Abstract: A method for video stabilization may include obtaining a target frame of a video; obtaining a plurality of first feature points associated with the target frame; determining whether the plurality of first feature points include at least one feature point relating to a moving object in the video; in response to a determination that the plurality of first feature points include at least one feature point relating to the moving object in the video, removing the at least one feature point relating to the moving object; performing video stabilization to the target frame based on remaining first feature points of the plurality of first feature points; determining, in the target frame, at least one supplement feature point; and designating the at least one supplement feature point and the remaining first feature points as second feature points associated with a frame immediately following the target frame in the video.

Abstract: An object pose estimation system, an execution method thereof and a graphic user interface are provided. The execution method of the object pose estimation system includes the following steps. A feature extraction strategy of a pose estimation unit is determined by a feature extraction strategy neural network model according to a scene point cloud. According to the feature extraction strategy, a model feature is extracted from a 3D model of an object and a scene feature is extracted from the scene point cloud by the pose estimation unit. The model feature is compared with the scene feature by the pose estimation unit to obtain an estimated pose of the object.

Abstract: The present disclosure provides a method of analyzing an image captured from an imaging device. The method includes receiving one or more images captured from a camera over a predetermined time period. Feature extraction and tracking are executed with respect to the received images. One or more features are detected as untracked features during the feature extraction and tracking. A geometrical prediction is executed with respect to the untracked features to predict one or more features and thereby achieving the one or more untracked features as the one or more tracked features.

Abstract: An image processing method and a device configured to implement the same are disclosed. The device comprises: a hybrid imaging device configured to obtain optical input; and a processing device in signal communication with the hybrid imaging device. The processing device comprises: a motion detection circuit that performs feature tracking based on a first component of an obtained optical input; a motion estimation circuit that performs motion compensation based on output of the motion detection unit; a frame reconstruction circuit that reconstructs image frame based on both the output of the motion estimation unit and a second component of the optical input; and an output unit that outputs image frame at a predetermined global frame rate.

Abstract: A method, computer system, and a computer program product for text detection is provided. The present invention may include training a text detection model. The present invention may include performing text detection on an inputted image using the trained text detection model. The present invention may include determining whether at least one of a plurality of bounding boxes generated using the inputted image has an aspect ratio above a threshold. The present invention may include based upon determining that at least one of the plurality of bounding boxes generated using the inputted image has the aspect ratio above the threshold, upscaling any text within the at least one bounding box and performing text detection on a new image using the trained text detection model. The present invention may include outputting an output image.

Abstract: An information processing apparatus includes a processor configured to: obtain plural images each including any of plural objects; and determine, based on an analysis result regarding the plural objects in the plural images, according to which of two or more objects, among the plural objects, included in an image the image is corrected.

Abstract: There is provided a program causing a computer to execute processing including: acquiring an endoscopic image of a subject from an endoscope; acquiring a three-dimensional medical image obtained by capturing an image of an internal body portion of the subject by means of X-ray CT, X-ray cone beam CT, MRI-CT, or an ultrasonic diagnosis apparatus configured to capture a three-dimensional image of an internal body portion of the subject; generating a virtual endoscopic image reconfigured from the three-dimensional medical image, based on the acquired endoscopic image; calculating distance image information in the endoscopic image based on the virtual endoscopic image and the endoscopic image; and outputting operation assistance information on an operation of the endoscope based on the distance image information and the three-dimensional medical image.

Abstract: An embodiment of the present disclosure discloses a training method for a multi-object tracking model and a multi-object tracking method. The multi-object tracking method comprises: constructing an object graph according to objects to be tracked in a current frame, wherein the vertexes of the object graph correspond to the objects to be tracked, and edge features of the edges between the two vertexes comprise an attribute relationship between the two vertexes; performing graph matching on the object graph and a tracklet graph to calculate matching scores between the object to be tracked and the tracked tracklet in the tracklet graph, wherein the vertexes of the tracklet graph correspond to tracked tracklets, and the edge features of the edges between the two vertexes comprise an attribute relationship between the two vertexes; and determining the matched tracklet of the object to be tracked according to the matching scores.

Abstract: The subject matter discloses a method of asset change detection using images, the method comprising steps executed by processing circuitry, the steps comprising: receiving at least one image of an asset captured by an image capturing device; receiving at least one attribute of a task of detecting a change in the asset using the received at least one image, at least one of the at least one attribute being one of the group consisting of: an attribute measured by a sensor, an attribute extracted from a website, an attribute retrieved from a database, an attribute input by a user, and an attribute encoded in computer code; selecting a reference image among a plurality of reference images of the asset according to at least one criterion based on the received at least one attribute of the task of detecting the change in the asset; computing an asset-difference pixel map, using the selected reference image and the image captured by the image capturing device; and detecting the change in the asset, using the computed

Abstract: A system is for in-situ monitoring and recording of fish health of fish in a fish cage. The system has at least one camera housing. The camera housing is provided with a camera group having at least two cameras arranged to take synchronized pictures for digital close-range photogrammetry. The system has a central data-processing unit, the central data-processing unit being arranged to calculate a three-dimensional model of an object photographed synchronously by the at least two cameras. The data-processing unit is arranged to report the number of structures deviating from the smooth surface of the object in the three-dimensional model.

