Dark watermark generation processing method and system based on deep learning

Abstract

The invention is suitable for the field of dark watermark generation processing, and provides a dark watermark generation processing method and system based on deep learning, and the system comprises a content attribute definition module, a watermark position recognition module, a watermark information generation module, a redundant information embedding module, and a content state monitoring and uploading locking module. According to the method, content type options are provided before the user uploads the content, and dark watermark embedding is carried out on the self-created content, so that the effectiveness of copyright protection is remarkably enhanced. And the dark watermark has the capability of actively responding to the uploading instruction, and can independently identify the uploading request and implement locking. The automatic identification and locking function does not depend on an external system for judgment any more, which means that the system can respond to potential infringement behaviors quickly in real time. Once an uploading request is detected, the locking logic in the redundant information is immediately started, uploading of the content is prohibited, and a user is reminded through warning information.

Description

Technical Field

[0001] The present invention belongs to the field of dark watermark generation and processing, and in particular to a dark watermark generation and processing method and system based on deep learning. Background Art

[0002] Dark watermark generation and processing technology is an advanced method for protecting digital content copyrights, aiming to provide stronger copyright protection for user-generated content. Unlike traditional visible watermarks, dark watermarks embed imperceptible information in the content, effectively identifying the rights of the original creator while maintaining the original quality of the content.

[0003] The background technology of dark watermark generation and processing is not only based on the theoretical foundation of information hiding and embedding algorithm, but also combines the needs of modern digital copyright protection, aiming to provide a more reliable protection mechanism for original content.

[0004] Currently, when users upload pictures, videos and other content on social platforms, they face significant copyright protection issues. Many contents are not related to copyright, and other users can forward, re-create or use them at will, while some contents are created by users themselves, such as paintings and self-created videos. Although such creative content can be downloaded, collected or used by other users, it should not be re-created or directly published on other platforms. Such behavior may lead to frequent infringement incidents and cause losses to the original creators.

[0005] In the prior art, some users try to add visible watermarks before uploading their own content. However, such watermarks have obvious defects. Visible watermarks are often easily removed by others through screenshots or using artificial intelligence technology, making it difficult to effectively protect the rights of the original creator. Such watermarks lack sufficient resistance and cannot provide effective protection when the content is illegally uploaded or shared. Summary of the invention

[0006] The purpose of the present invention is to provide a dark watermark generation and processing method based on deep learning, aiming to solve the technical problems existing in the prior art identified in the background technology.

[0007] The present invention is implemented as follows: a dark watermark generation and processing method based on deep learning, the method comprising: Obtain the content uploaded by the uploader, add a content type tag, obtain the uploader's attribute definition of the content, and set a watermark setting option for each attribute; Establish a deep learning model to extract content features of the uploaded content and identify the watermark embedding location; Generate a unique identifier and copyright information as watermark information for the content, and add a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; Add redundant information for verification and locking to the watermark information, and select the corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate content containing the watermark; When the content is uploaded again, the upload instruction is identified through redundant information, and the status instruction of the content is retrieved to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked.

[0008] As a further solution of the present invention, the steps of obtaining the content uploaded by the uploader, adding a content type tag, obtaining the attribute definition of the content by the uploader, and setting a watermark setting option for each attribute specifically include: Obtain the content uploaded by the uploader and automatically mark the content type based on file content analysis; Obtaining the uploader's attribute definition of the content, wherein the attribute definition includes two definitions: original and authorized; For the authorized attribute definition, only set the normal watermark option; for the original attribute definition, set the dark watermark option.

[0009] As a further solution of the present invention, the establishment of a deep learning model to extract content features of the uploaded content and identify the watermark embedding position specifically includes: Convert the uploaded content into a standard format and standardize the content; Establishing a deep learning model and marking a training data set, wherein the training data set includes content with a dark watermark and content without a dark watermark, and using the training data set to train the model; Use the trained deep learning model to process the uploaded content, extract the feature information of the content, and identify the watermark embedding location based on the extracted feature information.

