Abstract: Various implementations disclosed herein include devices, systems, and methods for per-pixel filtering. In some implementations, a method includes obtaining an image data frame. In some implementations, the image data frame includes a plurality of pixels. In some implementations, the method includes generating a respective pixel characterization vector for each of the plurality of pixels. In some implementations, each pixel characterization vector includes an object label indicating an object type that the corresponding pixel of the plurality of pixels represents. In some implementations, the method includes modifying corresponding pixel data of the plurality of pixels having a first object label. In some implementations, the method includes synthesizing a first modified image data frame that includes modified pixel data for the plurality of pixels having the first object label and unmodified pixel data for the plurality of pixels not having the first object label.

Abstract: A method is provided for unsupervised domain adaptation for video classification. The method learns a transformation for each target video clips taken from a set of target videos, responsive to original features extracted from the target video clips. The transformation corrects differences between a target domain corresponding to target video clips and a source domain corresponding to source video clips taken from a set of source videos. The method adapts the target to the source domain by applying the transformation to the original features extracted to obtain transformed features for the plurality of target video clips. The method converts the original and transformed features of same ones of the target video clips into a single classification feature for each of the target videos. The method classifies a new target video relative to the set of source videos using the single classification feature for each of the target videos.

Abstract: Described herein are methods and apparatus for increasing sensitivity of ultrasound imaging of fluid flow in an object of interest. Ultrasound imaging of blood perfusion can be performed without contrast enhancement. Embodiments include transforming a spatiotemporal echo data array into a three-dimensional perfusion data array having a spatial dimension, a slow-time dimension, and a frame-time dimension, and filtering the perfusion data array with an eigen passband clutter filter. The clutter filter can increase sensitivity and utility of ultrasound imaging of fluid flow. In some aspects, the method can yield blood flow signal power and perfusion values well separated from tissue clutter. In an example, enhancements to ischemic tissue perfusion maps in a murine model are shown.

Abstract: The present disclosure provides an object detection method, an object detection device, and a storage medium. In the present disclosure, with respect to each contour existing in an image to be detected, a contour diameter is calculated; a shape of the contour is determined based on side lengths of the circumscribed rectangular frame of each contour and the contour diameter; and an object in the image to be detected and parameter information of the object is determined based on the shape of each contour, thereby quickly and accurately detecting the shape of the object and reducing the labor cost and time of detection due to the low complexity of the detection.

Abstract: Embodiments of the present invention are directed to a computer-implemented method for detecting vegetation near a power line. A non-limiting example of the computer-implemented method includes detecting the spectral characteristics of a target location. The spectral characteristics can be used to detect a tree at the target location. The location of the power lines that are within a buffer distance of the tree can be detected. Given that the tree is within a buffer distance, a risk value can be calculated for the tree. The risk value can convey the likelihood that the tree will contact the power line. If the risk is above a threshold amount, a work order can be issued to direct a crew to trim or cut the tree to prevent it from contacting the power line.

Abstract: An image processing device converts an image that is a recognition object image to high-resolution, medium-resolution, and low-resolution images. The device sets the pixel of interest of the high-resolution image, and votes the co-occurrence in a gradient direction with offset pixels, the co-occurrence in the gradient direction pixels in the medium-resolution image, and the co-occurrence in the gradient direction pixels in the low-resolution image, to a co-occurrence matrix. The device creates such a co-occurrence matrix for each pixel combination and for each resolution. The device executes the process on each of the pixels of the high-resolution image, and creates a co-occurrence histogram wherein the elements of a plurality of co-occurrence matrices are arranged in a line. The device normalizes the co-occurrence histogram and extracts, as a feature quantity of the image, a vector quantity having as a component a frequency resulting from the normalization.

Abstract: Systems and methods for processing documents based on a cardinal graph convolution network by generating cardinal graph representations representing words as single nodes with edges connected between neighbouring nodes in four cardinal directions. Features tensors are generated for nodes of the cardinal graph representation and the cardinal directions are encoded to generate an adjacency tensor having node neighbour indices. Entries of the adjacency tensor are transformed into a one-hot encoding of the node neighbour indices. Neighbourhood feature tensors are created over node indices and the features in each block may be scaled, convolved and reduced into new feature tensors.

Abstract: A method and a system for reconstructing obstructed face portions are provided herein. The method may include the following steps: obtaining off-line 3D data, being 3D data of a head of a person not wearing a face-obstructing object, being an object which obstructs a portion of the face of the person; obtaining in real time, real-time 3D data, being 3D data of said head, wherein said person wears said face-obstructing object; applying a 3D transformation to at least a portion of the off-line 3D data, based on the real-time 3D data, to yield reconstructed real time 3D data, being real-time 3D data related to the obstructed face portions; and merging the reconstructed real time 3D data into the real-time 3D data. The system may implement the aforementioned steps over a computer processor.

Abstract: An image processing apparatus includes a first learner trained to detect an orientation of a face in an image including a human face for receiving an input of a target image including a human face and outputting first information associated with an orientation of a face included in the target image, at least one second learner trained, for the first information, to detect a position of a facial component in an image including a human face for receiving an input of the target image and outputting second information associated with a position of a facial component included in the target image, and a control unit that controls the first learner and the second learner. The control unit uses the first learner to obtain the first information from the target image, and uses the second learner corresponding to the first information to obtain the second information from the target image.

