognition results, and are matched with business category names stored in a database at the back end of a pre-auditing system, so that at least one successfully matched business category name, such as retail or wholesale, is obtained, and is automatically filled into a qualification auditing work order.

The target field in the qualification audit work order is rich in content, so that the selectable options of the target field are numerous, and related keywords extracted from text recognition content are automatically recommended in real time, so that the random selection of a user can be effectively avoided, the fields which are required to be actively input by the user are reduced, the signing cost is reduced, and the information acquisition quality is improved.

The merchant subscription qualification auditing method described in the embodiments of the present specification will be described in further detail below in connection with a more specific embodiment, but the detailed description is not limiting of the embodiments of the present specification.

In a more specific embodiment, the user needs to submit business information, store information, and contact information on a pre-audit page for pre-audit by uploading business license pictures and store sign pictures, as shown in FIG. 2.

After uploading original pictures of business license, a user firstly adopts a picture classification technology to carry out picture recognition so as to confirm whether the picture type is correct, and then uses a picture OCR technology to recognize text information in the picture; and then, based on the text recognition result, carrying out information verification with an authoritative data source provided by a corresponding third-party authority to judge the authenticity and validity of the business license information, including whether the business license information is forged, expired, legal and the like. And after the pre-examination of the business license picture passes, automatically recommending the business category field based on the text recognition result after the examination, thereby completing the pre-examination of the business information.

The more specific interactive pre-checking flow of business license pictures is shown in fig. 3, the pre-checking system firstly carries out picture type identification and checking on business license pictures submitted by users, a pre-trained classifier is adopted to judge whether the types of the submitted pictures are business license or not, if not, the checking is not qualified, and the users are prompted to re-submit in real time through a popup window; if the submitted business license is confirmed, further performing qualification validity judgment, namely judging whether the business license is truly valid or not from aspects of whether the business license is out of date, whether the business license is cancelled or revoked, whether legal persons are consistent, whether the business license is provided with a registration unit red seal and the like through picture text information verification and icon information verification, and if the information verification fails or the business license is judged to be invalid, prompting a user that a specific problem is located by a popup window, and guiding the user to resubmit; and if not, the business license is pre-audited to be qualified, and finally, information recommendation is carried out based on the keywords extracted from the picture text information, the industry category is automatically filled in, and the business information pre-audit is completed.

Then, uploading shop signboards by the user, firstly, utilizing a general picture identification technology, eliminating fake pictures obtained by means of non-real shooting, embezzlement and the like through fake detection, and filtering pictures with quality which does not meet the requirements, such as fuzzy, incomplete and overexposed shop signboards through quality detection; then adopting a picture classification technology to carry out special picture identification so as to confirm whether the picture type is correct; and then, recognizing text information in the picture by utilizing a picture OCR technology, extracting keywords related to store information from the text information and recommending the keywords to a user, thereby automatically filling in a store name, and then, supplementing and filling in a store address by the user to complete the pre-auditing of the store information.

Aiming at the more specific interactive pre-checking flow of the shop signboards, as shown in fig. 4, the pre-checking system firstly carries out fake detection on the shop signboards submitted by the user through icon information checking and other methods so as to identify the condition of stealing the pictures through means of screen capturing, screen shooting, PS and the like, for example, whether the pictures are forged or not is judged through watermark detection, whether the pictures are flipped from a computer is judged through mole pattern recognition, whether the pictures are flipped from a mobile phone is judged through frame recognition, and if the forged shop signboards are identified, the popup window prompts the user to upload real pictures again; otherwise, performing quality detection on the shop signboards by using a pre-trained picture quality evaluation model, if the picture quality problems of blurring, overexposure and the like seriously affecting the picture content are detected, pre-checking is unqualified, and a popup window prompts a user to upload a clearer picture; if the quality detection is passed, acquiring shop name information in the picture through OCR recognition, automatically filling in the shop name in the pre-checking page based on confidence ranking, and then filling in the shop address by the user to complete the pre-checking of the shop information.

The final contact person information can be filled in by the user by self, or the method can be used for pre-checking after uploading the identity card photo, and related information can be filled in automatically after checking is qualified, so that all pre-checking processes are completed, the page is submitted to a formal qualification checking link, the workload in the formal checking process can be greatly reduced, the checking efficiency is improved, the qualification checking effect is optimized, and the falsification and theft risks of the qualification information are avoided.

