Abstract: Embodiments are disclosed for expanding a seed scene using proposals from a generative model of scene graphs. The method may include clustering subgraphs according to respective one or more maximal connected subgraphs of a scene graph. The scene graph includes a plurality of nodes and edges. The method also includes generating a scene sequence for the scene graph based on the clustered subgraphs. A first machine learning model determines a predicted node in response to receiving the scene sequence. A second machine learning model determines a predicted edge in response to receiving the scene sequence and the predicted node. A scene graph is output according to the predicted node and the predicted edge.

Abstract: Methods, systems, and robots for multi-layer prepreg composite sheet layup. The robotic system may include a memory for storing a dataset including start and end point pairs of a mold of a 3D part that defines a layup sequence, a first robot or a first robot arm that is configured to conform a prepreg layer or sheet onto the mold of the 3D part, and a second robot or a second robot arm that is configured to hold or grasp the prepreg layer or sheet above the mold of the 3D part and stretch or relax the prepreg layer or sheet when the first robot or the first robot arm conforms the prepreg layer or sheet onto the mold. The robotic system may also include one or more processors connected to the first robot or the first robot arm and the second robot or the second robot arm.

Abstract: Unmanned Aerial Vehicles (UAVs) are provided with hammers having contact surfaces to produce acoustic signals in structures to be inspected. By selecting a suitable flight path, the contact surface can be dragged across or tapped against the structure to produce acoustic signals indicative of structure condition. Acoustic detectors are coupled to the UAV to produce detected acoustic signals that can be stored, communicated, and/or processed to access to arbitrary structure surfaces, including bottom surfaces of bridge decks and to locate delaminations.

Abstract: Described herein are techniques for using supervised machine learning to determine whether an image that has been posted to a feed of an online service includes a quotation. In some instance, supervised machine learning techniques are used to infer or predict an intent of a content poster in posting a content item to a feed of an online service. By better understanding the nature of the content being posted, various recommendations can be made during the time when an end-user is posting content, and thereafter.

Abstract: Piezoelectric vibrational energy harvesters (PVEHs) include a Macro Fiber Composite (MFC)) piezoelectric transducer coupled to a cantilever to harvest vibrational energy. One or more Ionic Polymer Metal Composite (IPMC) strips are situated to provide device tuning over a wide frequency range by applying variable contact force to the cantilever. Power consumption for tuning is sufficiently low that the tuning actuator (IPMCs) can be powered using harvested energy.

Abstract: Unmanned Aerial Vehicles (UAVs) are provided with hammers having contact surfaces to produce acoustic signals in structures to be inspected. By selecting a suitable flight path, the contact surface can be dragged across or tapped against the structure to produce acoustic signals indicative of structure condition. Acoustic detectors are coupled to the UAV to produce detected acoustic signals that can be stored, communicated, and/or processed to access to arbitrary structure surfaces, including bottom surfaces of bridge decks and to locate delaminations.

Abstract: A system and a method for searching an image within another image are disclosed. The method includes producing template edge images and target edge images, having image scales, based on determination of edge gradients of a template image and a target image in one or more directions. The template image indicates an image to be searched. The target image indicates another image within which the image needs to be searched. Further, images comprising correlation coefficient values are produced for each of the directions by computing correlation coefficients between the template edge images and the target edge images. At least one local peak is identified from each of the images comprising the correlation coefficient values. Spatial locations along with the correlation coefficients corresponding to the local peak are determined. Thereafter, a presence of the template image in the target image is identified based upon an intersection of the spatial locations.

Abstract: In the process of making a hollow product, especially a deformable toy ball, two halves made from on elastomeric material, a thermoplastic material, rubber, a blend of rubbers, or a thermoplastic elastomer are formed by injection molding with mating edges that engage with each other when the halves are assembled to form the product. The edge of one of the two halves is advantageously formed with a recess centered in the edge, in which a flange formed on the edge of the other of the two halves engages to provide a form-fitting connection between the two halves. The opposing edges are then bonded with each other by fusion or melting or by an adhesive applied to at least one of the edges prior to assembling the halves together. A deformable ball produced by the process is also described.

Abstract: A system and a method for searching an image within another image are disclosed. The method includes producing template edge images and target edge images, having image scales, based on determination of edge gradients of a template image and a target image in one or more directions. The template image indicates an image to be searched. The target image indicates another image within which the image needs to be searched. Further, images comprising correlation coefficient values are produced for each of the directions by computing correlation coefficients between the template edge images and the target edge images. At least one local peak is identified from each of the images comprising the correlation coefficient values. Spatial locations along with the correlation coefficients corresponding to the local peak are determined. Thereafter, a presence of the template image in the target image is identified based upon an intersection of the spatial locations.

Abstract: A system and a method for detecting presence of a living hold in a video stream are disclosed. The method comprises pre-processing N consecutive frames for eliminating noise. An edge score may be calculated for the N consecutive frames, and a clip threshold primary and a clip threshold secondary may be defined. The method further comprises identifying primary edges and secondary edges in the N consecutive frames. Successively, a percentage of fixed primary edges and percentage of fixed secondary edges may be determined. Presence of the living hold may be confirmed based on the value of the percentage of fixed primary edges and the percentage of fixed secondary edges.

