Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for utilizing a visual neural network to replace materials in a three-dimensional scene with visually similar materials from a source dataset. Specifically, the disclosed system utilizes the visual neural network to generate source deep visual features representing source texture maps from materials in a plurality of source materials. Additionally, the disclosed system utilizes the visual neural network to generate deep visual features representing texture maps from materials in a digital scene. The disclosed system then determines source texture maps that are visually similar to the texture maps of the digital scene based on visual similarity metrics that compare the source deep visual features and the deep visual features. Additionally, the disclosed system modifies the digital scene by replacing one or more of the texture maps in the digital scene with the visually similar source texture maps.

Abstract: A computerized method comprising acquiring an image of a physical environment comprising one or more physical entities, generating a virtual view based on the acquired image, and displaying the virtual view overlaid on the acquired image. The method further comprises receiving user input being indicative of a placement location within the virtual view, and receiving user input for creating a virtual bounding volume within the virtual view. The method further comprises manipulating the virtual bounding volume based on one or more intersections in 3D space between the virtual bounding volume and 3D data corresponding to the one or more physical entities, displaying the manipulated virtual bounding volume, and displaying at least one 3D object model that fits the manipulated virtual bounding volume.

Abstract: Various implementations disclosed herein include devices, systems, and methods for normal estimation using a directional measurement, such as a gravity vector. In various implementations, a device includes a non-transitory memory and one or more processors coupled with the non-transitory memory. In some implementations, a method includes identifying planar surfaces in an environment represented by an image. Each planar surface is associated with a respective orientation. A directional vector associated with the environment is determined. A subset of the planar surfaces that have a threshold orientation relative to the directional vector is identified. For each planar surface in the subset of the planar surfaces, a normal vector for the planar surface is determined based on the orientation of the planar surface and the directional vector.

Abstract: A multi-view interactive digital media representation (MVIDMR) of an object can be generated from live images of an object captured from a camera. Selectable tags can be placed at locations on the object in the MVIDMR. When the selectable tags are selected, media content can be output which shows details of the object at location where the selectable tag is placed. A machine learning algorithm can be used to automatically recognize landmarks on the object in the frames of the MVIDMR and a structure from motion calculation can be used to determine 3-D positions associated with the landmarks. A 3-D skeleton associated with the object can be assembled from the 3-D positions and projected into the frames associated with the MVIDMR. The 3-D skeleton can be used to determine the selectable tag locations in the frames of the MVIDMR of the object.

Abstract: Some embodiments are directed to an image director of a patient monitoring system to obtain calibration images of a calibration sheet or other calibration object at various orientations and locations. The images are then stored and processed to calculate camera parameters defining the location and orientation of the image detector and identifying internal characteristics of the image detector, and the information are stored. The patient monitoring system can be re-calibrated by using the image detector to obtain an additional image of a calibration sheet or calibration object. The additional image and the stored camera parameters are then used to detect any apparent change in the internal characteristics of the image detector (10)(S6-4).

Abstract: A packaging system and method includes an immersive reality system that renders a 3D representation of an item to be shipped and the shipping box in which the item is to be shipped. A server includes a database of items and item data and shipping box data, which is used to produce a 3D representation of the item and the shipping box. A GUI exhibits the items such that a user may selectively view, select, and/or manipulate the items. A computer enables a user to query the server to ship a selected item. A network node transmits selected item data and shipping box data to the immersive reality system. The 3D representations include position and orientation data enabling manipulation of the 3D representation of the selected item relative to the 3D representation of the shipping box into a desired position. The system ships the selected item in the desired position.

Abstract: An apparatus, such as a graphical processing unit (GPU), includes one or more processors configured to determine a plurality of first locality information of a received wave at a processing unit and to select a first processing element of a plurality of processing elements, the first processing unit having a plurality of second locality information from a previous wave that matches the plurality of first locality information to execute the received wave.

Abstract: A computer-aided design (CAD) system may support anti-constraint configuration and enforcement for CAD models that represent physical objects. Anti-constraints may specify given constraints not to be applied for components of the CAD model, and the CAD system may update components of the CAD model without applying the constraints specified in anti-constraints applicable to the CAD model components.

Abstract: Provided is a process including: receiving one or more write requests to write a plurality of values to a plurality of fields in one or more tuples of a relational database, different ones of the values corresponding to different ones of the fields, detecting duplicates of the values with steps for expediting detection of duplicates, and selecting a first subset of the values based on the first subset of values corresponding to fields in a first subset of the fields, the first subset of fields being designated as higher-security fields than a second subset of fields among the plurality of fields.

