Automated analysis and judgment system for third-party copyrighted works

Abstract

To provide a system that streamlines licensing businesses, primarily focusing on supervision, and ensures consistent quality. [Solution] A system that uses AI to automatically analyze and determine whether a third-party work created by a third party conforms to regulations established by the copyright holder regarding the original work, and includes a regulations database and a work database recording multiple original works, wherein the analysis and determination processing program causes a computer to analyze and determine whether a third-party work conforms to the regulations, and the computer outputs the analysis and determination results to an information processing terminal.

Description

Technical Field

[0001] The present invention relates to an automatic analysis and determination system for third-party works that can automate the supervision work in the licensing business of character contents such as anime and manga, and automate the authenticity appraisal of pirated versions and the like.

Background Art

[0002] In recent years, the popularity of Japanese anime and manga contents has been increasing worldwide, and it has grown to the point where it is said that about half of the anime market scale of about 3 trillion yen is the overseas market. In order to expand the Japanese content industry in the future and further enhance the popularity of Japanese-originated contents, the activation of licensing businesses such as merchandising is extremely important.

[0003] In particular, the supervision work of contents in the licensing business is a very important work to confirm whether the works submitted by the licensee comply with the regulations including the style guide of the original work (original work) and other criteria.

[0004] However, the conventional supervision work is carried out by visual inspection by the person in charge, and requires a high level of human resources to confirm whether there are any problems in the use of text and contents, and the working hours may also be long. In addition, the quality may vary depending on the person in charge. Also, the confirmation work between the licensee and the licensor is often carried out multiple times, which requires a lot of man-hours and takes a long time to finalize the case. Furthermore, since it is difficult to conduct appropriate supervision unless the person in charge of supervision correctly understands the regulations of the contents, it also takes time to train such supervision personnel.

[0005] Although each stakeholder has a sense of issue from the perspective of improving the efficiency of the licensing business, there is still no direct work efficiency improvement system in supervision work and the like.

[0006] Furthermore, while the popularity of Japanese anime and manga content is growing worldwide, various pirated and copyright-infringing works are rampant on the internet. Authentication of genuine and licensed works requires manual visual inspection, thus presenting the same challenges as those in the licensing business mentioned above. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Patent No. 6818217 [Patent Document 2] Japanese Patent Publication No. 2021-5140 [Patent Document 3] Patent No. 6592722 [Patent Document 4] Japanese Patent Publication No. 2019-197597 [Overview of the project] [Problems that the invention aims to solve]

[0008] The present invention aims to provide a system that can improve the efficiency and standardize the quality of licensing businesses, particularly those focused on supervision. Specifically, it aims to analyze and determine whether a work provided by a licensee (a third-party work) conforms to regulations related to the original work, using an AI-based learning model that has learned from the original work. This will significantly reduce the time required for supervision, standardize the quality of supervision work, and reduce human error.

[0009] Furthermore, this invention aims to strengthen the protection of intellectual property rights in the content industry by enabling accurate and rapid identification of copyright-infringing works, such as pirated copies, which are spreading mainly on the internet.

[0010] To achieve the above objective, the present invention is a system that uses AI to automatically analyze and determine whether a third-party work created by a third party conforms to the regulations concerning the original work formulated by the copyright holder. The aforementioned regulations include at least character images, colors, design standards, stories, and prohibited uses related to the original works, created based on the style guide and / or related materials, and a regulations database recording these as digital information, and a works database recording multiple original works as digital information, An AI-powered analysis and judgment processing program that performs processing including segmentation and vectorization on image data uploaded by an information processing terminal of the third-party copyrighted work to be analyzed and judged, and determines whether the third-party copyrighted work conforms to the regulations, The system includes the regulation database, the copyright database, and a computer storing the analysis and judgment processing program, The analysis and determination processing program causes the computer to analyze and determine whether the third-party copyrighted work conforms to the regulations, and the computer outputs the analysis and determination result to the information processing terminal. It is characterized by the following: [Effects of the Invention]

[0011] According to the system of the present invention, the supervision work for licensing businesses of content such as anime and manga can be automated and made more efficient using AI, resulting in a significant reduction in working time and a decrease in human error. Furthermore, because the analysis is performed based on regulations established by the original copyright holder, quality can be standardized, unnecessary communication between licensors and licensees can be reduced, and the overall efficiency of the licensing business can be promoted.

[0012] Furthermore, according to the system of the present invention, by automating the supervision work using an AI system, not only is the burden on the supervising worker reduced, but even those without specialized knowledge can make appropriate judgments, and the time and cost of training personnel for supervision work can also be reduced.

[0013] Furthermore, the system according to the present invention can analyze and determine whether a third-party copyrighted work is being used correctly in accordance with the regulations for the use of the original work. Therefore, by uploading product images suspected of being pirated, the system can automatically perform authenticity verification, and can also be used as a support function for legal and intellectual property departments.