When the pre-checking evaluation result of the same qualification picture is unqualified accumulated times, namely the accumulated times are refused for three times or more, entering a human checking spam mechanism link, prompting whether the data is correct or not to a user through a popup window, if the user does not recognize the checking result, filling in a supplementary explanation in the popup window through a text and picture form to further prove the authenticity and the effectiveness of the qualification material, and then submitting the qualification material to a human customer service department for secondary checking.

The merchant subscription qualification auditing method provided by the embodiment of the specification is based on an image-text multi-mode algorithm, and performs qualification pre-auditing in a qualification picture uploading link to realize real-time interactive pre-auditing, so that the qualification auditing efficiency is improved, and the auditing result is more accurate; in addition, the added spam manual auditing mechanism is beneficial to avoiding errors of algorithms, enhancing system robustness and optimizing user subscription experience.

In another embodiment of the present disclosure, a merchant subscription qualification auditing apparatus is provided. Fig. 5 is a block diagram schematically illustrating a structure of a merchant subscription qualification auditing apparatus according to an embodiment of the present disclosure.

As shown in fig. 5, includes:

a data acquisition module 20 configured to acquire a qualification material uploaded by a user, the qualification material including at least one qualification picture;

a pre-auditing module 22 configured to pre-audit at least one qualification picture to obtain a pre-audit evaluation result of each qualification picture;

the error prompt module 24 is configured to generate error prompt information aiming at the qualification picture when the pre-checking evaluation result of the qualification picture is unqualified, and feed the error prompt information back to the user; when the total times of unqualified pre-checking evaluation results of the qualification pictures reach a preset threshold, submitting the qualification pictures to a manual checking link;

the qualification audit work order generation module 26 is configured to generate a qualification audit work order when the pre-audit evaluation result of each qualification picture in the qualification material is qualified, and submit the qualification audit work order to the qualification audit link.

The user in the data acquisition module may be a merchant to be signed, the qualification material may be a qualification certificate such as a business license picture, a shop sign picture, an identity card picture and the like of the merchant, the merchant acquires the electronic version qualification picture through image acquisition equipment such as a camera, a smart phone, a scanner and the like, and the electronic version qualification picture is uploaded to the auditing system according to the prompt of the qualification auditing page.

The pre-auditing module is used for verifying the integrity, the authenticity and the validity of the qualification picture. On the one hand, the integrity pre-checking of the qualification picture, such as the picture quality pre-checking, can determine the pre-checking evaluation result according to whether the qualification picture is fuzzy, overexposed and whether the qualification picture has shielding and incomplete contents. On the other hand, the authenticity pre-check of the qualification picture can be carried out by identifying text content or local image content, extracting specific qualification information and comparing the specific qualification information with merchant qualification information recorded in a third-party authority, so as to judge whether the qualification is counterfeit, for example, the authenticity pre-check evaluation result can be determined based on the comparison result by identifying a official seal in a business license picture submitted by a merchant, analyzing the identification result and comparing the identification result with a standard provided by a business management department corresponding to the issued business license. In still another aspect, the validity of the qualification picture may also be evaluated by identifying text content in the picture, for example, by identifying text in a business license picture submitted by a merchant, and determining whether the business license is in a validity period, whether the business license is logged out or revoked, whether legal persons are consistent, etc. based on text information, so as to determine whether the business license is valid, and obtain a pre-review evaluation result.

In some embodiments, the pre-audit module includes a picture quality pre-audit module; the picture quality pre-checking module is specifically used for inputting a qualification picture into a pre-trained picture quality evaluation model to obtain a picture quality evaluation result of the qualification picture; and determining a pre-checking evaluation result of the qualification picture based on the picture quality evaluation result.

The picture quality pre-auditing module can evaluate whether the picture has quality problems such as incomplete, fuzzy or overexposure, and the corresponding picture quality evaluation result can be displayed in the form of picture quality evaluation scores. Before a user uploads a qualification picture, the qualification material itself may have defects due to improper storage, and the conditions of finger shielding, equipment shaking, too strong or too weak light and the like may also exist in the process of collecting the qualification material; after submitting the qualification picture, the picture may be distorted due to noise in the processes of compressing, transmitting, displaying, etc., which finally results in the problems of incomplete, fuzzy or overexposure of the submitted qualification picture.