Abstract: This application discloses methods, apparatus, and electronic devices for face pose estimation and three-dimensional face reconstruction. The face pose estimation method comprises: acquiring a two-dimensional face image for processing, constructing a three-dimensional face model corresponding to the two-dimensional face image, and determining a face pose of the two-dimensional face image based on face feature points of the three-dimensional face model and face feature points of the two-dimensional face image. With this approach, the face pose estimation is performed based on the three-dimensional face model corresponding to the two-dimensional face image, instead of only based on a three-dimensional average face model. As a result, a high accuracy pose estimation can be obtained even for a face with large-angle and exaggerated facial expressions. Thus, robustness of the pose estimation can be effectively improved.

Abstract: Systems and methods may be used to determine a fit for a hearing assistance device shell model. For example, a method may include receiving an image of anatomy of a patient including at least a portion of a canal aperture of an ear of the patient, generating a patient model of a portion of the anatomy of the patient, the patient model indicating at least one of a height or width of the canal aperture, and determining, using the patient model, a best fit model from a set of hearing assistance device shell models generated using a machine learning technique. The method may include outputting an identification of the best fit model.

Abstract: A system and method to collect feedback data from a patient wearing an interface such as a mask when using a respiratory pressure therapy device such as a CPAP device. The system includes a storage device including a facial image of the patient. An interface in communication with the respiratory pressure therapy device collect operational data from when the patient uses the interface. A patient interface collects subjective patient input data from the patient in relation to the patient interface. An analysis module correlates a characteristic of the interface with the facial image data, operational data and subjective patient input data.

Abstract: A CAD model comprises a set of at least three subentities. A candidate set of pairwise numerical constraints is obtained for the set of subentities, such that a first graph, representing the subentities of the set of subentities as nodes and the pairwise numerical constraints of the candidate set as edges, is connected. A minimal spanning subset of pairwise numerical constraints is obtained from the candidate set, such that a second graph, representing the subentities of the set of subentities as nodes and the pairwise numerical constraints of the minimal spanning subset as edges, is a spanning tree. A parameter set to parameterize the pairwise numerical constraints of the minimal spanning subset is determined. The parameter set and parametric constraints, based on the parameter set and the numerical values of the numerical constraints of the minimal spanning subset, are added to the CAD model.

Abstract: In various examples, live perception from sensors of a vehicle may be leveraged to detect and classify intersection contention areas in an environment of a vehicle in real-time or near real-time. For example, a deep neural network (DNN) may be trained to compute outputs—such as signed distance functions—that may correspond to locations of boundaries delineating intersection contention areas. The signed distance functions may be decoded and/or post-processed to determine instance segmentation masks representing locations and classifications of intersection areas or regions. The locations of the intersections areas or regions may be generated in image-space and converted to world-space coordinates to aid an autonomous or semi-autonomous vehicle in navigating intersections according to rules of the road, traffic priority considerations, and/or the like.

Abstract: The present disclosure generally relates to an avatar editing user interface. The avatar editing user interface includes a first view of an avatar at a first size. A request to enter an editing mode for a first feature of a plurality of avatar features of the avatar enables an editing mode for the first feature and causes a second view of the avatar at a second size to be displayed. The editing mode also shifts the avatar to emphasize the first feature relative to other avatar features.

Abstract: The disclosure notably relates to computer-implemented method for learning a neural network configured for inference, from a freehand drawing representing a 3D shape, of a solid CAD feature representing the 3D shape. The method includes providing a dataset including freehand drawings each representing a respective 3D shape, and learning the neural network based on the dataset. The method forms an improved solution for inference, from a freehand drawing representing a 3D shape, of a 3D modeled object representing the 3D shape.

Abstract: A method for virtual spectacle adjustment and a corresponding computer program and computing device are disclosed. First measurement points are defined on a 3D model of a person's head, and a model of a frame is adjusted on the basis of the first measurement points. Defining the first measurement points includes defining second measurement points on a parametric head model, adjusting the parametric head model to the 3D model of the person's head, and determining the first measurement points on the basis of the second measurement points and the adjustment. In this way, the second measurement points only have to be defined once on the parametric model so that the first measurement points can be defined for a plurality of different 3D models of different heads.

