Abstract: In an approach for object detection with missing annotations under visual inspection, a processor receives an image. A processor classifies the image being a not-good image using a pre-trained classifier. A not-good image means one or more defect objects being in the image. A processor, in response to classifying the image being the not-good image, detects the one or more defect objects in the not-good image. A processor masks the one or more defect objects in the not-good image. A processor inputs the masked image to train a detector.

Abstract: Systems and methods for generating text corpora comprising realistic optical character recognition (OCR) errors and training language models using the text corpora are provided. An example method comprises: generating, by a computer system, an initial set of images based on an input text corpus comprising text; overlaying, by the computer system, one or more simulated defects over the initial set of images to generate an augmented set of images; generating an output text corpus based on the augmented set of image; and training, using the output text corpus, a language model for optical character recognition.

Abstract: Images that are associated with an identification of a tracking target of a patient to receive radiation treatment may be received. The images may be sorted into a sequence based on a motion of the patient. The sorted images may be provided via a graphical user interface. The sequence of the sorted images that are based on the motion of the patient may be provided.

Abstract: A system and method for determining patient risk stratification is provided based on body composition derived from computed tomography images using segmentation with machine learning. The system may enable real-time segmentation for facilitating clinical application of body morphological analysis sets. A fully-automated deep learning system may be used for the segmentation of skeletal muscle cross sectional area (CSA). Whole-body volumetric analysis may also be performed. The fully-automated deep segmentation model may be derived from an extended implementation of a Fully Convolutional Network with weight initialization of a pre-trained model, followed by post processing to eliminate intramuscular fat for a more accurate analysis.

Abstract: Techniques for selectively associating frames with content entities and using such associations to dynamically generate web content related to the content entities. One embodiment performs a facial recognition analysis on frames of one or more instances of video content to identify a plurality of frames that each depict a first content entity. A measure of quality and a measure of confidence that the frame contains the depiction of the first content entity are determined for each of the identified plurality of frames. Embodiments select one or more frames from the identified plurality of frames, based on the measures of quality and the measures of confidence. The selected one or more frames are associated with the first content entity and web content associated with the first content entity is generated that includes a depiction of the selected one or more frames in association with an instance of video content.

Abstract: A method for image-processing is disclosed. The method includes obtaining first image-information of an image that is to be clustered in response to a first preset condition being met; clustering the image according to the first image-information and obtaining a first clustering-result; sending an image-clustering request to a server in response to a second preset condition being met, wherein the image-clustering request is configured to indicate the server to cluster the image which has been uploaded to the server and obtain a second clustering-result; and receiving the second clustering-result returned from the first server and updating at least one of the first clustering-result and the second clustering-result according to a preset rule.

Abstract: The present technology relates to the field of medical monitoring, and, in particular, to non-contact detecting and monitoring of patient breathing. Systems, methods, and computer readable media are described for calculating a change in depth of a region of interest (ROI) on a patient. In some embodiments, the systems, methods, and/or computer readable media can identify steep changes in depths. For example, the systems, methods, and/or computer readable media can identify large, inaccurate changes in depths that can occur at edge regions of a patient. In these and other embodiments, the systems, methods, and/or computer readable media can adjust the identified steep changes in depth before determining one or more patient respiratory parameters.

Abstract: A method for acquiring a blood flow volume and a blood flow velocity of a coronary artery is provided. The method comprises: acquiring image information of a coronary artery to obtain geometrical feature data of the coronary artery; obtaining, according to the geometrical feature data of the coronary artery, the total volume V of a reference lumen of the coronary artery; and calculating a blood flow volume Q at coronary artery ostia according to the following formula: Q = K * V 3 4 , wherein when the unit of Q is mm3/s, and when the unit of V is mm3, a value range of K is 5-9.5. A blood flow volume and a blood flow velocity of a coronary artery are obtained by means of image information of the coronary artery. Compared with a method, in the prior art, for estimating a blood flow volume of a coronary artery by means of the size of a cardiac muscle, this method is simpler, and can provide a more accurate boundary condition for image-based hemodynamic calculation.

Abstract: A system (100) includes a computer readable storage medium (122) with computer executable instructions (124), including: a predictor (126) configured to determine a baseline coronary state and a predicted coronary state from contrast enhanced cardiac computed tomography volumetric image data and a model of an effect of one or more substances on characteristics effecting the coronary state. The system further includes a processor (120) configured to execute the predictor to determine the baseline coronary state and the predicted coronary state from the contrast enhanced cardiac computed tomography volumetric image data and the model of the effect of one or more of the substances on the characteristics effecting the coronary state. The system further includes a display configured to display the baseline coronary state and the predicted coronary state.

Abstract: A computer-implemented method for classifying and presenting a contrast phase (CP) of a contrast enhanced computerized tomography (CECT) scan is provided. The method includes training an artificial intelligence (AI) algorithm utilizing a set of CPs labeled CECT data to associate a set of characteristics of the data with a probability associated with the CP. The method includes receiving a new set of unlabeled CECT data, and applying the AI algorithm to the new unlabeled CECT data to associate a first probability of a first CP and a second probability of a second CP. The method also includes providing a graphical representation including the first probability of the first CP and the second probability of the second CP.

