Cartoon book checking method based on thinking chain

Abstract

The invention discloses a comic book checking method based on a thinking chain, and belongs to the technical field of comic book checking, and the method comprises the steps: S1, data receiving: receiving comic book data to be checked, including image files and related metadata; s2, performing pretreatment; s3, cartoon content analysis; s4, cartoon content classification; s5, training an AI expert model; s6, AI expert model assessment: performing assessment on the AI expert model through cross validation and actual case test; s7, AI deployment and examination team establishment; s8, AI simulation check and comprehensive check are carried out; and S9, data feedback optimization and result output. The method comprises the following steps: establishing a special cartoon book data receiving platform, establishing data connection with each cartoon publishing house and author, obtaining original cartoon book data, performing format standardization processing on the received data, converting all image files into uniform formats and sizes, and performing definition optimization processing on images. Noisy points and blurred parts are removed, and it is ensured that the image quality meets the checking requirement.

Description

Technical Field

[0001] The present invention relates to the technical field of comic book proofreading, and in particular to a comic book proofreading method based on thought chain. Background Art

[0002] Traditional comic book proofreading relies primarily on manual page-by-page review, a time-consuming and labor-intensive approach prone to oversights. Because comic books typically contain a large amount of text and image information, proofreaders must simultaneously focus on the consistency and accuracy of both, making the proofreading process even more complex and challenging. Furthermore, existing automated proofreading tools often only process text content and are unable to effectively handle text and symbols in images, limiting their application. Therefore, we propose a comic book proofreading method based on thought chaining to address this issue. Summary of the Invention

[0003] The purpose of the present invention is to provide a comic book proofreading method based on thought chain to solve the problems raised in the above background technology.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] A comic book review method based on thought chain, comprising:

[0006] S1. Data reception: receiving comic book data to be reviewed, including image files and related metadata;

[0007] S2. Preprocessing: Standardize the image format to facilitate subsequent analysis and processing. For example, images of different resolutions are resized to a uniform size suitable for review and editing, and the image clarity is optimized to ensure that text and image details can be accurately recognized.

[0008] S3. Comic content analysis: Utilize image recognition and natural language processing technology to analyze elements in comics and identify characters, scenes, text bubbles, and other content.

[0009] S4. Comic content classification: Comics are classified based on factors such as theme, style (such as comedy, passionate, science fiction, etc.), and target audience. For example, by analyzing the image style and the emotional tendency of the text content, it can be determined whether it is a comedy or passionate comic.

[0010] S5. AI Expert Model Training: A large amount of data from different comic books of various genres is collected as training samples, including comics of various styles, themes, and quality levels. Deep learning algorithms and transfer learning techniques are used to train an AI expert model with comic proofreading capabilities. For example, convolutional neural networks are used to extract features from comic images, and recurrent neural networks are used to perform semantic analysis on text sequences in comics. During the training process, domain expert knowledge is incorporated to develop detailed proofreading rules and standards, such as the rationality of image composition, the matching of character expressions with context, and the accuracy of spelling and grammar.

[0011] S6. AI Expert Model Assessment: The AI expert model is assessed through cross-validation and real-world case testing to evaluate its performance in comic proofreading tasks, including metrics such as accuracy, recall, and F1 score.

[0012] S7. AI Allocation and Proofreading Team Formation: Based on the comic book classification results and proofreading requirements, appropriate AI expert models are intelligently selected and deployed. For example, for science fiction comics, models with a good understanding and recognition of scientific concepts and terminology are prioritized. For children's comics, models that prioritize engaging visuals and easy-to-understand text are selected. A diverse AI proofreading team is established, including AI models with different specialized expertise, such as image proofreaders, text proofreaders, and logic proofreaders. These models collaborate to review comic books from multiple perspectives.

