Abstract: A method for identifying a type of choroidal neovascularization in a retina of a human eye is disclosed, the method comprising receiving, in a processor, optical coherence tomography data of the eye; generating, in the processor, volume segments of the eye using the optical coherence tomography data and a neural network; and identifying, in the processor, the type of choroidal neovascularization in the eye, using the volume segments, the type of choroidal neovascularization comprising one or more of the following: classic choroidal neovascularization and occult choroidal neovascularization.

Abstract: One example provides a computing system comprising a depth sensor comprising a plurality of pixels, and a storage machine holding instructions executable by a logic machine to, for each pixel, make K phase measurements to form a set of noisy phase measurements, determine a location at which a projection line that passes through the set of noisy phase measurements in a K-dimensional phase space passes through a lower dimensional plane, the projection line being parallel to a noise free phase evolution line, compare the location to a plurality of independent terms of a predetermined matrix of points in the lower dimensional plane, locate a corresponding set of noiseless phase orders by using a selected set of independent terms to reference a look-up table, determine a distance value for the pixel based upon the corresponding set of noiseless phase orders, and output the distance value for the pixel.

Abstract: An image of a zone of inhibition (ZOI) plate is received. The image is processed to detect sutures within the image of the ZOI plate using computer vision, detect a contour of a ZOI within the image of the ZOI plate using computer vision, generate an overlay based on smoothening the contour of the ZOI and perform one or measurements on the image of the ZOI plate, wherein each of the one or more measurements comprises a distance between the sutures and the overlay.

Abstract: An image distortion correction method and an apparatus, where the method includes: performing optical distortion correction on a collected source image to obtain a first corrected image, where the first corrected image includes a background region and a portrait in which stretching deformation occurs, and where the portrait in which stretching deformation occurs includes at least a first human body region in which stretching deformation occurs and a second human body region in which stretching deformation occurs; and performing algorithm constraint correction on the first human body region, the second human body region, and the background region to obtain a second corrected image, where constraint terms used for the algorithm constraint correction respectively constrain the first human body region, the second human body region, and the background region.

Abstract: In some implementations, a device may receive an image that depicts an environment associated with a vehicle. The device may partition the image into a plurality of subsections. The device may analyze the plurality of subsections to determine respective subsection information, wherein subsection information, for an individual subsection, indicates: a probability score that the subsection includes a line segment associated with an object class, a position of a representative point of the line segment, and a direction of the line segment. The device may identify, based on the respective subsection information of the plurality of subsections, a line associated with the object class that is associated with a set of subsections of the plurality of subsections. The device may perform one or more actions based on identifying the line associated with the object class.

Abstract: Broadly speaking, the present techniques generally relate to machine learning models comprising neural network layers, in which the quantisation level of each layer of the model can be independently selected at run-time. In particular, the present application relates to a computer-implemented method for analysing input data on a device using a trained machine learning, ML, model, comprising independently selecting a quantisation level for each of a plurality of network layers of the model at runtime. The present application also relates to a computer-implemented method of training a machine learning model so that the quantisation level of each of the plurality of network layers is independently selectable at runtime. A single trained model with a single set of weights can therefore be deployed, with the quantisation of each layer selected at runtime to suit the capabilities of the device and available resource.

Abstract: A system and method for measuring circumference of human body are provided. The system includes a 3D sensor configured to obtain a 3D information of a human body with a garment on; a temperature sensor configured to obtain a thermal information of the human body with the garment on; a calibration unit configured to obtain a calibration parameter of the 3D sensor and the temperature sensor; a model generation unit configured to integrate the 3D information and the thermal information according to the calibration parameter to generate a 3D temperature model of the human body with the garment on; and a circumference computation unit configured to retrieve an original profile information corresponding to a target location from the 3D temperature model, and correct the original profile information according to a thermal compensation mechanism to obtain a real circumference of the human body corresponding to the target location.

Abstract: A method of identifying characters in images extracts features of a detection image including characters. Enhancement processing is performed on the detection image according to the features to obtain an enhanced image. Closed edges of the characters are detected in the enhanced image. First rectangular outlines of the characters are determined according to the closed edges. The first rectangular outlines are corrected to obtain second rectangular outlines. The characters are cropped from the detection image according to the second rectangular outlines. The method identifies characters in images accurately and rapidly.

Abstract: A computer-implemented method for modifying X-ray projection images of a subject region includes: generating a set of combined two-dimensional (2D) projections of a subject region, wherein each combined 2D projection includes one or more mask-bordering pixels and one or more mask-edge pixels; forming a three-dimensional (3D) matrix of the set of combined 2D projections; based on the 3D matrix, generating a linear algebraic system for determining pixel values for pixels indicated in a set of 2D projection metal masks, wherein a first change in slope of pixel value associated with a mask-edge pixel of a combined 2D projection is constrained to equal a second change in slope of pixel value associated with a mask-bordering pixel of a combined 2D projection; determining values for a variable vector of the linear algebraic system; and generating a set of inpainted 2D projections by modifying initial 2D projections with values for the variable vector.

