Abstract: An apparatus, computer readable storage medium and method of generating a diverse set of images from few-shot images, includes a parameter input receiving values for control parameters to control an extent to which each reference image impacts a newly generated image. The apparatus involves an image generation deep learning network for generating an image for each of the values for the control parameters. The deep learning network has an encoder, a transformer-based fusion block, and a decoder. The transformer-based fusion block includes a mapping network that computes meta-weights from features extracted from the reference images and the control parameters, and a cross-attention block to generate modulation weights based on the meta-weights. An output displays high-quality and diverse images generated based on the values for the control parameter.

Abstract: Glyph editing techniques through use of an adornment object are described. In one example, an input is received identifying a glyph and an adornment object in digital content displayed in a user interface. Glyph anchor points are obtained based on the glyph and adornment anchor points based on the adornment object. A link is generated between at least one said glyph anchor point and at least one said adornment anchor point. An edit input is received specifying an edit to a spatial property the glyph. The spatial property of the edit is propagated to a spatial property of the adornment object based on the link.

Abstract: Certain aspects and features of this disclosure relate to rendering images using target-augmented material maps. In one example, a graphics imaging application is loaded with a scene and an input material map, as well as a file for a target image. A stored, material generation prior is accessed by the graphics imaging application. This prior, as an example, is based on a pre-trained, generative adversarial network (GAN). An input material appearance from the input material map is encoded to produce a projected latent vector. The value for the projected latent vector is optimized to produce the material map that is used to render the scene, producing a material map augmented by a realistic target material appearance.

Abstract: An image is imported. The image is classified. A first algorithm from a plurality of algorithms, based on an image classification, is suggested via a graphical user interface (GUI). An input is received from a user, via the GUI, indicating that the user selects the suggested algorithm, or another algorithm, for vectorizing the image. The image is vectorized using the selected algorithm. An editable version of the vectorized image is presented to the user.

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for generating image mattes for detected objects in digital images without trimap segmentation via a multi-branch neural network. The disclosed system utilizes a first neural network branch of a generative neural network to extract a coarse semantic mask from a digital image. The disclosed system utilizes a second neural network branch of the generative neural network to extract a detail mask based on the coarse semantic mask. Additionally, the disclosed system utilizes a third neural network branch of the generative neural network to fuse the coarse semantic mask and the detail mask to generate an image matte. In one or more embodiments, the disclosed system also utilizes a refinement neural network to generate a final image matte by refining selected portions of the image matte generated by the generative neural network.

Abstract: Images placed in documents are enhanced based on the context in which the image is used. Context is determined according to document-specific indicators such as nearby text, headings, titles, and tables of content. A generative adversarial network (GAN) modifies the image according to the context to selectively emphasize relevant components of the image, which may include erasing or deleting irrelevant components. Relevant general-purpose images may be retrieved for use in the document and may be selectively enhanced according to usage of the general-purpose image in a given document.

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for generating three-dimensional meshes representing two-dimensional images for editing the two-dimensional images. The disclosed system utilizes a first neural network to determine density values of pixels of a two-dimensional image based on estimated disparity. The disclosed system samples points in the two-dimensional image according to the density values and generates a tessellation based on the sampled points. The disclosed system utilizes a second neural network to estimate camera parameters and modify the three-dimensional mesh based on the estimated camera parameters of the pixels of the two-dimensional image. In one or more additional embodiments, the disclosed system generates a three-dimensional mesh to modify a two-dimensional image according to a displacement input.

Abstract: In some implementations, a recommendation system may input text into a machine learning model that was trained using input specific to an organization associated with the text and was refined using input specific to a portion of the organization. The recommendation system may receive, from the machine learning model, a recommendation indicating one or more visual components, stored in a database associated with the organization, to use with the text. The machine learning model may use natural language processing and sentiment detection to parse the text. Accordingly, the recommendation system may receive the one or more visual components from the database and generate an initial draft including the text and the one or more visual components.

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for regenerative learning to enhance dense predictions. In one example method, an input image is accessed. A dense prediction output is generated based on the input image using a dense prediction machine learning (ML) model, and a regenerated version of the input image is generated. A first loss is generated based on the input image and a corresponding ground truth dense prediction, and a second loss is generated based on the regenerated version of the input image. One or more parameters of the dense prediction ML model are updated based on the first and second losses.