Abstract: Systems and methods are provided to create training data, validate, deploy and test artificial intelligence (AI) systems in a virtual development environment, incorporating virtual spaces, objects, machinery, devices, subsystems, and actual human action and behavior.

Abstract: Methods of and systems for characterization of a 3D point cloud are disclosed. The method comprises accessing a 3D point cloud, the 3D point cloud being a set of data points representative of the object, determining, based on the 3D point cloud, a 3D reconstructed object, determining, based on the 3D reconstructed object, a digital framework of the 3D point cloud, the digital framework being a ramified 3D tree structure, the digital framework being representative of a base structure of the object, morphing a 3D reference model of the object onto the 3D reconstructed object, the morphing being based on the digital framework; and determining, based on the morphed 3D reference model and the 3D reconstructed object, characteristics of the object.

Abstract: A system and method for creating a physical three-dimensional model based on a plurality of digital assets is disclosed. The system comprises a first computing device, a second computing device, and a third computing device. The first computing device is configured for sending, over a communications network to a second computing device that is provisioned and managed by a third party, a compressed top-down model, and a compressed three-dimensional base model to be stored on the second computing device. The third computing device is configured for a user to interact with a user interface provided by the second computing device to combine digital assets and wherein the system simultaneously generates a three-dimensional rendering of the composite of the digital assets. The system then creates a new digital asset stored on a blockchain network and generates physical model thereof based on the composite three-dimensional rendering.

Abstract: A method for simulating a scenario in a computing environment is described. The method includes determining a user intent of a user from an input received from the user for executing at least one intended task by the user, generating a feature set based on analyzing the user intent, and processing, based on the user intent, at least one data feed received from a database to select at least one plan of action for executing the at least one intended task. At least one plan of action is simulated in a graphical user interface as virtual or augmented reality based on the feature set to enable at least one of responding additionally received input from the user and the at least one intended task is performed according to the user intent.

Abstract: An information processing system, a program, and an information processing method that can improve convenience when operating a movable object within an application. An information processing system that provides an application that can be executed on an information processing terminal includes a detection unit that detects a three-dimensional motion of a finger of a user, and a motion control unit that controls a three-dimensional motion of a movable object within the application based on the three-dimensional motion of the finger of the user detected by the detection unit.

Abstract: In general, according to one embodiment, a stress analysis method comprising: dividing a surface of an object into a plurality of first rectangles each having a first size, on data; and acquiring a first type value for each of the first rectangles. The method further includes: specifying, from among the first rectangles, a plurality of second rectangles that have the first type value of a magnitude that falls within a first range and form a rectangle; and generating a stress model for a set of the second rectangles by using the second rectangles as an element.

Abstract: The disclosure relates to a computer-implemented method comprising inputting a representation of a 3D modeled object to an abstraction neural network which outputs a first set of a first number of first primitives fitting the 3D modeled object; and determining, from the first set, one or more second sets each of a respective second number of respective second primitives. The second number is lower than the first number. The determining includes initializing a third set of third primitives as the first set and performing one or more iterations, each comprising to merging one or more subsets of third primitives together each into one respective single fourth primitive, to thereby obtain a fourth set of fourth primitives. Each iteration further comprises setting the third set of a next iteration as the fourth set of a current iteration and setting the one or more second sets as one or more obtained fourth sets.

Abstract: A computer system obtains a digital 3-dimensional model of a body surface, such as a human face. Portions of the surface correspond to reference points in the model. The system obtains image data of a personal care device in use as well as image data of the surface on which the model is based. The system calculates a relative location of the device relative to the reference points based on the image data of the surface. The system determines a true position of the device with respect to the surface based on the image data of the device in use, known physical dimensions of the device, and its relative location, even when the device is partially occluded from view. The system may, based on the true position of the device, cause device to perform actions such as application of a cosmetic or administration of a skin therapy.

Abstract: Systems and techniques are provided for widening a hierarchical structure for ray tracing. For instance, a process can include obtaining a plurality of primitives of a scene object included in a first hierarchical acceleration data structure and determining one or more candidate hierarchical acceleration data structures each including the plurality of primitives. A cost metric can be determined for the one or more candidate hierarchical acceleration data structures and, based on the cost metric, a compressibility prediction associated with a candidate hierarchical acceleration data structure of the one or more candidate hierarchical acceleration data structures can be determined. An output hierarchical acceleration data structure can be generated based on the compressibility prediction.