[0013] S8. AI Simulated Proofreading and Comprehensive Proofreading: Utilizing an AI proofreading team, comic books undergo comprehensive, simulated human expert proofreading. From the perspective of image quality, they check whether the character design is reasonable, the color matching is harmonious, and the scene rendering is accurate. From the perspective of text content, they check whether the spelling is correct, the sentences are fluent, and the dialogue is consistent with the character's personality and plot development. From the perspective of logical relationships, they analyze whether the comic's storyline is coherent, whether the cause-effect relationship is reasonable, and whether the character's behavior motivation is clear, ensuring that every issue is accurately verified and corrected.

[0014] S9. Data feedback optimization and result output: Collect and process feedback data generated by the system during the proofreading process, including the judgment results of the AI model, proofreading time, error conditions, and other information; optimize and improve the proofreading system based on the feedback information. For example, if an AI model often makes misjudgments in the proofreading of a specific type of comics, analyze the reasons and adjust or retrain the model. Based on the proofreading results, output a proofreading report and correction suggestion report for the comic book. The report content includes a detailed error list, error type, correction suggestions, and an overall proofreading evaluation, providing a basis for decision-making for publishers.

[0015] Preferably, in said S2, the specific steps are as follows:

[0016] S201. When collecting and annotating large-scale comic data sets, not only common comic types are included, but also comics from different regions and eras are categorized and annotated to ensure data diversity and representativeness.

[0017] S202. Perform strict cleaning and preprocessing on the data to remove noise data and invalid annotations, thereby improving the quality of the training data.

[0018] S203. Annotate the image according to the characteristics of the comic, such as annotating key elements, text areas and their attributes, to provide more accurate information for subsequent training.

[0019] Preferably, in said S8, the specific steps are as follows:

[0020] S801. Enhanced image quality review: In addition to regular image quality checks, we now review the rationality of perspective switching and the effects of storyboards in comics.

[0021] S802. Use image analysis technology to check the lighting effects and texture details in the image to ensure that the image has sufficient visual appeal and artistic appeal;

[0022] S803. Expanded Text Content Review: In addition to basic spelling, grammar, and semantics checks, we also review the text's style and adaptability, and review the text layout in the comics to ensure that the text and images are in harmony and do not affect the reading experience.

[0023] S804. Deepen the review of logical relationships: Further analyze the logical structure of the comic story, check for plot holes, character behavior inconsistencies, and other issues, and sort out and review the timeline in the comic to ensure that the development of the story is reasonable and coherent in time.

[0024] Preferably, in said S7, the specific steps are as follows:

[0025] S701. Intelligent Allocation Strategy Upgrade: Introducing a reinforcement learning algorithm to continuously learn and optimize AI allocation strategies based on feedback and results from previous review tasks. For example, reinforcement learning is used to determine the optimal combination and order of different AI expert models for reviewing different types of comics. Taking into account the update frequency of comics and market feedback, the priority of AI expert models is dynamically adjusted. For popular and frequently updated comic series, AI models with stronger performance and faster response speeds are prioritized.

[0026] S702. Dynamic adjustment of team formation: Based on the special subject matter and temporary review needs of the comics, a temporary AI expert team is formed in real time. For example, for a comic with a cross-cultural background, an AI expert model with relevant cultural background knowledge is added in a timely manner, and a sharing and collaboration mechanism for AI expert models is established, allowing different AI expert models to call and support each other when needed, thereby improving the overall efficiency of the team.

[0027] Preferably, in S9, the specific steps are as follows:

[0028] S901. Conduct in-depth mining and analysis of feedback data, focusing not only on surface errors but also on underlying patterns and causes. For example, correlation analysis can be used to identify whether certain types of errors are associated with specific styles, themes, or authors.

[0029] S902. Targetedly optimize the AI expert model based on the analysis results. For example, if a model is found to be prone to misjudgment when processing comics of a certain style, specialized style adaptability training can be performed on the model.

[0030] S903. In addition to a detailed list of errors and correction suggestions, the review report should include evaluation indicators and charts for the overall quality of the comic. For example, charts could be used to show the distribution of scores for image quality, text accuracy, and logical rationality.

