Ai-based copyright protection device robust to content conversion, and control method thereof

Abstract

An AI-based copyright protection device robust to content conversion is disclosed. The AI-based copyright protection device robust to content conversion comprises: an input unit for receiving 2D content; and a processor for repeating, multiple times, a process of extracting distortion generated while converting the 2D content, so as to determine distortion information generated when the 2D content is converted multiple times, generating a watermark having a characteristic robust to the generated distortion on the basis of the distortion information, inserting the watermark into the 2D content, and finally converting the 2D content. Accordingly, it is possible to track copyright even after 2D content is converted into 3D content or is converted into various media formats, whereby copyright of original 2D content indiscriminately used in a metaverse space can be protected, copyright can be effectively protected in various multimedia environments, and rights of an original author can be protected.

Description

AI-based copyright protection device robust to content transformation and method for controlling the same

[0001] The present invention relates to an AI-based copyright protection device robust to content conversion and a control method thereof, and more specifically, to an AI-based copyright protection device robust to content conversion and a control method thereof that enables copyright information to be maintained without damage when 2D content is converted into various formats or compressed.

[0002] This invention is a study conducted by the Korea Electronics Technology Institute under the project title "Research on Neural Watermark Technology for Copyright Protection of Generative AI 3D Content" for the research project of the "Culture, Sports and Tourism R&D Support Program," with support from the Ministry of Culture, Sports and Tourism. (Project No. 2370000058, Project Unique Number RS-2024-00348469)

[0003] With the recent exponential advancement of information and communication technology, many people worldwide have been able to access multimedia content more easily through various online services. In particular, the market for multimedia content, such as movies and music, has experienced explosive growth in a non-face-to-face environment.

[0004] However, unlike the growth of the multimedia content market, the systems for protecting copyrights technically and institutionally have not been properly supplemented. Consequently, there is a problem in that the market for trading illegal copies of the many movies and music entering the multimedia content market is also expanding.

[0005] To identify such illegal copies, conventional copyright tracking systems did not account for situations where content was converted to different formats or compressed, resulting in cases where watermarks or digital signatures were damaged or failed to be recognized by detectors.

[0006] For example, as illustrated in FIG. 1, a conventional digital copyright tracking system has a problem in that when a watermark is inserted into original content and converted into another format such as a printed image, digital audio, video, or broadcast content, or compressed, the watermark inserted into the original content is damaged or altered and includes a watermark (11, 12, 13) different from the watermark inserted into the original content, so the watermark cannot be used to determine illegal copies or to determine whether it is the original.

[0007] In particular, this is emerging as a more serious problem in the metaverse space where various platforms interact.

[0008] Accordingly, there has been an increased need for robust copyright generation and tracking technology capable of maintaining copyright information for 2D content in complex digital environments such as the metaverse.

[0009] The objective of the present invention is to provide an AI-based copyright protection device robust to content conversion and a control method thereof, so as to maintain copyright information without damage when 2D content is converted into various formats or compressed.

[0010] An AI-based copyright protection device robust to content conversion according to one embodiment of the present invention for achieving such objectives includes an input unit that receives 2D content and a processor that repeats the process of extracting distortion occurring while converting the 2D content multiple times to determine distortion information occurring while converting the 2D content multiple times, and based on the distortion information, generates a watermark having robust characteristics against the distortion occurring and inserts it into the 2D content to finally convert it.

[0011] Additionally, an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention further includes an encoder unit and a decoder unit, and the processor can insert the generated watermark into the 2D content through the encoder unit and detect the generated watermark from the finally converted 2D content through the decoder unit.

[0012] In addition, the processor can convert the 2D content through a media conversion estimation model, and learn the distortion information and information regarding the generated watermark in the media conversion estimation model, and learn the conversion result of the 2D content and the effect of the conversion result on the watermark when the generated watermark is inserted into the 2D content during conversion.

[0013] In addition, the processor can generate the watermark based on the media conversion estimation model and insert it into the 2D content through the encoder unit.

[0014] In addition, the processor can detect the watermark from the finally converted 2D content through the decoder unit based on the media conversion estimation model.

[0015] In addition, the processor can convert the 2D content by performing at least one of format conversion, compression, resizing, and media method conversion on the 2D content.

