Method and System for Providing Advertisements in Archive Content Using Artificial Intelligence

Abstract

A method for providing an advertisement for archive content using artificial intelligence according to one technical aspect of the present invention is performed in an advertisement providing system and comprises: a step of receiving an archive content that has already been produced and distributed or is currently being distributed; a step of generating a preprocessed archive content by identifying and removing existing brand elements included in the archive content using an artificial intelligence-based object detection technology; and a step of generating an advertisement content by synthesizing an advertisement brand element for an advertised target brand to the preprocessed archive content.

Description

Technology Field

[0001] The present invention relates to the fields of advertising technology and image processing technology, and more specifically, to a method and system for generating and providing an advertising version optimized for platform-specific characteristics by using artificial intelligence to detect and remove existing brand elements within an image of archived content that has already been produced and distributed or is currently being distributed, and by naturally inserting advertising content, such as PPL (Product Placement). Background Technology

[0002] In general, advertising exposure through video content accounts for a significant portion of a company's marketing activities.

[0003] For example, broadcasters and OTT platforms enter into Product Placement (PPL) contracts during the production phase of dramas and variety shows to expose specific brands, while online platforms such as YouTube and TikTok generate advertising revenue by inserting pre-roll advertisements into videos produced by individual creators. Furthermore, advertisers expect to efficiently promote their brands by utilizing already proven popular content.

[0004] Existing PPL technology was primarily based on the premise of inserting advertisements during the production process. In other words, conventional technology could only composite advertising images during filming or post-production editing, and it was technically impossible to insert new advertisements or replace existing ones in content once production was complete. Consequently, it was difficult to generate additional revenue from archived content owned by broadcasters or older videos by individual creators once advertising contracts expired or they lost their relevance.

[0005] Furthermore, conventional advertising delivery methods failed to adequately consider the specific characteristics of each platform. Since the same ad video was distributed uniformly across all platforms, optimization reflecting specific features such as aspect ratios, user age groups, and interests for platforms like YouTube, Instagram, and TikTok was impossible. Consequently, the accuracy of ad targeting was low, and the effectiveness of the advertisements was inevitably limited.

[0006] Furthermore, conventional revenue structures had limitations due to a lack of transparency. Even when advertising revenue was generated, the distribution process among original creators, platforms, and advertisers often relied on manual procedures or unclear settlement rules, leading to a lack of trust between creators and advertisers.

[0007] For this reason, conventional technology has failed to provide the reuse of completed video content, differentiated advertising by platform, and transparent and automated revenue distribution; consequently, it has a fundamental limitation in that it cannot provide the level of sustainable monetization required by the content industry and individual creators. Prior art literature

[0008] Republic of Korea Published Patent Application No. 10-2025-0043926 (Date of publication: March 31, 2025) The problem to be solved

[0009] The present invention aims to solve the aforementioned problems, and an embodiment according to the present disclosure aims to enable the re-monetization of archive content, which was previously impossible, by detecting and removing existing brand elements from archive content that has already been produced and distributed using artificial intelligence, and then naturally synthesizing an advertisement (e.g., Product Placement (PPL)) into a new brand image.

[0010] One embodiment according to the present disclosure automatically recognizes not only visual objects such as brand logos, products, and signs within a video, but also multimodal elements such as voice mentions and subtitles, and removes them through inpainting and video restoration technology, thereby enabling the insertion of new advertisements without a sense of incongruity despite the expiration of existing advertising contracts or changes in trends.

[0011] An embodiment according to the present disclosure enables the provision of dynamic advertisements that reflect content context and timeliness, unlike conventional technology which was limited to single advertisement exposure, by automatically matching the most suitable brand through a comprehensive analysis of an advertiser database, real-time market trends, and audience characteristics by platform.

[0012] An embodiment according to the present disclosure analyzes the popularity and diffusion potential of a video to determine economic feasibility and optimizes the timing of ad insertion, thereby providing a new opportunity to immediately monetize even past content when it goes viral belatedly.

[0013] An embodiment according to the present disclosure enables the maximization of advertising effects for each platform by creating multiple versions of the same original video with differentiated aspect ratios, ad display methods, and inserted brands according to the characteristics of each platform, such as YouTube, TikTok, and Instagram.

[0014] However, the problems to be solved in this disclosure are not limited to those mentioned above, and may be expanded in various ways without departing from the spirit and scope of this disclosure. means of solving the problem

[0015] One technical aspect of the present invention proposes a method for providing advertisements for archive content using artificial intelligence. The method is performed in an advertisement providing system and comprises: receiving archive content that has already been produced and distributed or is currently being distributed; generating preprocessed archive content by identifying and removing existing brand elements included in the archive content using an artificial intelligence-based object detection technology; and generating advertisement content by synthesizing advertising brand elements for an advertising target brand into the preprocessed archive content.

[0016] In one embodiment, the advertising provision method may further include the step of generating advertising content for multiple platforms by performing aspect ratio conversion based on the generated advertising content so that it can be applied to multiple platforms.

[0017] In one embodiment, the step of generating the preprocessed archive content may include: identifying visual elements, voice elements, and text elements within the archive content, respectively, to detect existing brand elements based on multimodal; deleting the existing brand elements for a plurality of consecutive frames containing the detected existing brand elements; and inpainting the blank area created by the deletion process based on background information and light source information of the plurality of consecutive frames.

[0018] In one embodiment, the step of detecting existing brand elements based on the multimodal may include the step of identifying existing brand elements corresponding to the production time of the archive content by referring to a brand evolution vector learned based on the changes in the visual design of the existing brand over time.

[0019] In one embodiment, the step of deleting the existing brand element may include: a step of dividing and identifying the existing brand element in pixel units for each frame of the consecutive plurality of frames using a deep learning-based object segmentation network; and a step of deleting the existing brand element divided in pixel units while tracking its location and size for each frame of the consecutive plurality of frames.

