Method and apparatus for judging validity of character information, and electronic device

Abstract

The application relates to a method and device for judging the validity of character information and an electronic device. The method comprises: obtaining a character name to be judged and a corresponding video screenshot, wherein the character name to be judged is extracted from the video screenshot by optical character recognition technology or visual recognition technology; constructing a prompt word according to the character name to be judged, wherein the prompt word comprises a role definition, a text to be judged, a judgment task, a constraint condition and an output requirement; inputting the prompt word and the video screenshot into a target visual large model to enable the target visual large model to perform image understanding, semantic understanding and joint reasoning on the video screenshot and the character name to be judged according to the prompt word, and generate an output result comprising a judgment result and a judgment reason; and analyzing the output result to determine the validity of the character name to be judged and marking the character name to be judged. The application solves the technical problem that the existing method cannot accurately judge the validity of the character name in a video.

Description

Technical Field

[0001] This application relates to the field of video processing technology, and in particular to a method, apparatus, and electronic device for determining the validity of personal information. Background Technology

[0002] In the intelligent processing of video content such as micro-dramas and short dramas, character information recognition is a crucial foundation for content understanding and structuring. It typically requires automatically identifying and extracting character names from video screenshots or keyframes to support applications such as character relationship analysis, plot understanding, and content retrieval. However, due to the complexity of video footage and the diversity of subtitle styles, text extracted using optical character recognition (OCR) or visual recognition technologies often contains multiple types of information, such as character names, dialogue text, song lyrics, and scene cues. Only a small proportion of these are truly effective names used to identify characters.

[0003] Existing methods for determining the validity of personal names largely rely on rule matching or keyword filtering, using preset text format, position, or feature rules to filter the recognition results. These methods struggle to deeply understand the semantic meaning and contextual relationships of the text, failing to accurately distinguish subtle differences between character names and dialogue / lyrics. Furthermore, they lack the ability to comprehensively understand the visual content in video screenshots, making it difficult to combine information such as character appearance and subtitle layout for a comprehensive judgment. In addition, the rule systems are complex, maintenance costs are high, and they struggle to cover the constantly changing presentation styles of micro-dramas and short dramas, easily leading to a high false positive rate and low overall recognition accuracy. Summary of the Invention

[0004] This application provides a method, apparatus, and electronic device for determining the validity of personal information, in order to solve the technical problem that existing methods are unable to accurately determine the validity of personal names in videos.

[0005] In a first aspect, this application provides a method for determining the validity of person information, comprising: obtaining the name of the person to be determined and its corresponding video screenshot, wherein the name of the person to be determined is extracted from the video screenshot using optical character recognition technology or visual recognition technology; constructing prompt words based on the name of the person to be determined, wherein the prompt words include a role definition, text to be determined, a judgment task, constraints, and output requirements; inputting the prompt words and the video screenshot into a target visual big data model, so that the target visual big data model performs image understanding, semantic understanding, and joint reasoning on the video screenshot and the name of the person to be determined based on the prompt words, and generates an output result containing a judgment result and a judgment reason; parsing the output result to determine the validity of the name of the person to be determined, and marking the name of the person to be determined as valid.

[0006] Secondly, this application provides a device for judging the validity of person information, comprising: an acquisition module for acquiring the name of the person to be judged and its corresponding video screenshot, wherein the name of the person to be judged is extracted from the video screenshot using optical character recognition technology or visual recognition technology; a construction module for constructing prompt words based on the name of the person to be judged, wherein the prompt words include a role definition, text to be judged, judgment task, constraints, and output requirements; a reasoning module for inputting the prompt words and the video screenshot into a target visual model, so that the target visual model performs image understanding, semantic understanding, and joint reasoning on the video screenshot and the name of the person to be judged based on the prompt words, and generates an output result containing a judgment result and a judgment reason; and a determination module for parsing the output result to determine the validity of the name of the person to be judged and marking the name of the person to be judged as valid.