In some embodiments, the picture quality pre-review module inputs at least one qualification picture submitted by a user into a pre-trained picture quality evaluation model in which the qualification picture is first feature extracted for features of interest in the picture quality evaluation, e.g., whether the picture content is incomplete, occluded (slight incomplete, moderate incomplete, severe incomplete), or the degree of blurring of the picture (slight blurring, moderate blurring, severe blurring), or the light environment in which the picture is located (normal light, intense light, weak light, where intense light may cause the picture to be overexposed), each feature being represented by a specific value. The picture quality evaluation module analyzes and classifies the extracted qualification picture features to finally obtain quality evaluation scores corresponding to the qualification pictures, wherein the quality evaluation scores are higher as the picture quality is better. The picture quality pre-checking module compares the obtained picture quality evaluation result with a preset picture quality evaluation threshold, and if the picture quality evaluation result is higher than the threshold, the picture quality evaluation is qualified, and the pre-checking evaluation is also qualified; otherwise, the pre-checking is unqualified.

In the picture processing process, the quality of the picture plays a decisive role in the sufficiency and accuracy of information to be acquired, and in order to relieve distortion phenomena generated in the processes of acquiring, compressing, transmitting, displaying and the like of the picture, the picture quality pre-checking module pre-checks and filters the picture quality through the picture quality to obtain the qualification picture which does not reach the further pre-checking requirement, so that the qualification pre-checking efficiency can be improved.

In some embodiments, the pre-audit module includes a picture type audit module; the picture type auditing module is specifically used for inputting the qualification picture into a pre-trained classification model to obtain the prediction type of the qualification picture; and matching the prediction type with a preset target type, and determining a pre-checking evaluation result of the qualification picture based on the matching result.

Types of qualification pictures include, but are not limited to, business licenses, store signs, identity cards, and title cards. In the pre-audit page, the submitted positions of different types of qualification pictures are also different, and the picture type audit module requires the qualification pictures of the corresponding types to be uploaded to the correct positions, otherwise, the pre-audit is unqualified. For example, uploading a qualification picture at a position where a business license should be submitted, wherein the preset target type is the business license, inputting the qualification picture into a pre-trained classification model to obtain a prediction type of the qualification picture, if the uploaded qualification picture is a shop sign, the prediction type is the shop sign, and if the prediction type is failed to be matched with the target type, the pre-examination evaluation result is unqualified.

The classification model can be obtained by training the following method:

The picture type auditing module is a key module for further extracting and analyzing the content of the qualification pictures so as to adopt different auditing methods aiming at different types of qualification pictures in the subsequent pre-auditing process and improve the pre-auditing passing rate.

In some embodiments, the pre-audit module includes a picture text information audit module; the picture text information verification module is specifically used for inputting the qualification picture into a pre-trained classification model and determining the picture type of the qualification picture; extracting a text region image in the qualification picture, and performing optical character recognition on the text region image to obtain a text recognition result; determining a third party authority based on the picture type; and verifying the text recognition result based on the public data of the third-party authority, and determining a pre-verification evaluation result of the qualification picture based on the verification result.

Taking business license as an example, the picture text information verification module carries out OCR text recognition on the text region image of the picture text information verification module, and the obtained text recognition result comprises information such as merchant name, merchant type, merchant address, legal person, business deadline, operation range, registration unit and the like. A third party authority, typically an industrial and commercial administration corresponding to the merchant location, that issues and manages the business license may then be determined based on the registration entity information. And the picture text information verification module queries business license related information recorded in the business management department system according to the name of the business and compares the business license related information with text information such as business addresses, legal persons, business deadlines and the like identified in business license pictures submitted by the business to obtain verification results. The picture text information verification module mainly determines whether the business license is valid or not through verifying whether the business deadline is expired, whether the business license is logged off or is revoked or whether legal persons are consistent. If business license information recorded in the business management department system is verified to be consistent with information identified in business license pictures submitted from merchants, and the business license is valid according to the date, the verification result is qualified, and the pre-verification evaluation is also qualified; if the business license information recorded in the business management department system is inconsistent with the information identified in the business license picture submitted by the business, or the business license is verified to be invalid, the verification is failed, and the pre-verification evaluation is failed.

In some embodiments, the pre-audit module includes an icon information audit module; the icon information verification module is specifically used for inputting the qualification picture into a pre-trained classification model and determining the picture type of the qualification picture; determining an icon identification template corresponding to the picture type based on the picture type of the qualification picture, wherein the icon identification template comprises at least one target candidate frame, and the position of the target candidate frame is the position of an icon to be acquired; determining an icon area image in the qualification picture based on at least one target candidate frame of the icon recognition template; and matching the icon area image with a preset icon template, and determining a pre-checking evaluation result of the qualification picture based on the matching result.