[0014] In this invention, the regulations refer to standards established by the original copyright holder. However, these regulations do not merely mean that the characters are depicted identically to the original work. They can also include a world view that takes into account expressions, poses, text, background images, etc. that violate public order and morals, as well as the story of the content and the basic settings of each character (personality, hobbies, preferences, etc.). This allows the AI ​​to automatically determine aspects that are often overlooked by the person in charge of supervision in the past. [Brief explanation of the drawing]

[0015] [Figure 1] Conceptual diagram of the system configuration relating to the present invention [Figure 2] Diagram explaining the processing flow [Figure 3] Figure (1) showing an example of the output analysis results. [Figure 4] Figure (2) showing an example of the output analysis results. [Modes for carrying out the invention]

[0016] The embodiments of the present invention will be described in detail below. The embodiments described below are illustrative examples for illustrating the present invention and do not limit the present invention to these embodiments only. Various modifications are possible as long as they do not depart from the essence of the present invention.

[0017] In addition, in the present embodiment described below, a system that automatically analyzes and determines whether a work created by a licensee (licensee) who has received a license from the copyright owner (licensor) of the original work conforms to the original work and the regulations of the original work formulated by the licensor will be described. The present invention is composed of a computer as hardware and a processing program for causing the computer to execute processing. Each processing described below is executed by the computer by the processing program unless otherwise specified.

[0018] FIG. 1 is a conceptual diagram of the configuration of an automatic analysis and determination system 1 (hereinafter referred to as system 1) according to the invention of the present application. The system 1 includes an analysis and determination processing program 2, a regulation database 3 that records regulations related to original works (contents) formulated by the copyright owner, a work database 4 that records a plurality of original works, and an information processing terminal 5 used by the licensor or licensee.

[0019] The regulations are created based on style guides and / or related materials regarding the original work (content), and these regulations are recorded in the regulation database 3 as text data and / or image data.

[0020] Specific examples of style guides include the work name, copyright, character image, character name, color of each character, design criteria, the worldview of the content, etc. Specific examples of the worldview of the content include the story of the content, the setting, the relationship between each character, etc. In addition, related materials are unorganized materials in which various types of information related to the original work are described (recorded), rather than information organized like a style guide.

[0021] Furthermore, items that are unacceptable in terms of the content's worldview are also recorded in Regulation Database 3. Specific examples of unacceptable items in terms of the content's worldview include design expressions that contradict the story, setting, and relationships between characters, as well as content that violates public order and morals.

[0022] Furthermore, the copyright database 4 records not only multiple image data related to the content, but also image data of each character that has been segmented by hair, face, clothing, etc., as well as numerical data that has been processed as necessary for analysis, such as vectorization of the image data.

[0023] The analysis and judgment processing program 2 includes an image processing AI and an analysis AI, and uses pre-trained large-scale language and image recognition models that have been trained using various data recorded in the regulation database 3 and the copyright database 4 as training data, respectively, to analyze and judge images related to third-party works uploaded to the information processing terminal 5. In addition to the pre-trained models, RAG may also be used in conjunction with the program, which may acquire relevant information from the copyright database 4 as needed and utilize it for analysis.

[0024] The following describes the processing flow using this system 1 (see Figure 2). (Step 1) The licensor uploads images using the original content created by the licensee to whom they have granted permission, using the information processing terminal 5. Alternatively, the licensee may upload the images using the information processing terminal 5. The provision of this system is not limited to being offered as SaaS; it may also be used in the applicant's own operating environment, such as by embedding it as a module in the user's system.

[0025] (Step 2) The analysis and judgment processing program 2 uses image processing AI to segment the image data uploaded by the licensee into individual parts (hair, face, eyes, clothing, other details, etc.) and performs image processing such as vectorization on each part.

[0026] The image processing described above is not limited to segmentation or vectorization; other image processing necessary for analysis is also acceptable. Furthermore, if image processing is not necessary due to the performance of the analysis AI used in the third step described below, the image processing in the second step may be omitted.

[0027] (Step 3) Image data and processed data obtained through the second step of image processing are input to the analysis AI. The analysis AI uses a trained model that has learned various data recorded in the regulation database 3 and the copyright database 4 to analyze the uploaded image and output the analysis results. In addition, as mentioned above, RAG may be used in conjunction with the trained model, as needed to obtain relevant information from these copyright databases 4 and utilize it in the analysis.

[0028] Furthermore, image processing AI and analysis AI may be used with existing models via API integration, etc.

[0029] Figures 3 and 4 show examples of output items for the analysis results. The analysis items for the analysis AI are set in advance, and the AI ​​analyzes basic information such as the name of the target content, the name of the depicted character, and the product genre, as well as items such as the text written on the image, the colors of each character, the silhouettes of each character, the basic settings of each character, and consistency with the worldview of the content, based on various data recorded in the trained regulation database 3 and copyright database 4, and outputs comments in language. Note that important points to check may differ for each piece of content, so the output items may be changed for each piece of content.

[0030] The analysis of basic information by the analysis AI involves the AI ​​recognizing the depicted objects in the uploaded image data and analyzing and outputting the name of the content depicted in the image data, the name of the depicted character, the product genre, etc., based on various data recorded in the trained regulation database 3 and copyright database 4.

