Abstract: Embodiments of the present disclosure provides a fine-grained image recognition method and apparatus using graph structure represented high-order relation discovery, wherein the method includes: inputting an image to be classified into a convolutional neural network feature extractor with multiple stages, extracting two layers of network feature graphs in the last stage, constructing a hybrid high-order attention module according to the network feature graphs, and forming a high-order feature vector pool according to the hybrid high-order attention module, using each vector in the vector pool as a node, and utilizing semantic similarity among high-order features to form representative vector nodes in groups, and performing global pooling on the representative vector nodes to obtain classification vectors, and obtaining a fine-grained classification result through a fully connected layer and a classifier based on the classification vectors.

Abstract: A 3D anatomical model of one or more blood vessels of a patient may be obtained using CT angiography, while a 2D image of the blood vessels may be obtained based on fluoroscopy. The 3D model may be registered with the 2D image based on a contrast injection site identified on the 3D model and/or in the 2D image. A fused image may then be created to depict the overlaid 3D model and 2D image, for example, on a monitor or through a virtual reality headset. The injection site may be determined automatically or based on a user input that may include a bounding box drawn around the injection site on the 3D model, a selection of an automatically segmented area in the 3D model, etc.

Abstract: A method of generating a custom three-dimensional (3D) model of a patient bone from one or more 2D images is disclosed. The method includes obtaining a 2D image of a bone, optionally of a joint, and identifying a 3D bone template for a candidate or representative bone from a pre-aligned library of representative bones. The method further includes repositioning one or more views of the 3D model or 2D images (e.g., with respect to rotation angle or caudal angle). In an iterative process, another 3D bone model for another candidate bone can be identified based on the repositioning until an accuracy threshold is satisfied. When the accuracy threshold is satisfied, surface region(s) of the current 3D bone model can then be modified to generate the resulting 3D model for the patient bone. The process can then be repeated for other bone(s) associated with the joint of the patient.

Abstract: Techniques are described for generating bird's eye view (BEV) images and segmentation maps. According to one or more embodiments, a system is provided comprising a processor that executes computer executable components stored in at least one memory, comprising a machine learning component that generates a synthesized bird's eye view image from a stitched image based on removing artifacts from the stitched image present from a transformation process. The system further comprising a generator that produces the synthesized bird's eye view image and a segmentation map, and a discriminator that predicts whether the synthesized bird's eye view image and the segmentation map are real or generated.

Abstract: Techniques for automatic intelligent information extraction from an electronic document are disclosed. In one embodiment, a computerized method is disclosed comprising training a label prediction model to generate a set of label predictions, obtaining an electronic document, analyzing the electronic document and determining a set of features for each of a set of information items identified in the electronic document, obtaining model output from the label prediction model for each information item, the model output comprising, for a respective information item, a set of probabilities corresponding to a set of information classes, and generating an information extraction comprising a set of labels corresponding to the set of information items.

Abstract: Apparatuses, systems, and techniques to identify orientations of objects within images. In at least one embodiment, one or more neural networks are trained to identify an orientations of one or more objects based, at least in part, on one or more characteristics of the object other than the object's orientation.

Abstract: The systems, apparatuses and methods of the present invention set forth improvements to the problems of the current pairing or duplex paradigm, resulting in a dramatic increase in fiber transmission efficiency, accomplished explicitly by restructuring presently-aligned C-Band wavelengths into innovative DWDM transmit and receive formats, and through implementing photonic-wave changes, which directs Ethernet data flow onto new path adaptations. These improvements could reduce line haul expenses significantly, believed to reach a projected 50% less requirement/deployment of fiber strands. This saving would offer owner-operators substantial fiber strand cost reductions, affecting transportation rates of high-bandwidth digital payloads traversing over DWDM networks, and lower usage rates of cross-connections amid multiple equipment inter-exchanging throughout large data centers.

Abstract: A system for creating synthetic data for testing an autonomous system, comprising at least one hardware processor adapted to execute a code for: using a machine learning model to compute a plurality of depth maps based on a plurality of real signals captured simultaneously from a common physical scene, each of the plurality of real signals are captured by one of a plurality of sensors, each of the plurality of computed depth maps qualifies one of the plurality of real signals; applying a point of view transformation to the plurality of real signals and the plurality of depth maps, to produce synthetic data simulating a possible signal captured from the common physical scene by a target sensor in an identified position relative to the plurality of sensors; and providing the synthetic data to at least one testing engine to test an autonomous system comprising the target sensor.

Abstract: Techniques are disclosed for improving upon the usage of Light Detection and Ranging (LIDAR) supervision to perform image depth estimation. The techniques use a generator and adversary network to generate respective models that “compete” against one another to enable the generator model to output a desired output image that compensates for a LIDAR image having a structured or lined data pattern. The techniques described herein may be suitable for use by vehicles and/or other agents operating in a particular environment as part of machine vision algorithms that are implemented to perform autonomous and/or semi-autonomous functions.

Abstract: Disclosed is a pain assessment method using a deep learning model, the pain assessment method including operations of receiving, by an analysis device, an image indicating activity in a specific brain area of a subject animal and allowing the analysis device to input images of regions of interest in the image into a neural network model and assess the pain of the subject animal according to a result output by the neural network model.

