Adobe, Inc.

Arneh Jain

Mausoom Sarkar

Salil Taneja

Balaji Krishnamurthy

Milan Aggarwal

Surgan Jandial

Shripad Vilasrao Deshmukh

Abhinav Java

Puneet Mangla

Gaurav Ahuja

Silky Singh

Systems and methods for data augmentation are provided. One aspect of the systems and methods include receiving an image that is misclassified by a classification network; computing an augmentation image based on the image using an augmentation network; and generating an augmented image by combining the image and the augmentation image, wherein the augmented image is correctly classified by the classification network.

Systems and methods for data augmentation

A corrective noise system receives an electronic version of a fillable form generated by a segmentation network and receives a correction to a segmentation error in the electronic version of the fillable form. The corrective noise system is trained to generate noise that represents the correction and superimpose the noise on the fillable form. The corrective noise system is further trained to identify regions in a corpus of forms that are semantically similar to a region that was subject to the correction. The generated noise is propagated to the semantically similar regions in the corpus of forms and the noisy corpus of forms is provided as input to the segmentation network. The noise causes the segmentation network to accurately identify fillable regions in the corpus of forms and output a segmented version of the corpus of forms having improved fidelity without retraining or otherwise modifying the segmentation network.

Personalized form error correction propagation

Form structure similarity detection techniques are described. A content processing system, for instance, receives a query snippet that depicts a query form structure. The content processing system generates a query layout string that includes semantic indicators to represent the query form structure and generates candidate layout strings that represent form structures from a target document. The content processing system calculates similarity scores between the query layout string and the candidate layout strings. Based on the similarity scores, the content processing system generates a target snippet for display that depicts a form structure that is structurally similar to the query form structure. The content processing system is further operable to generate a training dataset that includes image pairs of snippets depicting form structures that are structurally similar. The content processing system utilizes the training dataset to train a machine learning model to perform form structure similarity matching.

Form structure similarity detection

In implementations of systems for assistive digital form authoring, a computing device implements an authoring system to receive input data describing a search input associated with a digital form. The authoring system generates an input embedding vector that represents the search input in a latent space using a machine learning model trained on training data to generate embedding vectors in the latent space. A candidate embedding vector included in a group of candidate embedding vectors is identified based on a distance between the input embedding vector and the candidate embedding vector in the latent space. The authoring system generates an indication of a search output associated with the digital form for display in a user interface based on the candidate embedding vector.

Assistive digital form authoring

Techniques for document segmentation. In an example, a document processing application segments an electronic document image into strips. A first strip overlaps a second strip. The application generates a first mask indicating one or more elements and element types in the first strip by applying a predictive model network to image content in the first strip and a prior mask generated from image content of the first strip. The application generates a second mask indicating one or more elements and element types in the second strip by applying the predictive model network to image content in the second strip and the first mask. The application computes, from a combined mask derived from the first mask and the second mask, an output electronic document that identifies elements in the electronic document and the respective element types.

Performing electronic document segmentation using deep neural networks

Electronic document segmentation using deep learning

A corpus of documents is processed using, for example, algorithms including deep learning and deep neural networks ('DNN'), to extract fragments across the corpus of documents. The extracted fragments can then be edited individually and referenced by a plurality of documents so that changes to the fragments are reflected universally across a corpus of documents efficiently. In one example case, a computer-implemented method is provided for extracting fragments in a digital document. The method includes indexing said document to generate a document element ID sequence; processing said document element ID sequence to generate at least one fragment candidate; processing said at least one fragment candidate to generate at least one respective fragment; and utilizing said at least one fragment to perform a reconstruction of said document.

Method to identify and extract fragments among large collections of digital documents using repeatability and semantic information

Portion-level digital rights management (DRM) in digital content is described. In one or more embodiments, a selection of a portion of the digital content is received at a computing device. Then, a policy is assigned to the selected portion by adding a markup element with an identifier to the selected portion. Based on the assigned policy, the selected portion is encrypted without encrypting another portion of the digital content. Subsequently, access to the selected portion is controlled based on the policy independently of the other portion. In this way, different portions of a single document can be protected with different policies. Different users may then have access to different portions of the digital content based on their user ID being associated with a particular policy, which improves security and management of distributable digital content.

Portion-level digital rights management in digital content

Digital form optimization techniques are disclosed which reduce the number of segments in empty digital forms that consumers of the empty digital forms need to complete. In some examples, a method may include determining potentially linkable segments in an empty digital form, determining a type of link to create for a potentially linkable segment of the potentially linkable segments, and providing a recommendation to create the determined type of link for the potentially linkable segment. The method may also include creating the determined type of link for the potentially linkable segment in response to a determination of an acceptance of the recommendation.

Growing community of inventors

Kerala, India

Arneh Jain