Adobe, Inc.

Lemon Inc.

Google Inc.

Peng Wang

Zhe Lin

Scott David Cohen

Xiaohui Shen

Radomir Mech

Brian Lynn Price

Jonathan Eisenmann

Bryan Russell

Xiaochen Lian

Hartwig Adam

Linjie Yang

Shupeng Zhang

Gerhard Florian Schroff

Liang-Chieh Chen

Xiaojie Jin

Shuo Cheng

Xueqing Deng

Jie Liao

Boheng Qiu

Georgios Papandreou

Xinyu Li

Glenn Engstrand

Alexander Hermans

Jian Sun

Angtian Wang

Heng Wang

Xianhang Li

Mingyu Ding

Systems and methods for performing temporal progressive learning for video processing are provided herein. Some examples include receiving a video that includes a plurality of frames, extracting a first subset of frames from the plurality of frames, and inputting the first subset of frames into a model that includes an encoder and a decoder. The examples further include comparing a first output of the model to the first subset of frames and updating the encoder, thereby training the encoder, and extracting a second subset of frames from the plurality of frames. The second subset of frames includes a number of frames that is larger than a number of frames in the first subset of frames. The examples further include inputting the second subset of frames into the model, comparing a second output of the model to the second subset of frames and updating the encoder, thereby further training the encoder.

Efficient video processing via temporal progressive learning

The techniques for dual-level semantic segmentation are provided. Data may be input to a first segmentation network. The input data comprises an image and label information associated with the image. The image may be captured at nighttime and may comprise a plurality of regions. At least one region among the plurality of regions may be determined based at least in part on output of the first segmentation network. The at least one region of the image may be cropped. The cropped at least one region may be input to a second segmentation network. A final output may be produced based on the output of the first segmentation network and output of the second segmentation network.

Dual-level model for segmentation

The present disclosure describes techniques for determining a bounding box. An image may be received. An X-frame, a Y-frame, and a normal frame may be estimated based on the image using a first neural network. At least one planar region may be detected from the image using a second neural network. A vanishing point detection may be performed on each of the at least one planar region. Output of the first neural network may be fused with results of the vanishing point detection. A depth value of each pixel in at least one plane corresponding to the at least one planar region may be determined based at least in part on a result of the fusing. A location of a bounding box may be determined based at least in part on the depth value of each pixel in the at least one plane.

Box detection for object attachment

The present disclosure describes techniques of rendering images using explicit object representation via rays tracing volume density aggregation. The techniques comprise reconstructing an object into a plurality of Gaussian ellipsoids; determining a volume density of each of the plurality of Gaussian ellipsoids along each of a plurality of viewing rays; determining a weight of each of the plurality of Gaussian ellipsoids based on the volume density; and synthesizing an image of the object using the determined weight on each pixel of the image to interpolate attributes of each of the plurality of Gaussian ellipsoids.

Renderer using explicit object represention via rays tracing volume density aggregation

Systems and methods for obtaining attention features are described. Some examples may include: receiving, at a projector of a transformer, a plurality of tokens associated with image features of a first dimensional space; generating, at the projector of the transformer, projected features by concatenating the plurality of tokens with a positional map, the projected features having a second dimensional space that is less than the first dimensional space; receiving, at an encoder of the transformer, the projected features and generating encoded representations of the projected features using self-attention; decoding, at a decoder of the transformer, the encoded representations and obtaining a decoded output; and projecting the decoded output to the first dimensional space and adding the image features of the first dimensional space to obtain attention features associated with the image features.

Lightweight transformer for high resolution images

A method and apparatus for converting a picture into a video, and a device and a storage medium. The method for converting a picture into a video includes: partitioning an original picture to obtain a foreground region and a background region; performing an iterative transformation for visual depth on the background region, and storing an image obtained through each transformation as a picture frame to obtain multiple frames of images; and splicing the multiple frames of images to obtain a target video. The iterative transformation includes at least two transformations for visual depth.

