Apple Inc.

Marco Zuliani

Elizabeth Caroline Cranfill

Francesco Rossi

Xiaojin Shi

Claus Molgaard

Todd Stephen Sachs

Frank Doepke

Bartlomiej W Rymkowski

Jianping Zhou

Vignesh Jagadeesh

George E Williams

Umesh Suresh Vaishampayan

Craig Michael Federighi

Xin Ru Wang

Andre Souza Dos Santos

Vinay Sharma

Kevin L Hunter

Gaurav Kapoor

Kit-Man Wan

Xiaohuan Corina Wang

Cecile Marie Foret

Grant R Paul

Behkish J Manzari

Nathan De Vries

Paul M Hubel

Steven D Baker

Ryan S Dixon

Rudolph Van Der Merwe

Srinivasan Venkatachary

Johnnie B Manzari

Marius Tico

Justin S Titi

James Nicholas Jones

Matthias Paulik

Neil Gareth Crane

Viktor Miladinov

Jeffrey A Brasket

Elliott B Harris

Vincent M Lane

Sebastien X Beysserie

Brandon J Corey

Xin Wang

Guillaume Borios

Alexey Marinichev

Vincent Y Wong

Etienne Belanger

Richard David Seely

Nicholas D Lupinetti

Brian Edward Walsh

Kellie L Albert

Brett M Keating

Ethan J Tira-Thompson

Geoffrey T Anneheim

Karl Hsu

Xiaoxing Li

Pulah J Shah

Benjamin Poulain

Omid Khalili

Michael Rousson

Yang Zhao

Matthias Dantone

Ron C Santos

Benjamin M Olson

Kayur Dushyant Patel

Atila Orhon

Thomas Deselaers

Chen Ye

Aya Siblini

Guillaume Tartavel

Ngoc Hong Pham

Albert Antony

Jeff Brasket

Brian P Keene

Francisco Alvaro Munoz

Rajen Chatterjee

Amittai Axelrod

Siyang Tang

Jennifer Pon Chen

Sebastien V Marineau-Mes

Guangyu Zhong

Adam Huff Bradford

Xishuo Liu

Mallika Priya Khullar

The subject technology provides a framework for evaluating activation functions of a neural network using lookup tables. In order to provide lookup table based activation functions with a desired precision within hardware constraints for the lookup tables, multiple lookup tables for each activation function can be provided. Each of the multiple lookup tables may correspond to a respective subrange of input values, within a full range of input values for the activation function.

Lookup table activation functions for neural networks

The subject technology provides receiving a neural network (NN) model to be executed on a target platform, the NN model including multiple layers that include operations and some of the operations being executable on multiple processors of the target platform. The subject technology further sorts the operations from the multiple layers in a particular order based at least in part on grouping the operations that are executable by a particular processor of the multiple processors. The subject technology determines, based at least in part on a cost of transferring the operations between the multiple processors, an assignment of one of the multiple processors for each of the sorted operations of each of the layers in a manner that minimizes a total cost of executing the operations. Further, for each layer of the NN model, the subject technology includes an annotation to indicate the processor assigned for each of the operations.

Compiling models for dedicated hardware

A method may include accessing a data processing architecture associated with a neural network to determine dependencies between intermediate data layers of the neural network; obtaining dimensions of the intermediate data layers in the neural network; calculating a minimum number of data storage portions for executing the neural network based on the dependencies; determining a memory allocation size for each respective data storage portion of the data storage portions based on the dimensions and dependencies; allocating memory on a storage device for each data storage portion in accordance with its respective determined memory allocation size.

Systems and methods of memory allocation for neural networks

A method includes receiving input data at a trained machine learning model that includes a common part and task-specific parts, receiving an execution instruction that identifies one or more processing tasks to be performed, processing the input data using the common part of the trained machine learning model to generate intermediate data, and processing the intermediate data using one or more of the task-specific parts of the trained machine learning model based on the execution instruction to generate one or more outputs.

Machine learning model with conditional execution of multiple processing tasks

The present disclosure generally relates to methods and user interfaces for managing visual content at a computer system. In some embodiments, methods and user interfaces for managing visual content in media are described. In some embodiments, methods and user interfaces for managing visual indicators for visual content in media are described. In some embodiments, methods and user interfaces for inserting visual content in media are described. In some embodiments, methods and user interfaces for identifying visual content in media are described. In some embodiments, methods and user interfaces for translating visual content in media are described.

