Abstract: Methods and systems are provided for accurately filling holes, regions, and/or portions of high-resolution images using guided upsampling during image inpainting. For instance, an image inpainting system can apply guided upsampling to an inpainted image result to enable generation of a high-resolution inpainting result from a lower-resolution image that has undergone inpainting. To allow for guided upsampling during image inpainting, one or more neural networks can be used. For instance, a low-resolution result neural network (e.g., comprised of an encoder and a decoder) and a high-resolution input neural network (e.g., comprised of an encoder and a decoder). The image inpainting system can use such networks to generate a high-resolution inpainting image result that fills the hole, region, and/or portion of the image.

Abstract: Machine learning is used to generate a first mipmap of a texture having a first compression based on a second mipmap of the same texture and having a second compression without using compression or decompression in generating the first mipmap. The first mipmap can then be used to render a computer graphics object.

Abstract: Embodiments of the present invention provide systems, methods, and computer storage media for editing images using a web-based intermediary between a user interface on a client device and an image editing neural network(s) (e.g., a generative adversarial network) on a server(s). The present image editing system supports multiple users in the same software container, advanced concurrency of projection and transformation of the same image, clubbing transformation requests from several users hosted in the same software container, and smooth display updates during a progressive projection.

Abstract: An apparatus includes circuitry configured to: select, from at least two images captured in different image-capturing directions and with image-capturing ranges overlapping with each other, an image to be at foreground as viewed from a virtual camera based on an orientation or an angle of view of the virtual camera and the image-capturing directions of the at least two images; map the at least two images onto a three-dimensional object to generate a virtual image, in which the at least two images overlap with each other, having a wider angle of view than the at least two images; and perform perspective projection on the virtual image using the virtual camera, to generate a plane image, based on the selected image to be at the foreground, as a display image.

Abstract: Systems and methods for entity recommendation can make use of rich data by allowing the items to be recommended and the recipients of the recommendation (e.g., users) to be modeled as “complex entities” composed of one or more static sub-entities and/or a dynamic component, and by utilizing information about multiple relationships between the sub-entities as reflected in bipartite graphs. Generating recommendations from such information may involve creating vector representations of the sub-entities based on the bipartite graphs (e.g., using graph-based convolutional networks), and combining these vector representations into representations of the items and users (or other recipients) to be fed into a classifier model.

Abstract: A computer-implemented method using a convolutional neural network is provided. The computer-implemented method using a convolutional neural network includes processing an input image through at least one channel of the convolutional neural network to generate an output image including content features of the input image morphed with style features of a reference style image. The at least one channel includes a down-sampling segment, a densely connected segment, and an up-sampling segment sequentially connected together. Processing the input image through the at least one channel of the convolutional neural network includes processing an input signal through the down-sampling segment to generate a down-sampling segment output; processing the down-sampling segment output through the densely connected segment to generate a densely connected segment output; and processing the densely connected segment output through the up-sampling segment to generate an up-sampling segment output.

Abstract: Automated three-dimensional (3D) building model estimation is disclosed. In an embodiment, a method comprises: obtaining, using one or more processors, an aerial image of a building based on an input address; obtaining, using the one or more processors, three-dimensional (3D) data containing the building based on the input address; pre-processing the aerial image and 3D data; reconstructing, using the one or more processors, a 3D building model from the pre-processed image and 3D data, the reconstructing including: predicting, using instance segmentation, a mask for each roof face; converting each roof face into a two-dimensional (2D) polygon; and projecting each 2D polygon into a 3D polygon representing the 3D building model.

Abstract: Embodiments of the present disclosure relate to computer storage, methods, and systems for the optimization of accessible color themes. Systems and methods are disclosed that leverage the use of confusion lines to identify and highlight relationships between colors that may be inaccessible (e.g., indistinguishable) for a person with a vision impairment, such as a color vision deficiency. In some embodiments, a graphical user interface is provided that, based on a selection of colors in a color wheel, visually indicates curves of confusion for each color in the selection of colors. Each curve of confusion visually indicates a confusion of colors for a type of vision impairment, such as a CVD.

Abstract: Embodiments of the present invention provide a novel solution that uses subjective end-user input to generate optimal image quality settings for an application. Embodiments of the present invention enable end-users to rank and/or select various adjustable application parameter settings in a manner that allows them to specify which application parameters and/or settings are most desirable to them for a given application. Based on the feedback received from end-users, embodiments of the present invention may generate optimal settings for whatever performance level the end-user desires. Furthermore, embodiments of the present invention may generate optimal settings that may be benchmarked either on a server farm or on an end-user's client device.

Abstract: A method for generating a classification model using a training data set. An iterative procedure for training an ANN model, in which an iteration includes selecting a small sample of training data from a source of training data, training the model using the sample, using the model in inference mode over a larger sample of the training data, and reviewing the results of the inferencing. The results can be evaluated to determine whether the model is satisfactory, and if it does not meet specified criteria, then cycles of sampling, training, inferencing and reviewing results (STIR cycles) are repeated in an iterative process until the criteria are met. A classification engine trained as described herein is provided.

