公开(公告)号：US11250548B2

公开(公告)日：2022-02-15

申请号：US16791939

申请日：2020-02-14

Applicant: Adobe Inc.

Inventor： Zhe Lin ,  Xin Lu ,  Xiaohui Shen ,  Jimei Yang ,  Jiahui Yu

IPC: G06T5/00 ,  G06N3/04 ,  G06N3/08 ,  G06T5/20

Abstract: Digital image completion using deep learning is described. Initially, a digital image having at least one hole is received. This holey digital image is provided as input to an image completer formed with a framework that combines generative and discriminative neural networks based on learning architecture of the generative adversarial networks. From the holey digital image, the generative neural network generates a filled digital image having hole-filling content in place of holes. The discriminative neural networks detect whether the filled digital image and the hole-filling digital content correspond to or include computer-generated content or are photo-realistic. The generating and detecting are iteratively continued until the discriminative neural networks fail to detect computer-generated content for the filled digital image and hole-filling content or until detection surpasses a threshold difficulty. Responsive to this, the image completer outputs the filled digital image with hole-filling content in place of the holey digital image's holes.

公开(公告)号：US10593043B2

公开(公告)日：2020-03-17

申请号：US15950087

申请日：2018-04-10

Applicant: Adobe Inc.

Inventor： Zhe Lin ,  Yibing Song ,  Xin Lu ,  Xiaohui Shen ,  Jimei Yang

IPC: G06T7/13 ,  G06K9/66 ,  G06K9/38 ,  G06K9/00 ,  G06K9/46 ,  G06T7/12 ,  G06N3/04 ,  G06N7/00 ,  G06N3/08 ,  G06T7/11

Abstract: Systems and methods are disclosed for segmenting a digital image to identify an object portrayed in the digital image from background pixels in the digital image. In particular, in one or more embodiments, the disclosed systems and methods use a first neural network and a second neural network to generate image information used to generate a segmentation mask that corresponds to the object portrayed in the digital image. Specifically, in one or more embodiments, the disclosed systems and methods optimize a fit between a mask boundary of the segmentation mask to edges of the object portrayed in the digital image to accurately segment the object within the digital image.

公开(公告)号：US10521705B2

公开(公告)日：2019-12-31

申请号：US15812695

申请日：2017-11-14

Applicant: Adobe Inc.

Inventor： Xin Lu ,  Zejun Huang ,  Jen-Chan Jeff Chien

IPC: G06K9/72 ,  G06K9/62 ,  G06K9/00 ,  H04N1/21 ,  G06F16/532 ,  G06F16/583

Abstract: The present disclosure is directed toward systems, methods, and non-transitory computer readable media that automatically select an image from a plurality of images based on the multi-context aware rating of the image. In particular, systems described herein can generate a plurality of probability context scores for an image. Moreover, the disclosed systems can generate a plurality of context-specific scores for an image. Utilizing each of the probability context scores and each of the corresponding context-specific scores for an image, the disclosed systems can generate a multi-context aware rating for the image. Thereafter, the disclosed systems can select an image from the plurality of images with the highest multi-context aware rating for delivery to the user. The disclosed system can utilize one or more neural networks to both generate the probability context scores for an image and to generate the context-specific scores for an image.

公开(公告)号：US20190295223A1

公开(公告)日：2019-09-26

申请号：US15928706

申请日：2018-03-22

Applicant: ADOBE INC.

Inventor： Xiaohui Shen ,  Zhe Lin ,  Xin Lu ,  Sarah Aye Kong ,  I-Ming Pao ,  Yingcong Chen

IPC: G06T5/00 ,  G06N3/08 ,  G06K9/62 ,  G06K9/66

Abstract: Methods and systems are provided for generating enhanced image. A neural network system is trained where the training includes training a first neural network that generates enhanced images conditioned on content of an image undergoing enhancement and training a second neural network that designates realism of the enhanced images generated by the first neural network. The neural network system is trained by determine loss and accordingly adjusting the appropriate neural network(s). The trained neural network system is used to generate an enhanced aesthetic image from a selected image where the output enhanced aesthetic image has increased aesthetics when compared to the selected image.

公开(公告)号：US20190205625A1

公开(公告)日：2019-07-04

申请号：US15856271

申请日：2017-12-28

Applicant: Adobe Inc.

Inventor： Yu Luo ,  Xin Lu ,  Jen-Chan Jeff Chien

IPC: G06K9/00 ,  G06K9/66 ,  G06N3/08

Abstract: Techniques are disclosed for a facial expression classification. In an embodiment, a multi-class classifier is trained using labelled training images, each training image including a facial expression. The trained classifier is then used to predict expressions for unlabelled video frames, whereby each frame is effectively labelled with a predicted expression. In addition, each predicted expression can be associated with a confidence score. Anchor frames can then be identified in the labelled video frames, based on the confidence scores of those frames (anchor frames are frames having a confidence score above an established threshold). Then, for each labelled video frame between two anchor frames, the predicted expression is refined or otherwise updated using interpolation, thereby providing a set of video frames having calibrated expression labels. These calibrated labelled video frames can then be used to further train the previously trained facial expression classifier, thereby providing a supplementally trained facial expression classifier.

公开(公告)号：US10290085B2

公开(公告)日：2019-05-14

申请号：US15379337

申请日：2016-12-14

Applicant: ADOBE INC.

Inventor： Zhe Lin ,  Xin Lu ,  Chao Yang

IPC: G06T5/00 ,  G06K9/62

Abstract: Image hole filling that account for global structure and local texture. One exemplary technique involves using both a content neural network and a texture neural network. The content neural network is trained to encode image features based on non-hole image portions and decode the image features to fill holes. The texture neural network is trained to extract image patch features that represent texture. The exemplary technique receives an input image that has a hole and uses the two neural networks to fill the hole and provide a result image. This is accomplished by selecting pixel values for the hole based on a content constraint that uses the content neural network to account for global structure and a texture constraint that uses the texture neural network to account for local texture. For example, the pixel values can be selected by optimizing a loss function that implements the constraints.