[0010] As a further solution of the present invention, the unique identifier and copyright information are generated for the content as watermark information, and a status instruction is added according to the attribute definition of the content, and the status instruction is used to indicate the upload status of the content, specifically including: Obtain the binding information actively uploaded by the user, and generate the basic watermark information of the content based on it, and divide the watermark information into ordinary watermark and dark watermark according to the attribute definition of the content; When the content defined by the original attribute is uploaded, a unique identifier is randomly generated for the content, and the generated unique identifier is associated with a dark watermark corresponding to the content; Set upload status instructions for the contents of the authorization attribute definition and the original attribute definition respectively, wherein the authorization attribute definition chooses not to set additional upload status instructions, and the original attribute definition chooses to restrict upload status instructions; The basic watermark information, the unique identifier, and the upload status instruction are associated to generate the final watermark information.

[0011] As a further solution of the present invention, the redundant information for verification and locking is added to the watermark information, and according to the content type tag, a corresponding embedding technology is selected to combine the redundant information with the content to generate the content containing the watermark, which specifically includes: According to the type mark of the content, an embedding technology is selected to add redundant information to the watermark information defined by the original attribute and embed the redundant information into the content, and the redundant information is encrypted; embedding a status monitoring code and a locking code in the redundant information, wherein the status monitoring code is used to check and identify the upload instruction, and the locking code is used to lock the current content; Define the signal parameters of the upload behavior, and automatically activate the lock code when the signal parameters of the upload behavior are identified; The lock code is used to change the availability of the current content in the upload interface while the upload operation is in progress.

[0012] As a further solution of the present invention, the embedding technology is selected to add redundant information to the watermark information defined by the original attribute and embed the content according to the content type tag, specifically: For image-type content: use steganography to embed watermark information containing redundant information into the least significant bit of the image; For video content: select specific frames and use embedding technology to embed watermark information containing redundant information into the video data stream; For document-type content: embed watermark information containing redundant information in the metadata or specific paragraphs of the document, and keep its information invisible in the document.

[0013] As a further solution of the present invention, when the content is uploaded again, the upload instruction is identified through redundant information, and the status instruction of the content is retrieved to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked, which specifically includes: For content with dark watermarks, parse the status instructions and identify the signal parameters of whether there is upload behavior by monitoring the code; If it exists, the lock code is activated to implement the upload lock operation of the content; Change the availability of the current content in the upload content list and generate a pop-up window to prohibit uploading.

[0014] As a further solution of the present invention, it also includes: When generating watermark information, an information channel will also be generated for the uploader for the watermark information; Feedback information is generated when the lock code is activated, and the information channel is used to receive the generated feedback information.

[0015] Another object of the present invention is to provide a dark watermark generation and processing system based on deep learning, the system comprising: The content attribute definition module is used to obtain the content uploaded by the uploader, add the content type tag, obtain the attribute definition of the content by the uploader, and set the watermark setting selection for each attribute; The watermark position identification module is used to establish a deep learning model, extract content features of the uploaded content, and identify the watermark embedding position; A watermark information generation module, used to generate a unique identifier and copyright information for the content as watermark information, and to add a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; The redundant information embedding module is used to add redundant information for verification and locking to the watermark information, and select the corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate the content containing the watermark; The content status monitoring and upload locking module is used to identify the upload instruction through redundant information when the content is uploaded again, and to retrieve the status instruction of the content to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked.

[0016] The beneficial effects of the present invention are: This method significantly enhances the effectiveness of copyright protection by providing content type options before users upload content and embedding dark watermarks into self-created content. First, by embedding watermark information directly into images, videos or documents, the copyright mark of the uploaded content is ensured, and the dark watermark can be retained even after other users download it. This mechanism effectively tracks and identifies the originality of the content, helping the original author protect his or her work.

[0017] Secondly, dark watermarks have the ability to actively respond to upload instructions, and can independently identify upload requests and implement locks. This automatic identification and locking function no longer relies on external systems for judgment, which means that the system can respond to potential infringements in real time and quickly. Once an upload request is detected, the locking logic in the redundant information will be immediately activated, prohibiting the upload of content, and reminding the user through a warning message. This process not only protects the rights and interests of the original creator, but also enhances users' trust in the platform and reduces unnecessary copyright disputes.