[0016] Meanwhile, a control method for an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention includes the steps of receiving 2D content, repeating the process of extracting distortion occurring while converting the 2D content multiple times, determining distortion information that occurs while converting the 2D content multiple times, generating a watermark having robust characteristics against the distortion occurring based on the distortion information, and inserting the generated watermark into the 2D content to finally convert it.

[0017] In addition, a control method for an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention may further include the step of inserting the generated watermark into the 2D content through an encoder unit and the step of detecting the generated watermark from the finally converted 2D content through a decoder unit.

[0018] In addition, a control method for an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention may further include the step of converting the 2D content through a model, wherein the media conversion estimation model learns information regarding distortion information and a generated watermark, and inserts the generated watermark into the 2D content to learn the conversion result of the 2D content and the effect of the conversion result on the watermark during conversion.

[0019] Here, the step of inserting into the 2D content may generate the watermark based on the media conversion estimation model and insert it into the 2D content through the encoder unit.

[0020] In addition, the step of detecting the generated watermark can detect the watermark from the finally converted 2D content through the decoder unit based on the media conversion estimation model.

[0021] Meanwhile, a computer-readable recording medium according to one embodiment of the present invention may include a program for executing on a computer a control method for an AI-based copyright protection device that is robust to content conversion.

[0022] According to various embodiments of the present invention as described above, copyright tracking is possible even after converting 2D content into 3D or converting it into various media formats, thereby protecting the copyright of original 2D content that is indiscriminately stolen in the metaverse space, and effectively protecting copyright and the rights of the original creator in various multimedia environments.

[0023] FIG. 1 is a drawing for explaining the prior art.

[0024] FIG. 2 is a diagram illustrating the configuration of an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention.

[0025] FIG. 3 is a diagram illustrating a copyright generation and learning process robust to media conversion according to an embodiment of the present invention.

[0026] FIG. 4 is a diagram illustrating the configuration of an AI-based copyright protection device robust to content conversion according to another embodiment of the present invention.

[0027] FIG. 5 is a flowchart illustrating a control method for an AI-based copyright protection device robust to content conversion according to an embodiment of the present invention.

[0028] Figure 6 is a diagram illustrating the specific configuration of an AI-based copyright protection device robust to content conversion shown in Figure 1.

[0029] FIG. 7 is a drawing relating to a software module stored in a storage unit according to an embodiment of the present invention.

[0030] The present invention will be described in more detail below with reference to the drawings. Furthermore, in describing the present invention, specific descriptions of related known functions or configurations are omitted if it is determined that such detailed descriptions would unnecessarily obscure the essence of the invention. Additionally, the terms described below are defined considering their functions in the present invention, and these may vary depending on the intentions or relationships of the user or operator. Therefore, their definitions should be based on the content throughout this specification.

[0031] FIG. 2 is a diagram illustrating the configuration of an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention.

[0032] Referring to FIG. 2, an AI-based copyright protection device (100) robust to content conversion according to one embodiment of the present invention may include an input unit (110) and a processor (120).

[0033] Here, the AI-based copyright protection device (100) that is robust against content conversion may be implemented as an electronic device, server, desktop, portable terminal device, etc. having a separate configuration, or may be implemented as embedded software or a software module.

[0034] Additionally, a processor (120) according to one embodiment of the present invention may be understood as a configuration unit comprising hardware and / or software for performing computing operations. For example, the processor (120) may read a computer program and perform data processing for machine learning. The processor (120) may process computational processes such as processing input data for machine learning, extracting features for machine learning, and calculating errors based on backpropagation. A processor (120) for performing such data processing may include a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), or a field programmable gate array (FPGA). Since the above-described types of processors (120) are merely examples, the types of processors (120) may be configured in various ways within a range understandable to those skilled in the art based on the contents of this disclosure.

[0035] Additionally, a storage unit (150) according to one embodiment of the present invention may be understood as a configuration unit comprising hardware and / or software for storing and managing data processed by an AI-based copyright protection device (100) that is robust to content conversion. That is, the storage unit (150) may store data of any form generated or determined by the processor (120) and data of any form received by the network unit.