[0020] In one embodiment, the step of inpainting based on background information and light source information of a plurality of consecutive frames may include: a step of performing a first-stage restoration process to restore the structural lines of the blank area to form a basic shape; and a step of performing a second-stage restoration process to perform texture inpainting based on color and lighting consistency by reflecting background information and light source information extracted from a plurality of consecutive frames on the area where the structural lines were restored by the first-stage restoration.

[0021] In one embodiment, the step of generating the advertising content may include: a step of restoring the three-dimensional spatial structure of the preprocessed archive content using monocular depth estimation; and a step of correcting and inserting the position, size, and rotation of the advertising brand element based on the restored spatial structure.

[0022] In one embodiment, the step of generating the advertising content may further include: a step of estimating the direction and intensity of lighting by analyzing the light source information of the preprocessed archive content; and a step of synthesizing the advertising brand element so that it is naturally inserted into the archive content by applying shadow and reflection effects of the advertising brand element according to the estimated lighting conditions.

[0023] Another technical aspect of the present invention proposes an advertising delivery system. The advertising delivery system comprises: a user terminal providing information on a brand to be advertised; and an advertising delivery server that receives the brand information to be advertised from the user terminal and generates advertising content by reflecting the brand information to be advertised in archive content that has already been produced and distributed or is currently being distributed. The advertising delivery server generates preprocessed archive content by identifying and removing existing brand elements included in the archive content using an artificial intelligence-based object detection technology, and generates advertising content by synthesizing the brand elements to be advertised into the preprocessed archive content. Effects of the invention

[0024] According to various embodiments of the present disclosure, by detecting and removing existing brand elements within archive content based on artificial intelligence and naturally inserting new advertisements (e.g., PPL), the present invention provides the effect of overcoming the problems of dependence on the production stage and the inability to re-monetize archive content inherent in conventional technology. Accordingly, past videos from broadcasters and OTT platforms, or old videos from individual creators, can also be utilized as new advertising media.

[0025] According to various embodiments of the present disclosure, the optimal advertising brand can be automatically matched by comprehensively analyzing the advertiser database, real-time market trends, and user characteristics by platform. Therefore, even the same content can be provided in differentiated advertising versions tailored to each platform, such as YouTube, TikTok, and Instagram, thereby significantly improving advertising targeting accuracy and advertising effectiveness.

[0026] According to various embodiments of the present disclosure, the economic feasibility of inserting advertisements can be evaluated by analyzing the current popularity and diffusion potential of content, and the timing of insertion can be optimized. As a result, videos that go viral late can be monetized immediately, and advertisers can be provided with timely advertising execution opportunities.

[0027] According to various embodiments of the present disclosure, advertising revenue can be transparently distributed among creators, platform operators, and advertisers through blockchain-based performance tracking and smart contract-based automatic settlement. Therefore, this has the effect of resolving the issue of opacity in the revenue structure, securing trust among stakeholders, and supporting the stable growth of the creative ecosystem.

[0028] According to various embodiments of the present disclosure, costs and resources required for new production can be reduced by recycling archived content. This contributes to the establishment of an eco-friendly media ecosystem and simultaneously provides a sustainable revenue model for individual creators and companies.

[0029] The effects obtainable from the present disclosure are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art to which the present disclosure belongs from the description below. Brief explanation of the drawing

[0030] FIG. 1 is a block diagram illustrating an advertising provision system according to one embodiment of the present invention. FIG. 2 is a flowchart illustrating a method for providing advertisements for archive content using artificial intelligence according to an embodiment of the present invention. FIG. 3 is a detailed flowchart illustrating the preprocessing process of archive content according to one embodiment of the present invention. FIG. 4 is a detailed flowchart illustrating the process of object splitting and tracking deletion of existing brand elements in archive content according to an embodiment of the present invention. FIG. 5 is a detailed flowchart illustrating a two-step inpainting process for restoring a blank area caused by the deletion of existing brand elements in archive content, according to an embodiment of the present invention. FIG. 6 is a detailed flowchart illustrating an advertisement synthesis process according to one embodiment of the present invention. FIG. 7 is a detailed flowchart illustrating the process of platform-customized conversion and result provision according to one embodiment of the present invention. Specific details for implementing the invention

[0031] Hereinafter, embodiments of the present disclosure are described in detail with reference to the drawings so that those skilled in the art can easily practice them. However, the present disclosure may be embodied in various different forms and is not limited to the embodiments described herein. In relation to the description of the drawings, the same or similar reference numerals may be used for identical or similar components. Furthermore, in the drawings and related descriptions, descriptions of well-known functions and configurations may be omitted for clarity and brevity.

[0032] The various embodiments of this document and the terms used therein are not intended to limit the technical features described in this document to specific embodiments, and should be understood to include various modifications, equivalents, or substitutions of said embodiments. In connection with the description of the drawings, similar reference numerals may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of said items unless the relevant context clearly indicates otherwise. In this document, phrases such as "A or B," "at least one of A and B," "at least one of A or B," "A, B or C," "at least one of A, B and C," and "at least one of A, B, or C" may each include any one of the items listed together in the corresponding phrase, or all possible combinations thereof. Terms such as "first," "second," or "first" or "second" may be used simply to distinguish said components from other said components and do not limit said components in any other aspect (e.g., importance or order). Where any (e.g., 1st) component is referred to as "coupled" or "connected" to another (e.g., 2nd) component, with or without the terms "functionally" or "communicationly," it means that said any component may be connected to said other component directly (e.g., via a wire), wirelessly, or through a third component.

[0033] As used in the various embodiments of this document, the term “module” may include a unit implemented in hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit, for example. A module may be a component formed integrally, or a minimum unit of said component or a part thereof that performs one or more functions.

[0034] Various embodiments of this document may be implemented as software (e.g., a program) comprising one or more instructions stored in a storage medium (e.g., memory) readable by a machine or device. For example, the processor of the machine or device may call at least one of the one or more instructions stored from the storage medium and execute it. This enables the machine to operate to perform at least one function according to the at least one called instruction. The one or more instructions may include code generated by a compiler or code that can be executed by an interpreter. The storage medium readable by a machine may be provided in the form of a non-transitory storage medium. Here, "non-transitory" simply means that the storage medium is a tangible device and does not contain a signal (e.g., electromagnetic waves), and this term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily in the storage medium.