Abstract: A mechanism is described for facilitating hemisphere cube map projection format imaging environments, according to one embodiment. A method of embodiments, as described herein, includes capturing, by a camera coupled to one or more processors, an image having image content, wherein the image content being represented by the image is omnidirectional such that the image content is mapped on a sphere while representing less than the sphere; mapping the image to a cubic representation based on six faces of a cube, wherein one or more of the six faces are classified as inactive regions such that they remain unoccupied or partially occupied; and arranging, based on the cubic representation, the six faces in a compact representation by avoiding inclusion of the inactive regions.

Abstract: Described herein is a technique in which a plurality of distortion meshes compensate for radial and chromatic aberrations created by optical lenses. The plurality of distortion meshes may include different lens specific parameters that allow the distortion meshes to compensate for chromatic aberrations created within received images. The plurality of distortion meshes may correspond to a red color channel, green color channel, or blue color channel to compensate for the chromatic aberrations. The distortion meshes may also include shaped distortions and grids to compensate for radial distortions, such as pin cushion distortions. In one example, the system uses a barrel-shaped distortion and a triangulation grid to compensate for the distortions created when the received image is displayed on a lens.

Abstract: A system and method of augmenting image data are described. In one embodiment, the method comprises receiving data of an image captured by a camera, the captured image including a region having a visible feature of an object, storing masking data defining a plurality of masks, each mask defining a respective masked portion of the region of the captured image, sampling pixel values at predefined locations of the captured image data, selecting at least one stored mask based on the sampled pixel values, modifying pixel values in the or each selected masked portion of the region of the captured image based on colourisation parameters, and outputting the captured image with the modified pixel values for display. In other embodiments pixel values of one or more identified regions of a face in a target image are modified based on the augmentation characteristics derived from corresponding identified regions of a face in a source image.

Abstract: A cognitive well log analysis using a computer includes receiving, by one or more processors, well log data from a plurality of well logs. One or more processors identify geometric features, specific values, and well attributes from the received well log data and embed the identified features to generate a plurality of intermediate vectors arranged based on a relevance for identifying petrophysical properties. The intermediate vectors are combined to create a fusion vector based on which the one or more processors identify reservoir layers.

Abstract: The present invention provides a method for reconstructing a 3D object based on dynamic graph network, first, obtaining a plurality of feature vectors from 2D image I of an object; then, preparing input data: predefining an initial ellipsoid mesh, obtaining a feature input X by filling initial features and creating a relationship matrix A corresponding to the feature input X; then, inputting the feature input X and corresponding relationship matrix A to a dynamic graph network for integrating and deducing of each vertex's feature, thus new relationship matrix is obtained and used for the later graph convoluting, which improves the initial graph information and makes the initial graph information adapted to the mesh relation of the corresponding object, therefore the accuracy and the effect of 3D object reconstruction have been improved; last, regressing the position, thus the 3D structure of the object is deduced, and the 3D object reconstruction is completed.

Abstract: In one aspect of the present application, a computer-implemented method for setting a local coordinate system of a tooth 3D digital model is provided, the method comprises: obtaining a first 3D digital model representing a first tooth, wherein the first 3D digital model is based on a world coordinate system; and setting a local coordinate system for the first 3D digital model using a first artificial neural network based on the first 3D digital model, where the first artificial neural network is a trained deep learning artificial neural network.

Abstract: The present principles relate to a method and device for gamut mapping from a first color gamut towards a second color gamut. The method comprises obtaining a preserved key color (prsv_colorK0) and a hue alignment angle (?K0) of the key color (colorK0) for each key color (colorK0) of at least 3 key colors to define a preserved gamut and a rotated gamut; and performing a hue mapping of a current color (color) from the first color gamut towards the second color gamut wherein in case the current color (color) is in a preserved gamut defined by the preserved key color (prsv_colorK0), the hue mapped current color is unchanged, and in case the current color is out of the preserved gamut, the hue mapped current color is calculated from the hue alignment angle (?K0) of two adjacent key colors of the current color (rotated gamut) and from the preserved area.

Abstract: A system is configured for converting an unstandardized architecture diagram into a braille language diagram is disclosed. The system receives the unstandardized architecture diagram which includes a plurality of architecture components. The system receives a standardized model that includes features to depict the architecture components of the unstandardized architecture diagram in a standard format. The system determines the architecture components, their connections, and their sequences from the unstandardized architecture diagram. The system determines the features to depict the architecture components of the unstandardized architecture diagram in the standard format. The system applies the identified features on the identified architecture components in the unstandardized architecture diagram. The system determines a standardized graphical representation of the unstandardized architecture diagram.

Abstract: Various implementations disclosed herein include devices, systems, and methods that complete content for a missing part of an image of an environment. For example, an example process may include obtaining an image including defined content and missing parts for which content is undefined, determining a spatial image transformation for the image based on the defined content and the missing parts of the image, altering the image by applying the spatial image transformation, and completing the altered image.

Abstract: A method includes processing, using at least one processor of an electronic device, each of multiple images using a photometric augmentation engine, where the photometric augmentation engine performs one or more photometric augmentation operations. The method also includes applying, using the at least one processor, multiple layers of a convolutional neural network to each of the images, where each layer generates a corresponding feature map. The method further includes processing, using the at least one processor, at least one of the feature maps using at least one feature augmentation engine between consecutive layers of the multiple layers, where the at least one feature augmentation engine performs one or more feature augmentation operations.