Abstract: Embodiments described herein provide systems and methods of subject-driven image generation. In at least one embodiment, a system receives, via a data interface, an image containing a subject, a text description of the subject in the image, and a text prompt relating to a different rendition of the subject. The system encodes, via an image encoder, the image into an image feature vector. The system encodes, via a text encoder, the text description int a text feature vector. The system generates, by a multimodal encoder, a vector representation of the subject based on the image feature vector and the text feature vector. The system generates, by a neural network based image generation model, an output image based on an input combining the text prompt and the vector representation.

Abstract: Methods, apparatus, systems, devices, and computer program products directed to augmenting reality with respect to real-world places, and/or real-world scenes that may include real-world places may be provided. Among the methods, apparatus, systems, devices, and computer program products is a method directed to augmenting reality via a device. The method may include capturing a real-world view that includes a real-world place, identifying the real-world place, determining an image associated with the real-world place familiar to a user of the device viewing the real-world view, and/or augmenting the real-world view that includes the real-world place with the image of the real-world place familiar to a user viewing the real-world view.

Abstract: System and methods for using a deep learning framework to customize animation of an in-game character of a video game. The system can be preconfigured with animation rule sets corresponding to various animations. Each animation can be comprised of a series of distinct poses that collectively form the particular animation. The system can provide an animation-editing interface that enables a user of the video game to make modifications to at least one pose or frame of the animation. The system can realistically extrapolate these modifications across some or all portions of the animation. In addition or alternatively, the system can realistically extrapolate the modifications across other types of animations.

Abstract: A method of providing a service for a conversation with a virtual character replicating a deceased person is provided. The method of the present disclosure includes predicting a response message of a virtual character replicating a deceased person in response to a message input by a user, generating a speech corresponding to an oral utterance of the response message on the basis of speech data of the deceased person and the response message, and generating a final video of the virtual character uttering the response message on the basis of a driving video guiding the movement of the virtual character and the speech.

Abstract: Some examples described herein relate to modifying avatars in user interfaces based on wearable device data using machine learning. For example, a system can receive, from a wearable device, a set of biological data with respect to a user wearing the wearable device. The user can be associated with an avatar of the user on a user interface. The system can provide the avatar of the user and the set of biological data as input to a trained machine-learning model. The trained machine-learning model can generate a modified avatar based on the input. The system can then receive the modified avatar as output from the trained machine-learning model. The system can modify the user interface to include the modified avatar. The modified avatar can be outputted for display on a client device.

Abstract: Systems and methods for avatar eye animation is provided. The method include obtaining a first gaze vector of a user associated with a first eye of the user and obtaining a second gaze vector of the user associated with a second eye of the user. The method further includes determining a first point in a three-dimensional (3D) virtual space that the user is looking toward. The method also includes creating a second point in the 3D virtual space corresponding to the first point in the 3D virtual space. The method then includes rendering a first eye and a second eye of an avatar of the user in the 3D virtual space based on the second point in the 3D virtual space such that the first eye and the second eye of the avatar are looking toward the second point in the 3D virtual space.

Abstract: An apparatus for displaying profile information in a virtual environment comprises a processor associated with a server. The processor is configured to generate an authorization token configured to assign a first avatar to a first user, wherein the authorization token is stored within the plurality of transfers of the blockchain record. The processor is further configured to receive session data associated with the first avatar, wherein the session data comprises at least one gesture for a session in a virtual environment and to compare the at least one gesture of the session data to one or more authorized gestures to identify a first authorized gesture. The processor is further configured to display profile information in the virtual environment that is stored in a first user profile associated with the first user in response to identifying the first authorized gesture.

Abstract: In an example, a method may include obtaining, from a data source, first data including multiple frames each including a human face. The method may include automatically detecting, in each of the multiple frames, one or more facial landmarks and one or more action units (AUs) associated with the human face. The method may also include automatically generating one or more semantic masks based at least on the one or more facial landmarks, the one or more semantic masks individually corresponding to the human face. The method may further include obtaining a facial hyperspace using at least the first data, the one or more AUs, and the semantic masks. The method may also include generating a synthetic image of the human face using a first frame of the multiple frames and one or more AU intensities individually associated with the one or more AUs.