[0035] According to one embodiment, the method according to the various embodiments disclosed herein may be provided by being included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)), or distributed online (e.g., download or upload) through an application store or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily created on a device-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0036] According to various embodiments, each component (e.g., module or program) of the components described above may include a singular or multiple entities, and some of the multiple entities may be separated and placed in other components. According to various embodiments, one or more of the components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Generally or additionally, multiple components (e.g., module or program) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each of the multiple components in the same or similar manner as those performed by the corresponding component among the multiple components prior to integration. According to various embodiments, operations performed by the module, program, or other components may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, omitted, or one or more other operations may be added.

[0037] In this disclosure, the term "processor" may refer to hardware capable of performing functions and operations according to each designation described herein, computer program code capable of performing specific functions and operations, or an electronic recording medium loaded with computer program code capable of performing specific functions and operations. According to an embodiment, the operation of the processor may be defined and / or interpreted as the operation of a knowledge graph adjustment device, but is not limited thereto. The processor may refer to a functional and / or structural combination of hardware for carrying out the technical concept of this disclosure and / or software for driving said hardware.

[0039] FIG. 1 is a block diagram illustrating an advertising provision system according to one embodiment of the present invention.

[0040] Under conventional technology, it was impossible to insert new advertisements into video content that had already been produced and distributed. Consequently, it was difficult to generate additional revenue from past content owned by broadcasters or OTT operators, or older videos uploaded by individual creators, after advertising contracts had expired. Furthermore, since conventional technology only allowed the insertion of advertisements (PPL) during the production phase, there was a limitation in that it could not utilize archive content that gained popularity belatedly.

[0041] In contrast, an advertising provision system (10) according to one embodiment of the present invention can create new advertising content by removing existing brand elements included in archive content through artificial intelligence-based object detection and inpainting technology, and synthesizing advertising target brand information (i.e., advertising target brand and related advertising brand elements). This enables the reuse of archive content and can provide a new revenue source for creators, advertisers, and platform operators alike.

[0042] In this specification, the term "archive content" means video content that has already been produced and distributed or is currently being distributed, and may include past programs of broadcasters and OTTs, videos produced and distributed by individual creators, etc.

[0043] In this specification, the term "advertising target brand information" may include the advertising target brand itself and related advertising brand elements (e.g., logo image, product image, 3D model, advertising text, etc.).

[0044] Specifically, as illustrated in FIG. 1, the advertising provision system (10) may include a user terminal (100) and an advertising provision server (200).

[0045] The user terminal (100) is a device that provides advertising target brand information (brand name, logo, product image, 3D model, etc.) from the advertiser or advertising business operator, and can be implemented as various electronic devices such as personal computers, laptops, smartphones, and tablets.

[0046] The user can use the user terminal (100) to transmit data related to advertising target brand information to the advertising provision server (200), and then receive the generated advertising content results.

[0047] The ad providing server (200) receives advertising target brand information from the user terminal (100), receives archive content that has already been produced and distributed, and performs preprocessing, ad synthesis, and platform-customized conversion.

[0048] For example, an ad delivery server (200) can generate preprocessed archive content by identifying existing brand elements included in archive content using artificial intelligence-based object detection technology and removing the elements through inpainting and image restoration techniques.

[0049] For example, an ad providing server (200) can analyze the three-dimensional spatial structure and light source information of preprocessed archive content and naturally synthesize advertising target brand elements to generate ad content.

[0050] For example, an ad providing server (200) can generate ad content for each platform by converting the generated ad content into an aspect ratio and format suitable for multiple platforms such as YouTube, TikTok, and Instagram.

[0051] Such an ad providing server (200) may include a processor (201) and memory (202). The processor (201) performs major operations such as the aforementioned preprocessing, ad synthesis, and platform conversion.

[0052] Such a processor (201) may include, as an example, at least one of a microprocessor, a central processing unit, a processor core, a multi-core processor, a multiprocessor, an ASIC (Application-Specific Integrated Circuit), or an FPGA (Field Programmable Gate Array), but is not limited thereto.

[0053] The memory (202) can store archive content, advertising target brand information, learning models and algorithms used for detection and synthesis, and final advertising content.

[0054] Memory (202) can store instructions (or programs) that can be executed by the processor (201). Memory (202) may include volatile memory or non-volatile memory. Volatile memory may be implemented as dynamic random access memory (DRAM), static random access memory (SRAM), thyristor RAM (T-RAM), zero capacitor RAM (Z-RAM), or twin transistor RAM (TTRAM). Non-volatile memory can be implemented as EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, MRAM (Magnetic RAM), Spin-Transfer Torque (STT)-MRAM, Conductive Bridging RAM (CBRAM), FeRAM (Ferroelectric RAM), PRAM (Phase change RAM), Resistive RAM (RRAM), Nanotube RRAM, Polymer RAM (PoRAM), Nano Floating Gate Memory (NFGM), holographic memory, Molecular Electronic Memory Device, or Insulator Resistance Change Memory.

[0055] In this way, the advertising provision system (10) according to one embodiment of the present invention enables the provision of archive content-based advertisements that conventional technology could not provide, and by automatically generating advertising content suitable for various platform environments, it can provide the effects of content recycling and the creation of new advertising revenue.

[0057] FIG. 2 is a flowchart illustrating a method for providing advertisements for archive content using artificial intelligence according to an embodiment of the present invention.

[0058] One embodiment illustrated in FIG. 2 is performed on an ad providing server (200).

[0059] In step S210, the ad providing server (200) can receive archive content that has already been produced and distributed or is currently being distributed.

[0060] At this time, the input source and path are not restricted, and, for example, old videos owned by broadcasters, popular videos owned by individual users, and licensed third-party video files can be received through the network. The file format can be configured to support various codecs and resolutions such as mp4, mov, and mkv.

