Abstract: An information processing method in which an object of interest is classified using node group information defining a node group having modeled a scheme of classification as a tree structure and having grouped nodes possessing a same parent node, comprises: setting depth information for determining whether to perform classification for a particular node group when sequentially traversing node groups from the parent node using the node group information to classify the object of interest; and classifying the object of interest by sequentially traversing node groups from the parent node using the node group information, and providing a classification result, wherein classifying the object of interest varies a depth up to which node groups are sequentially traversed from the parent node to classify the object of interest, in accordance with setting of the depth information.

Abstract: Systems and methods for video object segmentation are described. Embodiments of systems and methods may receive a referral expression and a video comprising a plurality of image frames, generate a first image mask based on the referral expression and a first image frame of the plurality of image frames, generate a second image mask based on the referral expression, the first image frame, the first image mask, and a second image frame of the plurality of image frames, and generate annotation information for the video including the first image mask overlaid on the first image frame and the second image mask overlaid on the second image frame.

Abstract: Herein is a probabilistic indexing technique for searching semi-structured text documents in columnar storage formats such as Parquet, using columnar input/output (I/O) avoidance, and needing minimal storage overhead. In an embodiment, a computer associates columns with text strings that occur in semi-structured documents. Text words that occur in the text strings are detected. Respectively for each text word, a bitmap, of a plurality of bitmaps, that contains a respective bit for each column is generated. Based on at least one of the bitmaps, some of the columns or some of the semi-structured documents are accessed.

Abstract: An information processing apparatus includes a processor configured to extract a mark specified in advance from an image of a document; and acquire a character string by performing character recognition on a region located in a particular direction with respect to a position of the mark, the direction being associated in advance with the mark.

Abstract: A system and method for sequence extraction using screenshot images to generate a robotic process automation workflow is disclosed. The system and method include capturing a plurality of screenshots of steps performed by a user on an application using a processor, storing the screenshots in memory, determining action clusters from the captured screenshots by randomly clustering actions into an arbitrary predefined number of clusters, wherein screenshots of different variations of a same action is labeled in the clusters, extracting a sequence from the clusters, and discarding consequent events on the screen from the clusters, and generating an automated workflow based on the extracted sequences.

Abstract: The present disclosure provides an object detection method and an object detection device. The object detection device includes: a heterogeneous processor and a memory, the heterogeneous processor including: a processing unit and a programmable logic unit, wherein the programmable logic unit is configured to receive a to-be-detected image, perform feature extraction on the to-be-detected image, and write an extracted feature into the memory; the processing unit is configured to read the feature from the memory, perform target object detection according to the feature, and output a detection result to the programmable logic unit; and the programmable logic unit is further configured to receive the detection result, generate prompt information according to the detection result, and output the prompt information.

Abstract: Discussed herein are devices, systems, and methods for autonomous, dynamic course of action (COA) generation and management. A method can include issuing a communication to one or more assets indicating operations of a first COA to be performed, receiving, by an intelligence, surveillance, and reconnaissance (ISR) device, data indicating an unexpected event, not accounted for in the first COA, has occurred, in response to the data indicating the unexpected event, identifying a second COA or a portion of a second COA that satisfies a mission of the first COA and accounts for the unexpected event, and issuing a second communication to the one or more assets indicating one or more operations of the second COA to be performed.

Abstract: A method and apparatus are provided for nulling a radio frequency (RF) beam for a high-altitude platform (HAP). A transmitter generates an RF signal. A primary antenna system generates an RF beam based on the RF signal. One or more processors determine a result indicating whether to modify the RF beam. When the result indicates to modify the RF beam, a detachable nulling subassembly generates nulling signals based on the RF signal to modify the RF beam generated by the primary antenna system.

Abstract: An example method comprises performing for each class from a plurality of classes: constructing binary training set for the class, the binary training set including labeled cases for that class from the main training set other labeled cases from the main training set; training classifier for the class on the binary training set; computing a local calibration threshold using scores of the labeled cases in the binary training set; and adjusting all scores of the label cases in the binary training set with the local calibration threshold to meet a global decision threshold. The method also comprises determining, with the processor, a global hierarchical calibration threshold by using the adjusted scores for all classes to optimize a performance measurement of all trained classifiers. The method further comprises classifying, with the processor, a new case by using a previously trained classifier, a local calibration threshold, and the global hierarchical calibration threshold.

Abstract: This disclosure relates generally to method for signal pre-processing based on a plurality of data driven models and a data dependent model transformation. The method includes (a) receiving, a raw signal as an input; (b) learning, a set of representational basis from the received raw signal, wherein the set of representational basis comprises a plurality of orthonormal vectors; (c) selecting, at least one orthonormal vector from the plurality of orthonormal vectors, (d) determining, a structure of the plurality of dictionary atoms, wherein structure of the plurality of dictionary atoms corresponds to a graph structure represented as a Laplacian matrix (L); (e) integrating, the graph structure as a structure of the set of representational basis to obtain a reconfigured data model; and (f) reconstructing, using the reconfigured data model to obtain a denoised signal, wherein at least one of constraints on a optimization problem corresponds to desired spectral and topological structure.

