Abstract: A method may include obtaining a plurality of frames of at least one video; determining an occurrence of a change from a first event of the at least one video to a second event of the video; determining a location in the plurality of frames; inserting one or more masked frames at the location between at least one frame of the first event and at least one frame of the second event; determining at least one transition component present in the at least one frame of the first event and the at least one frame of the second event; determining a motion of pixels for the at least one transition component across the plurality of frames of the first event and the plurality of frames of the second event; providing at least one transition effect in the one or more masked frames between the first event and the second event.

Abstract: An apparatus, method, and computer-readable medium for rendering a 3D image using a neural network for implicit representation of the image. UV coordinates of a texture map corresponding to a point on the image are calculated using a texture neural network. The image is rendered by applying a color value located at the UV coordinates of the texture map to the point.

Abstract: Described is a system (and method) for specialized data management when performing incremental backups to a cloud-based object storage. When backing up client data, the server may utilize incremental backups to conserve the amount of storage space required on the object storage. However, the server may implement a configuration that enables full reconstruction of the client data at the time of recovery. To provide such capabilities, the server may leverage a specialized metadata database. The metadata database may be an embedded database maintained by the server. The server may leverage the metadata database to provide further storage efficiencies by storing certain incremental backup data exclusively within the metadata database. Accordingly, the server may implement a data management configuration as part of a backup service that conserves storage usage, and as a result, storage costs when using on a third-party cloud-based object storage.

Abstract: Described is a system for managing backup data stored on a cloud-based object storage irrespective of the storage provider. The system may include a server (or gateway) that provides a backup service, and acts as an intermediary between a client device and object storage. To provide such capabilities, the system may manage an application programming interface (API) that provides a uniform set of methods to a client device irrespective of the destination object storage provider. A function library may include provider-specific functions that invoke corresponding storage-layer operations provided by the destination object storage. The server may also manage an operation pool to schedule operations and saturate bandwidth to the object storage. The server may also implement a specialized metadata database to provide further storage efficiencies by storing certain backup data exclusively within the metadata database.

Abstract: Example systems and methods provide input suggestions to a user to improve user experience on user devices. The input suggestions can be fill information from another app on device to the present app being used by user, information for performing a search (without the user having to copy-paste data or entering the data manually), responses to a message/notification received by the user, information/content/data to be shared between apps (without switching between apps), and emojis/GIFs that can be used by the user. The method includes analyzing one or more content of one or more screen displayed on device, generating at least one of a logical tree structure and a data mashup model of the one or more analyzed content for each screen, and providing a recommendation to a user. The recommendation can be a connected action or an input suggestion.

Abstract: The present disclosure describes methods and devices for generating a vector line drawing. A vector line drawing network may include a machine learning-based model that is trained to convert a raster image to a vector line drawing directly. The vector line drawing network may be trained end-to-end, using supervised learning, where only raster images are used as training data. A vector line drawing is generated stroke by stroke, over a series of time steps. In each time step, a dynamic drawing window is moved and scaled across the input raster image to sample a patch of the raster image, and a drawing stroke is predicted to draw a stroke in a corresponding patch in the canvas for the vector line drawing. The image patches are pasted in the canvas to assemble a final vector line drawing that corresponds to the input raster image.

Abstract: Approaches described and suggested herein relate to generating three-dimensional representations of objects to be used to render virtual reality and augmented reality effects on personal devices such as smartphones and personal computers, for example. An initial surface mesh of an object is obtained. A plurality of silhouette masks of the object taken from a plurality of viewpoints is also obtained. A plurality of depth maps are generated from the initial surface mesh. Specifically, the plurality of depth maps are taken from the same plurality of viewpoints from which the silhouette images are taken. A volume including the object is discretized into a plurality of voxels. Each voxel is then determined to be either inside the object or outside of the object based on the silhouette masks and the depth data. A final mesh is then generated from the voxels that are determined to be inside the object.

Abstract: The disclosure relates to an offline-online question answering system. In some aspects, the techniques described herein relate to a method including: receiving, by a processor, a query from a user; generating, by the processor, a query embedding representing the query; identifying, by the processor, at least one question corresponding to the query by comparing the query embedding to a plurality of embeddings of questions; and transmitting, by the processor, an answer corresponding to the at least one question to the user in response to the query.

Abstract: An apparatus, method, and computer-readable medium for rendering a 3D image using a neural network for implicit representation of the image. UV coordinates of a texture map corresponding to a point on the image are calculated using a texture neural network. The image is rendered by applying a color value located at the UV coordinates of the texture map to the point.

Abstract: Described is a system (and method) for providing multiple types of granular delete operations for a cloud-based object storage. The system may include a server that acts as an intermediary between a client device and an object storage that stores client data. The server may allow a client to perform a hard delete (or permanent) delete and a soft delete. The server may perform a specialized processing to synthetically create the soft delete feature using a permanent object delete operation provided by the object storage. The specialized processing may include manipulating retention periods associated with objects depending on whether the object is subject to a hard delete or a soft delete. As a result, the server may provide the ability for clients to perform a soft delete at an object level using existing storage level APIs that do not directly support such a feature.

Abstract: Described is a system for synchronizing expiration times for incremental backup data stored on a cloud-based object storage. More particularly, the system may provide a layer of intelligence when updating the expiration times associated with backup data to ensure that a full recovery of the client data to a point-in-time of any incremental backup may be performed. To provide such a capability, the system may maintain specialized metadata identifying expiration times for objects and a list of objects required to perform a full recovery to a point-in-time for each of the performed backups (e.g. full or incremental). The system may access this metadata to identify which objects stored by a previous backup are still referenced by a subsequent backup. Based on the identified objects, the system may synchronize object expiration times to ensure objects are not prematurely deleted from the object storage.

