Autodesk, Inc.

Monsters Aliens Robots Zombies Inc.

Thomas Davies

Michael Haley

Ara Danielyan

Morgan Fabian

Ali Mahdavi-Amiri

Jonathan Bronfman

Matthew Panousis

Aditya Sanghi

Lon Molnar

Debjoy Chowdhury

Anton Skourides

Ishrat Badami

Paul Birulin

Hooman Shayani

Nigel Jed Wesley Morris

Daniel Cohen-Or

Joseph George Lambourne

Ran Zhang

Pradeep Kumar Jayaraman

Alexander Ray Carlson

Divya Sivasankaran

Anup Bhalchandra Walvekar

Tarkeshwar Kumar Shah

Ahmed Moustafa Abdelhafez Hashem

A computer implemented system is described in various embodiments herein, the system includes a processor, a memory coupled to the processor and storing processor-executable instructions that, when executed, configure the processor to: receive an initial video data object V comprising a plurality of image frame data objects I; receive a set of time-synchronized viseme parameters corresponding to target audio data object A'; and process the initial video data object V and the set of time-synchronized viseme parameters using a machine learning network to generate an output video data object V′, wherein initial mouth regions in the initial video data object V have been replaced with replacement mouth regions generated based on the set of time-synchronized viseme parameters.

Systems and methods for improved lip dubbing

An image processing system comprising: a computer readable medium and at least one processor configured to provide a machine learning architecture for image processing. In particular, modified keyframes are used for training the machine learning architecture. The modifications are then automatically propagated to remaining frames requiring modification through interpolation or extrapolation through processing remaining frames through the trained machine learning architecture. The generated modified frames or frame portions can then be inserted into an original video to generate a modified video where the modifications have been propagated. Example usages include automatic computational approaches for aging/de-aging and addition/removal of tattoos or other visual effects.

Method and system for image processing

In various embodiments, a parameter domain graph application generates UV-net representations of 3D CAD objects for machine learning models. In operation, the parameter domain graph application generates a graph based on a B-rep of a 3D CAD object. The parameter domain graph application discretizes a parameter domain of a parametric surface associated with the B-rep into a 2D grid. The parameter domain graph application computes at least one feature at a grid point included in the 2D grid based on the parametric surface to generate a 2D UV-grid. Based on the graph and the 2D UV-grid, the parameter domain graph application generates a UV-net representation of the 3D CAD object. Advantageously, generating UV-net representations of 3D CAD objects that are represented using B-reps enables the 3D CAD objects to be processed efficiently using neural networks.

Techniques for generating UV-net representations of 3D CAD objects for machine learning models

One embodiment of the present invention sets forth a technique for adding dimensions to a target drawing. The technique includes generating a first set of node embeddings for a first set of nodes included in a target graph that represents the target drawing. The technique also includes receiving a second set of node embeddings for a second set of nodes included in a source graph that represents a source drawing, where one or more nodes included in the second set of nodes are associated with one or more source dimensions included in the source drawing. The technique further includes generating a set of mappings between the first and second sets of nodes based similarities between the first set of node embeddings and the second set of node embeddings, and automatically placing the one or more source dimensions within the target drawing based on the set of mappings.

Graph alignment techniques for dimensioning drawings automatically

In various embodiments, a training application generates a trained encoder that automatically generates shape embeddings having a first size and representing three-dimensional (3D) geometry shapes. First, the training application generates a different view activation for each of multiple views associated with a first 3D geometry based on a first convolutional neural network (CNN) block. The training application then aggregates the view activations to generate a tiled activation. Subsequently, the training application generates a first shape embedding having the first size based on the tiled activation and a second CNN block. The training application then generates multiple re-constructed views based on the first shape embedding. The training application performs training operation(s) on at least one of the first CNN block and the second CNN block based on the views and the re-constructed views to generate the trained encoder.

Shaped-based techniques for exploring design spaces

An image processing system comprising: a computer readable medium and at least one processor configured to provide a machine learning architecture for image processing. In particular, keyframes are selected for modification by a visual artist, and the modifications are used for training the machine learning architecture. The modifications are then automatically propagated to remaining frames requiring modification through interpolation or extrapolation through processing remaining frames through the trained machine learning architecture. The generated modified frames or frame portions can then be inserted into an original video to generate a modified video where the modifications have been propagated. Example usages include automatic computational approaches for aging/de-aging and addition/removal of tattoos or other visual effects.

In various embodiments, a training application generates a trained encoder that automatically generates shape embeddings having a first size and representing three-dimensional (3D) geometry shapes, First, the training application generates a different view activation for each of multiple views associated with a first 3D geometry based on a first convolutional neural network (CNN) block. The training application then aggregates the view activations to generate a tiled activation. Subsequently, the training application generates a first shape embedding having the first size based on the tiled activation and a second CNN block. The training application then generates multiple re-constructed views based on the first shape embedding. The training application performs training operation(s) on at least one of the first CNN block and the second CNN block based on the views and the re-constructed views to generate the trained encoder.

Growing community of inventors

Toronto, Canada

Thomas Davies