Abstract: In various examples, systems and methods are disclosed relating to generating a simulated environment and update a machine learning model to move each of a plurality of human characters having a plurality of body shapes, to follow a corresponding trajectory within the simulated environment as conditioned on a respective body shape. The simulated human characters can have diverse characteristics (such as gender, body proportions, body shape, and so on) as observed in real-life crowds. A machine learning model can determine an action for a human character in a simulated environment, based at least on a humanoid state, a body shape, and task-related features. The task-related features can include an environmental feature and a trajectory.

Abstract: The present disclosure provides an apparatus and method for generating a dancing avatar, that receives a latent code and map it using a neural network operation to obtain a plurality of genre-specific style codes for each of a plurality of dance genres, and decodes seed motion data and music data, which are motion data that must be referred to when generating an avatar's dance motion, using a genre-specific style code for a dance genre selected among the plurality of genre-specific style codes as a guide, thereby obtaining a dance vector representing a dance motion feature of the avatar in the selected dance genre. According to the present disclosure, it is possible to continuously generate various dance motions of the avatar in relation to previous dance motions, and freely change the dance genre according to user commands or music.

Abstract: An image processing apparatus includes at least one memory and at least one processor which function as an image acquiring unit configured to reduce a resolution of the three-dimensional image and acquire a three-dimensional image of an object, an initial parameter estimating unit configured to estimate an initial parameter in accordance with an initial value of an adjustable parameter to be utilized in obtaining a reference cross section, a cross sectional image acquiring unit configured to acquire one or more cross sectional images based on the three-dimensional image and the initial parameter, and an indicator estimating unit configured to estimate an indicator to be contributed to correcting the initial parameter, wherein the estimated indicator is in association with a specified cross sectional image out of the one or more cross sectional images.

Abstract: A pixel ray crossing-based multi-viewpoint MPI geometry generation method, device, and recording medium of the present disclosure, the method may comprise obtaining a multi-viewpoint image by original cameras which shoot a different viewpoint, obtaining a multi-plane image (MPI) based on the multi-viewpoint image, obtaining, based on the MPI, an atlas image in a 2D form, and obtaining a bitstream by encoding the atlas image.

Abstract: A computer-executable method for generating a side-by-side three-dimensional (3D) image includes the steps of creating a 3D mesh and estimating depth information of the raw image. The method further includes the steps of updating the left mesh area and the right mesh area of the 3D mesh based on the estimated depth information of the raw image and projecting each of the mesh vertices of the left mesh area onto a coordinate system of the side-by-side 3D image based on a left eye position, and projecting each of the mesh vertices of the right mesh area onto the coordinate system of the side-by-side 3D image based on a right eye position. The method further obtains the side-by-side 3D image by coloring the left mesh area and the right mesh area projected onto the coordinate system of the side-by-side 3D image based on the raw image.

Abstract: According to implementations of the present disclosure, there is provided a solution for completing textures of an object. In this solution, a complete texture map of an object is generated from a partial texture map of the object according to a texture generation model. A first prediction on whether a texture of at least one block in the complete texture map is an inferred texture is determined according to a texture discrimination model. A second image of the object is generated based on the complete texture map. A second prediction on whether the first image and the second image are generated images is determined according to an image discrimination model. The texture generation model, the texture and image discrimination models are trained based on the first and second predictions.

Abstract: In various embodiments, a scene reconstruction model generates three-dimensional (3D) representations of scenes. The scene reconstruction model maps a first red, blue, green, and depth (RGBD) image associated with both a first scene and a first viewpoint to a first surface representation of at least a first portion of the first scene. The scene reconstruction model maps a second RGBD image associated with both the first scene and a second viewpoint to a second surface representation of at least a second portion of the first scene. The scene reconstruction model aggregates at least the first surface representation and the second surface representation in a 3D space to generate a first fused surface representation of the first scene. The scene reconstruction model maps the first fused surface representation of the first scene to a 3D representation of the first scene.

Abstract: An information processing system includes: a first ray tracing unit configured to determine a two-dimensional ray trace from a transmission point at which a radio wave is transmitted, to a reception point at which the radio wave is received; a second ray tracing unit configured to determine a three-dimensional ray trace corresponding to the two-dimensional ray trace, using height information about the transmission point and the reception point; and a radio field intensity calculation unit that calculates the intensity of the radio wave at the reception point, using one or more three-dimensional ray traces determined by the second ray tracing unit.

