Adobe, Inc.

University of Surrey

John Philip Collomosse

Hailin Jin

Viswanathan Swaminathan

Zhe Lin

Jianming Zhang

Zhaowen Wang

Jason Wen Yong Kuen

Ritwik Sinha

Simon Jenni

Tong He

Alexander Black

Van Tu Bui

Brian Lynn Price

Baldo Antonio Faieta

Ning Xu

Saeid Motiian

Geoffrey Oxholm

Qing Liu

Alex Charles Filipkowski

Mai Long

Trisha Mittal

Md Mehrab Tanjim

Xin Yuan

Xiaoyang Li

Haotian Zhang

Yue Bai

Maksym Andriushchenko

Eric Nguyen

Yu Zeng

Shruti Agarwal

Embodiments of the present disclosure include obtaining a text prompt describing an element, layout information indicating a target region for the element, and a precision level corresponding to the element. Some embodiments generate a text feature pyramid based on the text prompt, the layout information, and the precision level, wherein the text feature pyramid comprises a plurality of text feature maps at a plurality of scales, respectively. Then, an image is generated based on the text feature pyramid. In some cases, the image includes an object corresponding to the element of the text prompt at the target region. Additionally, a shape of the object corresponds to a shape of the target region based on the precision level.

Semantic image synthesis

The present disclosure relates to systems, methods, and non-transitory computer readable media that utilize deep learning to identify regions of an image that have been editorially modified. For example, the image comparison system includes a deep image comparator model that compares a pair of images and localizes regions that have been editorially manipulated relative to an original or trusted image. More specifically, the deep image comparator model generates and surfaces visual indications of the location of such editorial changes on the modified image. The deep image comparator model is robust and ignores discrepancies due to benign image transformations that commonly occur during electronic image distribution. The image comparison system optionally includes an image retrieval model utilizes a visual search embedding that is robust to minor manipulations or benign modifications of images. The image retrieval model utilizes a visual search embedding for an image to robustly identify near duplicate images.

Identifying and localizing editorial changes to images utilizing deep learning

Techniques for video manipulation detection are described to detect one or more manipulations present in digital content such as a digital video. A detection system, for instance, receives a frame of a digital video that depicts at least one entity. Coordinates of the frame that correspond to a gaze location of the entity are determined, and the detection system determines whether the coordinates correspond to a portion of an object depicted in the frame to calculate a gaze confidence score. A manipulation score is generated that indicates whether the digital video has been manipulated based on the gaze confidence score. In some examples, the manipulation score is based on at least one additional confidence score.

Video manipulation detection

Techniques for latent space based steganographic image generation are described. A processing device, for instance, receives a digital image and a secret that includes a bit string. A pretrained encoder of an autoencoder generates an embedding of the digital image that includes latent code. A secret encoder is trained and utilized to generate an embedding of the secret to act as a latent offset to the latent code. The processing device leverages a pretrained decoder of the autoencoder to generate a steganographic image based on the embedding of the secret and the embedding of the digital image. The steganographic image includes the secret and is visually indiscernible from the digital image. Further, the processing device is configured to recover the secret from the steganographic image, such as by training and leveraging a secret decoder to extract the secret.

Latent space based steganographic image generation

Embodiments are disclosed for performing content authentication. A method of content authentication may include dividing a query video into a plurality of chunks. A feature vector may be generated, using a fingerprinting model, for each chunk from the plurality of chunks. Similar video chunks are identified from a trusted chunk database based on the feature vectors using a multi-chunk search policy. One or more original videos corresponding to the query video are then returned.

Scalable video fingerprinting for content authenticity

Embodiments of the present invention provide systems, methods, and computer storage media for guided visual search. A visual search query can be represented as a sketch sequence that includes ordering information of the constituent strokes in the sketch. The visual search query can be encoded into a structural search encoding in a common search space by a structural neural network. Indexed visual search results can be identified in the common search space and clustered in an auxiliary semantic space. Sketch suggestions can be identified from a plurality of indexed sketches in the common search space. A sketch suggestion can be identified for each semantic cluster of visual search results and presented with the cluster to guide a user towards relevant content through an iterative search process. Selecting a sketch suggestion as a target sketch can automatically transform the visual search query to the target sketch via adversarial images.

Guided content discovery in visual search

Systems and methods for image generation are described. Embodiments of the present disclosure receive a text phrase that describes a target image to be generated; generate text features based on the text phrase; retrieve a search image based on the text phrase; and generate the target image using an image generation network based on the text features and the search image.