[0007] As an optional example, the acquisition module includes: an extraction unit for extracting the video screenshot from the target video; a recognition unit for recognizing the text content in the video screenshot using the optical character recognition technology or the visual recognition technology, and using the recognized text content as the name of the person to be judged; and an establishment unit for establishing a mapping relationship between the name of the person to be judged and the corresponding video screenshot.

[0008] As an optional example, the above-mentioned construction module includes: a first construction unit, used to explicitly define the role in the above-mentioned prompt words, so as to instruct the above-mentioned target visual big model to judge the appearance information of the person in the above-mentioned video screenshot; an embedding unit, used to embed the name of the person to be judged into the above-mentioned prompt words, so as to serve as the text to be judged in the above-mentioned prompt words; a second construction unit, used to explain the judgment task and constraints in the above-mentioned prompt words; and a third construction unit, used to specify the output requirements in the above-mentioned prompt words, so as to instruct the above-mentioned target visual big model to return the judgment result and judgment reason in JSON format.

[0009] As an optional example, the above reasoning module includes: a judgment unit, used to input the above prompt words and the above video screenshots into the target visual big model, so that the target visual big model can analyze the text region, text type and text position in the above video screenshots through image understanding, understand the judgment task and constraints in the above prompt words through semantic understanding, combine the above video screenshots with the above name of the person to be judged through joint reasoning, judge whether the above name of the person to be judged meets the above constraints, and generate an output result containing the judgment result and the judgment reason in JSON format.

[0010] As an optional example, the above-mentioned determining module includes: a first determining unit, used to determine that the name of the person to be judged is valid when the field value corresponding to the judgment result in the above-mentioned output result indicates that it is valid; and a second determining unit, used to determine that the name of the person to be judged is invalid when the field value corresponding to the judgment result in the above-mentioned output result indicates that it is invalid.

[0011] As an optional example, the above-mentioned determining module includes: a first marking unit, used to mark the above-mentioned person name to be judged as valid and retain the above-mentioned person name to be judged when it is determined that the above-mentioned person name to be judged is valid.

[0012] As an optional example, the determination module includes: a second marking unit, used to mark the name of the person to be judged as invalid and delete the name of the person to be judged if it is determined that the name of the person to be judged is invalid.

[0013] Thirdly, this application provides a storage medium storing a computer program, wherein the computer program is executed by a processor to perform the method for determining the validity of the aforementioned person information.

[0014] Fourthly, this application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the method for determining the validity of the aforementioned person information through the computer program.

[0015] The technical solutions provided in this application have the following advantages compared with the prior art: This application employs a method for obtaining the name of the person to be judged and its corresponding video screenshot. The name of the person to be judged is extracted from the video screenshot using optical character recognition (OCR) or visual recognition technology. Based on the name of the person to be judged, a prompt word is constructed, which includes a role definition, the text to be judged, the judgment task, constraints, and output requirements. The prompt word and the video screenshot are input into a target visual model, enabling the target visual model to perform image understanding, semantic understanding, and joint reasoning on the video screenshot and the name of the person to be judged based on the prompt word, generating an output result containing a judgment result and a judgment reason. The output result is parsed to determine the validity of the name of the person to be judged, and the name of the person to be judged is marked as valid. This method, by obtaining the name of the person to be judged extracted from the video screenshot using OCR or visual recognition technology, constructing a prompt word containing a role definition, judgment task, constraints, and output requirements, and then matching the prompt word with the video screenshot... Figure 1The data is then input into a large-scale visual model with multimodal understanding capabilities. This model performs image understanding, semantic understanding, and joint reasoning on character names and image content to determine whether a character name represents valid character appearance information. This achieves intelligent verification of character name validity, accurately distinguishing character names from non-character text such as dialogue and lyrics, reducing false positive rates, and improving the accuracy and reliability of character information recognition in micro-dramas and short dramas. Ultimately, this solves the technical problem of existing methods' difficulty in accurately determining the validity of character names in videos. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0019] Figure 1 This is a flowchart of an optional method for determining the validity of personal information according to an embodiment of this application; Figure 2 This is a flowchart illustrating the specific implementation of an optional method for determining the validity of personal information according to an embodiment of this application. Figure 3 This is a schematic diagram of an optional device for determining the validity of personal information according to an embodiment of this application; Figure 4 This is a schematic diagram of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0022] According to a first aspect of the embodiments of this application, a method for determining the validity of personal information is provided, optionally, as follows: Figure 1 As shown, the above method includes: S102, obtain the name of the person to be judged and its corresponding video screenshot, wherein the name of the person to be judged is extracted from the video screenshot through optical character recognition technology or visual recognition technology; S104, construct prompt words based on the name of the person to be judged, wherein the prompt words include the character definition, the text to be judged, the judgment task, the constraints and output requirements; S106, Input the prompt words and video screenshots into the target visual big model, so that the target visual big model can perform image understanding, semantic understanding and joint reasoning on the video screenshots and the name of the person to be judged based on the prompt words, and generate an output result containing the judgment result and the judgment reason; S108, parse the output to determine the validity of the name of the person to be judged, and mark the validity of the name of the person to be judged.