[0031] S904. Provide a visual display of proofreading results, such as marking the error locations directly on the comic image, so that publishers can intuitively view and understand the problems.

[0032] The beneficial effects of the present invention are:

[0033] 1. In the present invention, the method for reviewing comic books based on thought chains, by building a multi-dimensional AI review team, conducts a comprehensive review from multiple key dimensions, such as image quality, text content, and logical relationships. For example, in terms of image quality, AI image review experts can use advanced image recognition technology to accurately identify subtle deviations in character design, such as disproportionate facial proportions and inharmonious color matching. In terms of text content, text review experts can accurately judge the spelling, grammar, semantics, and style adaptability of the text, and can effectively identify typos, grammatical errors, and expressions that do not match the comic style. This multi-dimensional review method can greatly improve the accuracy of review, avoid omissions and misjudgments that are prone to occur in traditional review, comprehensively guarantee the content quality of comic books, and provide readers with a better reading experience.

[0034] 2. In the present invention, the thought chain-based comic book review method, by introducing AI simulated review, can process massive amounts of data and complex information in a short period of time. During the initial overall review stage, the AI can quickly browse the entire content of the comic book, mark areas with potential problems, and provide clear direction for subsequent detailed review. The multi-round review strategy also makes the entire review process more orderly and efficient. Compared with traditional manual review, the thought chain-based review method can greatly shorten the review cycle, accelerate the publication process of comic books, enable works to be brought to market more quickly, and meet readers' needs. This is of great significance to the highly competitive comic publishing industry, and can help publishers seize market share and enhance their competitiveness immediately.

[0035] 3. In the present invention, the thought chain-based comic book proofreading method, through intelligent deployment of AI expert models, rationally allocates computing resources and human resources according to the characteristics and proofreading requirements of different comic types. For example, for comics with simpler visual styles and more common plots, a slightly lower-performance AI model can be deployed for proofreading; while for large-scale comic series or comics with unique artistic styles, more powerful computing resources and professional models can be deployed. This intelligent deployment method avoids waste of human resources, allowing publishers to invest more energy and resources in core aspects such as comic creation and promotion. At the same time, it reduces labor costs, improves economic benefits, and is conducive to the sustainable development of the comic publishing industry.

[0036] 4. In the present invention, the comic book review method based on thought chain provides certain guidance and norms for comic creators by learning and analyzing a large amount of comic data. During the review process, AI will evaluate comics according to pre-set rules and standards, such as the composition of the picture, the use of color, the expression of text, etc. These rules and standards can provide a reference for creators to help them better understand industry norms and readers' expectations. For example, in the picture quality review, AI can analyze the picture characteristics of excellent comic works, summarize some common rules and techniques about character design, scene drawing, etc., and feed this information back to the creators. This makes comic creation more rule-based, promotes the standardization and standardization of comic creation, and is conducive to improving the development level of the entire comic industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a flow chart of a comic book proofreading method based on thought chain proposed by the present invention. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0039] Reference Figure 1 , a comic book review method based on thought chain, including:

[0040] S1. Data reception: receiving comic book data to be reviewed, including image files and related metadata;

[0041] S2. Preprocessing: Standardize the image format to facilitate subsequent analysis and processing. For example, images of different resolutions are resized to a uniform size suitable for review and editing, and the image clarity is optimized to ensure that text and image details can be accurately recognized.

[0042] S3. Comic content analysis: Utilize image recognition and natural language processing technology to analyze elements in comics and identify characters, scenes, text bubbles, and other content.

[0043] S4. Comic content classification: Comics are classified based on factors such as theme, style (such as comedy, passionate, science fiction, etc.), and target audience. For example, by analyzing the image style and the emotional tendency of the text content, it can be determined whether it is a comedy or passionate comic.

