Abstract: Systems and methods are provided for registering images. The systems and methods perform operations comprising: receiving, at a first time point in a given radiation session, a first imaging slice corresponding to a first plane; encoding the first imaging slice to a lower dimensional representation; applying a trained machine learning model to the encoded first imaging slice to estimate an encoded version of a second imaging slice corresponding to a second plane at the first time point to provide a pair of imaging slices for the first time point; simultaneously spatially registering the pair of imaging slices to a volumetric image, received prior to the given radiation session, comprising a time-varying object to calculate displacement of the object; and generating an updated therapy protocol to control delivery of a therapy beam based on the calculated displacement of the object.

Abstract: A computer-implemented method that includes capturing a first image from a frame of video, analyzing the first image to create a set of artifacts, and classifying an artifact based on an attribute associated with the artifact to create an artifact class. The method also includes comparing the artifact class to a reference class, where, responsive to the comparing, a degree of correspondence is formed. The method continues with detecting a mismatch when the degree of correspondence is below a threshold degree of correspondence, generating a response indicating the mismatch, and causing, responsive to the mismatch, a content production system to change from a first setting to a second setting, the second setting causing a classification of the artifact to change to a second artifact class.

Abstract: A control method for an image processing apparatus that generates a plurality of pieces of album data corresponding to a plurality of albums includes automatically selecting at least one piece of image data out of a plurality of pieces of image data as common image data corresponding to an image to be placed in a common page, which is a page common to all of the plurality of albums, based on an evaluation that is based on a first factor, and automatically selecting at least one piece of image data out of the plurality of pieces of image data as individual image data corresponding to an image to be placed in an individual page, which is a page not common to all of the plurality of albums, based on an evaluation that is based on a second factor.

Abstract: Techniques disclosed herein relate to automatic pan-tilt-zoom adjustment to improve vital sign acquisition. In various embodiments, a vital sign acquisition camera (176) operable to pan, tilt, and zoom (“PTZ”) may capture (402) an image of a patient (100). The image may be analyzed (404) to detect a depicted position of the patient within an image coordinate space of the image. A desired position of the patient within the image coordinate space of the image may determined (406), and a difference in the image coordinate space between the depicted position and the desired position may be calculated (408). The difference may then be mapped (410) from the image coordinate space to a PTZ space. One or more PTZ parameters of the vital sign acquisition camera may be altered (412) based on the mapping. After altering the one or more PTZ parameters, the vital sign acquisition camera may acquire (414) one or more vital signs from the patient.

Abstract: The present disclosure provides a face recognition method and a mobile terminal. The method includes: detecting whether a mobile terminal switches from a stationary state to a raised state; activating a camera and capturing an image by the camera, when the mobile terminal switches from the stationary state to the raised state; judging whether the image matches a preset face template; and determining that recognition is successful when the image matches the preset face template.

Abstract: One embodiment provides a method for capturing information from a measuring instrument, including: capturing, using an image capture device, an image of information displayed on the measuring instrument; analyzing, using a processor, the information; detecting, based on the analyzing, a plurality of elements within the information; wherein the plurality of elements comprise: identification data associated with the measuring instrument and measurement data, the identification data being coded within a tag element; wherein the tag element is at least one of a quick response code, a two dimensional barcode, a barcode, and a service identification tag; the identification data comprising data relating to at least one of: unit type, notation type, device type, device identification, device location, device vendor, device manufacturer, and current user; extracting, using a processor, the plurality of elements from the image; and storing, in a storage device, the plurality of elements in a formatted file.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for multi-query object matching based on inverse model frequency. The methods, systems, and apparatus include actions of obtaining images of a sample object, obtaining models of known objects, determining an image score for each pair of the images and the models, determining an inverse model frequency for each image based on the image scores, determining a model score for each model based on the inverse model frequencies and the image scores, and selecting a particular known object of the known objects as the sample object based on the model scores.

Abstract: A method for deriving further and additional pictures from an original picture, for Artificial Intelligence (AI) training purposes, is applied in a device. The device establishes an original picture set and sets the original pictures as a training picture set for AI training. The original pictures are rotated or flipped or both to obtain amplification pictures. The original pictures are annotated, and each of the amplification pictures is annotated according to a preset conversion rule. The original pictures, the amplification pictures, the annotated original pictures, and the annotated amplification pictures are stored, for inclusion in the AI training picture set.

Abstract: A computed tomography (CT) image processing apparatus includes an image processor which sets two or more CT number ranges of interest defined by a window level and a window width for CT numbers of CT image data, and maps the CT numbers to display grayscale values of a display. The display displays the CT image data according to a mapping result. A gradient of a graph showing a relationship between the CT numbers and the display grayscale values in a CT number range included in the two or more CT number ranges of interest, is greater than a gradient in a CT number range not included in the two or more CT number ranges of interest. The graph has a zero or positive gradient over an entire section, or has a zero or negative gradient over the entire section.

Abstract: X-ray imaging and classification of volume defects within a three-dimensional structure includes identifying one or more volume defects within a three-dimensional structure of a sample and acquiring, with a transmission-mode x-ray diffraction imaging tool, one or more coherent diffraction images of the one or more identified volume defects. The process includes classifying the one or more volume defects within a volume of the three-dimensional structure based on the one or more coherent diffraction images, and training an additional optical or electron-based inspection tool based on the one or more classified defects.