Abstract: A mechanism for radiation detection is disclosed. An integrated circuit usable in detecting radiation includes a plurality of readout circuits is described. A readout circuit of the plurality of readout circuits includes an integration capacitor and an averaging capacitor. The integration capacitor is coupled with a pixel of a photodetector pixel array. The pixel has a pixel area. An available area less than the pixel area is usable for layout of the integration capacitor. The integration capacitor has a capacitor area less than the available area. The averaging capacitor has an averaging capacitance greater than the integration capacitance of the integration capacitor. In some aspects, the integrated circuit further includes at least one cascaded averaging circuit coupled with the averaging capacitor.

Abstract: A method for processing video data comprising a plurality of image frames, the method comprising: obtaining a forward model and a backward model of the plurality of image frames; processing the forward model and the backward model to propagate at least one label in the region or patch to at least one other image frame of the video sequence, using a probabilistic method for estimating the label in the at least one other image frame in forward and backward correspondences, wherein, during the processing, a pixel having a most likely label with a probability lower than a threshold value is assigned a predetermined generic label; and generating a labelled result for any given image frame by applying an image label difference, based on label uncertainty between the forward and backward correspondences, to the given image frame.

Abstract: The disclosure relates to technology for object detection in which a vision system receives training datasets including a set of two-dimensional (2D) images of the object from multiple views. A set of 3D models is reconstructed from the set of 2D images based on salient points of the object selected during reconstruction to generate one or more salient 3D models of the object that is an aggregation of the salient points of the object in the set of 3D models. A set of training 2D-3D correspondence data are generated between the set of 2D images in a first training dataset of the training datasets and the salient 3D model of the object generated using the first training dataset. A deep neural network is trained using the set of training 2D-3D correspondence data generated using the first training dataset for object detection and segmentation.

Abstract: A system for estimating a Lambertian equivalent reflectance for reflective band imagery is disclosed. In some embodiments, the system estimates an equivalent reflectance and performs atmospheric correction of reflective band imagery without user interaction and accounts for the effect of background reflectance mixing with individual target reflectances. Some of these embodiments use a dark pixel-based technique to improve the characterization of the atmosphere.

Abstract: A method and a labeling system for generating a label object for the symbolic description of an object of an environment of a mobile device, e.g., a robot or a vehicle. The label object includes at least one attribute of an object at a first point in time, from observations of this object. The method includes selecting, from the observations, a first observation recorded at a first point in time, a second observation recorded at a second point in time, the second point in time being a point in time before the first point in time, as well as a third observation recorded at a third point in time, the third point in time being a point in time after the first point in time; and ascertaining, by using the selected observations, the at least one attribute of the object.

Abstract: A captured image contains a region of interest (ROI) including a plurality of characters to be recognized as text, and non-ROI content to be excluded from the OCR. The captured image is preprocessed to detect and locate the ROI in the captured image, and to determine a boundary of the ROI, including transforming the captured image to a first feature descriptor representation (FDR), and performing a comparison between the first FDR and at least one ROI template that includes at least a second FDR of a representative ROI image. The preprocessing produces an output to be provided to an OCR engine to perform autonomous OCR processing of the ROI while ignoring the non-ROI content based on the determined boundary of the ROI.

Abstract: A method for implementing parametric models for scene representation to improve autonomous task performance includes generating an initial map of a scene based on at least one image corresponding to a perspective view of the scene, the initial map including a non-parametric top-view representation of the scene, implementing a parametric model to obtain a scene element representation based on the initial map, the scene element representation providing a description of one or more scene elements of the scene and corresponding to an estimated semantic layout of the scene, identifying one or more predicted locations of the one or more scene elements by performing three-dimensional localization based on the at least one image, and obtaining an overlay for performing an autonomous task by placing the one or more scene elements with the one or more respective predicted locations onto the scene element representation.

Abstract: A method for large volume holographic imaging is provided that may include determining projection operators within sub-volumes of a decomposed target volume, and determining a point aggregation operator for each sub-volume based on the projection operators. The method may further include receiving holographic field measurement data set captured for the target volume via the sensor array, generating a sub-volume interest value for each sub-volume by applying the holographic field measurement data set to each point aggregation operator, determining a sub-volume with a highest sub-volume interest value, and determining respective lower-tier sub-volume interest values for lower-tier sub-volumes of the sub-volume with the highest sub-volume interest value. The lower-tier sub-volumes may be defined by decomposing the sub-volume with the highest sub-volume interest value. Additionally, the method may include generating an image of the target volume based on the lower-tier sub-volume interest values.

Abstract: A method for a concept update, the method may include detecting that a certain signature of an object causes a false detection; the certain signature belongs to a concept structure that comprises multiple signatures; wherein the false detection comprises determining that the object is represented by the concept structure while the object is of a certain type that is not related to the concept structure; searching for an error inducing part of the certain signature that induced the false detection; and removing from the concept structure the error inducing part to provide an updated concept structure.

Abstract: A method for training thresholds controlling data flow in a plurality of cascaded classifiers for classifying malicious software, comprising: in each of a plurality of iterations: computing a set of scores, each for one of a set of threshold sequences, each threshold sequence is a sequence of sets of classifier output thresholds, each set of classifier output thresholds used to control a flow of data from a first cascaded classifier of the plurality of cascaded classifiers to a second cascaded classifier of the plurality of cascaded classifiers, each score computed when classifying, using the respective threshold sequence, each of a plurality of software objects as one of a set of maliciousness classes; computing a set of new threshold sequences by applying a genetic algorithm to the set of threshold sequences and the set of scores; and using the set of new threshold sequences in a consecutive iteration.