[0044] S5. AI Expert Model Training: A large amount of data from different comic books of various genres is collected as training samples, including comics of various styles, themes, and quality levels. Deep learning algorithms and transfer learning techniques are used to train an AI expert model with comic proofreading capabilities. For example, convolutional neural networks are used to extract features from comic images, and recurrent neural networks are used to perform semantic analysis on text sequences in comics. During the training process, domain expert knowledge is incorporated to develop detailed proofreading rules and standards, such as the rationality of image composition, the matching of character expressions with context, and the accuracy of spelling and grammar.

[0045] S6. AI expert model assessment: The AI expert model is assessed through cross-validation and real-world case testing to evaluate its performance in comic proofreading tasks, including accuracy, recall, F1 value, and other indicators.

[0046] S7. AI Allocation and Proofreading Team Formation: Based on the comic book classification results and proofreading requirements, appropriate AI expert models are intelligently selected and deployed. For example, for science fiction comics, models with a good understanding and recognition of scientific concepts and terminology are prioritized. For children's comics, models that prioritize engaging visuals and easy-to-understand text are selected. A diverse AI proofreading team is established, including AI models with different specialized expertise, such as image proofreaders, text proofreaders, and logic proofreaders. These models collaborate to review comic books from multiple perspectives.

[0047] S8. AI Simulated Proofreading and Comprehensive Proofreading: Utilizing an AI proofreading team, comic books undergo comprehensive, simulated human expert proofreading. From the perspective of image quality, they check whether the character design is reasonable, the color matching is harmonious, and the scene rendering is accurate. From the perspective of text content, they check whether the spelling is correct, the sentences are fluent, and the dialogue is consistent with the character's personality and plot development. From the perspective of logical relationships, they analyze whether the comic's storyline is coherent, whether the cause-effect relationship is reasonable, and whether the character's behavior motivation is clear, ensuring that every issue is accurately verified and corrected.

[0048] S9. Data feedback optimization and result output: Collect and process feedback data generated by the system during the proofreading process, including the judgment results of the AI model, proofreading time, error conditions, and other information; optimize and improve the proofreading system based on the feedback information. For example, if an AI model often makes misjudgments in the proofreading of a specific type of comics, analyze the reasons and adjust or retrain the model. Based on the proofreading results, output a proofreading report and correction suggestion report for the comic book. The report content includes a detailed error list, error type, correction suggestions, and an overall proofreading evaluation, providing a basis for decision-making for publishers.

[0049] In this embodiment, in S2, the specific steps are as follows:

[0050] S201. When collecting and annotating large-scale comic data sets, not only common comic types are included, but also comics from different regions and eras are categorized and annotated to ensure data diversity and representativeness.

[0051] S202. Perform strict cleaning and preprocessing on the data to remove noise data and invalid annotations, thereby improving the quality of the training data.

[0052] S203. Annotate the image according to the characteristics of the comic, such as annotating key elements, text areas and their attributes, to provide more accurate information for subsequent training.

[0053] In this embodiment, in S8, the specific steps are as follows:

[0054] S801. Enhanced image quality review: In addition to regular image quality checks, we now review the rationality of perspective switching and the effects of storyboards in comics.

[0055] S802. Use image analysis technology to check the lighting effects and texture details in the image to ensure that the image has sufficient visual appeal and artistic appeal;

[0056] S803. Expanded Text Content Review: In addition to basic spelling, grammar, and semantics checks, we also review the text's style and adaptability, and review the text layout in the comics to ensure that the text and images are in harmony and do not affect the reading experience.

[0057] S804. Deepen the review of logical relationships: Further analyze the logical structure of the comic story, check for plot holes, character behavior inconsistencies, and other issues, and sort out and review the timeline in the comic to ensure that the development of the story is reasonable and coherent in time.

