Systems and methods include obtaining a Uniform Resource Locator (URL) for a site on the Internet; analyzing the URL with a Machine Learning (ML) model to determine whether or not the site is suspicious for phishing; responsive to the URL being suspicious for phishing, loading the site to determine whether or not an associated brand of the site is legitimate or not; and, responsive to the site being not legitimate for the brand, categorizing the URL for phishing and performing a first action based thereon. The systems and methods can further include, responsive to the URL being not suspicious for phishing or the site being legitimate for the brand, categorizing the URL as legitimate and performing a second action based thereon.

Techniques are provided for assessing uniqueness of information using string-based collection frequency techniques. One method comprises obtaining multiple collections of documents from at least one data source; determining a collection frequency for a given character string based on a number of the collections comprising the given character string relative to a total number of the collections; assigning a uniqueness rating to the given character string based at least in part on a comparison of the collection frequency of the given character string to a collection frequency of one or more additional character strings in one or more of the plurality of collections; and performing an automated action using the given character string based on the assigned uniqueness rating. The automated action may comprise protecting the given character string and/or identifying the given character string as important information satisfying one or more importance criteria.

A system automatically identifies objects in an environment based on a walkthrough video and an annotated floorplan of the environment. The annotated floorplan indicates locations and types of objects that are expected to be in the environment. The system receives the walkthrough video and generates a 3D model of the environment. The system applies a machine learning model to the walkthrough video to identify regions within frames where objects are captured. After identifying the regions within frames of the walkthrough video that include objects, the system modifies corresponding regions of the 3D model to include the identified objects. For each of the identified objects, the system determines a likelihood of the identified object being present at a location in the environment based on a comparison of the modified 3D model and the annotated floorplan.

The present disclosure is directed to extracting text from form-like documents. In particular, a computing system can obtain an image of a document that contains a plurality of portions of text. The computing system can extract one or more candidate text portions for each field type included in a target schema. The computing system can generate a respective input feature vector for each candidate for the field type. The computing system can generate a respective candidate embedding for the candidate text portion. The computing system can determine a respective score for each candidate text portion for the field type based at least in part on the respective candidate embedding for the candidate text portion. The computing system can assign one or more of the candidate text portions to the field type based on the respective scores.

A system and method for semantic segmentation with a soft cross-entropy loss is provided. The system inputs a first color image to an input layer of a semantic segmentation network for a multi-class classification task. The semantic segmentation network generates, at an auxiliary stride, a first feature map as an output of an auxiliary layer of the semantic segmentation network based on the input first color image. The system extracts the generated first feature map from the auxiliary layer and computes a probability map as a set of soft labels over a set of classes of the multi-class classification task, based on the extracted first feature map. The system further computes an auxiliary cross-entropy loss between the computed probability map and a ground truth probability map for the auxiliary stride and trains the semantic segmentation network for the multi-class classification task based on the computed auxiliary cross-entropy loss.

Provided are methods, systems, and devices for generating semantic objects and an output based on the detection or recognition of the state of an environment that includes objects. State data, based in part on sensor output, can be received from one or more sensors that detect a state of an environment including objects. Based in part on the state data, semantic objects are generated. The semantic objects can correspond to the objects and include a set of attributes. Based in part on the set of attributes of the semantic objects, one or more operating modes, associated with the semantic objects can be determined. Based in part on the one or more operating modes, object outputs associated with the semantic objects can be generated. The object outputs can include one or more visual indications or one or more audio indications.

Text-to-visual machine learning embedding techniques are described that overcome the challenges of conventional techniques in a variety of ways. These techniques include use of query-based training data which may expand availability and types of training data usable to train a model. Generation of negative digital image samples is also described that may increase accuracy in training the model using machine learning. A loss function is also described that also supports increased accuracy and computational efficiency by losses separately, e.g., between positive or negative sample embeddings a text embedding.

Embodiments described herein transforms a complex and usually unstructured table to a relational table based on the header pattern. Specifically, the original complex table is expanded into a single dimensional relational database format, in which each cell corresponds to one or more corresponding categories or subcategories from the original header. The transformed one-dimensional relational table is then populated with the corresponding cell values from the original table. In this way, data from the original complex and unstructured data table can be stored at a relational database.

An information processing apparatus includes a processor. The processor is configured to identify, from a character string recognition result for a form, a form feature that indicates at least a field in which the form is used or an attribute of a filling-out person filling out the form, accumulate past correction tendencies for character string recognition results for forms having respective identified form features, and obtain a correction tendency for a form having a form feature that is the same as the identified form feature from among the accumulated correction tendencies, and perform control to display a candidate correct expression for the character string recognition result for the form in accordance with the obtained correction tendency.

Method to customize an application associated with a television experience based on the recognition of users located in front of a display and in the field of view of a camera, comprising the following steps: an initialization step during which each user is enrolled in a database of a computer system and is defined by a profile referenced by a profile ID and comprising the user name, biometric data and additional personal data, a finding step during which a wide image, acquired by said camera is scanned to isolate at least one user's faces, to define a marking area surrounding it, to memorize the position of said marking areas, a matching step to extract the biometric data from said marking area, to match them with the biometric data of the profiles stored in the database, and to assign the detected profile ID with the marking area.

While subsequent identification is requested by the application, the following steps are executed acquiring a wide image with the camera, extracting the areas of said image according to the marking areas, extracting for a particular marking area the biometric data of a face, starting the comparison of the extracted biometric data with the biometric data of the profile ID related to this marking area, and in absence of match, continuing with the other biometric data of the database until one profile is found, transmitting the found profile ID to the application.