[0018] In addition, the format design of redundant information makes watermark information more systematic and standardized. By clearly identifying the watermark status, locking logic and warning information, the system can clearly convey the intention and measures of copyright protection. This structured information management method not only improves the efficiency and accuracy of the system, but also facilitates subsequent content management. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of a dark watermark generation and processing method based on deep learning provided by an embodiment of the present invention; Figure 2 A flowchart of obtaining content uploaded by an uploader, adding a content type tag, obtaining the uploader's attribute definition of the content, and setting a watermark setting option for each attribute provided by an embodiment of the present invention; Figure 3 A flowchart of establishing a deep learning model, extracting content features of uploaded content, and identifying watermark embedding locations provided by an embodiment of the present invention; Figure 4 A flowchart of generating a unique identifier and copyright information as watermark information for content provided by an embodiment of the present invention, and adding a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; Figure 5 A flowchart of adding redundant information for verification and locking to watermark information provided by an embodiment of the present invention, and selecting a corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate content containing the watermark; Figure 6 A flowchart of an embodiment of the present invention for identifying an upload instruction through redundant information when content is uploaded again, and retrieving a status instruction of the content to determine whether the content is uploadable, and if the content is not uploadable, locking the upload operation of the content; Figure 7 A structural block diagram of a dark watermark generation and processing system based on deep learning provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0021] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0022] Figure 1 A flowchart of a dark watermark generation method based on deep learning provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes: S100, obtaining the content uploaded by the uploader, adding a content type tag, obtaining the attribute definition of the content by the uploader, and setting a watermark setting option for each attribute; like Figure 2 As shown, the process of obtaining the content uploaded by the uploader, adding a content type tag, obtaining the attribute definition of the content by the uploader, and setting a watermark setting option for each attribute specifically includes: S110, obtaining the content uploaded by the uploader, and automatically marking the content type based on the file content analysis; S120, obtaining the attribute definition of the content by the uploader, wherein the attribute definition includes two definitions: original and authorized; S130, only setting the common watermark option for the authorized attribute definition; setting the dark watermark option for the original attribute definition.

[0023] In this step, the system receives the file content uploaded by the user, including but not limited to images, videos, and documents. During the upload process, the system guides the user to conveniently select files through a friendly user interface. In terms of automatic content type analysis, the system uses deep learning algorithms to analyze uploaded files and can automatically identify file types, such as pictures, videos, documents, etc.

[0024] Next, the system guides users to define the attributes of the content they upload. The attribute definition includes two options: "original" and "authorized". Original content means that the content has copyright effects and cannot be uploaded again after downloading, avoiding the occurrence of infringement and fundamentally eliminating infringement; while for authorized content, it means that other users are allowed to upload again, re-create, etc. The system allows users to easily select the appropriate attribute through a drop-down menu or radio button.

[0025] Based on the user's attribute definition, the system automatically sets the watermark option. For content that is selected as "original", the system will enable the dark watermark option to ensure that the visibility of the content is not affected when the watermark is embedded; for content that is selected as "authorized", the system will only set a normal watermark to protect the user's legal right of use. After the user selects the attributes and completes the upload, the system provides feedback information to confirm the selected content type and attribute definition, and informs the user of the next steps. The user can confirm and modify at this stage to ensure the accuracy of the information.

[0026] Through highly automated content type analysis, the workload of manual input by users is greatly reduced, and the user experience is improved. Secondly, clear copyright statements enable users to clearly indicate their rights, thereby enhancing the copyright protection of original content. In addition, flexible watermark strategies automatically select watermark types according to content attributes, so that different types of content have targeted protection measures, thereby optimizing the use of resources. By setting dark watermarks and introducing status instructions, the system reduces the secondary creation and infringement of original content by others from the source. In this process, the system provides necessary education and guidance when uploading content, enhances users' awareness of the importance of copyright, and thus clarifies the boundaries of reasonable use of other people's works. At the same time, using deep learning technology to analyze content features, the system can more accurately identify and protect uploaded original content, enhancing the effectiveness of copyright protection. Finally, through effective copyright protection measures, the quality of social platform content is improved, a healthy creative environment is promoted, and users' sense of trust is enhanced.