Abstract: The present disclosure provides a prediction method for a healthy radius of a blood vessel path, a prediction method for candidate stenosis of a blood vessel path, and a blood vessel stenosis degree prediction device. The prediction method for a healthy radius includes: obtaining a blood vessel radius of the blood vessel path; by a processor, detecting a radius peak of the blood vessel radius of the blood vessel path; and by the processor, predicting the healthy radius of the blood vessel path by performing a regression on the radius peak of the blood vessel radius. The blood vessel stenosis degree prediction device can, in certain embodiments, automatically determine the candidate stenosis and detect the degree of stenosis for the candidate stenosis range, significantly reduce the computation load, improve the detection efficiency and effectively avoid missed detection.

Abstract: Systems and methods for retraining a trained machine learning model are provided. One or more input medical images are received. Measures of interest for a primary task and a secondary task are predicted from the one or more input medical images using a trained machine learning model. The predicted measures of interest for the primary task and the secondary task are output. User feedback on the predicted measure of interest for the secondary task is received. The trained machine learning model is retrained for predicting the measures of interest for the primary task and the secondary task based on the user feedback on the output for the secondary task.

Abstract: A method for designing a patient-customized EA or selecting an existing EA that fits the patient best includes segmenting shapes of SOIs of the cochlea in a pre-operative CT image using a shape model; defining a 3D curve of interest within the shape model of the SOIs as a sequence of points-; automatically transforming the defined 3D curve to the pre-operative CT image so as to obtain a structure curve in the cochlea; determining a length and curvatures of the structure curve at the sequence of points; and designing a patient-customized EA or selecting an existing EA based on the determined length and curvatures of the structure curve such that after the EA shape model, which estimates the resting state shape of the EA, is rigidly registered to the structure curve in the cochlea, the EA shape model has a registration error smaller than a preset value.

Abstract: A machine trains a first neural network using a first set of images. Training the first neural network comprises computing a first set of weights for a first set of neurons. The machine, for each of one or more alpha values in order from smallest to largest, trains an additional neural network using an additional set of images. The additional set of images comprises a homographic transformation of the first set of images. The homographic transformation is computed based on the alpha value. Training the additional neural network comprises computing an additional set of weights for an additional set of neurons. The additional set of weights is initialized based on a previously computed set of weights. The machine generates a trained ensemble neural network comprising the first neural network and one or more additional neural networks corresponding to the one or more alpha values.

Abstract: Described are systems, methods, and apparatus for generating motion extracted images having a high dynamic range (“HDR”) based on image data obtained from one or more image sensors at different times. The implementations described herein may be used with a single image sensor or camera that obtains images at different exposures sequentially in time. The images may be processed to detect an object moving within the field of view and pixel information corresponding to that moving object extracted. The non-extracted image data may then be combined to produce a motion extracted HDR image that is substantially devoid of the moving object.

Abstract: A hardware system is configured for, and a method of, generating detail-rich gradient-based disparity maps in real-time using an automated gradient-based disparity map classification process that is scalable, can be used under different environment conditions with little to no restrictions, and whose level of precision can be adjusted in a scalable manner. Highly accurate cross-spectral stereo matching methods may be used for search and rescue operations and work at day time and night time using current and past visual and full infrared imaging to generate, classify, and identify scenes in real-time with minimum constraints. Such system and methods may be used to improve operations of existing search and rescue equipment.

Abstract: A method for retrieving an object which becomes lost receives an image of an object from a camera device and analyzes the image to obtain features of the object. The features information comprises category of the object and other details such as serial number. Supplementary information of the object is input on a displayer, the supplemental information comprises time and location of a lost property office or other responsible authority in receiving an object apparently lost. Information according to the features of the object and the supplementary information is broadcast. A device facilitating the retrieval of such object is also disclosed.

Abstract: Techniques for estimating depth for a video stream captured by a monocular image sensor are disclosed. A sequence of image frames are captured by the monocular image sensor. A first neural network is configured to process at least a portion of the sequence of image frames to generate a depth probability volume. The depth probability volume includes a plurality of probability maps corresponding to a number of discrete depth candidate locations over a range of depths defined for the scene. The depth probability volume can be updated using a second neural network that is configured to generate adaptive gain parameters to integrate the DPVs over time. A third neural network is configured to refine the updated depth probability volume from a lower resolution to a higher resolution that matches the original resolution of the sequence of image frames. A depth map can be calculated based on the depth probability volume.

Abstract: A computer-implemented method of anonymizing at least one feature in a scene comprises receiving one or more images representing the scene and applying a sorting algorithm to the image(s) to identify features within the image(s) and classify the features according to at least one predetermined feature category to produce at least one classified feature in the image(s). An annotation scheme, specifying at least one obscuring annotation type to be applied to at least one corresponding feature category, is applied to the image(s). The image(s) are then modified by, for each classified feature in the image(s) having an annotation indicating that said classified feature is in a feature category corresponding to a particular obscuring annotation type to be applied, applying the particular obscuring annotation type to that classified feature in the image(s) so that at least one identifying aspect of that classified feature is obscured within the image(s).

Abstract: An operation analysis apparatus includes: an acquisition unit configured to acquire a plurality of pieces of operation information indicating operations of a plurality of persons to be analyzed; a classification unit configured to classify the plurality of pieces of operation information into a plurality of groups; and an analysis unit configured to analyze the operations of the persons to be analyzed based on the pieces of operation information that have been classified into the plurality of groups. The plurality of pieces of operation information each include skeleton information that indicates positions of a plurality of body parts that are associated with the skeleton of each of the plurality of persons to be analyzed, and the classification unit classifies the plurality of pieces of operation information using the skeleton information included in the operation information.