[0031] The analysis of text in uploaded image data involves an analysis AI extracting text from the image data and comparing it with the character names, copyright notices, and dialogue expressions recorded in the trained regulation database 3 to analyze and output any inappropriate wording or errors.

[0032] The color analysis for each character is performed by an analysis AI that extracts the individual colors (RGB values ​​and color palettes) of each part of the character, such as hair, eyes, clothing, and accessories, from the uploaded image data and the data after image processing. The AI ​​then compares these colors to the respective reference values ​​based on various data recorded in the pre-trained regulation database 3 and copyright database 4, and outputs the results. For example, if there is a discrepancy between the color of a character's eyes and those in the pre-trained regulation database 3 and copyright database 4, the analysis comments will indicate the discrepancy and suggest corrections.

[0033] The silhouette analysis of each character involves the analysis AI extracting the outline of each character from the uploaded image data and processed image data, and then comparing it to standard values ​​for the character's silhouette (e.g., overall silhouette, facial proportions, etc.) from various data recorded in the trained regulation database 3 and copyright database 4. The analysis output is then performed, and the ratios between characters are also analyzed by comparing the extracted outlines to standard values. For example, if there is a discrepancy between the character's facial outline and the various data recorded in the trained regulation database 3 and copyright database 4, the analysis comments will indicate the discrepancy and suggest corrections.

[0034] The analysis of consistency between each character's basic settings and the content's worldview involves an AI analyzing the uploaded image data to recognize the depicted objects. It then analyzes and outputs whether the elements of each character and other depicted objects, as well as the design composition, are consistent with each character's basic settings and the content's worldview. For example, if there is a regulation that enemies cannot be placed side by side, the AI ​​will suggest the need to change designs where enemy characters are placed next to each other. Similarly, if a character who dislikes sweets is shown holding sweets, the AI ​​will suggest the need to change designs that contradict the character's basic settings.

[0035] The results of each of the above analysis processes are output to the information processing terminal 5. Furthermore, the interface allows for adjustments so that the licensor can add comments to identify any NG items that were missing from the regulations, or correct any errors in the AI's judgment. The added comments are recorded as regulations in the regulation database 3, and when the next analysis / judgment processing program 2 is used, the analysis AI performs the analysis with the past analysis comments reflected.

[0036] Furthermore, the output information and additional comments may be made available as external files such as PDF or Excel as part of the analysis results.

[0037] Furthermore, this system can be used not only for supervision work performed under a normal license, but also for analyzing whether or not images suspected of being pirated are being used legitimately, thus being utilized in determining the authenticity of pirated works.

[0038] Furthermore, increasing the database used to train the analytical AI enables a variety of applications. For example, by using new training data—such as existing license information, licensed designs, and product sales information—it becomes possible not only to evaluate consistency but also, as a byproduct, to suggest more effective designs. The basic structure of this system can also be utilized in various processes within the licensing business.

[0039] As described above, the system according to the present invention enables licensees and licensors to perform content supervision work efficiently, accurately, and appropriately, thereby reducing working time and human error. Furthermore, it is possible to provide appropriate correction instructions based on AI judgment for third-party copyrighted works that do not conform to the supervision regulations, thereby enabling the standardization of quality.

[0040] Furthermore, the system relating to this invention makes it possible to efficiently and accurately authenticate whether a product is a genuine or counterfeit version, in response to the widespread copyright infringement of pirated copies on the internet, thereby strengthening the protection of intellectual property rights.

[0041] The system according to the present invention not only enables efficient supervision work, but also contributes to the overall efficiency of the licensing business, thereby strengthening the protection of the content industry and contributing to its further development. [Explanation of Symbols]

[0042] 1. Automated Analysis and Judgment System 2. Analysis and Judgment Processing Program 3. Regulation Database 4. Copyright Database 5. Information Processing Terminal

Claims

1. A system that uses AI to automatically analyze and determine whether a third-party work created by a third party conforms to the regulations established by the copyright holder regarding the original work, The aforementioned regulations include at least character images, colors, design standards, stories, and prohibited uses related to the original works, created based on the style guide and / or related materials, and a regulations database recording these as digital information, and a works database recording multiple original works as digital information, An AI-powered analysis and judgment processing program that performs processing including segmentation and vectorization on image data uploaded by an information processing terminal of the third-party copyrighted work to be analyzed and judged, and determines whether the third-party copyrighted work conforms to the regulations, The system includes the regulation database, the copyright database, and a computer storing the analysis and judgment processing program, The analysis and determination processing program causes the computer to analyze and determine whether the third-party copyrighted work conforms to the regulations, and the computer outputs the analysis and determination result to the information processing terminal. An automated analysis and determination system for third-party copyrighted works, characterized by the following features.

2. The AI ​​performs analysis and judgment processing using a trained model that has learned from the data recorded in the regulation database and the data recorded in the copyright database. The automated analysis and determination system for third-party copyrighted works as described in feature 1.

3. The aforementioned analysis and judgment processing program is executed on the information processing terminal in a cloud environment or SaaS environment. An automated analysis and determination system for third-party copyrighted works as described in 1 or 2.

Images