[0061] In step S220, the ad providing server (200) can generate preprocessed archive content by identifying and removing existing brand elements included in the archive content using artificial intelligence-based object detection technology.

[0062] In one embodiment, existing brand elements can be detected on a multimodal basis by identifying visual elements, voice elements, and text elements within archive content, respectively.

[0063] For example, visual analysis can detect logos, product appearances, and color patterns, speech recognition can extract brand name mentions, and OCR can recognize subtitles and signboard text to integrate candidates.

[0064] In one embodiment, existing brand elements corresponding to the production time of archive content can be identified by referring to a brand evolution vector learned based on the temporal changes of the visual design of an existing brand. For example, the 2018 package design and the 2024 design can be distinguished to reduce the time-span discrepancy.

[0065] In one embodiment, an existing brand element can be divided and identified in pixel units for each frame of a series of consecutive frames using a deep learning-based object segmentation network, and the location and size of the existing brand element divided in pixel units for each frame of a series of consecutive frames can be tracked and deleted. Such frame tracking and deletion processing may be advantageous for suppressing flicker and jitter during playback.

[0066] In one embodiment, a second-stage restoration process consisting of structure line restoration and texture inpainting is applied to the blank area created by the deletion process, and the consistency of color and lighting can be corrected by reflecting background information and light source information extracted from a plurality of consecutive frames. For example, after removing a signboard, the background brick pattern and shadow can be corrected so that they naturally connect between frames.

[0067] In step S230, the ad providing server (200) can generate ad content by synthesizing ad brand elements for the target brand to the preprocessed archive content.

[0068] In one embodiment, the three-dimensional spatial structure of preprocessed archive content can be restored using monocular depth estimation, and the position, size, and rotation of advertising brand elements can be corrected and inserted based on the restored spatial structure. For example, a sticker advertisement attached to a flat surface on a desk can have its distortion and scale automatically adjusted to match the camera viewpoint and perspective.

[0069] In one embodiment, the direction and intensity of lighting are estimated by analyzing the light source information of the preprocessed archive content, and shadow and reflection effects of the advertising brand element are applied according to the estimated lighting conditions, thereby allowing the advertising brand element to be synthesized so that it is naturally inserted into the archive content. For example, weak shadows and diffuse reflections can be applied in indoor low-light scenes, and strong directional shadows can be applied in outdoor scenes with strong sunlight.

[0070] In step S240, the ad providing server (200) can generate ad content for multiple platforms by performing aspect ratio conversion based on the generated ad content so that it can be applied to multiple platforms. For example, the same source can be converted to 16:9 (YouTube), 9:16 (TikTok), and 1:1 (Instagram), and if necessary, the maximum length, bitrate, and additional subtitle specifications for each platform can be automatically adjusted.

[0071] In this way, an embodiment illustrated in FIG. 2 can systematically implement the configuration and operation described in claims 1 to 8 through a series of procedures leading to archive content input (S210), preprocessing through AI-based identification, removal, and inpainting of existing brand elements (S220), creation of advertising content through 3D and light source matching synthesis of advertising brand elements (S230), and aspect ratio conversion by platform (S240).

[0072] As such, one embodiment illustrated in FIG. 2 can generate new advertising content by detecting and removing existing brand elements from already produced and distributed archive content based on artificial intelligence, and by naturally synthesizing the brand elements to be advertised in accordance with the three-dimensional spatial structure and light source information. This can be utilized to provide new revenue generation opportunities to advertisers, platform operators, and creators by reusing various archive content such as old broadcast works, OTT content, and videos by individual creators.

[0074] FIG. 3 is a detailed flowchart illustrating the preprocessing process of archive content according to one embodiment of the present invention.

[0075] One embodiment of FIG. 3 is a process of generating preprocessed archive content in an ad providing server (200).

[0076] In step S310, the ad providing server (200) may receive archive content to be preprocessed. The archive content is video that has already been produced and distributed or is currently being distributed, and may include, for example, old broadcast works, OTT platform content, or videos by individual creators.

[0077] In step S320, the ad delivery server (200) can detect existing brand elements on a multimodal basis by identifying visual elements, voice elements, and text elements within the archive content, respectively.

[0078] In one embodiment, visual analysis can detect logo, product, signboard candidate, etc., on an object basis using an object detector—e.g., YOLO, Faster R-CNN, SSD, etc. In addition, the detected objects can be subdivided into pixel-unit masks using an instance segmenter such as Mask R-CNN, YOLACT, SOLOv2, or SAM (Segment Anything).

[0079] In one embodiment, voice analysis can generate transcription results for each time interval using an ASR (Automatic Speech Recognition) model such as Whisper, Conformer-Transducer, or wav2vec 2.0, detect brand names and product names in real time using a Keyword Spotting (KWS) model, and record the utterance time and reliability. According to the embodiment, the accuracy of brand mention detection can be improved by separating background sound and dialogue by applying speaker segmentation (speaker diarization, x-vector based) and sound source separation (Demucs, etc.).

[0080] In one embodiment, for text analysis, an OCR pipeline is applied to scene text, and subtitles, signboards, and poster text can be recognized by combining a detector (CRAFT, DBNet) and a recognizer (CRNN, TRBA, TrOCR, etc.). The recognized text can be morphologically normalized and matched with a brand dictionary and a thesaurus, and misspellings and spelling variations can be corrected using edit distance or pronunciation similarity.

[0081] In one embodiment, modal fusion can be performed using weighted summation or logistic regression-based late fusion, or by multimodal similarity calculation using image-text combined embeddings. Candidates derived from visual, voice, and text data are aligned along a time axis, and if high confidence scores overlap in the same interval, it can be determined that brand detection is confirmed.

[0082] In one embodiment, the ad providing server (200) can identify existing brand elements corresponding to the time of production of archive content by referring to a brand evolution vector learned based on the changes in the visual design of the existing brand over time.

[0083] In other words, a brand evolution vector refers to a time-series vector that represents the patterns of change in a specific brand's logo, product design, and positioning over time by embedding them in a multidimensional space.