Abstract: Systems and methods are disclosed for assessing cardiovascular disease and treatment effectiveness based on adipose tissue. One method includes identifying a vascular bed of interest in a patient's vasculature; receiving a medical image of the patient's identified vascular bed of interest; identifying adipose tissue in the received medical image; receiving a geometric vascular model comprising a representation of the patient's identified vascular bed of interest; and computing an inflammation index associated with the geometric vascular model, using the identified adipose tissue.

Abstract: Disclosed techniques relate to identifying subjects in digital images. In various embodiments, a first camera (276, 376, 476, 976) may acquire (1002) digital images (420) depicting a subject in an area. Portion(s) of the digital image(s) that depict a face of the subject may be detected (1004) as detected face image(s). Features of the detected face image(s) may be compared (1006) with features of subject reference templates associated with subject(s) in a subject reference database (412). The subject reference database stores subject reference templates that are associated with multiple subjects and that are generated based on data received from a second camera (256, 456). Based on the comparing, an identity of the subject may be determined (1014) and the features of the detected face image(s) may be stored (1016) in the subject reference database in association with the subject, e.g., as an auxiliary template feature vector for later use.

Abstract: Systems and methods according to one or more embodiments are provided for classifying and counting objects that pass through monitored zones in a field of view of an imaging device. An imaging device receives a series of images of a target scene having at least one monitored zone. An image processor extracts a target line of image values from each image, and adds the target line to a first spatio-temporal image. An edge detector and a cluster analysis module analyzes the first spatio-temporal image to identify objects associated with the at least one zone. An object classifier classifies in object and stores tracking information.

Abstract: A vehicular driving assistance system includes a camera disposed at a vehicle and a control having an image processor that processes image data captured by the camera to determine lane markings demarcating a traffic lane in which the vehicle is traveling. The control, responsive to a map input to the control, estimates a path of travel for the vehicle to maintain the vehicle in the traffic lane in which the vehicle is traveling in situations where the lane markings demarcating the traffic lane in which the vehicle is traveling are not readily determinable. The system updates the estimated path of travel ahead of the vehicle even when no lane markings are determined present on the road ahead of the vehicle. The control adjusts processing of captured image data responsive at least in part to a driving situation that the vehicle is in.

Abstract: It is possible to provide an inspection apparatus having a function of automatically checking a validity of an appearance inspection result without manual operation. An inspection apparatus for performing an appearance inspection using a plurality of images obtained by imaging an inspection target includes a determination unit that estimates an inspection result on each of the plurality of images based on a predetermined relationship model between the image and the inspection result and calculates a comprehensive inspection result based on reliability of each of the inspection results on the plurality of images.

Abstract: Vasculature modeling systems and methods are disclosed that generate an enhanced 3D model based on a combination of two dimensional, 2D, imaging data of a region of interest and 3D imaging data of the region of interest. A hemodynamic simulation is performed using the enhanced 3D model to derive at least one hemodynamic parameter based on the hemodynamic simulation.

Abstract: An automated measurement device and method for coronary artery disease scoring is disclosed. An example device includes a processor configured to obtain a computerized model of a plurality of vascular segments of a patient and create an unstenosed computerized model from the computerized model by virtually enlarging at least some locations of the vascular segments of the computerized model. The processor also determines vascular state scoring tool (“VSST”) scores based on characteristics of vascular locations along the vascular segments. The processor further determines a severity of stenosis for the vascular locations based on comparisons of first blood flow parameter values at the vascular locations in the computerized model to corresponding second blood flow parameter values at the same vascular locations in the unstenosed computerized model. A user interface of the device displays the severity of stenosis in conjunction with the VSST scores for the vascular locations.

Abstract: A method of measuring land deformation over time using interferograms derived from synthetic aperture radar data. The method includes: acquiring radar images covering an area at different points in time; deriving interferograms from pairs of the images, each interferogram measuring phase difference between pixels of a respective pair of images; for each pixel of the interferograms: determining an average coherence value over all of the interferograms; and if the average value is less than a threshold, determining an adjusted average coherence value equal to or above the threshold by excluding one or more of the interferograms below the threshold, provided the number of remaining interferograms is not less than a preset minimum number for each pixel of each interferogram for which the average or adjusted average coherence value is above the threshold, deriving vertical movement from the phase difference; and deriving the map of land deformation from the vertical movement.

Abstract: A method of dynamically updating a feature database that contains features corresponding to a known target object includes providing an image, extracting a first set of features from within the captured image, and comparing the first set of features to the features stored in the feature database. If it is determined that the target object is present in the image then at least one of the extracted features of the first set that are not already included in the feature database are added to the feature database.

Abstract: In one embodiment, a method includes receiving one or more querying images associated with a container of a pharmaceutical product, each of the one or more querying images being based on a particular angle of the container of the pharmaceutical product, calculating one or more confidence scores associated with one or more defect indications, respectively for the container of the pharmaceutical product, by processing the one or more querying images using a target machine-learning model, and determining a defect indication for the container of the pharmaceutical product from the one or more defect indications based on a comparison between the one or more confidence scores and one or more predefined threshold scores, respectively.