[0036] For example, the storage unit (150) may include at least one type of storage medium among flash memory type, hard disk type, multimedia card micro type, card type memory, RAM (random access memory), SRAM (static random access memory), ROM (read-only memory), EEPROM (electrically erasable programmable read-only memory), PROM (programmable read-only memory), magnetic memory, magnetic disk, and optical disk. Additionally, the storage unit (150) may include a database system that controls and manages data in a predetermined system. Since the above-described types of the storage unit (150) are merely examples, the types of the storage unit (150) may be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0037] The storage unit (150) can structure and organize data, combinations of data, and program code executable by the processor (120) that are necessary for the processor (120) to perform calculations. For example, the storage unit (150) may store an algorithm or program related to a media conversion estimation model. Additionally, the storage unit (620) may store a media conversion estimation model, program code that operates to perform learning of the media conversion estimation model, program code that operates the media conversion estimation model to receive information regarding distortion information and watermarks of 2D content and perform inference according to the purpose of use of an AI-based copyright protection device (100) robust to content conversion, and processed data generated as the program code is executed.

[0038] Meanwhile, the input unit (110) can receive 2D content, and the input unit (110) can receive data in the form of data stored in a separate memory or storage space or downloaded online via a wired communication method or a wireless communication method, and the input unit (110) can also perform the function of a communication unit or communication module that performs a wired communication method or a wireless communication method.

[0039] And, the processor (120) can determine distortion information that occurs as the 2D content is converted multiple times by repeating the process of extracting distortion that occurs while converting the 2D content multiple times, and can finally convert the 2D content by creating a watermark that has robust characteristics against distortion based on the distortion information and inserting it into the 2D content.

[0040] Specifically, the processor (120) can repeat the process of extracting distortions that occur while converting 2D content multiple times, and can extract and store a first distortion that occurs while performing compression or deformation by receiving 2D content as input, and can re-extract and store a second distortion that occurs while receiving the distorted 2D content as input again and performing compression or deformation.

[0041] The processor (120) can determine distortion information that occurs when 2D content is transformed into various forms multiple times by repeating the process described above several times. For example, distortion information that occurs when 2D content is transformed into various forms multiple times may appear as a change in data order, breakage, overlap, loss, etc., and the processor (120) can organize and determine distortion information that occurs when 2D content is transformed multiple times by repeating the process of extracting transformation and distortion of 2D content several times as described above.

[0042] And, the processor (120) can generate a watermark having robust characteristics that are not affected by distortions such as changes in data order, breakage, overlap, or loss when 2D content is transformed into various forms as described above, based on the determined distortion information.

[0043] And, the processor (120) can insert the generated watermark into the 2D content to finally convert it.

[0044] As a result, the watermark inserted into the finally converted 2D content is not affected by changes in data order, corruption, overlap, or loss as described above, even if the 2D content is converted into various formats or media, so the watermark inserted into the finally converted 2D content is not damaged or lost.

[0045] FIG. 3 is a diagram illustrating a copyright generation and learning process robust to media conversion according to an embodiment of the present invention.

[0046] Referring to FIG. 2, a copyright generation and learning process robust to media conversion according to one embodiment of the present invention is illustrated, and the media conversion distortion estimation algorithm illustrated on the left shows a process of repeating the process of extracting distortion parameters (230) appearing in the media-converted 3D content (220) multiple times when 3D content (210) with a watermark embedded in it is input as input data and media conversion is performed.

[0047] Here, the specification of the present invention mainly describes cases where 2D content is input and converted into various data formats, media formats, or 3D content, but cases where 3D content is input and converted into other 3D content formats can also be included, and the content can include all types of content regardless of whether it is 2D or 3D.

[0048] And, the media conversion distortion relearning process (240) illustrated on the right includes a process of extracting distortion parameters (230) appearing in the media-converted 3D content (220) when a 3D content (210) with a watermark embedded is input and media conversion is performed, repeating this process multiple times, inserting the generated watermark into the 3D content based on the determined distortion information to convert it, and detecting the watermark (241) from the converted 3D content through an extractor.

[0049] In this way, by reflecting the distortion information determined while converting the content into a derivative work, a derivative work with a watermark robust to various distortions can be generated, and by training a detector that reflects the distortion information, it becomes possible to detect and track the watermark from the media-converted 2D content.