[0027] S200, establishing a deep learning model, extracting content features of the uploaded content, and identifying the watermark embedding position; like Figure 3 As shown, the deep learning model is established to extract content features of the uploaded content and identify the watermark embedding position, which specifically includes: S210, converting the uploaded content into a standard format and performing standardization processing on the content; S220, establishing a deep learning model and marking a training data set, wherein the training data set includes content with a dark watermark and content without a dark watermark, and using the training data set to train the model; S230, using the trained deep learning model to process the uploaded content, extract feature information of the content, and identify the watermark embedding position based on the extracted feature information.

[0028] In this step, the system converts the uploaded content into a standard format to ensure that all content is processed in a unified format. This process includes format conversion of images, videos, and documents, so that different types of content can be effectively processed by deep learning models.

[0029] Next, the system will normalize the content. This means that before being input into the model, the content will be adjusted to the same size, color mode, and other characteristics to ensure consistency and accuracy in model training. Normalization can prevent model recognition errors caused by differences in input data.

[0030] The system then builds a deep learning model and labels the training data set. The training data set contains content with and without dark watermarks, which will be used to train the model to learn the characteristics and distribution of watermarks. The process of labeling the training data set includes classifying each content and accurately marking the specific location of the watermark. This process requires a large amount of sample data to ensure that the model training can cover a variety of styles and formats.

[0031] When training the model using a labeled training data set, the system continuously optimizes the model's parameters through a back-propagation algorithm to improve the accuracy of watermark recognition. After training, the model will have the ability to extract content features and identify watermark embedding locations.

[0032] During the uploaded content processing phase, the system uses the trained deep learning model to process the content and extract the feature information of the content. Based on the extracted feature information, the model can accurately identify the embedding position of the watermark and provide the necessary information for subsequent watermark embedding.

[0033] S300, generating a unique identifier and copyright information as watermark information for the content, and adding a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; like Figure 4 As shown, the unique identifier and copyright information are generated for the content as watermark information, and a status instruction is added according to the attribute definition of the content. The status instruction is used to indicate the upload status of the content, specifically including: S310, obtaining binding information actively uploaded by the user, generating basic watermark information of the content based on the binding information, and dividing the watermark information into ordinary watermark and dark watermark according to the attribute definition of the content; S320, when the content defined by the original attribute is uploaded, a unique identifier is randomly generated for the content, and the generated unique identifier is associated with a dark watermark corresponding to the content; S330, respectively setting an upload status instruction for the contents of the authorization attribute definition and the original attribute definition, wherein the authorization attribute definition chooses not to set an additional upload status instruction, and the original attribute definition chooses to restrict the upload status instruction; S340, associating the basic watermark information, the unique identifier, and the upload status instruction to generate final watermark information.

[0034] In this step, the binding information actively uploaded by the user will be obtained, which may include the user's identity information, contact information, and other metadata related to the content. The purpose of this step is to associate the user's identity with the content he or she uploaded, so as to facilitate subsequent copyright tracing and management.

[0035] Based on the acquired binding information, the system generates basic watermark information for the content. The generation of watermark information includes the definition of content attributes, especially the classification of watermark information into ordinary watermarks and dark watermarks. When content with original attribute definition is uploaded, the system will randomly generate a unique identifier. This identifier not only ensures that each content has a unique identity, but also associates it with the dark watermark corresponding to the content, so that the copyright information of the content can be clearly identified in subsequent use.

[0036] When the attribute of the content is defined as authorized, the system sets the upload status instruction to "no additional restrictions", which means that other users are allowed to freely use the content within the scope of authorization. For content defined as original, the system will set a restricted upload status instruction to indicate that the content is not allowed to be re-uploaded or re-created. This status instruction setting helps protect the legitimate rights and interests of the original creator and prevent unauthorized infringement.

[0037] Finally, the system associates the basic watermark information, unique identifier and upload status instructions to generate the final watermark information. This information will be embedded in the content to ensure that each uploaded content has a clear copyright identification and usage status, forming a complete copyright protection chain.

[0038] This step obtains the user's binding information and generates basic watermark information, so that each uploaded content is closely associated with the user's identity, which provides an important basis for future copyright tracing. The randomly generated unique identifier ensures the uniqueness of each original content, enhances the accuracy of content identification, and reduces the risk of confusion.