Abstract: Embodiments of this application provide a method for obtaining depth information and an electronic device, which is applied to the field of image processing technologies and can help an electronic device improve accuracy of obtaining depth information. The electronic device includes a first camera, a second camera, and an infrared projector. The electronic device receives a instruction to obtain depth information of a target object; transmits infrared light with a light spot by an infrared projector; collects first image information of the target object by using the first camera; collects second image information by using the second camera, where the first and second image information include a feature of the target object and a texture feature formed by infrared light; and calculates depth information of the target object based on the first and second image information, the first length, lens focal lengths of the first camera and the second camera.

Abstract: The present disclosure relates to systems, methods, and non-transitory computer readable media for training a generative inpainting neural network to accurately generate inpainted digital images via object-aware training and/or masked regularization. For example, the disclosed systems utilize an object-aware training technique to learn parameters for a generative inpainting neural network based on masking individual object instances depicted within sample digital images of a training dataset. In some embodiments, the disclosed systems also (or alternatively) utilize a masked regularization technique as part of training to prevent overfitting by penalizing a discriminator neural network utilizing a regularization term that is based on an object mask. In certain cases, the disclosed systems further generate an inpainted digital image utilizing a trained generative inpainting model with parameters learned via the object-aware training and/or the masked regularization.

Abstract: A method determines the region where a vehicle can travel without changing clarity of the position information of the object detected by a sensor. An in-vehicle control device includes an object detection unit detecting the position of an object from image information captured by an image pickup device, an object information storage unit stores a pre-processing grid map including a position of a detected object set as an object occupied region and a position where no object has been detected is as an object unoccupied region, an information processing unit generates a determination grid map in which part of the object unoccupied region of the map is replaced with the object occupied region, and a road surface region determination unit that generates an automatic driving grid map in which a closed space surrounded by the object occupied region of the determination grid map is set as a road surface region.

Abstract: An image processing method includes: receiving an input image; performing a low-frequency image regulating operation to regulate the local intensity of the image of pixel unit(s) according to low-frequency information of the image of pixel unit(s) of the input image; performing a high-frequency image regulating operation to improve the details of the image of pixel unit(s) according to high-frequency information of the image of pixel unit (s) of the input image; and, generating an output image according to the input image, the low-frequency image regulating operation, and the high-frequency image regulating operation.

Abstract: An image processing device according to one embodiment estimates first optical flow information of a pixel unit on the basis of consecutive frames, and estimates a polynomial model corresponding to the optical flow information. The image processing device estimates second optical flow information of holes included in the frames on the basis of the polynomial model, and inpaints an input image on the basis of the first optical flow information and/or the second optical flow information.

Abstract: In examples, image data representative of an image of a field of view of at least one sensor may be received. Source areas may be defined that correspond to a region of the image. Areas and/or dimensions of at least some of the source areas may decrease along at least one direction relative to a perspective of the at least one sensor. A downsampled version of the region (e.g., a downsampled image or feature map of a neural network) may be generated from the source areas based at least in part on mapping the source areas to cells of the downsampled version of the region. Resolutions of the region that are captured by the cells may correspond to the areas of the source areas, such that certain portions of the region (e.g., portions at a far distance from the sensor) retain higher resolution than others.

Abstract: Embodiments of this specification provide methods and apparatuses for processing images, devices, and storage media. A method includes: obtaining an image for processing; generating a first mask image corresponding to the image based on a machine-learning model and determining whether the image has a foreground object; in response to determining that the image has the foreground object, setting pixels corresponding to the foreground object to a first grayscale value range, and setting pixels corresponding to non-foreground objects in the first mask image to a second grayscale value range; determining an outline corresponding to the foreground object in the image according to a grayscale value range difference between the first grayscale value range of the foreground object and the second grayscale value range of the non-foreground objects in the first mask image; and sending, to a client device for display, the image with the determined outline for interaction by a user.

Abstract: An embodiment method includes performing first convolutional filtering on a first tensor constructed using a current frame and reference frames (or digital world reference images) of the current frame in a video, to generate a first estimated image of the current frame having a higher resolution than an image of the current frame. The method also includes performing second convolutional filtering on a second tensor constructed using the first estimated image and estimated reference images of the reference frames, to generate a second estimated image of the current having a higher resolution than the image of the current frame. The estimated reference images of the reference frames are reconstructed high resolution images of the reference images.

Abstract: Apparatuses and computer-implemented methods for compression and extraction of three-dimensional (3D) volumetric imaging data, including in particular medical and dental imaging data, that may receive the 3D volumetric image data and preparing compressed 3D volumetric image data for compression by analyzing a relevant region of the 3D volumetric image data. The method may further include compressing, into compressed 3D volumetric image data, the pre-processed static 3D volumetric image data using a video compression scheme. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: A detection unit configured to detect a detection target from an image to be processed by detection processing capable of adjusting at least one setting, a correction cost estimation unit configured to estimate a correction cost needed for correction needed in a case where a result of detection by the detection unit is corrected, an adjustment cost estimation unit configured to estimate an adjustment cost needed in a case where the setting of the detection processing is adjusted to the detection target, and a display control unit configured to perform control to display the correction cost and the adjustment cost on a display unit in a comparable manner are included.