Honda Motor Co., Ltd.

Nvidia Corporation

Google Inc.

Honda Giken Kogyo Kabushiki Kaisha

University of Florida Research Foundation, Incorporated

Ming-Hsuan Yang

Jan Kautz

Jongwoo Lim

Ming-Yu Liu

Deqing Sun

David Ross

Jeffrey Ho

Varun Jampani

Ruei-Sung Lin

Behzad Dariush

Chia-Kai Liang

Yi-Ting Chen

Alessandro Bissacco

Yi-Hsuan Tsai

Nakul Agarwal

Takahiro Ohashi

Yichang Shih

Siargey Pisarchyk

Ryhor Karpiak

Wei-Chih Tu

Wei-Sheng Lai

Jiang Wang

Shengyang Dai

Ce Liu

Peyman Milanfar

Yijun Li

Rakesh Gupta

Shalini De Mello

Feng Yang

Sifei Liu

Jinwei Gu

Ananth Ranganathan

Jason Meltzer

Yinxiao Li

Qifei Wang

Pengchong Jin

Hang Su

Joshua Foster Slocum

Taihong Xiao

Kuang-Chih Lee

Rui Li

Gang Hua

Xiaojie Jin

Jinwei Yuan

Zhimin Fan

Example aspects of the present disclosure are directed to systems and methods which feature a machine-learned video super-resolution (VSR) model which has been trained using a bi-directional training approach. In particular, the present disclosure provides a compression-informed (e.g., compression-aware) super-resolution model that can perform well on real-world videos with different levels of compression. Specifically, example models described herein can include three modules to robustly restore the missing information caused by video compression. First, a bi-directional recurrent module can be used to reduce the accumulated warping error from the random locations of the intra-frame from compressed video frames. Second, a detail-aware flow estimation module can be added to enable recovery of high resolution (HR) flow from compressed low resolution (LR) frames. Finally, a Laplacian enhancement module can add high-frequency information to the warped HR frames washed out by video encoding.

Compression-informed video super-resolution

Apparatus and methods related to image processing are provided. A computing device can determine a first image area of an image, such as an image captured by a camera. The computing device can determine a warping mesh for the image with a first portion of the warping mesh associated with the first image area. The computing device can determine a cost function for the warping mesh by: determining first costs associated with the first portion of the warping mesh that include costs associated with face-related transformations of the first image area to correct geometric distortions. The computing device can determine an optimized mesh based on optimizing the cost function. The computing device can modify the first image area based on the optimized mesh.

Real time perspective correction on faces

A method includes obtaining a first plurality of feature vectors associated with a first image and a second plurality of feature vectors associated with a second image. The method also includes generating a plurality of transformed feature vectors by transforming each respective feature vector of the first plurality of feature vectors by a kernel matrix trained to define an elliptical inner product space. The method additionally includes generating a cost volume by determining, for each respective transformed feature vector of the plurality of transformed feature vectors, a plurality of inner products, wherein each respective inner product of the plurality of inner products is between the respective transformed feature vector and a corresponding candidate feature vector of a corresponding subset of the second plurality of feature vectors. The method further includes determining, based on the cost volume, a pixel correspondence between the first image and the second image.

Learnable cost volume for determining pixel correspondence

A method includes filtering a point cloud transformation of a 3D object to generate a 3D lattice and processing the 3D lattice through a series of bilateral convolution networks (BCL), each BCL in the series having a lower lattice feature scale than a preceding BCL in the series. The output of each BCL in the series is concatenated to generate an intermediate 3D lattice. Further filtering of the intermediate 3D lattice generates a first prediction of features of the 3D object.

Bilateral convolution layer network for processing point clouds

A system and method for providing unsupervised domain adaption for spatio-temporal action localization that includes receiving video data associated with a source domain and a target domain that are associated with a surrounding environment of a vehicle. The system and method also include analyzing the video data associated with the source domain and the target domain and determining a key frame of the source domain and a key frame of the target domain. The system and method additionally include completing an action localization model to model a temporal context of actions occurring within the key frame of the source domain and the key frame of the target domain and completing an action adaption model to localize individuals and their actions and to classify the actions based on the video data. The system and method further include combining losses to complete spatio-temporal action localization of individuals and actions.

System and method for providing unsupervised domain adaptation for spatio-temporal action localization

A method, computer readable medium, and system are disclosed for identifying residual video data. This data describes data that is lost during a compression of original video data. For example, the original video data may be compressed and then decompressed, and this result may be compared to the original video data to determine the residual video data. This residual video data is transformed into a smaller format by means of encoding, binarizing, and compressing, and is sent to a destination. At the destination, the residual video data is transformed back into its original format and is used during the decompression of the compressed original video data to improve a quality of the decompressed original video data.