Abstract: Systems and methods for creating images of an environment includes controlling at least one camera to acquire imaging data from the environment and selecting, from the imaging data, a three-dimensional-two-dimensional correspondence as a control point for use in a perspective-n-point problem to determine a position and orientation of the at least one camera from n known correspondences between three-dimensional object points and their two-dimensional image projections in the environment. The method also includes reprojecting at least a selected number of the projections, determining a reprojection error for each of the projections, and performing a weight assignment of reprojection errors to distinguish the inliers form outliers. These steps of the method are repeated to apply the weight assignment to outliers in a decreased fashion during iterations to reduce an impact of outliers in the real-time display of the environment.

Abstract: A method for denoising an image. The method includes training a fully convolutional neural network (FCN) and generating a reconstructed image by applying the FCN on the image. Training the FCN includes generating an nth training image of a plurality of training images, initializing the FCN with a plurality of initial weights, and repeating a first iterative process. The nth training image includes a training array In. The first iterative process includes extracting an nth denoised training image from the FCN, generating a plurality of updated weights, and replacing the plurality of initial weights with the plurality of updated weights. The nth denoised training image may be extracted by applying the FCN on the nth training image. In an exemplary embodiment, the nth denoised training image may include a denoised array În. The plurality of updated weights is generated by minimizing a loss function including ?n=1N|In?În|.

Abstract: Disclosed herein is a method for receiving point cloud data, including receiving a bitstream containing the point cloud data, and decoding the point cloud data. Disclosed herein is a method for transmitting point cloud data, including encoding the point cloud data, and transmitting a bitstream containing the point cloud data.

Abstract: Using various embodiments, methods and systems for generating three-dimensional (3D) digital smart objects for use in various 3D environments to collect data related to user interaction or viewability with the 3D digital smart object are described. In one embodiment, a system can generate a 3D digital smart object by presenting a 3D placeholder to a publisher or developer of the 3D environment, receive an asset, receive asset data including asset category and asset type associated to the asset and a minimum and maximum asset polygon count value. The system can also receive standard asset size information of the asset including an X-axis, Y-axis, and Z-axis dimension of the asset, and receive an asset positioning information, including an anchor location and an asset orientation of the asset. The data received can then be stored to generate the 3D digital smart object that can be placed within a 3D environment.

Abstract: A point cloud data processing method and device, a computer device and a storage medium are provided. The method includes: acquiring point cloud data, and constructing a corresponding neighboring point set for each of data points in the point cloud data; calculating Hausdorff distances between the neighboring point set and a pre-constructed kernel point cloud to obtain a distance matrix; calculating a convolution of the neighboring point set with the distance matrix and a network weight matrix in a Hausdorff convolution layer in an encoder, to obtain high-dimensional point cloud features, the encoder and a decoder being two parts in a deep learning network; and reducing feature dimension of the high-dimensional point cloud features through the decoder, so that a classifier performs semantic classification on the point cloud data according to object point cloud features obtained by the dimension reduction.

Abstract: A system for creating an animation including a non-transitory memory storing an executable code, a hardware processor executing the executable code to receive a first scene input including a plurality of scene elements from an input device, create a three-dimensional (3D) digital representation of each of the plurality of scene elements in the first scene input, transmit the 3D digital representation of the plurality of scene elements in the first scene input for physical production, receive a first scene capture depicting the physical 3D reproductions of the plurality of scene elements of the first scene input arranged to recreate the first scene input from a recording device, process the first scene capture for display, and transmit the first scene capture for display.

Abstract: Systems and methods are provided. A physical measurement of the core loss increase associated with a physical deformation of a material of the device is obtained. A data structure describing a model of the device is accessed. A first edge of the model of the device associated with a physical deformation of the device is identified. A finite element mesh is generated to include a single layer mesh comprising a plurality of mesh elements associated with the first edge of the finite element mesh. A core loss value is assigned to each of the plurality of mesh elements. Each of the core loss values representative of the physical measurement of the core loss increase of the material as a result of the physical deformation of the material. An electromagnetic model is generated by performing a finite element analysis based on the finite element mesh and the single layer mesh.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a storage device, for electric grid asset detection are enclosed. An electric grid asset detection method includes: obtaining overhead imagery of a geographic region that includes electric grid wires; identifying the electric grid wires within the overhead imagery; and generating a polyline graph of the identified electric grid wires. The method includes replacing curves in polylines within the polyline graph with a series of fixed lines and endpoints; identifying, based on characteristics of the fixed lines and endpoints, a location of a utility pole that supports the electric grid wires; detecting an electric grid asset from street level imagery at the location of the utility pole; and generating a representation of the electric grid asset for use in a model of the electric grid.