Abstract: A computer implemented method converts ray data for a ray into a ray representative, wherein the ray representative is a compressed representation of the ray data, and wherein the ray data comprises three direction components and three position components for the ray. The method comprises identifying which of the three direction components of the ray data has the greatest magnitude, and defining the axis of the identified direction component as the major axis of the ray. The method further comprises determining a translated position on the ray at which the position component along the major axis is zero, and rescaling the three direction components of the ray so that the magnitude of the direction component along the major axis is one. The ray representative comprises: (i) the two position components of the translated position along the axes which are not the major axis, and (ii) the two rescaled direction components along the axes which are not the major axis.

Abstract: According to one embodiment, a method, computer system, and computer program product for identifying induced deformation of a 3D object is provided. The embodiment may include receiving an unaltered three-dimensional (3D) rendering of an object and attribute information of the object. The embodiment may include identifying one or more influencing factors of forecasted local deformation of one or more portions of the 3D rendering based on the attribute information. The embodiment may include creating, via a generative adversarial network (GAN), a 3D rendering of the object showing the forecasted local deformation. The embodiment may include identifying induced deformation of one or more other portions of the 3D rendering caused by the forecasted local deformation. The embodiment may include creating, via the GAN, a 3D rendering of the object showing the identified induced deformation.

Abstract: Systems, apparatuses and methods may provide for technology that generates, by a first neural network, an initial set of model weights based on input data and iteratively generates, by a second neural network, an updated set of model weights based on residual data associated with the initial set of model weights and the input data. Additionally, the technology may output a geometric model of the input data based on the updated set of model weights. In one example, the first neural network and the second neural network reduce the dependence of the geometric model on the number of data points in the input data.

Abstract: A deep neural network based hair rendering system is presented to model high frequency component of furry objects. Compared with existing approaches, the present method can generate photo-realistic rendering results. An acceleration method is applied in our framework, which can speed up training and rendering processes. In addition, a patch-based training scheme is introduced, which significantly increases the quality of outputs and preserves high frequency details.

Abstract: Systems and methods described herein support enhanced computer vision capabilities which may be applicable to, for example, autonomous vehicle operation. An example method includes generating a latent space and a decoder based on image data that includes multiple images, where each image has a different viewing frame of a scene. The method also includes generating a volumetric embedding that is representative of a novel viewing frame of the scene. The method includes decoding, with the decoder, the latent space using cross-attention with the volumetric embedding, and generating a novel viewing frame of the scene based on an output of the decoder.

Abstract: Embodiments of the disclosure provide a method, apparatus, electronic device, and storage medium for control based on extended reality. The control method based on extended reality includes: obtaining a real environment image and spatial positional relationship of the real environment, obtaining a real environment image and a spatial positional relationship of a real environment, and mapping the real environment to an extended reality space based on the real environment image and the spatial positional relationship; setting a virtual light source in the extended reality space; and rendering a light effect for the real environment mapped to the extended reality space based on a light source parameter of the virtual light source, and displaying the rendered extended reality space. Embodiments of the present disclosure can realize personalized environmental lighting display for users.

Abstract: The present invention sets forth a technique for generating two-dimensional (2D) renderings of a three-dimensional (3D) scene from an arbitrary camera position under arbitrary lighting conditions. This technique includes determining, based on a plurality of 2D representations of a 3D scene, a radiance field function for a neural radiance field (NeRF) model. This technique further includes determining, based on a plurality of 2D representations of a 3D scene, a radiance field function for a “one light at a time” (OLAT) model. The technique further includes rendering a 2D representation of the scene based on a given camera position and illumination data. The technique further includes computing a rendering loss based on the difference between the rendered 2D representation and an associated one of the plurality of 2D representations of the scene. The technique further includes modifying at least one of the NeRF and OLAT models based on the rendering loss.

