Abstract: Systems and methods for detecting objects in a video are provided. A method can include inputting a video comprising a plurality of frames into an interleaved object detection model comprising a plurality of feature extractor networks and a shared memory layer. For each of one or more frames, the operations can include selecting one of the plurality of feature extractor networks to analyze the one or more frames, analyzing the one or more frames by the selected feature extractor network to determine one or more features of the one or more frames, determining an updated set of features based at least in part on the one or more features and one or more previously extracted features extracted from a previous frame stored in the shared memory layer, and detecting an object in the one or more frames based at least in part on the updated set of features.

Abstract: A method for assessing test adequacy of deep neural networks based on element decomposition is provided. The network testing is divided into black box testing and white box testing, of which key elements are decomposed and defined. Network parameters including a weight matrix and a bias vector are extracted. Importance values of neurons in individual layers of the deep neural network are calculated and clustered, and an importance value hot map of neurons in each layer is generated based on clustering results. Mutation testing, and index calculation and evaluation are performed.

Abstract: A system and method for positioning oilfield tubulars on a drilling floor (2), having an automated tubular handling system, to permit alignment and automated make up and break out of threading operations between a stationary tubular (8) and a moving tubular (34), utilizing image information from radially offset cameras (54, 56), processing the image information to recognize the tubulars within the image and develop position information for said tubulars, combining the position information from the two camera systems to develop three dimensional information for each of the tubulars, developing instructions for the automated tubular handling system to bring the stationary and moving tubulars into vertical alignment and lowering the moving tubular into threaded contact, and engaging an automated hydraulic torque wrench to make up the threaded connection.

Abstract: A system and method for positioning oilfield tubulars on a drilling floor (2), having an automated tubular handling system, to permit alignment and automated make up and break out of threading operations between a stationary tubular (8) and a moving tubular (34), utilizing image information from radially offset cameras (54, 56), processing the image information to recognize the tubulars within the image and develop position information for said tubulars, combining the position information from the two camera systems to develop three dimensional information for each of the tubulars, developing instructions for the automated tubular handling system to bring the stationary and moving tubulars into vertical alignment and lowering the moving tubular into threaded contact, and engaging an automated hydraulic torque wrench to make up the threaded connection.

Abstract: A method for automatically detecting the presence of a contrast agent in an x-ray image includes acquiring a preliminary x-ray image. A background image is estimated. The contrast agent is administered. A plurality of image frames is acquired. The background image is subtracted from each image frame. An image having a highest image intensity is selected. A predefined shape model is fitted to the selected image using a semi-global optimization strategy. The fitting of the shape model is used to fit the shape model to each of the subtracted images. A feature value is calculated for each image frame based on pixel intensities of each pixel fitted to the shape model for the corresponding subtracted image. An image frame of peak contrast is determined by selecting the image frame with the greatest feature value.

Abstract: A method for fusing 2D fluoroscopic images with 2D angiographic images for real-time interventional procedure guidance includes using a 2-dimensional angiographic image to detect injection of a contrast agent into an anatomical structure, where the contrast agent first appears in a detected image frame, subtracting a background image from the detected image frame and binarizing the subtracted image frame to segment the anatomical structure from the subtracted image frame, registering the segmented anatomical structure with a predefined shape model of the anatomical structure to generate an anatomical mask, generating an anatomical probability map from the anatomical mask using intensity information from the subtracted image frame, where the anatomical probability map expresses a probability of a pixel in the subtracted image frame belonging to the anatomical structure, and fusing the angiographic image with one or more subsequently acquired 2-dimensional fluoroscopic images of the anatomical structure using t

Abstract: A method for automatically detecting the presence of a contrast agent in an x-ray image includes acquiring a preliminary x-ray image. A background image is estimated. The contrast agent is administered. A plurality of image frames is acquired. The background image is subtracted from each image frame. An image having a highest image intensity is selected. A predefined shape model is fitted to the selected image using a semi-global optimization strategy. The fitting of the shape model is used to fit the shape model to each of the subtracted images. A feature value is calculated for each image frame based on pixel intensities of each pixel fitted to the shape model for the corresponding subtracted image. An image frame of peak contrast is determined by selecting the image frame with the greatest feature value.

Abstract: A method for fusing 2D fluoroscopic images with 2D angiographic images for real-time interventional procedure guidance includes using a 2-dimensional angiographic image to detect injection of a contrast agent into an anatomical structure, where the contrast agent first appears in a detected image frame, subtracting a background image from the detected image frame and binarizing the subtracted image frame to segment the anatomical structure from the subtracted image frame, registering the segmented anatomical structure with a predefined shape model of the anatomical structure to generate an anatomical mask, generating an anatomical probability map from the anatomical mask using intensity information from the subtracted image frame, where the anatomical probability map expresses a probability of a pixel in the subtracted image frame belonging to the anatomical structure, and fusing the angiographic image with one or more subsequently acquired 2-dimensional fluoroscopic images of the anatomical structure using t