[0023] Optionally, in this embodiment, a method for determining the validity of person information is provided, applied in intelligent processing scenarios of video content such as micro-dramas and short dramas. This method is used to verify the validity of person names identified from video footage, thereby improving the accuracy and reliability of person information recognition. Specific implementation steps are as follows: Figure 2 As shown, the first step is to obtain the name of the person to be identified and its corresponding video screenshot. Specifically, keyframes or screenshots containing the character's appearance information are selected from the micro-drama or short drama video, and text recognition processing is performed on the video screenshots. The name of the person to be identified can be extracted from the subtitle area of ​​the video screenshot using Optical Character Recognition (OCR) technology, or it can be obtained by recognizing the character's image and its corresponding text identifier in the image using visual recognition technology. Through the above methods, the identified name of the person is associated with the corresponding video screenshot, providing image context information for subsequent validity judgment.

[0024] Secondly, prompt words are constructed based on the names of the characters to be judged. These prompt words guide the target visual model in performing the task of judging the validity of character information. Specifically, this includes the character definition, the text to be judged, the judgment task, constraints, and output requirements. The character definition explicitly tells the model that it needs to judge the appearance information of characters in screenshots from micro-dramas or short dramas; the text to be judged is the name of the character to be judged; the judgment task instructs the model to determine whether the name of the character to be judged belongs to the character information shown in the video screenshot; the constraints limit the judgment criteria, such as whether the text is related to the character's appearance area or whether it conforms to the semantic features of the character identifier; and the output requirements specify that the model should output the judgment results and judgment basis in a structured format.

[0025] Then, the prompt words and video screenshots are input into the target visual big data model, enabling the model to reason about the video screenshots and output a judgment result and reasoning. The target visual big data model is a multimodal understanding model capable of processing both image and text information simultaneously. Upon receiving the prompt words and video screenshots, the model first performs image understanding on the video screenshots, analyzing the text regions, text positions, text types, and their spatial relationship with the character images; simultaneously, it performs semantic understanding on the prompt words, parsing the judgment task and constraints. Based on this, the target visual big data model performs joint reasoning on the image understanding and semantic understanding results, comprehensively judging whether the name of the character to be judged meets the feature requirements of the character's appearance information, and generating an output containing the judgment result and reasoning. The output is presented in a pre-defined structured format.

[0026] Finally, the output is parsed to determine the validity of the names to be judged. Specifically, a judgment result field is extracted from the output to indicate whether the name is valid information; a judgment reason field is also extracted to record the judgment basis provided by the model. Based on the judgment result field, the names are marked as valid. Valid names are retained, while invalid names are filtered or removed, thus completing the process of judging the validity of the person information.

[0027] Optionally, this embodiment introduces a large visual model with multimodal understanding capabilities to jointly analyze the visual information in video screenshots with the semantic information of character names. This can accurately distinguish character names from non-character information such as dialogue text and lyrics, effectively reducing the false judgment rate. At the same time, structured prompts guide the model to perform judgment tasks, avoiding complex rule design and improving the system's versatility and scalability. The output of structured judgment results facilitates subsequent automated processing and system integration, significantly improving the overall accuracy and stability of character information recognition in micro-dramas and short dramas.