[0084] In one embodiment, the ad delivery server (200) can embed a dataset of collected yearly logo images, product packaging, ad video captures, catalog images, etc., into feature vectors using a Vision Transformer (ViT) or CNN encoder, and perform metric learning by applying a time-aware embedding model so that vectors between consecutive years are close together and vectors between distant years are gradually separated. Subsequently, candidate elements detected in archived content are embedded into feature vectors in the same manner to calculate similarity with a brand evolution vector corresponding to the production year of the content, and can be determined as brand elements with confirmed time-appropriateness only when the similarity is above a threshold. Through this, even for the same brand, the latest logos or packaging that are inconsistent with the production year can be excluded, and only elements that match the actual time can be reliably identified.

[0085] In one embodiment, the ad providing server (200) may use a brand-related dataset collected yearly or at a specific time to generate or learn a brand evolution vector. The dataset may include logo images, product packaging images, ad video captures, catalog images, etc.

[0086] For example, brand image It collected for each year t can be embedded into a feature vector ft with d real components using a vision transformer (ViT) or a convolutional neural network (CNN), as shown in [Equation 1] below.

[0087] [Mathematical Formula 1]

[0088]

[0089] Here, Enc() represents the trained image encoder, and θ represents the set of parameters.

[0090] Afterwards, the ad providing server (200) can learn the feature vector ft and the corresponding year t together to reflect time information. To do this, a time-aware embedding model can be applied to perform metric learning so that the brand features at each time point form a continuous trajectory.

[0091] In other words, adjacent year data of the same brand can be learned to be close in the vector space, while data from distant years can be learned to gradually move further apart. Through this process, a time series trajectory reflecting changes in the brand's design is formed, and this time series trajectory can be defined as a brand evolution vector.

[0092] In one embodiment, the ad providing server (200) can determine whether a candidate element is an actual brand element at that time by extracting a detection candidate vector fc from archive content and calculating the similarity between the detection candidate vector fc and the brand evolution vector ft* corresponding to the production year.

[0093] Here, the detection candidate vector fc refers to the result of embedding candidate elements into a feature space when candidate elements likely to be specific brands are discovered through visual, voice, and text analysis within the archive content.

[0094] This is a process that not only checks whether candidate elements resemble a specific brand but also verifies whether they align with the production timeline.

[0095] In one embodiment, the ad serving server (200) can use a brand evolution vector to exclude the latest designs that are inconsistent with the time of production, even if they are of the same brand. For example, if a candidate detected in archive content produced in 2010 has a higher similarity to a logo vector that was rebranded in 2022, the detection can be considered a false positive and removed.

[0096] In one embodiment, time-series embeddings (time-aware metric learning) are learned using year-labeled logo, package, and catalog data, and cosine similarity with candidate logo embeddings is calculated based on the production time of the content to exclude candidates with low time-appropriateness.

[0097] In step S330, the ad providing server (200) can segment and identify existing brand elements in pixel units for each frame of a series of consecutive frames using a deep learning-based object segmentation network. This allows the object to be subdivided into pixel-unit masks using an instance segmenter such as Mask R-CNN, YOLACT, SOLOv2, or SAM (Segment Anything), as described in the example in step S320.

[0098] In step S340, the ad providing server (200) can delete existing brand elements by tracking their position and size in pixel units across a series of consecutive frames. For example, the logo of a moving product packaging can be tracked frame by frame and naturally removed.

[0099] In one embodiment, a combination of a brand element tracker and an optical flow can be used to stabilize the object trajectory and consistently apply a frame-by-frame deletion mask based on the tracking results.

[0100] As such, one embodiment illustrated in FIG. 3 can generate preprocessed archive content by receiving archive content as input, detecting brand elements through multimodal analysis, identifying elements appropriate for the production time using brand evolution vectors, and reliably deleting brand elements through object segmentation and frame tracking, which can be used as a preliminary preparation process for the advertisement synthesis stage.

[0102] FIG. 4 is a detailed flowchart illustrating the process of object splitting and tracking deletion of existing brand elements in archive content according to an embodiment of the present invention.

[0103] One embodiment of FIG. 4 relates to an object splitting and deletion processing process for generating preprocessed archive content, which is performed in an ad providing server (200).

[0104] In step S410, the ad serving server (200) may receive a candidate region containing existing brand elements within the archive content. This candidate region may be the result detected through the previously described multimodal detection and brand evolution vector reference process.

[0105] In step S420, the ad providing server (200) can divide candidate regions into pixel units using a deep learning-based object segmentation network.

[0106] In one embodiment, the ad providing server (200) can apply an instance segmenter such as Mask R-CNN, SOLOv2, YOLACT, SAM (Segment Anything) to finely mask brand logos, product appearances, signboard text, etc., in pixel units.

[0107] In one embodiment, the segmented mask is used as a reference in a subsequent deletion step and can be applied to Conditional Random Field (CRF)-based boundary refinement, morphological operations (such as dilation), and connectivity component analysis to remove small noise or incorrectly detected regions.

[0108] In step S430, the ad providing server (200) can track the location and size of the divided object in a series of consecutive frames.

[0109] In one embodiment, if there are occlusion sections, the trajectory can be interpolated or broken objects can be reconnected using a Kalman filter or ReID (Re-identification) embedding.

[0110] In step S440, the ad providing server (200) can process the deletion of the tracked object area.

[0111] In one embodiment, the ad providing server (200) removes objects by applying a frame-by-frame deletion mask, and then fills in the deleted parts through an inpainting process described with reference to FIG. 5.

[0112] For example, if the logo of a moving beverage can appears in different locations in each frame, visual discontinuity can be prevented by consistently deleting the same area in each frame based on the tracking results.

[0113] In one embodiment, forward-backward consistency is applied to detect tracking errors, and incorrectly deleted areas can be automatically corrected.

[0114] In this way, the ad providing server (200) can generate a result video in which object splitting and tracking deletion are completed.

[0115] The resulting image, after object segmentation and trace deletion are completed, can be naturally filled in through structure line restoration and texture correction during the subsequent inpainting process to be completed as preprocessed archive content.