Abstract: Techniques related to validating an image based 3D model of a scene are discussed. Such techniques include detecting an object within a captured image used to generate the scene, projecting the 3D model to a view corresponding to the captured image to generate a reconstructed image, and comparing image regions of the captured and reconstructed images corresponding to the object to validate the 3D model.

Abstract: In some examples, a system identifies sub-portions of a database query, assigns identifiers to the identified sub-portions, and adds the identifiers to a data structure. The system generates a fingerprint representing the database query based on applying a fingerprint function on the data structure including the identifiers.

Abstract: The present invention provides systems and methods for generating a 3D product mesh model and product dimensions from user images. The system is configured to receive one or more images of a user's body part, extract a body part mesh having a plurality of body part key points, generate a product mesh from an identified subset of the body part mesh, and generate one or more product dimensions in response to the selection of one or more key points from the product mesh. The system may output the product mesh, the product dimensions, or a manufacturing template of the product. In some embodiments, the system uses one or more machine learning modules to generate the body part mesh, identify the subset of the body part mesh, generate the product mesh, select the one or more key points, and/or generate the one or more product dimensions.

Abstract: In one embodiment, a method includes receiving a user input from a first user from a client system associated with the first user, determining one or more products of interest to the first user based on one or more machine-learning models and the user input, accessing social data associated with the one or more products of interest from a data store, wherein the social data comprises social signals or content objects associated with one or more second users, each second user being within a threshold degree of separation of the first user within a social graph, generating one or more social summaries associated with the one or more products of interest based on the accessed social data, and sending instructions for presenting the one or more social summaries of the one or more products of interest to the client system.

Abstract: In variants, the method can include: determining a timeseries of measurements of a geographic region; determining a set of object representations from the timeseries of measurements; and determining a timeseries of object versions based on relationships between the object representations.

Abstract: Provided are a method and apparatus for creating a composite image, by which the state of a target may be effectively expressed, and a computer program stored in a recording medium to execute the method. The composite image creating method, performed by a computing apparatus, of synthesizing an input image to a target image, includes identifying information of the input image by obtaining the input image, generating a projected image based on information about a position in units of sub-pixels for the target image of the input image, by using information of the input image, generating a reduced image by reducing the projected image at a ratio corresponding to the target image, and synthesizing the reduced image to the target image.

Abstract: Apparati, methods, and computer readable media for inserting identity information from a source image (static image or video) (301) into a destination video (302), while mimicking motion of the destination video (302). In an apparatus embodiment, an identity encoder (304) is configured to encode identity information of the source image (301). When source image (301) is a multi-frame static image or a video, an identity code aggregator (307) is positioned at an output of the identity encoder (304), and produces an identity vector (314). A driver encoder (313) is coupled to the destination (driver) video (302), and has two components: a pose encoder (305) configured to encode pose information of the destination video (302), and a motion encoder (315) configured to separately encode motion information of the destination video (302). The driver encoder (313) produces two vectors: a pose vector (308) and a motion vector (316).

Abstract: Provided is a process including: receiving one or more write requests; selecting a first subset of the values as corresponding to higher-security fields; segmenting a first value in the first subset; instructing a first computing device to store a first subset of segments among the plurality of segments in memory; and instructing a second computing device to store a second subset of segments among the plurality of segments in memory.

Abstract: Systems and methods for segmenting scan data are disclosed. The methods include creating a graph from scan data representing a plurality of points in an environment associated with a ground and one or more objects, where the graph includes a plurality of vertices corresponding to the plurality of points in the environment, a first terminal vertex associated with the ground label, and a second terminal vertex associated with the non-ground label. A unary potential being the cost of assigning a vertex to a ground label or a non-ground label is assigned to each vertex, and a pairwise potential is assigned to each pair of neighboring vertices in the graph as a measure of a cost of assigning different labels. The methods include using the unary the pairwise potentials to identify labels for each point and segmenting the scan data to identify points associated with the ground.