[0028] As an optional example, obtaining the name of the person to be identified and its corresponding video screenshot includes: Extract video screenshots from the target video; The text content in video screenshots is identified using optical character recognition or visual recognition technology, and the identified text content is used as the name of the person to be identified. Establish a mapping relationship between the name of the person to be identified and the corresponding video screenshot.

[0029] Optionally, in this embodiment, firstly, video screenshots for character information identification are extracted from the target video. The target video can be a micro-drama, short drama, or other video content containing character appearance information. One or more frames of video are extracted from the target video according to preset time intervals, keyframe detection, or character appearance detection, and used as input data for subsequent character information identification. Subsequently, text recognition processing is performed on the video screenshots. Specifically, optical character recognition technology can be used to detect and recognize the subtitle or text areas in the video screenshots to extract the text content displayed on the screen; or, visual recognition technology can be used to recognize the character images and their associated text tags in the video screenshots to obtain text information related to the character. Through the above recognition methods, text content that may be used to identify the character can be extracted from complex video images, and the recognized text content can be used as the name of the character to be identified.

[0030] After obtaining the names of the individuals to be judged, a mapping relationship is established between these names and their corresponding video screenshots. This mapping relationship represents the one-to-one correspondence between each individual's name and its source video screenshot, ensuring that the validity of the individual's name can be determined in subsequent processing by considering the specific image context.

[0031] As an optional example, constructing prompts based on the name of the person to be identified includes: Clearly define the roles in the prompts to instruct the target visual model to determine the appearance information of the characters in the video screenshots; The name of the person to be judged is embedded in the prompt words, which serve as the text to be judged for the prompt words; The prompts should describe the judgment task and constraints. The output requirements are specified in the prompt, instructing the target visual large model to return the judgment result and the reason for the judgment in JSON format.

[0032] Optionally, in this embodiment, the role is first clearly defined when constructing the prompt words. This role is used to indicate the current task role of the target visual model, making it understand that it needs to perform a task of judging the presence information of people in video screenshots, rather than a simple image description or text generation task. By clearly defining the model's role, the model can be effectively guided to focus on the scene of recognizing people's information and judging its validity, avoiding the generation of output results that are irrelevant to the task.

[0033] Secondly, the name of the person to be judged is embedded in the prompt words, serving as the text to be judged within the prompt words. By explicitly presenting the name of the person to be judged in the prompt words, the target visual model can clearly identify the specific object that needs to be judged. Thus, when analyzing video screenshots, the focus is on whether the text corresponds to the information about the person's appearance shown in the scene, improving the targeting and accuracy of the judgment.

[0034] Subsequently, the prompt further clarifies the judgment task and constraints. The judgment task explicitly requires the target visual model to determine whether the name of the person to be judged belongs to a real and valid person in the video screenshot; the constraints are used to limit the judgment criteria, such as whether the text is located in the person's appearance information area, whether it conforms to the semantic features of the person's name, and whether it matches the image of the person in the picture. By setting clear task descriptions and constraints, the model has a unified and controllable judgment criterion when performing inference.

[0035] Finally, the output requirements are specified in the prompts to instruct the target visual model to return the judgment results in a preset structured format. Preferably, the target visual model is required to output the results in JSON format, including a judgment field to identify the validity of the person's name and a supporting field to explain the reasoning behind the judgment. The prompts constructed in this way not only effectively guide the target visual model to complete the multimodal joint judgment task, but also facilitate the subsequent automatic parsing and processing of the model's output results by the system, improving the overall system's practicality and scalability.

[0036] As an optional example, the prompt words and video screenshots are input into the target visual model, enabling the target visual model to perform image understanding, semantic understanding, and joint reasoning on the video screenshots and the name of the person to be judged based on the prompt words, generating an output containing the judgment result and the reasoning for the judgment, including: The prompt words and video screenshots are input into the target visual big model, which then uses image understanding to analyze the text regions, text types, and text positions in the video screenshots, uses semantic understanding to understand the judgment task and constraints in the prompt words, and uses joint reasoning to combine the video screenshots with the name of the person to be judged to determine whether the name of the person to be judged meets the constraints. Finally, it generates an output result in JSON format that includes the judgment result and the reasoning behind the judgment.