[0050] Meanwhile, in copyright protection systems, encoders and decoders play a major role in inserting and detecting watermarks through AI models.

[0051] In this regard, an AI-based copyright protection device (100) robust to content conversion according to another embodiment of the present invention may further include an encoder unit and a decoder unit.

[0052] FIG. 4 is a diagram illustrating the configuration of an AI-based copyright protection device robust to content conversion according to another embodiment of the present invention.

[0053] Referring to FIG. 4, an AI-based copyright protection device (100) robust to content conversion according to another embodiment of the present invention may include an input unit (110), a processor (120), an encoder unit (130), and a decoder unit (140).

[0054] And, the processor (120) can insert the generated watermark into the 2D content through the encoder unit (130) and detect the generated watermark from the finally converted 2D content through the decoder unit (140).

[0055] And, the processor (120) converts 2D content through a media conversion estimation model, and learns information about distortion information and generated watermarks in the media conversion estimation model, and inserts the generated watermarks into the 2D content to learn the conversion result of the 2D content and the effect of the conversion result on the watermarks when converting.

[0056] That is, the processor (120) builds an AI-based media conversion estimation model by adding a model that estimates media conversion to the AI ​​model, and through this, can simulate and predict various conversions that the content may undergo.

[0057] In addition, the processor (120) can train the media conversion estimation model with simulation and prediction results regarding various conversions of content, and when a watermark generated according to the training results is inserted into 2D content through the encoder unit (130), the process of inserting the watermark considering various conversions and distortions of content can also be trained in the media conversion estimation model.

[0058] And, the processor (120) can insert a watermark generated according to the above-described learning result into 2D content through the encoder unit (130), and the watermark inserted in this way can be maintained without data distortion even after various media conversions of the content.

[0059] That is, the processor (120) can generate a watermark based on a media conversion estimation model and insert it into 2D content through the encoder unit (130).

[0060] And, since the processor (120) learns the expected conversion through the media conversion estimation model, it can accurately detect the watermark from the content converted through the decoder unit (140).

[0061] That is, the processor (120) can detect a watermark from 2D content that has undergone various conversions, that is, has been finally converted through the decoder unit (140) based on a media conversion estimation model.

[0062] Meanwhile, the processor (120) can convert 2D content by performing at least one of format conversion, compression, resizing, and media method conversion on the 2D content, and it is obvious that it may include various conversion methods in addition to the conversion methods described above.

[0063] In addition, format conversion or media method conversion may include various format or method changes, such as changing specifications, converting audio data into video data, or converting still image data into dynamic image data.

[0064] FIG. 5 is a flowchart illustrating a control method for an AI-based copyright protection device robust to content conversion according to an embodiment of the present invention.

[0065] Referring to FIG. 5, a control method for an AI-based copyright protection device robust to content conversion according to an embodiment of the present invention may include the steps of receiving 2D content (S510), repeating the process of extracting distortion occurring while converting 2D content multiple times (S520), determining distortion information occurring while converting 2D content multiple times (S530), generating a watermark having robust characteristics against distortion based on distortion information (S540), and inserting the generated watermark into 2D content to finally convert it (S550).

[0066] Here, a control method for an AI-based copyright protection device robust to content conversion according to another embodiment of the present invention may further include the steps of inserting a generated watermark into 2D content through an encoder unit and detecting a watermark generated from the finally converted 2D content through a decoder unit.

[0067] In addition, a control method for an AI-based copyright protection device robust to content conversion according to another embodiment of the present invention may further include the step of converting 2D content through a media conversion estimation model, training the media conversion estimation model with distortion information and information regarding a generated watermark, and inserting the generated watermark into the 2D content to learn the conversion result of the 2D content and the effect of the conversion result on the watermark during conversion.

[0068] In addition, the step of inserting the generated watermark into 2D content through the encoder unit can generate a watermark based on a media conversion estimation model and insert it into 2D content through the encoder unit.

[0069] In addition, the step of detecting the generated watermark can detect the watermark from the 2D content finally converted through the decoder unit based on a media conversion estimation model.