Retrieval-based text-to-image generation with visual-semantic contrastive representation

The present disclosure relates to systems, methods, and non-transitory computer readable media that utilize a deep visual fingerprinting model with parameters learned from robust contrastive learning to identify matching digital images and image provenance information. For example, the disclosed systems utilize an efficient learning procedure that leverages training on bounded adversarial examples to more accurately identify digital images (including adversarial images) with a small computational overhead. To illustrate, the disclosed systems utilize a first objective function that iteratively identifies augmentations to increase contrastive loss. Moreover, the disclosed systems utilize a second objective function that iteratively learns parameters of a deep visual fingerprinting model to reduce the contrastive loss. With these learned parameters, the disclosed systems utilize the deep visual fingerprinting model to generate visual fingerprints for digital images, retrieve and match digital images, and provide digital image provenance information.

Adversarially robust visual fingerprinting and image provenance models

A visual search system facilitates retrieval of provenance information using a machine learning model to generate content fingerprints that are invariant to benign transformations while being sensitive to manipulations. The machine learning model is trained on a training image dataset that includes original images, benign transformed variants of the original images, and manipulated variants of the original images. A loss function is used to train the machine learning model to minimize distances in an embedding space between benign transformed variants and their corresponding original images and increase distances between the manipulated variants and their corresponding original images.

Robust content fingerprinting for image attribution

Digital content layout encoding techniques for search are described. In these techniques, a layout representation is generated (using machine learning automatically and without user intervention) that describes a layout of elements included within the digital content. In an implementation, the layout representation includes a description of both spatial and structural aspects of the elements in relation to each other. To do so, a two-pathway pipeline that is configured to model layout from both spatial and structural aspects using a spatial pathway, and a structural pathway, respectively. In one example, this is also performed through use of multi-level encoding and fusion to generate a layout representation.

Digital content layout encoding for search

The present disclosure relates to systems, methods, and non-transitory computer readable media that utilize deep learning to map query videos to known videos so as to identify a provenance of the query video or identify editorial manipulations of the query video relative to a known video. For example, the video comparison system includes a deep video comparator model that generates and compares visual and audio descriptors utilizing codewords and an inverse index. The deep video comparator model is robust and ignores discrepancies due to benign transformations that commonly occur during electronic video distribution.

Determining video provenance utilizing deep learning

Techniques and systems are provided for configuring neural networks to perform certain image manipulation operations. For instance, in response to obtaining an image for manipulation, an image manipulation system determines the fitness scores for a set of neural networks resulting from the processing of a noise map. Based on these fitness scores, the image manipulation system selects a subset of the set of neural networks for cross-breeding into a new generation of neural networks. The image manipulation system evaluates the performance of this new generation of neural networks and continues cross-breeding this neural networks until a fitness threshold is satisfied. From the final generation of neural networks, the image manipulation system selects a neural network that provides a desired output and uses the neural network to generate the manipulated image.

Image processing network search for deep image priors

Systems, methods, and non-transitory computer-readable media are disclosed for utilizing an encoder-decoder architecture to learn a volumetric 3D representation of an object using digital images of the object from multiple viewpoints to render novel views of the object. For instance, the disclosed systems can utilize patch-based image feature extraction to extract lifted feature representations from images corresponding to different viewpoints of an object. Furthermore, the disclosed systems can model view-dependent transformed feature representations using learned transformation kernels. In addition, the disclosed systems can recurrently and concurrently aggregate the transformed feature representations to generate a 3D voxel representation of the object. Furthermore, the disclosed systems can sample frustum features using the 3D voxel representation and transformation kernels. Then, the disclosed systems can utilize a patch-based neural rendering approach to render images from frustum feature patches to display a view of the object from various viewpoints.

Utilizing voxel feature transformations for view synthesis

The present disclosure relates to systems, methods, and non-transitory computer readable media for accurately and flexibly identifying digital images with similar style to a query digital image using fine-grain style determination via weakly supervised style extraction neural networks. For example, the disclosed systems can extract a style embedding from a query digital image using a style extraction neural network such as a novel two-branch autoencoder architecture or a weakly supervised discriminative neural network. The disclosed systems can generate a combined style embedding by combining complementary style embeddings from different style extraction neural networks. Moreover, the disclosed systems can search a repository of digital images to identify digital images with similar style to the query digital image. The disclosed systems can also learn parameters for one or more style extraction neural network through weakly supervised training without a specifically labeled style ontology for sample digital images.

Growing community of inventors

Ripley, United Kingdom

John Philip Collomosse