[0116] As such, an embodiment illustrated in FIG. 4 can reliably remove existing brand elements from archive content by receiving detected existing brand elements, dividing them into pixel units, tracking their location and size in consecutive frames, and consistently deleting the tracked areas, which can be utilized to provide base data for a subsequent inpainting process.

[0118] FIG. 5 is a detailed flowchart illustrating a two-step inpainting process for restoring a blank area caused by the deletion of existing brand elements in archive content, according to an embodiment of the present invention.

[0119] One embodiment of FIG. 5 relates to a procedure for correcting blank areas through structure line restoration and texture inpainting, which is performed in an ad providing server (200).

[0120] In step S510, the ad providing server (200) can receive archive content in which existing brand elements are deleted and a blank area is set. This blank area is an area created through the object splitting and tracking deletion process described above.

[0121] In step S520, the ad providing server (200) can perform a first-step restoration process to restore the structure line of the blank area to form a basic shape.

[0122] In one embodiment, the ad providing server (200) can apply an edge detection network (HED, PiDiNet, etc.) or a structural line prediction model (EdgeConnect step 1) to extract the geometric structure of the background connected to the deleted object (wall line, road boundary line, desk corner, etc.) and reconstruct the morphological skeleton of the empty area based on this.

[0123] In one embodiment, the ad providing server (200) can naturally connect broken line segments by applying a GAN-based line completion technique to the detected edges, and this can be particularly effective in scenes containing buildings or regular patterns.

[0124] In one embodiment, when the ad providing server (200) restores the structure line of the blank area, it may apply not only simple line segment detection but also a multi-scale structure line restoration technique.

[0125] For example, the ad providing server (200) can simultaneously secure global consistency and local detail by distinguishing between large-scale structures (building walls, road boundaries) and small-scale structures (desk corners, object outlines), restoring them at different network layers, and finally fusing them.

[0126] In step S530, the ad providing server (200) can perform a second-stage restoration process of performing texture inpainting on the blank area where the structure line restoration is completed.

[0127] In one embodiment, background information and light source information are extracted from a plurality of consecutive frames to fill a texture while maintaining consistency in color, brightness, and illuminance.

[0128] For example, if a logo on a wall is removed, the pattern can be naturally restored using brick patterns extracted from adjacent frames, and the lighting direction and intensity can be corrected through a light source estimation network (SfSNet, DeepLight, etc.) to ensure visual consistency with the surrounding environment.

[0129] In one embodiment, a video inpainting model (LaMa, ProPainter, E2FGVI, etc.) is applied to ensure temporal continuity between frames, and a color transfer technique can be used to reduce the color difference between the corrected area and the surrounding area.

[0130] In one embodiment, the ad providing server (200) may apply a light source and material adaptive restoration technique that is more advanced than simple color correction during the texture inpainting process.

[0131] For example, the ad providing server (200) can estimate BRDF-based reflection characteristics to correct the difference between reflected light and transmitted light on surfaces of different materials such as metal, glass, and fabric, and can synthesize a texture so that the deleted area matches the light source conditions of the actual scene through NeRF-based light field representation.

[0132] In addition, temporal loss can be additionally applied to ensure temporal consistency for multiple consecutive frames.

[0133] [Mathematical Formula 2]

[0134]

[0135] [Equation 2] represents the temporal loss Ltemp to ensure temporal consistency for a series of consecutive frames, and by correcting the inpainting result It of the current frame and the result It-1 of the previous frame as warping to minimize the difference, temporal consistency can be ensured and inter-frame shaking can be reduced.

[0136] In step S540, the ad providing server (200) can complete the preprocessing of the archive content by outputting the result of the completed inpainting.

[0137] In one embodiment, the ad providing server (200) can perform reliability-based mask correction. For example, if the deletion mask is inconsistently calculated across multiple frames, the degree of alignment is calculated as a result of synthesizing multiple frames, and additional inpainting can be performed on areas with low reliability using a supplementary learning auxiliary network.

[0138] The completed content can be used as input for inserting advertising brand elements during the subsequent advertising synthesis process.

[0139] In this way, one embodiment illustrated in FIG. 5 sequentially performs structural line restoration and texture inpainting on the blank area created after deleting existing brand elements, thereby naturally connecting the deleted part with the surrounding scene. Through this, the preprocessed archive content is corrected so that it is indistinguishable from the actual scene, thereby ensuring high naturalness and reliability during the advertising synthesis stage.

[0141] FIG. 6 is a detailed flowchart illustrating an advertisement synthesis process according to one embodiment of the present invention.

[0142] One embodiment of FIG. 6 relates to a process of creating advertising content by inserting advertising target brand elements into preprocessed archive content, which is performed at an advertising provider server (200).

[0143] In step S610, the ad providing server (200) can receive preprocessed archive content. The preprocessed archive content is a video in which existing brand elements have been deleted and restored as previously described, and is content that is ready for synthesizing ad brand elements.

[0144] In step S620, the ad providing server (200) can select an ad brand element to be synthesized from the ad target brand information. The ad target brand information may include the ad target brand itself and related ad brand elements (logo image, product image, 3D object, ad text, etc.), and the element to be synthesized may vary depending on the ad contract, video context, user profile, etc.

[0145] In step S630, the ad providing server (200) can restore the three-dimensional spatial structure of the preprocessed archive content using a depth estimation model. In one embodiment, a depth map of the scene can be generated using MiDaS, DPT, and NeRF series models, and the actual distance and scale of the insertion target location can be estimated through the restored depth information.

[0146] In step S640, the ad providing server (200) can insert an ad brand element by correcting its position, size, and rotation based on the restored spatial structure. For example, when inserting an ad for a beverage can on a desk, the perspective of the can can be corrected by considering the planar angle of the desk and the camera viewpoint, and the size can be automatically adjusted to match the ratio with the actual scene.