The icon to be verified is a certificate capable of indicating the validity of the qualification picture, such as a official seal in a business license or a shop logo in a shop sign. For different types of qualification pictures, the positions and the sizes of the icons are also different, for example, business licenses can be divided into enterprise business licenses and individual business licenses, and corresponding registration unit official stamps are positioned at different positions; the position of the official seal is also different according to whether the business license is a vertical version or a horizontal version. Therefore, the icon information verification module presets corresponding icon recognition templates based on different types of qualification pictures, and marks the positions of the icons to be recognized by the target candidate frames, so that the icon information verification is performed in a targeted manner.

In some more specific embodiments, the icon information verification module performs feature extraction on the icon region image after collecting the target icon, performs feature matching with the icon features in the preset icon template, specifically may be set to calculate the similarity between the icon features in the qualification picture and the icon features in the preset icon template, preset a feature matching threshold and compare with the similarity calculation result, if the similarity calculation result is higher than the threshold, the matching is successful, otherwise the matching is failed; and when the icon area image of the qualification picture is successfully matched with the preset icon template, the qualification picture pre-examination evaluation result is qualified.

It should be noted that the pre-audit modules in the above embodiments may be executed individually or in combination in a suitable manner, and are not necessarily executed in the order illustrated and described in the present specification. For example, the pre-checking module may include three modules, namely, a picture quality pre-checking module, a picture type checking module and a picture text information checking module, and preferably, the pre-checking module may also include all modules, namely, a picture quality pre-checking module, a picture type checking module, a picture text information checking module and an icon information checking module, and when the checking results of all modules are qualified, the pre-checking evaluation is qualified. Based on a plurality of image-text multi-mode algorithms, the more comprehensive and careful the pre-examination flow, the more accurate and reliable the pre-examination evaluation result.

The pre-checking evaluation of each qualification picture needs to pass through one or more aspects of pre-checking modules, when a certain checking module fails, the error prompt module displays clear and comprehensive error prompt information in real time in the page to describe the reject reason, the user is guided to submit the qualification materials again after correcting the error, the interactive pre-checking is effectively realized, and the conversion rate of the checking page and the passing rate of the qualification materials after submitting are improved.

For the same qualification picture, when the accumulated reject times reach a preset threshold, in order to avoid the influence of algorithm misjudgment on user experience, the error prompt module provides a manual checking entrance for secondary checking, and user subscription experience is effectively improved.

In some embodiments, the preset threshold may be set to three times, when the pre-checking evaluation result of the same qualification picture is that the cumulative number of times of failure reaches three or more, the error prompting module prompts to confirm whether the data is correct to the user through the popup window, if the user does not recognize the checking result, the supplementary explanation in the popup window may be filled in through text and/or picture form to further prove the authenticity and effectiveness of the qualification material, and then submitted to the manual customer service department for secondary checking.

In some embodiments, the qualification audit worksheet generating module is specifically configured to extract, for each qualified picture that is qualified by the pre-audit evaluation result, a text area image of the qualified picture; performing optical character recognition on the text region image to obtain a text recognition result; extracting keywords from the text recognition result, and matching the extracted keywords with preset options of target fields in the qualification audit worksheets; generating at least one selectable item of the target field based on the matching result, and recommending the selectable item to the user; and determining a target field based on a selection result of the user on the selectable items to obtain the qualification audit work order.

The qualification auditing work order generation module extracts basic information related to the qualification and matches the basic information with target fields by carrying out real-time text recognition on the content of the qualification picture, wherein the target fields comprise, but are not limited to, business categories, store names or contact names. Recommending the matching result to the user according to the confidence score, and automatically filling in a target field in the qualification audit work order based on the matching result. Taking business category filling as an example, aiming at business license pictures with qualified pre-auditing evaluation results, text information in the pictures is extracted through OCR recognition, keywords related to the business categories, such as business names and business ranges, are filtered out from the text recognition results, and are matched with business category names stored in a database at the back end of a pre-auditing system, so that at least one successfully matched business category name, such as retail or wholesale, is obtained, and is automatically filled into a qualification auditing work order.

The target field in the qualification audit work order is rich in content, the target field is multiple in selectable options, and the qualification audit work order generation module can effectively avoid random selection of a user, reduce fields which are required to be actively input by the user, reduce signing cost and improve information acquisition quality by recommending relevant keywords extracted from text identification content in real time and automatically.