Abstract: Disclosed are a point cloud model processing method and apparatus, and a readable storage medium. The method includes: acquiring a target image pair and first and second neighborhood images corresponding to the target image pair from first and second image sets photographed for a target scene with different visual effects; performing a calculation to obtain a merge parameter based on the relationship among the target image pair, the first neighborhood image, the second neighborhood image, a first point cloud model, and a second point cloud model; and merging the first point cloud model and the second point cloud model based on the merge parameter to obtain a target point cloud model for reconstructing a three-dimensional structure of the target scene.

Abstract: A computer-implemented method includes establishing a connection between a point cloud browser and a transfer agent. The method further includes establishing a connection between the transfer agent and a target application. The method further includes in response to selection of one or more points in the point cloud browser, generating one or more objects in the target application.

Abstract: An information processing apparatus is provided. Three-dimensional shape information indicating a three-dimensional shape of an object is acquired. The three-dimensional shape information is for generating a virtual viewpoint image of the object. Image quality of the object in a captured image of the object obtained by an image capturing apparatus is specified. Information indicating the image quality of the object and the three-dimensional shape information of the object are output.

Abstract: Methods and systems for generating 3D models of surfaces that accurately reconstruct both the global structure of an object and its local features are described. The methods and systems generally operated by fusing point features from the point cloud data with voxel features extracted from voxelization procedures. Furthermore, the methods and systems utilize voxelization at multiple spatial resolutions. The use of point-voxel fusion and multiple spatial resolutions may permit the extraction of both global and local geometric features, increasing the accuracy of 3D modeling of objects.

Abstract: A rule set or scene grammar can be used to generate a scene graph that represents the structure and visual parameters of objects in a scene. A renderer can take this scene graph as input and, with a library of content for assets identified in the scene graph, can generate a synthetic image of a scene that has the desired scene structure without the need for manual placement of any of the objects in the scene. Images or environments synthesized in this way can be used to, for example, generate training data for real world navigational applications, as well as to generate virtual worlds for games or virtual reality experiences.

Abstract: Systems and methods are provided for pitch determination. An example method includes obtaining an image depicting a structure, the image being captured via a user device positioned proximate to the structure. The image is segmented to identify, at least, a roof facet of the structure. An eave vector and a rake vector which are associated with the roof facet are determined. A normal vector of the roof facet is calculated based on the eave vector and the rake vector, and compared to a vector indicating a vertical direction such as gravity. The angle made out by the normal and a gravity vector may be utilized to calculate the pitch of the roof facet.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating an output that characterizes a scene at a current time step. In one aspect, one of the systems include: a voxel neural network that generates a current early-stage feature representation of the current point cloud, a fusion subsystem that generates a current fused feature representation at the current time step; a backbone neural network that generates a current late-stage feature representation at the current time step, and an output neural network that generate an output that characterizes a scene at the current time step.

Abstract: A method includes scanning a user's nipple via a user computing device to generate a scan image and applying a machine learning engine to the scan image to identify the user's nipple. The method also includes generating an output scan image via the machine learning engine, where the output scan image includes features identifying the user's nipple within the output scan image. The method also includes applying a genetic algorithm to the output scan image to generate a 3D image of the user's nipple, where the genetic algorithm employs at least one genetic process and where the 3D image of the user's nipple is a baby bottle nipple profile. The method also includes transmitting the 3D image of the user's nipple to a second user computing device for 3D printing of a custom baby bottle nipple, where the custom baby bottle nipple is a 3D replication of the user's nipple.

Abstract: The embodiments relate to a method and a technical equipment for implementing the method. The method comprises (210) creating a scene structure for a three-dimensional media content, wherein the scene structure comprises three-dimensional data for objects of the three-dimensional media content; (220) determining dependency information for the objects, which dependency information indicates an external factor on which an object is dependent on; (230) storing a scene description defining the objects and their dependency information into a bitstream structure; (240) and transferring a generated scene description to a renderer.

Abstract: Some embodiments provide a mapping application that has a novel way of displaying traffic congestion along roads in the map. The mapping application in some embodiments defines a traffic congestion representation to run parallel to its corresponding road portion when the map is viewed at a particular zoom level, and defines a traffic congestion representation to be placed over its corresponding mad portion when the map is viewed at another zoom level. The mapping application in some embodiments differentiates the appearance of the traffic congestion representation that signifies heavy traffic congestion from the appearance of the traffic congestion representation that signifies moderate traffic congestion. In some of these embodiments, the mapping application does not generate a traffic congestion representation for areas along a road that are not congested.