Abstract: The disclosure describes one or more embodiments of systems, methods, and non-transitory computer-readable media that modify vertex positions of a 3D-object mesh to align the vertices with discontinuities indicated by a sharpness map for a displacement map. For example, the disclosed systems access a sharpness map that encodes discontinuities of a displacement map to preserve sharp features. During sharp tessellation, for a given vertex of the 3D-object mesh, the disclosed systems iterate over neighboring vertices to search for a nearest intersection between one of the incident edges of the vertex and neighboring vertices and a discontinuity line from the sharpness map. Then, the vertex is moved within a threshold distance of the nearest discontinuity line intersection along the incident edge. The disclosed systems similarly reposition other vertices of the 3D-object mesh to generate a modified 3D-object mesh that includes vertices that align with discontinuities present in a displacement map.

Abstract: Apparati, methods, and computer readable media for inserting identity information from a source image (1) into a destination image (2), while mimicking illumination of the destination image (2). In an apparatus embodiment, an identity encoder (4) is configured to encode just identity information of the source image (1) and to produce an identity vector (7), where the identity encoder (4) does not encode any pose information or illumination information of the source image (1). A driver encoder (12) has two components: a pose encoder (5) configured to encode pose information of the destination image (2) and an illumination encoder (6) configured to separately encode illumination information of the destination image (2), and to produce two vectors: a pose vector (8) and an illumination vector (9). A neural network generator (10) is coupled to the identity encoder (4) and to the driver encoder (12), and has three inputs: the identity vector (7), the pose vector (8), and the illumination vector (9).

Abstract: A manifold voxel mesh or surface mesh is manufacturable by carving a single block of material and a non-manifold mesh is not manufacturable. Conventional techniques for constructing or extracting a surface mesh from an input point cloud often produce a non-manifold voxel mesh. Similarly, extracting a surface mesh from a voxel mesh that includes non-manifold geometry produces a surface mesh that includes non-manifold geometry. To ensure that the surface mesh includes only manifold geometry, locations of the non-manifold geometry in the voxel mesh are detected and converted into manifold geometry. The result is a manifold voxel mesh from which a manifold surface mesh of the object may be extracted.

Abstract: An illustrative 3D modeling system generates a first voxelized representation of an object with respect to a voxel space and a second voxelized representation of the object with respect to the voxel space. Based on a first normal of a first voxel included in the first voxelized representation and a second normal of a second voxel included in the second voxelized representation, the 3D modeling system identifies a mergeable intersection between the first and second voxels. Based on the first and second voxelized representations, the 3D modeling system generates a merged voxelized representation of the object with respect to the voxel space. The merged voxelized representation includes a single merged voxel generated, based on the identified mergeable intersection, to represent both the first and second voxels. Corresponding methods and systems are also disclosed.

Abstract: Methods for lossy and lossless pre-processing of image data. In one embodiment, a method for lossy pre-processing image data, where the method may include, at a computing device: receiving the image data, where the image data includes a model having a mesh, the mesh includes vertices defining a surface, the vertices including attribute vectors, and the attribute vectors including values. The method also including quantizing the values of the attribute vectors to produce modified values, where a precision of the modified values is determined based on a largest power determined using a largest exponent of the values, encoding pairs of the modified values into two corresponding units of information. The method also including, for each pair of the pairs of the modified values, serially storing the two corresponding units of information as a data stream into a buffer, and compressing the data stream in the buffer.

Abstract: The present disclosure includes methods and systems for identifying and manipulating a segment of a three-dimensional digital model based on soft classification of the three-dimensional digital model. In particular, one or more embodiments of the disclosed systems and methods identify a soft classification of a digital model and utilize the soft classification to tune segmentation algorithms. For example, the disclosed systems and methods can utilize a soft classification to select a segmentation algorithm from a plurality of segmentation algorithms, to combine segmentation parameters from a plurality of segmentation algorithms, and/or to identify input parameters for a segmentation algorithm. The disclosed systems and methods can utilize the tuned segmentation algorithms to accurately and efficiently identify a segment of a three-dimensional digital model.

Abstract: A virtual image generation method and apparatus, an electronic device and a storage medium which relates to the field of artificial intelligence technologies such as augmented reality, computer vision and deep learning. A specific implementation scheme involves: acquiring base coefficients corresponding to key points of a target face based on a target face picture; generating a structure of a virtual image of the target face based on a mapping relationship of spatial alignment between a preset virtual model and a standard model, a base of the standard model and the base coefficients corresponding to the key points of the target face; and performing texture filling on the structure of the virtual image based on textures of the target face picture, to obtain the virtual image of the target face.