Abstract: This disclosure describes one or more implementations of systems, non-transitory computer-readable media, and methods that perform language guided digital image editing utilizing a cycle-augmentation generative-adversarial neural network (CAGAN) that is augmented using a cross-modal cyclic mechanism. For example, the disclosed systems generate an editing description network that generates language embeddings which represent image transformations applied between a digital image and a modified digital image. The disclosed systems can further train a GAN to generate modified images by providing an input image and natural language embeddings generated by the editing description network (representing various modifications to the digital image from a ground truth modified image). In some instances, the disclosed systems also utilize an image request attention approach with the GAN to generate images that include adaptive edits in different spatial locations of the image.

Abstract: A system having multiple devices that can host different versions of an artificial neural network (ANN) as well as different versions of a feature dictionary. In the system, encoded inputs for the ANN can be decoded by the feature dictionary, which allows for encoded input to be sent to a master version of the ANN over a network instead of an original version of the input which usually includes more data than the encoded input. Thus, by using the feature dictionary for training of a master ANN there can be reduction of data transmission.

Abstract: The preset invention aims to provide a method of presenting a learning condition which enables an improvement in the accuracy of image analysis. An information processing apparatus for machine learning is provided which includes a true/false information generating unit which generates true/false information of an image analysis result, a reliability determining unit which determines reliability related to analysis in image analysis processing, and a learning condition output unit which presents a learning condition, based on the true/false information and the reliability.

Abstract: A system includes a housing and a first circuit board positioned inside the housing. The housing has a top panel, a bottom panel, a left side panel, a right side panel, a front panel, and a rear panel. The front panel is at an angle relative to the bottom panel in which the angle is in a range from 30 to 150°. The first circuit board has a length, a width, and a thickness, in which the length is at least twice the thickness, the width is at least twice the thickness, and the first circuit board has a first surface defined by the length and the width. The first surface of the first circuit board is at a first angle relative to the bottom panel in which the first angle is in a range from 30 to 150°. The first surface of the first circuit board is substantially parallel to the front panel or at a second angle relative to the front panel in which the second angle is less than 60°.

Abstract: An approach is provided for lane width estimation from incomplete lane marking detections of a road lane. The approach, for example, involves generating one or more perpendicular lines respectively from location centers of one or more first lane marking detections. A respective lane marking detection represents at least a portion of a boundary of the road lane as a line delimited by two location data points in accordance with detections by at least one sensor device onboard at least one vehicle. The approach also involves identifying second lane marking detections that each respectively intersect one of the one or more perpendicular lines. The approach further involves selecting one or more candidate lane widths based on one or more respective distances from the location centers to the second lane marking detections. The approach further involves determining an estimated lane width of the road lane based on the one or more candidate lane widths.

Abstract: An object orientation indicator for indicating the orientation of an object, such as the vertical positioning of an object to be inserted or being inserted into a surface. The object orientation indicator is configured to determine or indicate when an object being inserted into a surface parallel or non-parallel to the ground is vertically level, i.e. perpendicular to the ground. The object orientation indicator may also be configured to determine or indicate if a preexisting vertical object inserted into a surface parallel or non-parallel to the ground remains in its originally inserted orientation. The object orientation indicator may include a housing assembly and a display panel.

Abstract: A trusted graph data node classification method includes: (1) inputting a topological graph and node features, and calculating a discrete Ricci curvature of the discrete topological graph; (2) preprocessing the curvature and the node features; (3) mapping the curvature, reconstructing original features, and performing a semi-supervised training on graph data containing adversarial examples; and (4) performing a classification on unlabeled nodes. The new method uses a discrete curvature to extract topological information, and uses a residual network to reconstruct node feature vectors without knowing the technical details of the adversarial examples, and without using a large number of adversarial examples for adversarial training. Hence, the system effectively defends against attacks from adversarial examples on the graph data, outperforms the existing mainstream models in terms of accuracy when used in data without adversarial examples, and is thus a trusted node classification system.

Abstract: A method of determining a region of interest in an image of tissue of an individual by an apparatus including processing circuitry may include executing, by the processing circuitry, instructions that cause the apparatus to partition an image of tissue of an individual into a set of areas, identify a tissue type of each area of the image, and apply a classifier to the image to determine a region of interest, the classifier being configured to determine regions of interest based on the tissue types of the set of areas of the image.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for predicting locations of utility assets. One of the methods includes receiving an input image of an area in a first geographical region; generating, from the input image and using a generative adversarial network, a corresponding reference image; and generating, by an object detection model and from the reference image, an output that identifies respective locations of one or more utility assets with reference to the input image.

Abstract: The present disclosure discloses a method, apparatus and system for customer group analysis, and a storage medium. The method includes: obtaining video images of a customer group passing by a display apparatus; recognizing and tracking head images of the customer group in the video images, and determining behavior characteristics of individuals in the customer group; and performing statistical analysis on the behavior characteristics of the customer group corresponding to the display apparatus, so as to update a quantity of individuals corresponding to each behavior characteristic.