Abstract: A user device detects, in a field of view of the camera, a first side of a document, and determines first information associated with the first side of the document. The user device selects a first image resolution based on the first information and captures, by the camera, a first image of the first side of the document according to the first image resolution. The user device detects, in the field of view of the camera, a second side of the document, and determines second information associated with the second side of the document. The user device selects a second image resolution based on the second information, and captures, by the camera, a second image of the second side of the document according to the second image resolution. The user device performs an action related to the first image and the second image.

Abstract: The present invention is directed to providing an image processing apparatus and an image processing method capable of easily obtaining a combined image and a developed image in a case where a construction formed of a plurality of planes is divisionally imaged.

Abstract: A perception device includes: a first neural network that performs a common process associated with perception of an object and thus outputs results of the common process; a second neural network that receives an output of the first neural network and outputs results of a first perception process of perceiving the characteristics of the object with a first accuracy; and a third neural network that receives the output of the first neural network and intermediate data which is generated by the second neural network in the course of the first perception process and outputs results of a second perception process of perceiving the characteristics of the object with a second accuracy which is higher than the first accuracy.

Abstract: A LED controller includes a power distribution module and an interface to an external data bus. An image frame buffer is connected to the interface to receive image data. A separate logic module is connected to the interface and configured to modify image frame buffer output signals sent to an LED pixel array connected to the image frame buffer. The LED pixel array can project light according to a pattern and intensity defined at least in part by the image held in the image frame buffer.

Abstract: In a method for identifying a robot arm responsible for creating a scratch on a wafer, at least one scratch mark on a wafer is detected. A first scratch dimension of the at least one scratch mark is determined. The determined first scratch dimension is compared to a plurality of first robot arm dimensions to generate a plurality of first comparing results, wherein the first comparing results respectively correspond to a plurality of robot arms. One of the robot arms is identified based on the first comparing results.

Abstract: Convolutional neural networks (CNNs) of a set of CNNs are evaluated using a test set of images (electron micrographs) associated with a selected particle type. A preferred CNN is selected based on the evaluation and used for processing electron micrographs of test samples. The test set of images can be obtained by manual selection or generated using a model of the selected particle type. Upon selection of images using the preferred CNN in processing additional electron micrographs, the selected images can be added to a training set or used as an additional training set to retrain the preferred CNN. In some examples, only selected layers of the preferred CNN are retrained. In other examples, two dimensional projections of based on particles of similar structure are used for CNN training or retraining.

Abstract: An image processing apparatus includes first and second acquisition units, a determination unit, and an estimation unit. The first acquisition unit is configured to acquire shape information of an object. The second acquisition unit is configured to acquire a plurality of pieces of image data. The determination unit is configured to determine a pixel position corresponding to a position at which an orientation of a surface is the same as or similar to an orientation of a surface at a position of interest on the object, as a pixel position for estimating a reflection characteristic of the object at the position of interest. The estimation unit is configured to estimate the reflection characteristic of the object at the position of interest by using a pixel value at the pixel position determined by the determination unit.

Abstract: The apparatus includes an image data obtainer, a candidate region extractor, a candidate line extractor, an overlap degree determiner, and a clip image region extractor. The candidate region extractor extracts, as a candidate region, a region containing an object detectable from the image data. The candidate line extractor extracts, as a candidate line, a line that is at least either a line segment or an arc included in the image data. The overlap degree determiner determines whether the degree of overlap between a closed line forming the outline of the candidate region extracted and the candidate line extracted is greater than or equal to a preset predetermined first percentage value. If the overlap degree determiner determines that the degree of overlap is greater than or equal to the first percentage value, the clip image region extractor 19 extracts the candidate region as a clip image.

Abstract: A system that includes a computer processor having a plurality of input data devices, a plurality of output data devices, and a plurality of sensors; and a mechanical assembly integrated with the computer processor to reposition a piece of wood lumber based on software code executing in the computer processor. In some embodiments, the system performs a method that includes eliciting and receiving into the computer processor data parameters from a first human user; obtaining incoming data points about the lumber from the plurality of sensors; processing and storing the data parameters; comparing the incoming data points to the data parameters to obtain comparison results; and, based on the comparison results, (1) rejecting the lumber to a preprogrammed position, (2) feeding the lumber into a saw assembly as positioned, or (3) repositioning the lumber to a more optimal position prior to feeding the lumber to the saw assembly.

Abstract: Methods, apparatuses and computer programs for image processing, in particular image correction, are provided. A sequence of images is processed in this case. Fluctuations, for example as a result of air schlieren, may be recognized and compensated for by identifying a fluctuation signature.

Abstract: A system comprises an eyewear device that includes a frame, a temple connected to a lateral side of the frame, an infrared emitter, and an infrared camera. The infrared emitter and the infrared camera are connected to the frame or the temple to emit a pattern of infrared light. The system includes a processor coupled to the eyewear device, a memory accessible to the processor, and programming in the memory. Execution of the programming by the processor configures the system to perform functions, including functions to emit, via the infrared emitter, a pattern of infrared light on an eye of a user of the eyewear device; capture, via the camera, reflection variations in the pattern of infrared light on the eye of the user; and identify a user of the eyewear device based on the reflection variations of the emitted pattern of infrared light on the eye of the user.