Abstract: In example embodiments, a superelevation tool extracts and displays superelevation data from a 3D roadway model by accessing roadway meshes and a horizonal alignment from the 3D roadway model, extracting a plurality of template drops from the one or more roadway meshes at locations along the horizonal alignment to produce an ordered list of template drops and processing the template drops of the ordered list to identify top-facing roadway edges in each template drop that represent top pavement surface of the roadway at the location of the template drop, iteratively searching for a superelevation candidate and detecting superelevation data from the superelevation candidate at least in part by comparing cross-slopes of the top-facing roadway edges of consecutive template drops in the ordered list, wherein a superelevation candidate includes at least two or more template drops having cross-slopes that are locked, and providing a visualization of the detected superelevation data.

Abstract: Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for generating and editing object track labels for objects detected in video data. One of the methods includes obtaining a video segment comprising multiple image frames associated with multiple time points; obtaining object track data specifying a set of object tracks; providing, for presentation to a user, a user interface for modifying the object track data, the user interface displaying object timeline representations of the object tracks; receiving one or more user inputs that indicate one or more modifications to the object timeline representations; updating the object timeline representations displayed in the timeline display area; and updating the object track data according to the updated object timeline representations.

Abstract: A method for detecting and tracking target object in a captured video using convolutional neural network (CNN) is provided. The method includes: inputting image data into a detecting model to generate detection results, wherein the detecting model is constructed by the CNN; inputting the image data into tracking models to generate tracking results; performing detection score enhancement operation according to the detection results and the tracking results to obtain enhanced detection results; matching the enhanced detection results and the tracking results by a matching operation; processing matched results and unmatched target detection results and unmatched target tracking results; and selectively updating the tracking models using tracking reliability estimation according to the matched results.

Abstract: Provided is an image processing device with which it is possible to automate the setting of a detection parameter that is suitable for obtaining a desired detection result.

Abstract: In one aspect, a method includes receiving an image of a target subject; determining a direction in response to the receipt of the image, the direction being one in which the target subject was likely to move during a time period in the past or is likely to move during a time period in the future; determining a target area within which another image of the target subject can be expected to appear based on the determined direction; and determining if a portion of a subsequent image is outside the determined target area to identify if the subsequent image is one relating to the target subject, wherein the subsequent image is one taken during the time period in the past or during the time period in the future.

Abstract: Disclosed is an image analysis system which includes a first analyzer and a second analyzer. The first analyzer generates first features by encoding images through a first model and adjusts a weight of the first model based on the first features. The second analyzer generates second features based on the first features by encoding the images through a second model, classifies the second features into classes, respectively, and adjusts a weight of the second model based on mutual information and a correlation between a first class among the classes which features corresponding to an original image group from among the second features are classified as and a second class among the classes which features corresponding to an augmentation group of the original image group from among the second features are classified as.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for eye gesture recognition. In one aspect, a method includes obtaining an electrical signal that represents a measurement, by a photodetector, of an optical signal reflected from an eye and determining a depth map of the eye based on phase differences between the electrical signal generated by the photodetector and a reference signal. Further, the method includes determining gaze information that represents a gaze of the eye based on the depth map and providing output data representing the gaze information.

Abstract: Systems/techniques that facilitate self-supervised deblurring are provided. In various embodiments, a system can access an input image generated by an imaging device. In various aspects, the system can train, in a self-supervised manner based on a point spread function of the imaging device, a machine learning model to deblur the input image. More specifically, the system can append to the model one or more non-trainable convolution layers having a blur kernel that is based on the point spread function of the imaging device. In various aspects, the system can feed the input image to the model, the model can generate a first output image based on the input image, the one or more non-trainable convolution layers can generate a second output image by convolving the first output image with the blur kernel, and the system can update parameters of the model based on a difference between the input image and the second output image.

Abstract: Devices, systems, and methods for machine learning model generation. A method can include generating image chips of an image. The image chips can each provide a view of a different extent of an object in the image. Based on an object definition that indicates respective features of the object and a location of the respective features along a length of the object, it can be determined whether any of the image chips include any of the respective features. Each image chip of the image chips can be labeled to include an indication of any of the features included in the image chip resulting in labelled image chips. The method can include training an ensemble classifier based on the labelled image chips resulting in a trained ensemble classifier.

Abstract: A system, apparatus, method and computer program product are provided for determining a mobile device integrity status. Images of a mobile device captured by the mobile device and using a reflective surface are processed with various trained models, such as neural networks, to verify authenticity, detect damage, and to detect occlusions. A mask may be generated to enable identification of concave occlusions or blocked corners of an object, such as a mobile device, in an image. Images of the front and/or rear of a mobile device may be processed to determine the mobile device integrity status such as verified, not verified, or inconclusive. A user may be prompted to remove covers, remove occlusions, and/or move the mobile device closer to the reflective surface. A real-time response relating to the mobile device integrity status may be provided. The trained models may be trained to improve the accuracy of the mobile device integrity status.

Abstract: An image of a financial instrument can be received and analyzed. One or more properties of the financial instrument can be determined based on the image. A uniqueness rating of the financial instrument can be determined based on the property or properties. Terms of the financial instrument are identified, and the financial instrument and identified terms are recorded in a digital storage base.