[0037] Optionally, in this embodiment, firstly, the constructed prompt words are matched with the corresponding video screenshots. Figure 1 The data is then input into the target visual model. The target visual model is a multimodal processing model that can simultaneously receive and process image and text information, thus providing a unified reasoning basis for judging the validity of person information.

[0038] During model processing, the target vision big data model first performs image understanding analysis on the video screenshots. Specifically, the model analyzes the visual features of the video screenshots to identify text regions, text types, and the spatial relationship of the text within the frame. For example, the model can determine whether the text is located in a common display area for character appearance information, and whether the presentation of the text conforms to the visual characteristics of character name identifiers, thus providing visual basis for subsequent judgments.

[0039] Meanwhile, the target vision big data model performs semantic understanding processing on the prompt words. By parsing the role definitions, judgment tasks, and constraints in the prompt words, the model can accurately understand the judgment target and judgment criteria that need to be completed, and clarify whether the name of the person to be judged should be regarded as the appearance information of the person shown in the video screenshot. Through semantic understanding, the model follows the preset judgment logic during the reasoning process, avoiding the generation of outputs irrelevant to the task.

[0040] After completing image understanding and semantic understanding, the target visual big data model further executes a joint inference process. This joint inference process comprehensively analyzes the image understanding results and semantic understanding results, combining the visual content in the video screenshot with the semantic features of the name of the person to be judged, to determine whether the name of the person to be judged meets the constraints set in the prompt, and determines its validity accordingly. Finally, the target visual big data model generates the output results in JSON format according to the output requirements specified in the prompt. The output results include a judgment result field indicating the judgment conclusion and a judgment reason field explaining the basis for the judgment.

[0041] As an optional example, parsing the output to determine the validity of the person's name to be judged includes: If the field value corresponding to the judgment result in the output result indicates that it is valid, then the name of the person to be judged is determined to be valid. If the field value corresponding to the judgment result in the output result indicates that it is invalid, then the name of the person to be judged is determined to be invalid.

[0042] Optionally, in this embodiment, the structured output result returned by the target visual model is first received. The output result is preferably structured data conforming to a preset format, such as JSON format, used to uniformly express the model's judgment conclusion on the name of the person to be judged and its corresponding judgment basis, thereby facilitating subsequent automated parsing and processing by the system.

[0043] During the parsing process, the judgment result field, representing the judgment conclusion, is first extracted from the output. This field indicates the target visual model's conclusion regarding the validity of the person's name; its value can be preset as identifier information to distinguish between valid and invalid states. When the judgment result field indicates validity, the person's name is determined to be valid, meaning it matches the person's appearance information shown in the video screenshot and meets the preset judgment conditions.

[0044] Accordingly, when the field value of the judgment result field indicates invalidity, the name of the character to be judged is determined to be invalid character information, indicating that the character name does not meet the judgment criteria for character appearance information and may belong to dialogue text, song lyrics text, or other non-character identifier text. In this case, the name of the character to be judged can be marked as invalid for subsequent filtering or removal processing.

[0045] In some implementations, the judgment reason field in the output results can also be parsed simultaneously to obtain the judgment basis given by the target visual model. The judgment basis can be used to assist in analyzing why a person's name is judged as valid or invalid, providing a reference for system optimization, model evaluation, or manual review. Through structured parsing and validity marking of the output results, automated verification of the person's name to be judged is achieved, ensuring that only true and valid person information is retained, thereby improving the overall accuracy and reliability of the person information recognition results.

[0046] As an optional example, validating the names of people to be judged includes: If the name of the person to be judged is determined to be valid, mark the name as valid and retain the name.

[0047] Optionally, in this embodiment, after determining the validity of the person's name, a process of marking the name as valid is also included. This validity mark clearly distinguishes between valid names and names that fail the determination, thus providing a reliable basis for subsequent person information management, content analysis, and system applications.