[0058] In this embodiment, in S7, the specific steps are as follows:

[0059] S701. Intelligent Allocation Strategy Upgrade: Introducing a reinforcement learning algorithm to continuously learn and optimize AI allocation strategies based on feedback and results from previous review tasks. For example, reinforcement learning is used to determine the optimal combination and order of different AI expert models for reviewing different types of comics. Taking into account the update frequency of comics and market feedback, the priority of AI expert models is dynamically adjusted. For popular and frequently updated comic series, AI models with stronger performance and faster response speeds are prioritized.

[0060] S702. Dynamic adjustment of team formation: Based on the special subject matter and temporary review needs of the comics, a temporary AI expert team is formed in real time. For example, for a comic with a cross-cultural background, an AI expert model with relevant cultural background knowledge is added in a timely manner, and a sharing and collaboration mechanism for AI expert models is established, allowing different AI expert models to call and support each other when needed, thereby improving the overall efficiency of the team.

[0061] In this embodiment, in S9, the specific steps are as follows:

[0062] S901. Conduct in-depth mining and analysis of feedback data, focusing not only on surface errors but also on underlying patterns and causes. For example, correlation analysis can be used to identify whether certain types of errors are associated with specific styles, themes, or authors.

[0063] S902. Targetedly optimize the AI expert model based on the analysis results. For example, if a model is found to be prone to misjudgment when processing comics of a certain style, specialized style adaptability training can be performed on the model.

[0064] S903. In addition to a detailed list of errors and correction suggestions, the review report should include evaluation indicators and charts for the overall quality of the comic. For example, charts could be used to show the distribution of scores for image quality, text accuracy, and logical rationality.

[0065] S904. Provide a visual display of proofreading results, such as marking the error locations directly on the comic image, so that publishers can intuitively view and understand the problems.

[0066] In this embodiment, when in use, a dedicated comic book data receiving platform is established, data connections are established with major comic book publishers and authors, the original comic book data is obtained, the format of the received data is standardized, all image files are converted to a unified format and size, the image clarity is optimized, noise and blur are removed, and the image quality meets the review requirements. Advanced image recognition technology is used to identify and extract elements in the comics. For example, an object detection algorithm is used to identify elements such as characters, scenes, and props in the comics. Based on the recognition results and pre-set classification rules, the comics are classified. By building a classification model, the comics are automatically classified according to the theme and style.

[0067] A large amount of comic book data of different types is collected, including comics of various styles, themes and quality levels. Deep learning algorithms and transfer learning technology are used to train AI expert models with comic proofreading capabilities. During the training process, domain expert knowledge and proofreading rules are introduced. The AI expert model is assessed through cross-validation and actual case testing to evaluate its performance in comic proofreading tasks. According to the classification results and proofreading requirements of comic books, appropriate AI expert models are intelligently selected and deployed to build a diversified AI proofreading team, whose members include AI models with different professional directions such as image proofreading experts, text proofreading experts, and logic proofreading experts. The AI proofreading team is used to conduct comprehensive simulated human expert proofreading of comic books. Inspections are conducted from multiple dimensions, including image quality, text content, and logical relationships. A multi-round proofreading strategy is adopted, with a preliminary overall proofreading first, and then an in-depth detailed proofreading of the issues initially marked. Feedback data generated by the system during the proofreading process is collected and processed, including the judgment results of the AI model, proofreading time, error conditions, etc. The proofreading system is optimized and improved based on the feedback information, including adjusting the parameters of the AI expert model, optimizing the deployment strategy, etc. The proofreading report and correction suggestion report of the comic book are output based on the proofreading results to provide a basis for decision-making for the publisher.