Abstract: A method, a 3D map generation system, and a device for generating 3D map of surroundings of a vehicle. In the system, conditions of the surroundings are detected. Further, cameras and sensing devices are operated based on the conditions to generate a 3D data representation of the surroundings. A combined 3D map of the surroundings is generated using a combination of the 3D data representation obtained from the cameras and the sensing devices.

Abstract: Text-to-image generation generally refers to the process of generating an image from one or more text prompts input by a user. While artificial intelligence has been a valuable tool for text-to-image generation, current artificial intelligence-based solutions are more limited as it relates to text-to-3D content creation. For example, these solutions are oftentimes category-dependent, or synthesize 3D content at a low resolution. The present disclosure provides a process and architecture for high-resolution text-to-3D content creation.

Abstract: In various embodiments, a training application trains a machine learning model to generate three-dimensional (3D) representations of two-dimensional images. The training application maps a depth image and a viewpoint to signed distance function (SDF) values associated with 3D query points. The training application maps a red, blue, and green (RGB) image to radiance values associated with the 3DI query points. The training application computes a red, blue, green, and depth (RGBD) reconstruction loss based on at least the SDF values and the radiance values. The training application modifies at least one of a pre-trained geometry encoder, a pre-trained geometry decoder, an untrained texture encoder, or an untrained texture decoder based on the RGBD reconstruction loss to generate a trained machine learning model that generates 3D representations of RGBD images.

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for generating three-dimensional meshes representing two-dimensional images for editing the two-dimensional images. The disclosed system utilizes a first neural network to determine density values of pixels of a two-dimensional image based on estimated disparity. The disclosed system samples points in the two-dimensional image according to the density values and generates a tessellation based on the sampled points. The disclosed system utilizes a second neural network to estimate camera parameters and modify the three-dimensional mesh based on the estimated camera parameters of the pixels of the two-dimensional image. In one or more additional embodiments, the disclosed system generates a three-dimensional mesh to modify a two-dimensional image according to a displacement input.

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for generating three-dimensional meshes representing two-dimensional images for editing the two-dimensional images. The disclosed system utilizes a first neural network to determine density values of pixels of a two-dimensional image based on estimated disparity. The disclosed system samples points in the two-dimensional image according to the density values and generates a tessellation based on the sampled points. The disclosed system utilizes a second neural network to estimate camera parameters and modify the three-dimensional mesh based on the estimated camera parameters of the pixels of the two-dimensional image. In one or more additional embodiments, the disclosed system generates a three-dimensional mesh to modify a two-dimensional image according to a displacement input.

Abstract: Methods are provided for generating training data in a form of a plurality of frames of facial animation, each of the plurality of frames represented as a three-dimensional (3D) mesh comprising a plurality of vertices. The training data is usable to train an actor-specific actor-to-mesh conversion model which, when trained, receives a performance of the actor captured by a head-mounted camera (HMC) set-up and infers a corresponding actor-specific 3D mesh of the performance of the actor. The methods may involve performing a blendshape optimization to obtain a blendshape-optimized 3D mesh and performing a mesh-deformation refinement on the blendshape-optimized 3D mesh to obtain a mesh-deformation-optimized 3D mesh. The training data may be generated on the basis of the mesh-deformation-optimized 3D mesh.

Abstract: Embodiments of the invention are directed to a computer-implemented method that includes accessing, using a processor system, a three-dimensional (3D) model of a device-under-design (DUD). The processor system is used to receive a first design operation associated with the 3D model of the DUD. A collaboration dependency model is used to make a prediction of a dependency relationship between the first design operation and a second design operation.

Abstract: An object of the present disclosure is to provide an annotation apparatus that can improve workability of annotation. An annotation apparatus (1) according to an example aspect of the present disclosure includes a cluster generation unit (11) configured to generate a plurality of clusters by grouping point cloud data corresponding to three-dimensional position information about a measurement target, a presentation order determination unit (12) configured to determine a presentation order of the plurality of clusters based on similarity between the generated plurality of clusters, a cluster presentation unit (13) configured to present the plurality of clusters in order based on the determined presentation order, and a label assignment unit (14) configured to assign a predetermined label to each of the clusters presented in the order.