[0048] Specifically, after analyzing the output of the target visual model and determining that the name of the person to be judged is valid, the name is marked as valid person information. Validity marking can be achieved by setting corresponding status fields, label information, or flag bits in the person name data structure to indicate that the person name has passed validity verification and meets the judgment criteria for person appearance information. Through this marking method, the system can quickly identify and distinguish valid person information in subsequent processing.

[0049] After validity marking is completed, the names of the characters to be judged that have been marked as valid are retained. Retention may include storing valid character names in a character information database, using them to build character lists or character relationship structures, or using them as input data for subsequent content understanding, plot analysis, and recommendation services. By retaining only the character names that have passed the validity judgment and been marked as valid, irrelevant text or incorrect recognition results are avoided from being introduced into subsequent processing, thereby reducing system noise.

[0050] In some implementations, the validity marker can also be associated with the video screenshot or video timestamp corresponding to the person's name for subsequent backtracking or verification. Through the aforementioned validity marker and retention mechanism, the person's information has a clear validity status throughout the entire processing flow, which helps improve the accuracy, stability, and scalability of processing person information in video content such as micro-dramas and short dramas.

[0051] As an optional example, validating the names of people to be judged includes: If the name of the person to be judged is determined to be invalid, mark the name of the person to be judged as invalid and delete the name of the person to be judged.

[0052] Optionally, in this embodiment, after determining the validity of the name of the person to be judged, the invalid name can be marked and deleted. By clearly marking and promptly cleaning up invalid names, irrelevant or erroneous information can be effectively prevented from entering the subsequent processing flow, thereby improving the overall quality of the person information recognition results.

[0053] Specifically, after analyzing the output of the target visual model and determining that the name of the person to be judged is invalid, the name is marked as invalid person information. Invalidity marking can be achieved by setting corresponding status fields, flag bits, or label information to indicate that the name has failed validity verification and does not meet the judgment criteria for person appearance information. Through this marking method, the system can clearly distinguish between invalid and valid person names, preventing them from being mistakenly used in subsequent business processing.

[0054] After invalidity marking is completed, the names of the individuals to be judged are deleted. Deletion may include removing the names from the candidate name set or clearing them from the person information cache, person list, and related data structures, thereby preventing them from participating in subsequent processes such as person information analysis, person relationship building, or content recommendation. Timely deletion of invalid names can significantly reduce the proportion of invalid data in the system and reduce noise interference.

[0055] In some implementations, the deletion process can also be combined with log recording or statistical information to record the source, frequency of occurrence, and reasons for judgment of invalid person names, so as to optimize and adjust the identification model or judgment strategy. By marking and deleting invalid person names, the person information processing flow becomes more rigorous, which helps to improve the accuracy and reliability of person information identification and management in video content such as micro-dramas and short dramas.

[0056] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0057] According to another aspect of the embodiments of this application, a device for determining the validity of person information is also provided, such as... Figure 3 As shown, it includes: The acquisition module 302 is used to acquire the name of the person to be judged and its corresponding video screenshot, wherein the name of the person to be judged is extracted from the video screenshot through optical character recognition technology or visual recognition technology. Module 304 is used to construct prompt words based on the name of the person to be judged. The prompt words include the character definition, the text to be judged, the judgment task, the constraints, and the output requirements. The reasoning module 306 is used to input the prompt words and video screenshots into the target visual big model, so that the target visual big model can perform image understanding, semantic understanding and joint reasoning on the video screenshots and the name of the person to be judged based on the prompt words, and generate an output result containing the judgment result and the judgment reason; The determination module 308 is used to parse the output results to determine the validity of the name of the person to be judged and to mark the validity of the name of the person to be judged.

[0058] As an optional example, the modules to be acquired include: The extraction unit is used to extract video screenshots from the target video. The recognition unit is used to identify the text content in a video screenshot using optical character recognition technology or visual recognition technology, and to use the identified text content as the name of the person to be identified. A unit is created to establish a mapping relationship between the name of the person to be judged and the corresponding video screenshot.