[0068] The above describes in detail the comic book proofreading method based on thought chaining provided by the present invention. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above examples is only intended to help understand the method and core concept of the present invention. It should be noted that for those skilled in the art, various improvements and modifications can be made to the present invention without departing from the principles of the present invention, and such improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A comic book proofreading method based on thought chain, characterized by: include: S1. Data reception: receiving comic book data to be reviewed, including image files and related metadata; S2. Preprocessing: Standardize the image format to facilitate subsequent analysis and processing. For example, images of different resolutions are resized to a uniform size suitable for review and editing, and the image clarity is optimized to ensure that text and image details can be accurately recognized. S3. Comic content analysis: Utilize image recognition and natural language processing technology to analyze elements in comics and identify characters, scenes, text bubbles, and other content. S4. Comic content classification: Comics are classified according to factors such as theme, style (such as funny, passionate, science fiction, etc.), and target audience; S5. AI expert model training: Collect a large amount of comic book data of different types as training samples, including comics of various styles, themes, and quality levels; S6. AI expert model assessment: The AI expert model is assessed through cross-validation and real-world case testing to evaluate its performance in comic proofreading tasks, including accuracy, recall, F1 value, and other indicators. S7. AI deployment and proofreading team formation: Intelligently select and deploy appropriate AI expert models based on comic book classification results and proofreading requirements; and build a diverse AI proofreading team. S8. AI Simulated Proofreading and Comprehensive Proofreading: Utilize the AI proofreading team to conduct comprehensive simulated human expert proofreading of comic books, checking the image quality for reasonable character design, harmonious color matching, and accurate scene drawing. In terms of text content, we review whether the text is spelled correctly, whether the sentences are coherent, and whether the dialogue is consistent with the character's personality and plot development. In terms of logical relationships, we analyze whether the comic's storyline is coherent, whether the cause and effect relationships are reasonable, and whether the character's motivations are clear, ensuring that every issue is accurately verified and corrected. S9. Data feedback optimization and result output: Collect and process the feedback data generated by the system during the proofreading process, optimize and improve the proofreading system based on the feedback information, and output the proofreading report and correction suggestion report of the comic book based on the proofreading results. The report content includes a detailed error list, error type, correction suggestion and overall proofreading evaluation, providing a basis for decision-making for the publisher.

2. The comic book proofreading method based on thought chain according to claim 1 is characterized in that: In S2, the specific steps are as follows: S201. When collecting and annotating large-scale comic data sets, not only common comic types are included, but also comics from different regions and eras are categorized and annotated to ensure data diversity and representativeness. S202. Perform strict cleaning and preprocessing on the data to remove noise data and invalid annotations, thereby improving the quality of the training data. S203. Annotate the image according to the characteristics of the comic, such as annotating key elements, text areas and their attributes, to provide more accurate information for subsequent training.

3. The comic book proofreading method based on thought chain according to claim 1 is characterized in that: In said S8, the specific steps are as follows: S801. Enhanced image quality review: In addition to regular image quality checks, we now review the rationality of perspective switching and the effects of storyboards in comics. S802, using image analysis technology to check the light and shadow effects and texture details in the picture; S803. Expanded text content review: In addition to basic spelling, grammar, and semantics checks, we also review the text for style adaptability. S804. Deepen the review of logical relationships: Further analyze the logical structure of the comic story to check for plot holes, contradictions in character behavior, and other issues.

4. The comic book proofreading method based on thought chain according to claim 1 is characterized in that: In said S7, the specific steps are as follows: S701, Intelligent Allocation Strategy Upgrade: Introducing a reinforcement learning algorithm to continuously learn and optimize AI allocation strategies based on feedback and results from previous review tasks; S702. Dynamic adjustment of team formation: Based on the special subject matter of the comics and temporary review needs, a temporary AI expert team is formed in real time.

5. The comic book proofreading method based on thought chain according to claim 1 is characterized in that: In said S9, the specific steps are as follows: S901. Conduct in-depth mining and analysis of feedback data, focusing not only on surface error information but also on analyzing the underlying patterns and causes of errors. S902. Optimize the AI expert model in a targeted manner based on the analysis results; S903. In the review report, in addition to a detailed list of errors and correction suggestions, add evaluation indicators and charts to show the overall quality of the comics; S904. Provide a visual display of proofreading results, such as marking the error locations directly on the comic image, so that publishers can intuitively view and understand the problems.

Images