Abstract: The present disclosure describes methods and devices for generating a vector line drawing. A vector line drawing network may include a machine learning-based model that is trained to convert a raster image to a vector line drawing directly. The vector line drawing network may be trained end-to-end, using supervised learning, where only raster images are used as training data. A vector line drawing is generated stroke by stroke, over a series of time steps. In each time step, a dynamic drawing window is moved and scaled across the input raster image to sample a patch of the raster image, and a drawing stroke is predicted to draw a stroke in a corresponding patch in the canvas for the vector line drawing. The image patches are pasted in the canvas to assemble a final vector line drawing that corresponds to the input raster image.

Abstract: Embodiments herein disclose a method and electronic device for predicting multi-modal drawings. The method includes: receiving, by the electronic device, at least one of a text input and strokes of a drawing and determining, by the electronic device, features associated with the text input and features associated with the strokes of the drawing. The method includes classifying, by the electronic device, the features associated with the text input and the features associated with the strokes of the drawing into one of a dominant feature and a non-dominant feature and performing, by the electronic device, early concatenation or late concatenation of the features based on the classification; classifying, by the electronic device, the strokes of the drawing based on the concatenation into a category using a deep neural network (DNN) model; and predicting, by the electronic device, primary drawings corresponding to the category.

Abstract: Images of various views of objects can be captured. An object mesh structure can be created based at least in part on the object images. The object mesh structure represents the three-dimensional shape of the object. Alpha masks indicating which pixels are associated with the object can be used to refine the object mesh structure. A request can be made to view the object from an arbitrary viewpoint which differs from the viewpoints associated with the captured images. A subset of the captured images can be used to create a synthetic image. Different weights can be assigned to the captured image to render a synthetic image that represents the view from the arbitrary viewpoint selected. The input images for the synthetic image can be prefetched, or loaded into memory before the arbitrary view is requested. The images can also be cached for future use or to avoid reloading them for another synthetic image.

Abstract: Described is a system for rule-based re-tiering of backup data stored on a cloud-based object storage. More specifically, the system may re-tier objects based on one or more storage rules such that the system retains the ability to efficiently perform a full restore of backup data even when incremental backups are performed to a cloud-based object storage. To provide such a capability, the system may maintain a specialized metadata database that stores information indicating the backup time for each backup, and a list of objects required to perform a full restore to each of the backup times. Accordingly, when initiating a re-tiering based on one or more storage rules, the system may intelligently select candidate objects for re-tiering by leveraging the metadata database to ensure that objects that may still need to be referenced are not unnecessarily moved to a lower storage tier.

Abstract: Present disclosure relates to a method and a user equipment (UE) for enhancing multimodal input content. The UE obtains text-blocks associated with multimodal input and determines a plurality of features for each of the text-blocks using a natural language processing (NLP) model and a neural network (NN) model. Further, the UE identifies usage pattern of the plurality of features based on historic data associated with the usage of the plurality of features using NN model. Thereafter, the UE generates personalized text for each of the textblocks based on the plurality of features and usage pattern. The UE inserts personalized emoticons based on context of text-blocks and lingo of each word using user lingo database. Thus, the UE eliminates manual correction of texts and enhances the readability of message/text entered by user.

Abstract: A method for intelligently reading displayed contents by an electronic device is provided. The method includes obtaining a screen representation based on a plurality of contents displayed on a screen of the electronic device. The method includes extracting a plurality of insights comprising at least one of intent, importance, emotion, sound representation and information sequence of the plurality of contents from the plurality of contents based on the screen representation. The method includes generating audio emulating the extracted plurality of insights.

Abstract: Described is a system for intelligent re-tiering of backup data stored on a cloud-based object storage. More specifically, the system may re-tier objects such that the system retains the ability to efficiently perform a full restore of backup data even when incremental backups are performed to a cloud-based object storage. To provide such a capability, the system may maintain a specialized metadata database that stores information indicating the backup time for each backup, and a list of objects required to perform a full restore to each of the backup times. Accordingly, when using a threshold time (e.g. expiry) to select object candidates for re-tiering, the system may leverage the metadata database to ensure that objects that may still need to be referenced are not unnecessarily moved to a lower storage tier.

Abstract: A system including persistent storage for storing an object repository and a user data backup. The system also includes a manager that obtains fragmented user data associated with an object that is potentially duplicative of an object stored in the object repository; makes a determination that the object is duplicative based on a copy of the object stored in the object repository; in response to making the determination, adds the fragmented user data to a user data backup without adding the object to the backup; associates the fragmented user data in the user data backup with the copy of the object stored in the object repository; and stores the user data backup in the persistent storage.

Abstract: Example systems and methods provide input suggestions to a user to improve user experience on user devices. The input suggestions can be fill information from another app on device to the present app being used by user, information for performing a search (without the user having to copy-paste data or entering the data manually), responses to a message/notification received by the user, information/content/data to be shared between apps (without switching between apps), and emojis/GIFs that can be used by the user. The method includes analyzing one or more content of one or more screen displayed on device, generating at least one of a logical tree structure and a data mashup model of the one or more analyzed content for each screen, and providing a recommendation to a user. The recommendation can be a connected action or an input suggestion.