[0039] Secondly, the clear distinction between ordinary watermarks and dark watermarks enables the system to flexibly respond to different types of content protection needs. The restricted upload status instructions for original content effectively protect the rights of creators and prevent their works from being used or re-created at will. The loose management of authorized content encourages users to share and use content within the legal scope, promoting the activity of the platform.

[0040] By effectively associating basic watermark information, unique identifiers, and upload status instructions, the system has built a complete copyright protection mechanism. This not only enhances users' trust in the platform and improves user experience, but also establishes a good content management environment for social platforms. In addition, this method also helps reduce the occurrence of infringement incidents and provides a safer creative space for original authors, thereby encouraging more users to participate in content creation and promoting the healthy development of the platform.

[0041] S400, adding redundant information for verification and locking to the watermark information, and selecting a corresponding embedding technology to combine the redundant information with the content according to the content type tag to generate content containing the watermark; like Figure 5 As shown, the redundant information for verification and locking is added to the watermark information, and according to the content type tag, a corresponding embedding technology is selected to combine the redundant information with the content to generate the content containing the watermark, which specifically includes: S410, according to the type mark of the content, using an embedding technology to add redundant information to the watermark information defined by the original attribute and embed the redundant information into the content, and the redundant information is encrypted; S420, embedding a status monitoring code and a locking code in the redundant information, wherein the status monitoring code is used to check and identify the upload instruction, and the locking code is used to lock the current content; S430, defining a signal parameter of an upload behavior, and automatically activating a lock code when the signal parameter of the upload behavior is identified; S440, the lock code is used to change the availability of the current content in the upload interface when the upload operation is in progress.

[0042] This step selects the appropriate embedding technology based on the type tag of the content. When processing watermark information defined by the original attribute, the system will use a watermark embedding algorithm to effectively embed the watermark into the image, video or document without significantly affecting the quality of the content.

[0043] Next, the redundant information is encrypted to enhance its security. This process prevents malicious users from obtaining or modifying the watermark information through simple extraction. The encrypted redundant information is embedded in the content, ensuring that only authorized users or systems can decrypt and verify the relevant information.

[0044] The system will also embed status monitoring codes and locking codes in the redundant information. The status monitoring code is used to check and identify upload instructions in real time to ensure that the system can effectively monitor the usage status of the content. Once the signal parameters of the upload behavior are detected, the system will automatically activate the locking code. The purpose of the locking code is that when the content is identified as not uploadable, it will change the availability of the current content during the upload operation to ensure that the content will not be uploaded by mistake or used infringingly.

[0045] In addition, the lock code can be applied at multiple levels, for example, it can trigger the system's warning mechanism to notify users that the content has been locked, or take corresponding measures. This mechanism can effectively prevent unauthorized content from being uploaded again, thereby improving the security and copyright protection of original content.

[0046] Therefore, the dark watermark mentioned above is not only embedded in the content, but also can actively respond to upload instructions. This watermark should have automatic recognition and locking functions, allowing the dark watermark to actively identify upload instructions and implement locking without relying on external system judgment. By deeply binding redundant information with content and realizing a self-identification mechanism, it can effectively prevent content from being illegally uploaded and protect copyright.

[0047] In the embodiment of the present invention, the embedding technology is selected to add redundant information to the watermark information defined by the original attribute and embed the content according to the type mark of the content, specifically: For image-type content: use steganography to embed watermark information containing redundant information into the least significant bit of the image; For video content: select specific frames and use embedding technology to embed watermark information containing redundant information into the video data stream; For document-type content: embed watermark information containing redundant information in the metadata or specific paragraphs of the document, and keep its information invisible in the document.

[0048] S500, when the content is uploaded again, the upload instruction is identified through redundant information, and the status instruction of the content is retrieved to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked.

[0049] like: #Determine the redundant information format Redundant information = { "Watermark Status":"Original", "Lock Logic": "If an upload request is detected, lock and prompt the user", "Warning message": "This file cannot be uploaded due to copyright protection" } like Figure 6As shown, when the content is uploaded again, the upload instruction is identified through redundant information, and the status instruction of the content is retrieved to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked, which specifically includes: S510, for the content with dark watermark, parse the status instruction and identify the signal parameter of whether there is upload behavior by monitoring the code; S520, if it exists, activating the lock code and implementing the upload lock operation of the content; S530: Change the availability of the current content in the upload content list, and generate a pop-up window to prohibit uploading.