Abstract: In the process of making a hollow product, especially a deformable toy ball, two halves made from on elastomeric material, a thermoplastic material, rubber, a blend of rubbers, or a thermoplastic elastomer are formed by injection molding with mating edges that engage with each other when the halves are assembled to form the product. The edge of one of the two halves is advantageously formed with a recess centered in the edge, in which a flange formed on the edge of the other of the two halves engages to provide a form-fitting connection between the two halves. The opposing edges are then bonded with each other by fusion or melting or by an adhesive applied to at least one of the edges prior to assembling the halves together. A deformable ball produced by the process is also described.

Abstract: Described herein are reinforcing members for use in fabricating reinforced concrete, as well as methods of fabricating such reinforcing members. A reinforcing member can include a central reinforcement element and a supplementary cementitious material coated on the central reinforcement element. A method can including applying the supplementary cementitious material to the central reinforcement element, placing concrete around the reinforcing member, and allowing the concrete to cure. The supplementary cementitious material can be a pozzolan.

Abstract: In the process of making a hollow product, especially a deformable toy ball, two halves made from an elastomeric material, a thermoplastic material, rubber, a blend of rubbers, or a thermoplastic elastomer are formed by injection molding with mating edges that engage with each other when the halves are assembled to form the product. The edge of one of the two halves is advantageously formed with a recess centered in the edge, in which a flange formed on the edge of the other of the two halves engages to provide a form-fitting connection between the two halves. The opposing edges are then bonded with each other by fusion or melting or by an adhesive applied to at least one of the edges prior to assembling the halves together. A deformable ball produced by the process is also described.

Abstract: A method and system detection of inherent noise present within a video source prior to digital video compression is disclosed. A noise image is extracted by subtracting a current image from its filtered version. Each pixel of the extracted noise image is normalized based on a determined principal edge image and the analog noise pixels are accumulated to generate an intermediate noise confidence value. Analog noise may be detected based on an analog noise confidence value generated based on the intermediate noise confidence value and a ringing metric, a blockiness metric, a motion vector cost of the current image, a blurriness exception weight, a flashiness exception weight, and a pan blur exception weight. The method may further comprise detection of high frequency noise based on determining a high frequency noise confidence value that may be based on a high frequency noise value and a frequency component with highest magnitude.

Abstract: A method and system for ghost detection in an image are described. Initially, a mode for detecting ghosting artifact is determined for deciding whether to perform analog or inter field ghosting analysis on the image. Based on the determined mode, a plurality of fields is determined to generate a field overlay image. The image is updated based on the field overlay image and an inter-field average of absolute difference is computed. A significant edge image, a principal edge image, and a delta gradient image are generated. A confidence score indicative of a likelihood of ghosting artifact in the image, is computed based on at least the inter-field average of absolute difference, a first and a second count of pixels in the delta gradient image, a third count of pixels in the principal edge image, and a count of rows of the significant edge image that correspond to the ghosting artifact.

Abstract: A method and system for determining blur in digital video due to high quantization adopted during encoding of an image. The image is preprocessed to filter the noise and an edge image is extracted from the preprocessed image. A gradient sign change is determined for each pixel of the edge image. A ringing metric is determined based on the gradient magnitude and the gradient sign change. Further, a blockiness metric is determined based on a block grid periodicity and offset that is determined based on block grid analysis of the edge image. Further, a perceptual blurriness metric is determined using a mean and kurtosis computed based on a gradient histogram analysis of the block grid analysed edge image. A composite blurring metric is determined based on the blockiness metric, the perceptual blurriness metric, and ringing metric.

Abstract: Methods and apparatuses for compressive sensing that enable efficient recovery of features in an input signal based on acquiring a few measurements corresponding to the input signal. One method of compressive sensing includes folding an image to generate first and second folds, and recovering a feature of the image based on the first and second folds without reconstructing the image. One example of a compressive sensing apparatus includes a lens, a focal plane array coupled to the lens and configured to generate first and second folds based on the image, and a decoder configured to receive the first and second folds and to recover a feature of the image without reconstructing the image. The feature may be a local geometric feature or a corner. Compressive sensing methods and apparatuses for determining translation and rotation between two images are also disclosed.

Abstract: The present invention relates to computation of digital fingerprint of a video sequence. The invention presents systems and methods for quick identification of shot boundaries and extraction of fingerprints by processing one or more specific frames. The systems and methods are applied on uncompressed video or compressed video having inter-frame or intra-frame compression. The methods comprises of comparing two frames of the video having a gap in between and identifying a specific frame present in between the two frames such that the specific frame may have a shot boundary. Shot boundaries are calculated for the entire video and then a fingerprint is generated using all the shot boundaries present in the video.

Abstract: The present invention relates to computation of digital fingerprint of a video sequence. The invention presents systems and methods for quick identification of shot boundaries and extraction of fingerprints by processing one or more specific frames. The systems and methods are applied on uncompressed video or compressed video having inter-frame or intra-frame compression. The methods comprises of comparing two frames of the video having a gap in between and identifying a specific frame present in between the two frames such that the specific frame may have a shot boundary. Shot boundaries are calculated for the entire video and then a fingerprint is generated using all the shot boundaries present in the video.

Abstract: Methods and apparatuses for compressive sensing that enable efficient recovery of features in an input signal based on acquiring a few measurements corresponding to the input signal. One method of compressive sensing includes folding an image to generate first and second folds, and recovering a feature of the image based on the first and second folds without reconstructing the image. One example of a compressive sensing apparatus includes a lens, a focal plane array coupled to the lens and configured to generate first and second folds based on the image, and a decoder configured to receive the first and second folds and to recover a feature of the image without reconstructing the image. The feature may be a local geometric feature or a corner. Compressive sensing methods and apparatuses for determining translation and rotation between two images are also disclosed.