[0147] In step S650, the ad providing server (200) can estimate the direction and intensity of the light by analyzing the light source information of the preprocessed archive content. In one embodiment, a deep learning-based light source estimation network (DeepLight, SfSNet, etc.) can be applied, or a light source vector can be calculated by analyzing shadow and highlight patterns within the image.

[0148] Afterward, the ad providing server (200) can synthesize the ad brand elements so that they are naturally inserted into the archive content by applying shadow and reflection effects to the ad brand elements according to the estimated lighting conditions. For example, strong directional shadows can be generated in outdoor sunlight scenes, and diffuse reflection and weak shadows can be applied in indoor lighting scenes to visually match the actual scene.

[0149] In one embodiment, the ad delivery server (200) can analyze contextual information of archived content to select an ad target brand element or adjust the insertion location. For example, it can be configured to synthesize a sports drink advertisement in a sports game scene and a food and beverage advertisement in a kitchen scene. This context-based ad synthesis can enhance the advertising effect without hindering the viewer's immersion by increasing the association between the content and the advertisement.

[0150] In one embodiment, the ad providing server (200) can synthesize different ad brand elements even for the same archive content by considering the characteristics of the user base for each platform. For example, on YouTube, it can synthesize home appliance advertisements that are considered for long viewing, on TikTok, it can synthesize fashion and beauty advertisements that have short attention spans, and on Instagram, it can synthesize simple logo-centered advertisements suitable for a square aspect ratio. Through this, even for the same content, optimized ad effects tailored to the characteristics of each platform and user preferences can be obtained.

[0151] In one embodiment, the ad providing server (200) can perform ad synthesis when it is determined that there is economic feasibility by analyzing the popularity, view count trends, exposure potential, etc. of the content. For example, ad synthesis can be applied immediately to videos that have recently seen a rapid increase in view counts, while synthesis can be withheld for videos with a low probability of exposure. This reduces unnecessary computational costs and increases the efficiency of the advertiser.

[0152] Finally, the ad providing server (200) can output ad content with ad brand elements inserted. The output ad content can then be provided to various distribution channels after undergoing a platform-specific aspect ratio conversion process.

[0153] As such, an embodiment illustrated in FIG. 6 can generate advertising content that is naturally fused with the actual scene by receiving preprocessed archive content, selecting advertising target brand elements, performing 3D spatial reconstruction and light source analysis through monocular depth estimation, and synthesizing advertising brand elements with corrected position, size, rotation, and lighting conditions. Furthermore, by applying context-based advertising selection, platform-specific differentiated synthesis, and economic judgment logic, advertising effectiveness and operational efficiency can be maximized.

[0155] FIG. 7 is a detailed flowchart illustrating the process of platform-customized conversion and result provision according to one embodiment of the present invention.

[0156] One embodiment of FIG. 7 relates to a procedure for converting advertising content to suit a plurality of platform environments and outputting it in a form that can be distributed, which is performed by an advertising provider server (200).

[0157] In step S710, the ad providing server (200) can receive ad content in which ad synthesis is completed. The ad content is the result generated through the previously described preprocessing and ad synthesis processes.

[0158] In step S720, the ad providing server (200) can determine the necessary conversion specifications by analyzing the characteristics of multiple platforms.

[0159] In one embodiment, since major platforms such as YouTube, TikTok, and Instagram have different aspect ratios and video length limits, the ad providing server (200) can set conversion rules by referring to the requirements specifications of each platform.

[0160] In one embodiment, YouTube may require a standard suitable for a 16:9 widescreen and long videos, TikTok may require a standard suitable for a 9:16 verticalscreen and short-duration focused viewing, and Instagram may require a standard suitable for a 1:1 squarescreen and short feed-type content.

[0161] The ad providing server (200) manages the platform-specific resolution, frame rate, and bitrate requirements as metadata and can automatically convert the input ad content to each standard.

[0162] In step S730, the ad delivery server (200) can generate ad content for multiple platforms by performing aspect ratio conversion to convert the same ad content to be suitable for various platforms.

[0163] In one embodiment, the ad serving server (200) can reconstruct the screen so that the ad brand element is not cut off by applying semantic-aware cropping, auto-reframing, and an important object preservation algorithm instead of simple cropping.

[0164] For example, the ad delivery server (200) can automatically adjust the entire ad brand product to be placed in the center in the YouTube version, enlarge the brand elements to fit the vertical screen in the TikTok version, and place the logo in the center of the square frame in the Instagram version.

[0165] In one embodiment, the ad providing server (200) can process different ad brand elements between ad content for multiple platforms for the same content by reflecting platform-specific characteristics.

[0166] For example, even for the same home appliance advertising brand, detailed advertisements for home appliances can be inserted in the YouTube version (tens of minutes) to suit the characteristics of long viewing, image-based concept advertisements in the TikTok version, and simple logo-centered advertisements in the Instagram version.

[0167] As another example, the ad providing server (200) can enhance the advertising effect by referring to the content context analysis results and inserting sports drink advertisements in sports scenes and food and beverage advertisements in kitchen scenes.

[0168] Finally, the ad providing server (200) can generate platform-specific conversion results. The outputted ad content can be uploaded directly to the platform by the user, or the ad providing server (200) can automatically distribute it via API.

[0169] In one embodiment, the ad delivery server (200) may include logic for determining economic feasibility. For example, the ad delivery server (200) can automatically adjust the conversion and distribution priority by analyzing view count trends, expected exposure potential, and ad unit prices per platform. Alternatively, if specific archive content is rapidly gaining popularity on YouTube, content for that platform may be prioritized for creation and immediate distribution, while content with stagnant view counts may have its conversion delayed or not distributed. This allows for efficient management of computational resources and advertising costs.

[0170] In one embodiment, the ad delivery server (200) can utilize a blockchain-based smart contract to track platform-specific distribution performance (views, click-through rates, etc.) and automatically distribute ad revenue among the advertiser, platform, and creator. This ensures transparency and reliability in the process of providing platform-specific ads.