Method and apparatus for converting picture into video, and device and storage medium

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for instance segmentation. In one aspect, a system generates: (i) data identifying one or more regions of the image, wherein an object is depicted in each region, (ii) for each region, a predicted type of object that is depicted in the region, and (iii) feature channels comprising a plurality of semantic channels and one or more direction channels. The system generates a region descriptor for each of the one or more regions, and provides the region descriptor for each of the one or more regions to a segmentation neural network that processes a region descriptor for a region to generate a predicted segmentation of the predicted type of object depicted in the region.

Instance segmentation

Techniques for achieving consistency across data layers associated with multiple microservices. An always strictly consistent microservice coordination system provides immediate consistency for data stored across multiple microservices. A strictly successful eventually consistent multiple microservice coordination system provides eventual consistency for data stored across multiple microservices.

Reducing frontend complexity for multiple microservices with consistent updates

Techniques for planar region-guided estimates of 3D geometry of objects depicted in a single 2D image. The techniques estimate regions of an image that are part of planar regions (i.e., flat surfaces) and use those planar region estimates to estimate the 3D geometry of the objects in the image. The planar regions and resulting 3D geometry are estimated using only a single 2D image of the objects. Training data from images of other objects is used to train a CNN with a model that is then used to make planar region estimates using a single 2D image. The planar region estimates, in one example, are based on estimates of planarity (surface plane information) and estimates of edges (depth discontinuities and edges between surface planes) that are estimated using models trained using images of other scenes.

Planar region guided 3D geometry estimation from a single image

In techniques of combined composition and change-based models for image cropping, a composition application is implemented to apply one or more image composition modules of a learned composition model to evaluate multiple composition regions of an image. The learned composition model can determine one or more cropped images from the image based on the applied image composition modules, and evaluate a composition of the cropped images and a validity of change from the image to the cropped images. The image composition modules of the learned composition model include a salient regions module that iteratively determines salient image regions of the image, and include a foreground detection module that determines foreground regions of the image. The image composition modules also include one or more imaging models that reduce a number of the composition regions of the image to facilitate determining the cropped images from the image.

Combined composition and change-based models for image cropping

Joint depth estimation and semantic labeling techniques usable for processing of a single image are described. In one or more implementations, global semantic and depth layouts are estimated of a scene of the image through machine learning by the one or more computing devices. Local semantic and depth layouts are also estimated for respective ones of a plurality of segments of the scene of the image through machine learning by the one or more computing devices. The estimated global semantic and depth layouts are merged with the local semantic and depth layouts by the one or more computing devices to semantically label and assign a depth value to individual pixels in the image.

Joint depth estimation and semantic segmentation from a single image

In techniques for iterative saliency map estimation, a salient regions module applies a saliency estimation technique to compute a saliency map of an image that includes image regions. A salient image region of the image is determined from the saliency map, and an image region that corresponds to the salient image region is removed from the image. The salient regions module then iteratively determines subsequent salient image regions of the image utilizing the saliency estimation technique to recompute the saliency map of the image with the image region removed, and removes the image regions that correspond to the subsequent salient image regions from the image. The salient image regions of the image are iteratively determined until no salient image regions are detected in the image, and a salient features map is generated that includes each of the salient image regions determined iteratively and combined to generate the final saliency map.

Iterative saliency map estimation

In techniques for image foreground detection, a foreground detection module is implemented to generate varying levels of saliency thresholds from a saliency map of an image that includes foreground regions. The saliency thresholds can be generated based on an adaptive thresholding technique applied to the saliency map of the image and/or based on multi-level segmentation of the saliency map. The foreground detection module applies one or more constraints that distinguish the foreground regions in the image, and detects the foreground regions of the image based on the saliency thresholds and the constraints. Additionally, different ones of the constraints can be applied to detect different ones of the foreground regions, as well as to detect multi-level foreground regions based on the saliency thresholds and the constraints.

Growing community of inventors

Los Angeles, CA, United States of America

Peng Wang