[0059] As an optional example, the building blocks include: The first building unit is used to explicitly define the role in the prompt words so as to instruct the target visual big model to judge the appearance information of the characters in the video screenshot; The embedding unit is used to embed the name of the person to be judged into the prompt word, so as to serve as the text to be judged for the prompt word; The second building block is used to describe the judgment task and constraints in the prompt words; The third building block is used to specify the output requirements in the prompt words, instructing the target visual large model to return the judgment result and the reason for the judgment in JSON format.

[0060] As an optional example, the inference module includes: The judgment unit is used to input prompt words and video screenshots into the target visual big model, so that the target visual big model can analyze the text region, text type and text position in the video screenshot through image understanding, understand the judgment task and constraints in the prompt words through semantic understanding, and combine the video screenshot with the name of the person to be judged through joint reasoning to determine whether the name of the person to be judged meets the constraints, and generate the output result containing the judgment result and the reasoning in JSON format.

[0061] As an optional example, the identified modules include: The first determining unit is used to determine that the name of the person to be judged is valid if the field value corresponding to the judgment result in the output result indicates that it is valid. The second determining unit is used to determine that the name of the person to be judged is invalid when the field value corresponding to the judgment result in the output result indicates that it is invalid.

[0062] As an optional example, the identified modules include: The first marking unit is used to mark the name of the person to be judged as valid and retain the name of the person to be judged when it is determined that the name of the person to be judged is valid.

[0063] As an optional example, the identified modules include: The second marking unit is used to mark the name of the person to be judged as invalid and delete the name of the person to be judged if it is determined that the name of the person to be judged is invalid.

[0064] For other examples of this embodiment, please refer to the examples above, which will not be repeated here.

[0065] Figure 4 This is a schematic diagram of an optional electronic device according to an embodiment of this application, such as... Figure 4 As shown, it includes a processor 402, a communication interface 404, a memory 406, and a communication bus 408. The processor 402, communication interface 404, and memory 406 communicate with each other via the communication bus 408. Memory 406 is used to store computer programs; When processor 402 executes a computer program stored in memory 406, it performs the following steps: Obtain the name of the person to be identified and its corresponding video screenshot. The name of the person to be identified is extracted from the video screenshot using optical character recognition technology or visual recognition technology. Construct prompt words based on the name of the person to be judged, where the prompt words include the character definition, the text to be judged, the judgment task, the constraints, and the output requirements; The prompt words and video screenshots are input into the target visual big model, so that the target visual big model can perform image understanding, semantic understanding and joint reasoning on the video screenshots and the name of the person to be judged based on the prompt words, and generate an output result containing the judgment result and the reason for the judgment. The output is parsed to determine the validity of the names of the people to be judged, and the validity of the names of the people to be judged is marked.

[0066] Optionally, in this embodiment, the communication bus can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single thick line, but this does not indicate that there is only one bus or one type of bus. The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0067] The memory may include RAM, or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0068] As an example, the memory 406 described above may include, but is not limited to, the acquisition module 302, construction module 304, reasoning module 306, and determination module 308 from the aforementioned device for determining the validity of person information. Furthermore, it may include, but is not limited to, other module units from the aforementioned device for determining the validity of person information, which will not be elaborated upon in this example.

[0069] The processor mentioned above can be a general-purpose processor, including but not limited to: CPU (Central Processing Unit), NP (Network Processor), etc.; it can also be DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0070] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments, and will not be repeated here.

[0071] Those skilled in the art will understand that Figure 4 The structure shown is for illustrative purposes only. The device that implements the above method for determining the validity of personal information can be a terminal device, such as a smartphone (e.g., Android phone, iOS phone), tablet computer, PDA, mobile Internet Devices (MID), PAD, etc. Figure 4 This does not limit the structure of the aforementioned electronic devices. For example, the electronic device may also include components that are more... Figure 4 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 4 The different configurations shown.

[0072] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, ROM, RAM, disk or optical disk, etc.

[0073] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, wherein a computer program is stored in the computer program, and the computer program is executed by a processor to perform the steps in the above-described method for determining the validity of person information.