[0070] Meanwhile, as described above, a computer-readable recording medium may be provided that records a program for executing on a computer a control method for an AI-based copyright protection device robust to content conversion according to one embodiment of the present invention.

[0071] Figure 6 is a diagram illustrating the specific configuration of an AI-based copyright protection device robust to content conversion shown in Figure 1.

[0072] Referring to FIG. 6, an AI-based copyright protection device (100) robust to content conversion may include an input unit (110), a processor (120), and a storage unit (150).

[0073] The processor (120) controls the overall operation of the AI-based copyright protection device (100) which is robust to content conversion.

[0074] Specifically, the processor (120) includes RAM (121), ROM (122), main CPU (123), graphics processing unit (124), first to n interfaces (125-1 to 125-n), and a bus (126).

[0075] RAM (121), ROM (122), main CPU (123), graphics processing unit (124), first to n interfaces (125-1 to 125-n), etc. can be connected to each other via a bus (126).

[0076] The first to n interfaces (125-1 to 125-n) are connected to the various components described above. One of the interfaces may be a network interface connected to an external device through a network.

[0077] The main CPU (123) accesses the storage unit (150) and performs booting using the O / S stored in the storage unit (150). Then, it performs various operations using various programs, content, data, etc. stored in the storage unit (150).

[0078] In particular, the main CPU (123) can determine distortion information that occurs as the 2D content is converted multiple times by repeating the process of extracting distortion that occurs while converting the 2D content multiple times, and can finally convert the 2D content by creating a watermark that has robust characteristics against distortion based on the distortion information and inserting it into the 2D content.

[0079] A set of instructions for booting the system is stored in the ROM (122). When a turn-on command is input and power is supplied, the main CPU (123) copies the O / S stored in the storage unit (150) to the RAM (121) according to the instructions stored in the ROM (122), and executes the O / S to boot the system. When booting is complete, the main CPU (123) copies various application programs stored in the storage unit (150) to the RAM (121), and executes the application programs copied to the RAM (121) to perform various operations.

[0080] The graphics processing unit (124) generates a screen containing various objects such as icons, images, and text using a calculation unit (not shown) and a rendering unit (not shown). The calculation unit (not shown) calculates attribute values ​​such as coordinate values, shape, size, and color for each object to be displayed according to the layout of the screen based on a received control command. The rendering unit (not shown) generates a screen of various layouts containing objects based on the attribute values ​​calculated by the calculation unit (not shown).

[0081] In particular, the graphics processing unit (124) can implement objects generated by the main CPU (123) into a GUI (Graphic User Interface), icon, user interface screen, etc.

[0082] Meanwhile, the operation of the above-described processor (120) can be performed by a program stored in the storage unit (150).

[0083] The storage unit (150) stores various data, such as an O / S (Operating System) software module for operating an AI-based copyright protection device (100) that is robust against content conversion, and various multimedia content.

[0084] In particular, the storage unit (150) may include a software module for finally converting the 2D content by repeating the process of extracting distortion that occurs while converting the 2D content multiple times to determine distortion information that occurs while converting the 2D content multiple times, and generating a watermark that has robust characteristics against distortion that occurs based on the distortion information and inserting it into the 2D content.

[0085] FIG. 7 is a drawing relating to a software module stored in a storage unit according to an embodiment of the present invention.

[0086] Referring to FIG. 7, the storage unit (150) may store programs such as a distortion extraction module (151), a distortion information determination module (152), a watermark generation module (153), and a content final conversion module (154).

[0087] Meanwhile, the operation of the processor (120) described above can be performed by a program stored in the storage unit (150). Below, the detailed operation of the processor (120) using the program stored in the storage unit (150) will be explained in detail.

[0088] The distortion extraction module (151) can repeat the process of extracting distortion that occurs while converting 2D content multiple times.

[0089] Also, the distortion information judgment module (152) can determine distortion information that occurs as 2D content is converted multiple times.

[0090] Additionally, the watermark generation module (153) can generate a watermark that has robust characteristics against distortions that occur based on distortion information.

[0091] Additionally, the content final conversion module (154) can finally convert the generated watermark by inserting it into the 2D content.

[0092] Meanwhile, a non-transitory computer-readable medium storing a program that sequentially performs the control method according to the present invention may be provided.