One embodiment of the present disclosure further provides a computer readable storage medium having a computer program stored thereon, where the computer program when executed by a processor implements the merchant subscription qualification auditing method described above.

One embodiment in the present specification also provides an electronic device, including:

one or more processors; and

and a memory associated with the one or more processors, the memory for storing program instructions that, when read and executed by the one or more processors, perform the steps of the merchant subscription qualification audit method described above.

As shown in FIG. 6, in a typical configuration, a computer includes one or more processors (CPUs), an input/output interface, a network interface, and memory.

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, read only compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic disk storage, quantum memory, graphene-based storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by the computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

It should be noted that the above-mentioned embodiments are merely examples of the present invention, and it is obvious that the present invention is not limited to the above-mentioned embodiments, and many similar variations are possible. All modifications attainable or obvious from the present disclosure set forth herein should be deemed to be within the scope of the present disclosure.

Claims

1. A merchant subscription qualification auditing method includes:

2. The method of claim 1, the pre-audit flow comprising picture quality pre-audit; based on a preset pre-checking flow, pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

3. The method of claim 1, the pre-review process comprising a picture type review; based on a preset pre-checking task flow, pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

4. The method of claim 1, the pre-audit flow comprising picture text information verification; based on a preset pre-checking task flow, pre-checking the at least one qualification picture to obtain a pre-checking evaluation result of each qualification picture, wherein the pre-checking evaluation result comprises the following specific steps:

determining a third party authority based on the picture type;

5. The method of claim 1, wherein the pre-audit flow comprises an icon information verification, the icon information verification specifically comprising:

6. The method of claim 1, the qualification audit worksheet comprising the at least one qualification picture and at least one target field corresponding to the at least one qualification picture; the generating qualification audit worksheet specifically comprises the following steps:

7. A merchant subscription qualification auditing apparatus, comprising:

8. The apparatus of claim 7, the pre-audit module comprising a picture quality pre-audit module; the picture quality pre-auditing module is specifically used for inputting the qualification picture into a pre-trained picture quality evaluation model to obtain a picture quality evaluation result of the qualification picture; and determining a pre-checking evaluation result of the qualification picture based on the picture quality evaluation result.

9. The apparatus of claim 7, the pre-audit module comprising a picture type audit module; the picture type auditing module is specifically used for inputting the qualification picture into a pre-trained classification model to obtain the prediction type of the qualification picture; and matching the prediction type with a preset target type, and determining a pre-checking evaluation result of the qualification picture based on a matching result.

10. The apparatus of claim 7, the pre-audit module comprising a picture text information audit module; the picture text information verification module is specifically used for inputting the qualification picture into a pre-trained classification model and determining the picture type of the qualification picture; extracting a text region image in the qualification picture, and performing optical character recognition on the text region image to obtain a text recognition result; determining a third party authority based on the picture type; and verifying the text recognition result based on the public data of the third-party authority, and determining a pre-verification evaluation result of the qualification picture based on the verification result.

11. The apparatus of claim 7, the pre-audit module comprising an icon information audit module; the icon information verification module is specifically used for inputting the qualification picture into a pre-trained classification model and determining the picture type of the qualification picture; determining an icon identification template corresponding to the picture type based on the picture type of the qualification picture, wherein the icon identification template comprises at least one target candidate frame, and the position of the target candidate frame is the position of an icon to be acquired; determining an icon area image in the qualification picture based on at least one target candidate frame of the icon recognition template; and matching the icon area image with a preset icon template, and determining a pre-checking evaluation result of the qualification picture based on a matching result.

12. The device of claim 7, wherein the qualification audit work order generation module is specifically configured to extract, for each qualified picture for which the pre-audit evaluation result is qualified, a text region image of the qualified picture; performing optical character recognition on the text region image to obtain a text recognition result; extracting keywords from the text recognition result, and matching the extracted keywords with preset options of a target field in the qualification audit worksheet; generating at least one selectable item of the target field based on the matching result, and recommending the selectable item to a user; and determining the target field based on the selection result of the user on the selectable item to obtain the qualification audit work order.

13. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of any of claims 1 to 6.

14. An electronic device, comprising:

one or more processors; and

a memory associated with the one or more processors for storing program instructions that, when read for execution by the one or more processors, perform the steps of the method of any of claims 1 to 6.