[0074] Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0075] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0076] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0077] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0078] In the several embodiments provided in this application, it should be understood that the disclosed client can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between units or modules, and may be electrical or other forms.

[0079] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0080] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0081] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for determining the validity of personal information, characterized in that, include: Obtain the name of the person to be identified and its corresponding video screenshot, wherein the name of the person to be identified is extracted from the video screenshot using optical character recognition technology or visual recognition technology; Based on the name of the person to be judged, a prompt word is constructed, wherein the prompt word includes the character definition, the text to be judged, the judgment task, the constraints, and the output requirements; The prompt words and the video screenshots are input into the target visual big model, so that the target visual big model can perform image understanding, semantic understanding and joint reasoning on the video screenshots and the name of the person to be judged based on the prompt words, and generate an output result containing the judgment result and the judgment reason; The output is parsed to determine the validity of the name of the person to be judged, and the name of the person to be judged is marked as valid.

2. The method according to claim 1, characterized in that, Obtaining the name of the person to be judged and its corresponding video screenshot includes: Extract the video screenshot from the target video; The optical character recognition technology or the visual recognition technology is used to identify the text content in the video screenshot, and the identified text content is used as the name of the person to be judged. Establish a mapping relationship between the names of the individuals to be identified and the corresponding video screenshots.

3. The method according to claim 1, characterized in that, Based on the name of the person to be identified, the following prompt words are constructed: The prompt word clearly defines the role so as to instruct the target visual big model to determine the appearance information of the characters in the video screenshot; The name of the person to be judged is embedded in the prompt word to serve as the text to be judged for the prompt word; The prompt text describes the judgment task and constraints; The output requirements are specified in the prompt, instructing the target visual large model to return the judgment result and the reason for the judgment in JSON format.

4. The method according to claim 1, characterized in that, The prompt words and the video screenshot are input into the target visual model, so that the target visual model performs image understanding, semantic understanding, and joint reasoning on the video screenshot and the name of the person to be judged based on the prompt words, and generates an output result containing the judgment result and the judgment reasoning, including: The prompt words and the video screenshots are input into the target visual big model, so that the target visual big model can analyze the text regions, text types and text positions in the video screenshots through image understanding, understand the judgment task and constraints in the prompt words through semantic understanding, and combine the video screenshots with the name of the person to be judged through joint reasoning to determine whether the name of the person to be judged meets the constraints. The output results containing the judgment results and judgment reasons are generated in JSON format.

5. The method according to claim 1, characterized in that, Parsing the output to determine the validity of the name of the person to be judged includes: If the field value corresponding to the judgment result in the output result indicates that it is valid, the name of the person to be judged is determined to be valid; If the field value corresponding to the judgment result in the output result indicates that it is invalid, the name of the person to be judged is determined to be invalid.

6. The method according to claim 1, characterized in that, The validity marking of the names of the persons to be judged includes: If the name of the person to be judged is determined to be valid, the name of the person to be judged is marked as valid and the name of the person to be judged is retained.

7. The method according to any one of claims 1 to 6, characterized in that, The validity marking of the names of the persons to be judged includes: If it is determined that the name of the person to be judged is invalid, the name of the person to be judged is marked as invalid and deleted.

8. A method for determining the validity of personal information, characterized in that, include: The acquisition module is used to acquire the name of the person to be judged and its corresponding video screenshot, wherein the name of the person to be judged is extracted from the video screenshot through optical character recognition technology or visual recognition technology; The construction module is used to construct prompt words based on the name of the person to be judged, wherein the prompt words include the character definition, the text to be judged, the judgment task, the constraints and the output requirements; The reasoning module is used to input the prompt words and the video screenshots into the target visual big model, so that the target visual big model can perform image understanding, semantic understanding and joint reasoning on the video screenshots and the name of the person to be judged based on the prompt words, and generate an output result containing the judgment result and the judgment reason; The determination module is used to parse the output results to determine the validity of the name of the person to be judged, and to mark the validity of the name of the person to be judged.

9. A computer-readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to perform the method described in any one of claims 1 to 7.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to execute the method described in any one of claims 1 to 7 through the computer program.

Images