[0093] A non-transient readable medium refers to a medium that stores data semi-permanently and can be read by a device, rather than a medium that stores data for a short period of time, such as a register, cache, or memory. Specifically, the various applications or programs described above may be stored and provided on non-transient readable media such as CDs, DVDs, hard disks, Blu-ray discs, USBs, memory cards, and ROMs.

[0094] In addition, although the bus is not shown in the aforementioned block diagram illustrating an AI-based copyright protection device robust to content conversion, a processor such as a CPU or microprocessor may be further included to determine distortion information that occurs as the 2D content is converted multiple times by repeating the process of extracting distortion that occurs while converting 2D content multiple times, and to generate a watermark that has robust characteristics against the distortion that occurs based on the distortion information and insert it into the 2D content to finally convert it.

[0095] Furthermore, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. It is understood that various modifications can be made by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention.

Claims

1. In an AI-based copyright protection device robust to content transformation, An input unit for receiving 2D content; and An AI-based copyright protection device robust to content conversion, comprising: a processor that repeats the process of extracting distortions occurring while converting the 2D content multiple times to determine distortion information occurring as the 2D content is converted multiple times, and based on the distortion information, generates a watermark having robust characteristics against the occurring distortions and inserts it into the 2D content to finally convert it.

2. In Paragraph 1, Encoder section; and It further includes a decoder section; The above processor is, An AI-based copyright protection device robust to content conversion, wherein the generated watermark is inserted into the 2D content through the encoder unit, and the generated watermark is detected from the finally converted 2D content through the decoder unit.

3. In Paragraph 2, The above processor is, An AI-based copyright protection device robust to content conversion, wherein the 2D content is converted using a media conversion estimation model, the media conversion estimation model is trained with information regarding distortion information and a generated watermark, and the generated watermark is inserted into the 2D content to learn the conversion result of the 2D content and the effect of the conversion result on the watermark during conversion.

4. In Paragraph 3, The above processor is, An AI-based copyright protection device robust to content conversion, which generates the watermark based on the above media conversion estimation model and inserts it into the 2D content through the above encoder unit.

5. In Paragraph 4, The above processor is, An AI-based copyright protection device robust to content conversion, which detects the watermark from the finally converted 2D content through the decoder unit based on the media conversion estimation model.

6. In Paragraph 5, The above processor is, An AI-based copyright protection device robust to content conversion, wherein the 2D content is converted by performing at least one of format conversion, compression, resizing, and media method conversion on the 2D content.

7. A method for controlling an AI-based copyright protection device robust to content transformation, Step of receiving 2D content; A step of repeating the process of extracting distortions occurring while converting the above 2D content multiple times; A step of determining distortion information that occurs as the above 2D content is converted multiple times; A step of generating a watermark having robust characteristics against the distortion occurring based on the distortion information above; and A method for controlling an AI-based copyright protection device robust to content conversion, comprising the step of inserting the generated watermark into the 2D content to finally convert it.

8. In Paragraph 7, A step of inserting the generated watermark into the 2D content through an encoder unit; and A control method for an AI-based copyright protection device robust to content conversion, further comprising the step of detecting the generated watermark from the finally converted 2D content through the decoder unit.

9. In Paragraph 8, A control method for an AI-based copyright protection device robust to content conversion, further comprising the step of converting the 2D content through a media conversion estimation model, wherein the media conversion estimation model is trained with information regarding distortion information and a generated watermark, and the generated watermark is inserted into the 2D content to learn the conversion result of the 2D content and the effect of the conversion result on the watermark during conversion.

10. In Paragraph 9, The step of inserting into the above 2D content is, A control method for an AI-based copyright protection device robust to content conversion, wherein the watermark is generated based on the above media conversion estimation model and inserted into the 2D content through the above encoder unit.

11. In Paragraph 10, The step of detecting the generated watermark above is, A control method for an AI-based copyright protection device robust to content conversion, wherein the watermark is detected from the finally converted 2D content through the decoder unit based on the above media conversion estimation model.

12. A computer-readable recording medium having a program stored on it for executing on a computer a control method for an AI-based copyright protection device robust to content conversion as described in any one of paragraphs 7 through 11.

Images