User interfaces for managing visual content in media

Training a network for image processing with temporal consistency includes obtaining un-annotated frames from a video feed. A pretrained network is applied to the first frame of first frame set comprising a plurality of frames to obtain a first prediction, wherein the pretrained network is pretrained for a first image processing task. A current version of the pretrained network is applied to each frame of the first frame set to obtain a first prediction. A content loss term is determined, based on the first prediction and a current prediction for the frame, based on the current network. A temporal consistency loss term is also determined based on a determined consistency of pixels within each frame of the first frame set. The pretrained network may be refined based on the content loss term and the temporal term to obtain a refined network.

Method for improving temporal consistency of deep neural networks

Artistic styles extracted from source images may be applied to target images to generate stylized images and/or video sequences. The extracted artistic styles may be stored as a plurality of layers in one or more neural networks, which neural networks may be further optimized, e.g., via the fusion of various elements of the networks' architectures. The artistic style may be applied to the target images and/or video sequences using various optimization methods, such as the use of a first version of the neural network by a first processing device at a first resolution to generate one or more sets of parameters (e.g., scaling and/or biasing parameters), which parameters may then be mapped for use by a second version of the neural network by a second processing device at a second resolution. Analogous multi-processing device and/or multi-network solutions may also be applied to other complex image processing tasks for increased efficiency.

Enhanced image processing techniques for deep neural networks

Artistic styles extracted from one or more source images may be applied to one or more target images, e.g., in the form of stylized images and/or stylized video sequences. The extracted artistic style may be stored as a plurality of layers in a neural network, which neural network may be further optimized, e.g., via the fusion of various elements of the network's architectures. An optimized network architecture may be determined for each processing environment in which the network will be applied. The artistic style may be applied to the obtained images and/or video sequence of images using various optimization methods, such as the use of scalars to control the resolution of the unstylized and stylized images, temporal consistency constraints, as well as the use of dynamically adjustable or selectable versions of Deep Neural Networks (DNN) that are responsive to system performance parameters, such as available processing resources and thermal capacity.

Flexible resolution support for image and video style transfer

The invention relates to systems, methods, and computer readable media for responding to a user snapshot request by capturing anticipatory pre-snapshot image data as well as post-snapshot image data. The captured information may be used, depending upon the embodiment, to create archival image information and image presentation information that is both useful and pleasing to a user. The captured information may automatically be trimmed or edited to facilitate creating an enhanced image, such as a moving still image. Varying embodiments of the invention offer techniques for trimming and editing based upon the following: exposure, brightness, focus, white balance, detected motion of the camera, substantive image analysis, detected sound, image metadata, and/or any combination of the foregoing.

Method and apparatus for finding and using video portions that are relevant to adjacent still images

Real-time selection of DNN style transfer networks from DNN sets

Systems and methods for improving automatic selection of keeper images from a commonly captured set of images are described. A combination of image type identification and image quality metrics may be used to identify one or more images in the set as keeper images. Image type identification may be used to categorize the captured images into, for example, three or more categories. The categories may include portrait, action, or 'other.' Depending on the category identified, the images may be analyzed differently to identify keeper images. For portrait images, an operation may be used to identify the best set of faces. For action images, the set may be divided into sections such that keeper images selected from each section tell the story of the action. For the 'other' category, the images may be analyzed such that those having higher quality metrics for an identified region of interest are selected.

Automated selection of keeper images from a burst photo captured set

Managing a secure session includes detecting a login event at an electronic device using a first login method to initiate a secure session, capturing an initial image at a same time as the login event, capturing initial sensor data at the same time as the login event, monitoring for changes in the sensor data during the secure session, maintaining the secure session based on the initial sensor data and the monitored changes from the initial sensor data, and during the secure session, permitting access to the electronic device using reidentification.

System and method for person reidentification

Methods and systems for an image construction component capable of generating pixel information for certain regions of an image based on other, existing regions of the image. For example, the image construction component may identify a target block of pixels for which to generate pixel information and then use pixel information for pixels surrounding the target block of pixels in order to identify similar image information within pixels in another part of the image. These identified pixels may then be used in defining the pixel information of the target block of pixels and also used in blending the target block of pixels with the defined pixels surrounding the target block of pixels.

Reconstruction of missing regions of images

Techniques are disclosed herein for applying an artistic style extracted from one or more source images, e.g., paintings, to one or more target images. The extracted artistic style may then be stored as a plurality of layers in a neural network. In some embodiments, two or more stylized target images may be combined and stored as a stylized video sequence. The artistic style may be applied to the target images in the stylized video sequence using various optimization methods and/or pixel- and feature-based regularization techniques in a way that prevents excessive content pixel fluctuations between images and preserves smoothness in the assembled stylized video sequence. In other embodiments, a user may be able to semantically annotate locations of undesired artifacts in a target image, as well as portion(s) of a source image from which a style may be extracted and used to replace the undesired artifacts in the target image.

Style transfer-based image content correction

Growing community of inventors

Los Gatos, CA, United States of America

Marco Zuliani