[0050] In this step, when the content containing the dark watermark is uploaded again, the content with the dark watermark is first parsed to extract the status instructions therein. This process involves reading the status monitoring code in the watermark information to understand the upload status of the current content. The status instructions may include multiple states such as "uploadable", "unable to upload", and "review required" to ensure that the system can handle different situations accordingly. Through the monitoring code, the system identifies whether there are signal parameters for upload behavior, which can be achieved by analyzing the user's operation behavior (such as clicking the upload button, dragging files, etc.). The system will continue to monitor these operations to ensure that the user's upload intention can be captured in time. If the system detects the signal parameters of the upload behavior, it will automatically activate the locking code and trigger a series of operations to ensure that the content is temporarily locked or restricted during the upload process. In the upload content list, the system will update the availability of the current content. If the content is marked as unavailable for upload, the system will display the content as "upload prohibited" or "content locked" on the user interface for the user to understand. When the content is locked, the system will pop up a window to remind the user that the content cannot be uploaded due to copyright reasons. This notice can include more information, such as the original creator of the content, relevant copyright information, and instructions on how to resolve the issue (such as contacting platform support, modifying the content, etc.).

[0051] By monitoring upload behavior, the system can respond immediately when users try to upload protected content, preventing potential infringement and avoiding copyright disputes. When content is locked, the system can promptly notify users through pop-up windows to let them know the current status of the content. This transparent communication method can effectively improve the user experience and reduce user confusion and dissatisfaction with the system. Step S400 provides a dynamic content management method that can adjust the availability of content in real time according to the upload status of the content to ensure that the platform is always in a legal and compliant state. By implementing the upload lock operation, the system provides an additional layer of protection for the original creators to prevent their works from being used without authorization, which not only enhances the original creators' sense of trust, but also promotes the creation of more high-quality content. An effective copyright protection mechanism enhances users' trust in the platform and improves the platform's credibility. This sense of trust will attract more creators to participate and promote the positive development of the platform.

[0052] The dark watermark generation and processing method based on deep learning proposed in the present invention also includes: When generating watermark information, an information channel will also be generated for the uploader for the watermark information; Feedback information is generated when the lock code is activated, and the information channel is used to receive the generated feedback information.

[0053] Figure 7 The structural block diagram of the dark watermark generation and processing system based on deep learning provided by the embodiment of the present invention is as follows: Figure 7 As shown, the system comprises: The content attribute definition module 100 is used to obtain the content uploaded by the uploader, add a content type tag, obtain the attribute definition of the content by the uploader, and set a watermark setting option for each attribute; The watermark position identification module 200 is used to establish a deep learning model, extract content features of the uploaded content, and identify the watermark embedding position; The watermark information generating module 300 is used to generate a unique identifier and copyright information as watermark information for the content, and to add a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; The redundant information embedding module 400 is used to add redundant information for verification and locking to the watermark information, and select a corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate content containing the watermark; The content status monitoring and upload locking module 500 is used to identify the upload instruction through redundant information when the content is uploaded again, and retrieve the status instruction of the content to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked.

[0054] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0055] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0056] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0057] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A dark watermark generation and processing method based on deep learning, characterized in that: The method comprises: Obtain the content uploaded by the uploader, add a content type tag, obtain the uploader's attribute definition of the content, and set a watermark setting option for each attribute; Establish a deep learning model to extract content features of the uploaded content and identify the watermark embedding location; Generate a unique identifier and copyright information as watermark information for the content, and add a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; Add redundant information for verification and locking to the watermark information, and select the corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate content containing the watermark; When the content is uploaded again, the upload instruction is identified through redundant information, and the status instruction of the content is retrieved to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked.

2. The method according to claim 1, characterized in that The process of obtaining the content uploaded by the uploader, adding a content type tag, obtaining the attribute definition of the content by the uploader, and setting a watermark setting option for each attribute specifically includes: Obtain the content uploaded by the uploader and automatically mark the content type based on file content analysis; Obtaining the uploader's attribute definition of the content, wherein the attribute definition includes two definitions: original and authorized; For the authorized attribute definition, only set the normal watermark option; for the original attribute definition, set the dark watermark option.