Using residual video data resulting from a compression of original video data to improve a decompression of the original video data

Methods, and systems, including computer programs encoded on computer storage media for neural network architecture search. A method includes defining a neural network computational cell, the computational cell including a directed graph of nodes representing respective neural network latent representations and edges representing respective operations that transform a respective neural network latent representation; replacing each operation that transforms a respective neural network latent representation with a respective linear combination of candidate operations, where each candidate operation in a respective linear combination has a respective mixing weight that is parameterized by one or more computational cell hyper parameters; iteratively adjusting values of the computational cell hyper parameters and weights to optimize a validation loss function subject to computational resource constraints; and generating a neural network for performing a machine learning task using the defined computational cell and the adjusted values of the computational cell hyper parameters and weights.

Resource constrained neural network architecture search

A system and method for providing unsupervised domain adaption for spatio-temporal action localization that includes receiving video data associated with a surrounding environment of a vehicle. The system and method also include completing an action localization model to model a temporal context of actions occurring within the surrounding environment of the vehicle based on the video data and completing an action adaption model to localize individuals and their actions and to classify the actions based on the video data. The system and method further include combining losses from the action localization model and the action adaption model to complete spatio-temporal action localization of individuals and actions that occur within the surrounding environment of the vehicle.

Segmentation is the identification of separate objects within an image. An example is identification of a pedestrian passing in front of a car, where the pedestrian is a first object and the car is a second object. Superpixel segmentation is the identification of regions of pixels within an object that have similar properties. An example is identification of pixel regions having a similar color, such as different articles of clothing worn by the pedestrian and different components of the car. A pixel affinity neural network (PAN) model is trained to generate pixel affinity maps for superpixel segmentation. The pixel affinity map defines the similarity of two points in space. In an embodiment, the pixel affinity map indicates a horizontal affinity and vertical affinity for each pixel in the image. The pixel affinity map is processed to identify the superpixels.

Training a neural network to predict superpixels using segmentation-aware affinity loss

Photorealistic image stylization concerns transferring style of a reference photo to a content photo with the constraint that the stylized photo should remain photorealistic. Examples of styles include seasons (summer, winter, etc.), weather (sunny, rainy, foggy, etc.), lighting (daytime, nighttime, etc.). A photorealistic image stylization process includes a stylization step and a smoothing step. The stylization step transfers the style of the reference photo to the content photo. A photo style transfer neural network model receives a photorealistic content image and a photorealistic style image and generates an intermediate stylized photorealistic image that includes the content of the content image modified according to the style image. A smoothing function receives the intermediate stylized photorealistic image and pixel similarity data and generates the stylized photorealistic image, ensuring spatially consistent stylizations.

Photorealistic image stylization using a neural network model

A spatial linear propagation network (SLPN) system learns the affinity matrix for vision tasks. An affinity matrix is a generic matrix that defines the similarity of two points in space. The SLPN system is trained for a particular computer vision task and refines an input map (i.e., affinity matrix) that indicates pixels the share a particular property (e.g., color, object, texture, shape, etc.). Inputs to the SLPN system are input data (e.g., pixel values for an image) and the input map corresponding to the input data to be propagated. The input data is processed to produce task-specific affinity values (guidance data). The task-specific affinity values are applied to values in the input map, with at least two weighted values from each column contributing to a value in the refined map data for the adjacent column.

Learning affinity via a spatial propagation neural network

Segmentation is the identification of separate objects within an image. An example is identification of a pedestrian passing in front of a car, where the pedestrian is a first object and the car is a second object. Superpixel segmentation is the identification of regions of pixels within an object that have similar properties An example is identification of pixel regions having a similar color, such as different articles of clothing worn by the pedestrian and different components of the car. A pixel affinity neural network (PAN) model is trained to generate pixel affinity maps for superpixel segmentation. The pixel affinity map defines the similarity of two points in space. In an embodiment, the pixel affinity map indicates a horizontal affinity and vertical affinity for each pixel in the image. The pixel affinity map is processed to identify the superpixels.

A visual tracker tracks an object in a sequence of input images. A tracking module detects a location of the object based on a set of weighted blocks representing the object's shape. The tracking module then refines a segmentation of the object from the background image at the detected location. Based on the refined segmentation, the set of weighted blocks are updated. By adaptively encoding appearance and shape into the block configuration, the present invention is able to efficiently and accurately track an object even in the presence of rapid motion that causes large variations in appearance and shape of the object.

Online articulate object tracking with appearance and shape

Growing community of inventors

Sunnyvale, CA, United States of America

Ming-Hsuan Yang