Adobe, Inc.

Mehmet Yumer

Vladimir Kim

Kalyan Krishna Sunkavalli

Elya Shechtman

Radomir Mech

Aaron Phillip Hertzmann

Duygu Ceylan Aksit

Emiliano Gambaretto

Xiaohui Shen

Oliver Wang

Jimei Yang

Paul John Asente

Nathan Aaron Carr

Sunil Sharadchandra Hadap

Jonathan Eisenmann

Zhixin Shu

Vojtech Krs

Chen-Hsuan Lin

Eunbyung Park

Xiao Yang

Gavin Stuart Peter Miller

Niloy Jyoti Mitra

Yangtuanfeng Wang

Qingnan Zhou

Minhyuk Sung

Zimo Li

Marc-André Gardner

Li Yi

Guilin Liu

Haibin Huang

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.

Utilizing soft classifications to select input parameters for segmentation algorithms and identify segments of three-dimensional digital models

Systems and techniques for estimating illumination from a single image are provided. An example system may include a neural network. The neural network may include an encoder that is configured to encode an input image into an intermediate representation. The neural network may also include an intensity decoder that is configured to decode the intermediate representation into an output light intensity map. An example intensity decoder is generated by a multi-phase training process that includes a first phase to train a light mask decoder using a set of low dynamic range images and a second phase to adjust parameters of the light mask decoder using a set of high dynamic range image to generate the intensity decoder.

High dynamic range illumination estimation

The present disclosure relates to an image composite system that employs a generative adversarial network to generate realistic composite images. For example, in one or more embodiments, the image composite system trains a geometric prediction neural network using an adversarial discrimination neural network to learn warp parameters that provide correct geometric alignment of foreground objects with respect to a background image. Once trained, the determined warp parameters provide realistic geometric corrections to foreground objects such that the warped foreground objects appear to blend into background images naturally when composited together.

Image composites using a generative neural network

Mixing segmentation algorithms utilizing soft classifications to identify segments of three-dimensional digital models

Disclosed systems and methods categorize text regions of an electronic document into document object types based on a combination of semantic information and appearance information from the electronic document. A page segmentation application executing on a computing device provides a textual feature representation and a visual feature representation to a neural network. The application identifies a correspondence between a location of the set of pixels in the electronic document and a location of a particular document object type in an output page segmentation. The application further outputs a classification of the set of pixels as being the particular document object type based on the identified correspondence.

Semantic page segmentation of vector graphics documents

Embodiments are directed towards providing a target view, from a target viewpoint, of a 3D object. A source image, from a source viewpoint and including a common portion of the object, is encoded in 2D data. An intermediate image that includes an intermediate view of the object is generated based on the data. The intermediate view is from the target viewpoint and includes the common portion of the object and a disoccluded portion of the object not visible in the source image. The intermediate image includes a common region and a disoccluded region corresponding to the disoccluded portion of the object. The disoccluded region is updated to include a visual representation of a prediction of the disoccluded portion of the object. The prediction is based on a trained image completion model. The target view is based on the common region and the updated disoccluded region of the intermediate image.

Generating novel views of a three-dimensional object based on a single two-dimensional image

Certain embodiments detect human-object interactions in image content. For example, human-object interaction metadata is applied to an input image, thereby identifying contact between a part of a depicted human and a part of a depicted object. Applying the human-object interaction metadata involves computing a joint-location heat map by applying a pose estimation subnet to the input image and a contact-point heat map by applying an object contact subnet to the to the input image. The human-object interaction metadata is generated by applying an interaction-detection subnet to the joint-location heat map and the contact-point heat map. The interaction-detection subnet is trained to identify an interaction based on joint-object contact pairs, where a joint-object contact pair includes a relationship between a human joint location and a contact point. An image search system or other computing system is provided with access to the input image having the human-object interaction metadata.

Interaction detection model for identifying human-object interactions in image content

Various embodiments enable curves to be drawn around 3-D objects by intelligently determining or inferring how the curve flows in the space around the outside of the 3-D object. The various embodiments enable such curves to be drawn without having to constantly rotate the 3-D object. In at least some embodiments, curve flow is inferred by employing a vertex position discovery process, a path discovery process, and a final curve construction process.

Drawing curves in space guided by 3-D objects

A digital medium environment is described to dynamically modify or extend an existing path in a user interface. An un-parameterized input is received that is originated by user interaction with a user interface to specify a path to be drawn. A parameterized path is fit as a mathematical ordering representation of the path to be drawn as specified by the un-parametrized input. A determination is made as to whether the parameterized path is to extend or modify the existing path in the user interface. The existing path is modified or extended in the user interface using the parameterized path in response to the determining that the parameterized path is to modify or extend the existing path.

Dynamic path modification and extension

A digital medium environment is described to generate a three dimensional facial expression from a blend shape and a facial expression source. A semantic type is detected that defines a facial expression of the blend shape. Transfer intensities are assigned based on the detected semantic type to the blend shape and the facial expression source, respectively, for individual portions of the three dimensional facial expression, the transfer intensities specifying weights given to the blend shape and the facial expression source, respectively, for the individual portions of the three dimensional facial expression. The three dimensional facial expression is generated from the blend shape and the facial expression source based on the assigned transfer intensities.

Three dimensional facial expression generation

Image composites using a generative adversarial neural network

Three-dimensional segmentation of digital models utilizing soft classification geometric tuning

Techniques are disclosed for performing manipulation of facial images using an artificial neural network. A facial rendering and generation network and method learns one or more compact, meaningful manifolds of facial appearance, by disentanglement of a facial image into intrinsic facial properties, and enables facial edits by traversing paths of such manifold(s). The facial rendering and generation network is able to handle a much wider range of manipulations including changes to, for example, viewpoint, lighting, expression, and even higher-level attributes like facial hair and age&#x2014;aspects that cannot be represented using previous models.

Neural face editing with intrinsic image disentangling

Methods and systems are provided for using a single image of an indoor scene to estimate illumination of an environment that includes the portion captured in the image. A neural network system may be trained to estimate illumination by generating recovery light masks indicating a probability of each pixel within the larger environment being a light source. Additionally, low-frequency RGB images may be generated that indicating low-frequency information for the environment. The neural network system may be trained using training input images that are extracted from known panoramic images. Once trained, the neural network system infers plausible illumination information from a single image to realistically illumination images and objects being manipulated in graphics applications, such as with image compositing, modeling, and reconstruction.

Illumination estimation from a single image

Growing community of inventors

San Jose, CA, United States of America

Mehmet Yumer