3. The method according to claim 2, characterized in that The deep learning model is established to extract content features of the uploaded content and identify the watermark embedding position, specifically including: Convert the uploaded content into a standard format and standardize the content; Establishing a deep learning model and marking a training data set, wherein the training data set includes content with a dark watermark and content without a dark watermark, and using the training data set to train the model; Use the trained deep learning model to process the uploaded content, extract the feature information of the content, and identify the watermark embedding location based on the extracted feature information.

4. The method according to claim 3, characterized in that The unique identifier and copyright information are generated for the content as watermark information, and a status instruction is added according to the attribute definition of the content. The status instruction is used to indicate the upload status of the content, specifically including: Obtain the binding information actively uploaded by the user, and generate the basic watermark information of the content based on it, and divide the watermark information into ordinary watermark and dark watermark according to the attribute definition of the content; When the content defined by the original attribute is uploaded, a unique identifier is randomly generated for the content, and the generated unique identifier is associated with a dark watermark corresponding to the content; Set upload status instructions for the contents of the authorization attribute definition and the original attribute definition respectively, wherein the authorization attribute definition chooses not to set additional upload status instructions, and the original attribute definition chooses to restrict upload status instructions; The basic watermark information, the unique identifier, and the upload status instruction are associated to generate the final watermark information.

5. The method according to claim 4, characterized in that The method of adding redundant information for verification and locking to the watermark information and selecting a corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate the content containing the watermark specifically includes: According to the type mark of the content, an embedding technology is selected to add redundant information to the watermark information defined by the original attribute and embed the redundant information into the content, and the redundant information is encrypted; embedding a status monitoring code and a locking code in the redundant information, wherein the status monitoring code is used to check and identify the upload instruction, and the locking code is used to lock the current content; Define the signal parameters of the upload behavior, and automatically activate the lock code when the signal parameters of the upload behavior are identified; The lock code is used to change the availability of the current content in the upload interface while the upload operation is in progress.

6. The method according to claim 5, characterized in that The method of adding redundant information to the watermark information defined by the original attribute and embedding the content according to the type mark of the content is as follows: For image-type content: use steganography to embed watermark information containing redundant information into the least significant bit of the image; For video content: select specific frames and use embedding technology to embed watermark information containing redundant information into the video data stream; For document-type content: embed watermark information containing redundant information in the metadata or specific paragraphs of the document, and keep its information invisible in the document.

7. The method according to claim 5, characterized in that When the content is uploaded again, the upload instruction is identified through redundant information, and the status instruction of the content is retrieved to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked, which specifically includes: For content with dark watermarks, parse the status instructions and identify the signal parameters of whether there is upload behavior by monitoring the code; If it exists, the lock code is activated to implement the upload lock operation of the content; Change the availability of the current content in the upload content list and generate a pop-up window to prohibit uploading.

8. The method according to claim 5, characterized in that Also includes: When generating watermark information, an information channel will also be generated for the uploader for the watermark information; Feedback information is generated when the lock code is activated, and the information channel is used to receive the generated feedback information.

9. A dark watermark generation and processing system based on deep learning, characterized in that: The system comprises: The content attribute definition module is used to obtain the content uploaded by the uploader, add the content type tag, obtain the attribute definition of the content by the uploader, and set the watermark setting selection for each attribute; The watermark position identification module is used to establish a deep learning model, extract content features of the uploaded content, and identify the watermark embedding position; A watermark information generation module, used to generate a unique identifier and copyright information as watermark information for the content, and to add a status instruction according to the attribute definition of the content, wherein the status instruction is used to indicate the upload status of the content; The redundant information embedding module is used to add redundant information for verification and locking to the watermark information, and select the corresponding embedding technology to combine the redundant information with the content according to the content type mark to generate the content containing the watermark; The content status monitoring and upload locking module is used to identify the upload instruction through redundant information when the content is uploaded again, and to retrieve the status instruction of the content to determine whether it is uploadable content. If it is not uploadable content, the upload operation of the content is locked.

Images