[0171] In one embodiment, the ad delivery server (200) can perform performance tracking and automatic settlement functions after the platform-customized conversion. Specifically, after the ad content is distributed, performance indicators such as views, click-through rates, and dwell time can be collected and recorded in an on-chain blockchain network or an off-chain database. Subsequently, a predefined smart contract can be executed using the performance indicators as a trigger to automatically distribute revenue among the advertiser, platform operator, and creator. This procedure ensures transparency in revenue distribution, increases trust among stakeholders, and prevents unnecessary disputes.

[0172] As such, one embodiment illustrated in FIG. 7 receives advertising content with completed advertising synthesis, performs aspect ratio conversion and differentiation of advertising elements according to platform-specific requirements and user characteristics, and includes economic judgment based on performance data and blockchain-based revenue distribution, thereby enabling optimized advertising effects on various platforms such as YouTube, TikTok, and Instagram even with the same archive content, and can enable efficient and reliable monetization for advertisers, platforms, and creators alike.

[0174] The ad serving server described above may be implemented through a computing system comprising one or more processors and storage devices. For example, each component of the ad serving server may be implemented as a software program or microservice running on one or more servers or cloud-based computing instances, and data transfer between these modules may be performed through message queues, databases, or API interfaces. Furthermore, it is evident that the components of the present invention may be implemented as a series of processing logic or services running on physical or virtualized computing resources.

[0175] The device described above may be implemented as a hardware component, a software component, and / or a combination of a hardware component and a software component. For example, the device and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. In addition, other processing configurations, such as parallel processors, are also possible.

[0176] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0177] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0178] Although the embodiments have been described above with reference to limited examples and drawings, those skilled in the art can make various modifications and variations from the description above. For example, suitable results can be achieved even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0179] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

[0180] Although specific embodiments have been described in the detailed description of this document, it will be obvious to those skilled in the art that various modifications are possible within the scope of this document.

Claims

Claim 1 An advertising delivery method performed in an advertising delivery system comprises: a step of receiving archive content that has already been produced and distributed or is currently being distributed; a step of generating preprocessed archive content by identifying and removing existing brand elements included in the archive content using an AI-based object detection technology; and a step of generating advertising content by synthesizing advertising brand elements for an advertising target brand into the preprocessed archive content; wherein the step of generating the preprocessed archive content includes: a step of detecting existing brand elements based on a multimodal basis by identifying visual elements, voice elements, and text elements within the archive content, respectively; a step of deleting the existing brand elements for a plurality of consecutive frames containing the detected existing brand elements; and a step of inpainting the blank area generated by the deletion process based on background information and light source information of the plurality of consecutive frames; wherein the step of detecting existing brand elements based on a multimodal basis includes: a step of identifying existing brand elements corresponding to the production time of the archive content by referring to a brand evolution vector learned based on the epochal changes of the visual design of the existing brand. A method for providing advertisements of archive content using artificial intelligence, comprising, wherein the brand evolution vector is a time-series vector that expresses the changes in a specific brand's logo, product design, and positioning over time by embedding them in a multidimensional space. Claim 2 A method for providing advertisements of archive content using artificial intelligence, wherein, in claim 1, the method for providing advertisements further comprises the step of generating advertisement content for multiple platforms by performing aspect ratio conversion based on the generated advertisement content so that it can be applied to multiple platforms. Claim 3 delete Claim 4 delete Claim 5 A method for providing advertisements of archived content using artificial intelligence, wherein the step of deleting the existing brand element in claim 1 comprises: a step of dividing and identifying the existing brand element in pixel units for each frame of the plurality of consecutive frames using a deep learning-based object segmentation network; and a step of deleting the existing brand element divided in pixel units while tracking its location and size for each frame of the plurality of consecutive frames. Claim 6 A method for providing advertisements for archive content using artificial intelligence, wherein the step of inpainting based on background information and light source information of a plurality of consecutive frames in claim 1 comprises: a step of performing a first-stage restoration process to restore structural lines of the blank area to form a basic shape; and a step of performing a second-stage restoration process to perform texture inpainting based on color and lighting consistency by reflecting background information and light source information extracted from a plurality of consecutive frames on the area where structural lines were restored by the first-stage restoration. Claim 7 A method for providing an advertisement of archive content using artificial intelligence, wherein the step of generating the advertisement content comprises: a step of restoring the three-dimensional spatial structure of the preprocessed archive content using monocular depth estimation; and a step of correcting and inserting the position, size, and rotation of the advertisement brand element based on the restored spatial structure. Claim 8 A method for providing advertisements in archive content using artificial intelligence, wherein the step of generating the advertisement content further comprises: a step of estimating the direction and intensity of lighting by analyzing light source information of the preprocessed archive content; and a step of synthesizing the advertisement brand element so that it is naturally inserted into the archive content by applying shadow and reflection effects of the advertisement brand element according to the estimated lighting conditions. Claim 9 User terminal providing advertising target brand information; and includes an advertising providing server that receives advertising target brand information from the user terminal and generates advertising content by reflecting the advertising target brand information in archive content that has already been produced and distributed or is currently being distributed; wherein the advertising providing server generates preprocessed archive content by identifying and removing existing brand elements included in the archive content using AI-based object detection technology, and generates advertising content by synthesizing advertising target brand elements into the preprocessed archive content; wherein, in generating the preprocessed archive content, the advertising providing server identifies visual elements, voice elements, and text elements within the archive content respectively to detect existing brand elements based on a multimodal basis, and for a series of consecutive frames containing the detected existing brand elements, processes the deletion of the existing brand elements, and then inpaints the blank areas created by the deletion process based on the background information and light source information of the series of consecutive frames; wherein, in detecting existing brand elements based on the multimodal basis, the server identifies existing brand elements corresponding to the production time of the archive content by referring to a brand evolution vector learned based on the changes in the visual design of the existing brand over time, and wherein the brand evolution vector includes the logo, product design, and An advertising delivery system that provides archive content using artificial intelligence, characterized by being a time series vector in which the pattern of positioning change over time is embedded in a multidimensional space. Claim 10 A computer-readable recording medium combined with hardware, storing a computer program for performing the method of any one of claims 1, 2, 5 through 8.

Images