Abstract: A method for use in processing of image data stream is described. The method comprising: providing input data comprising at least one sequence of image data pieces indicative of a region of interest. Processing one or more image frames for detection of one or more foreground pixels regions. Collecting adjacent foreground pixels and determining one or more blobs associated with one or more suspected foreground objects in the region of interest Selecting one or more blobs and generating image portion indicative of one or more suspected foreground objects. And processing said one or more image portions for detection of one or more foreground objects and corresponding one or more object classes. For generating output data comprising class probability for one or more objects detected in one or more frame images of said image data stream.

Abstract: The present invention provides systems and methods for simultaneous localization and mapping from video with adversarial shape prior learning in real-time. For example, an unsupervised direct and dense SLAM may learn a geometry prior from data. Given a video sequence, a depth map of a target frame, as well as the target frame and the camera motions between the target frame and all the remaining frames may be output. Further, by fusing a camera motion estimate with a positional sensor's output, positional drift and the need for loop closure can be avoided.

Abstract: A method includes determining, from a first image of a mouth, a first region of the first image comprising a representation of teeth; generating one or more parametric functions associated with tooth color based on color data from the first region; receiving image data comprising new contours of the mouth, wherein one or more of the teeth have a different position in the image data than in the first image; determining a second region comprising the teeth in the image data; and generating a new image based on the image data and the one or more parametric functions, wherein a shape of the teeth is based on the image data and a color of the teeth is based on applying the one or more parametric functions to the second region of the image data.

Abstract: A face recognition technology that enables high-speed determination of erroneous detection of face regions and high-precision face recognition by means of a first face recognition processing unit of relatively low accuracy for detecting a face region of a person on each captured image captured by a monitor camera, and a second face recognition processing unit for performing face recognition of a person more accurately with respect to the detected face region transmitted from the first face recognition processing unit. The second face recognition. processing unit determines that the first face recognition processing unit has erroneously detected when the second face recognition processing unit is not able to recognize a face region on the detected face region transmitted from the first face recognition processing unit, and the first face recognition processing unit calculates and self-learns a barycentric coordinate of the detected face region as a point mask position. Thereafter, the first face recognition.

Abstract: The present disclosure relates to a video processing method, and relates to the technical field of multimedia. The method can include: acquiring a source image from a source video, and obtaining a first object by recognizing the source image; adding an object identifier of the first object to the source image; and generating a target video based on the source image with a same object identifier.

Abstract: An objective of the present disclosure is to enable estimation of an attention target of communication parties even if there are a plurality of objects having the same appearance in a space. An attention target estimation device 10 according to the present disclosure acquires a first-person point-of-view video IMi captured from a perspective of a person and a line-of-sight position gi of a person when the first-person point-of-view video IMi is captured, identifies positions in a 3D space of objects 31, 32, and 33 extracted from the first-person point-of-view video IMi, and determines an object close to the line-of-sight position gi of the person among the objects 31, 32, and 33 included in the first-person point-of-view video IMi as the attention target of the person.

Abstract: The present invention provides a learning image generation apparatus (100) including: a background image acquisition unit (111) that acquires a background image; a background camera posture information acquisition unit (112) that acquires posture information of a background camera which generates a background image; an object continuous image acquisition unit (113) that acquires an object continuous image including an object; a synthesis position determination unit (121) that determines a synthesis position on the background image of the object included in each of a plurality of object still images included in the object continuous image based on the posture information of the background camera; and an image synthesis unit (131) that synthesizes the objects included in each of the plurality of object still images with the background image to generate a synthesis continuous image based on the background image, the object continuous image, the synthesis position.

Abstract: An investigation assist system includes a plurality of servers and an integration server communicatively connected to a terminal and the plurality of servers. In response to reception of a video captured by a plurality of cameras, each of the plurality of servers performs a video analysis of an object with respect to an incident, the plurality of servers processing different objects, respectively. Based on an input of a plurality of different object feature elements from the terminal, the integration server sends a search request for corresponding objects to the respective servers corresponding to the object feature elements, receives and integrates search results of the corresponding objects from the respective servers, and causes the terminal to display an integrated search result.

Abstract: The present disclosure provides embodiments of methods, systems, and apparatuses for detecting wet spots on machined surfaces of wood workpieces. Images of laser spots on a workpiece may be processed to determine area and aspect ratio values of the laser spots. Wet spots may be detected on the workpiece based at least on the area and aspect ratio values, and optionally based in part on color image data. A facility may use wet spot detection in grade determination and/or to classify wood pieces as ‘wet’ or ‘dry’ for the determination of appropriate drying conditions.

Abstract: The disclosure relates to systems and methods for image processing. A trained deep learning model may be determined. An input image may be acquired. A processing result may be generated by processing the input image based on the trained deep learning model. The trained deep learning model may be obtained according to a process including: acquiring a preliminary deep learning model; acquiring a sample image; generating a plurality of sub-images based on the sample image; and training the preliminary deep learning model based on the plurality of sub-images to obtain the trained deep learning model.

Abstract: Systems, methods, and computer-readable storage media for digital fingerprinting of media content. The system does this by normalizing and pre-processing the media content into a standard format, then generating fingerprints for image, audio, and/or video aspects of the media. These fingerprints are mathematical vectors generated using one or more techniques, such as perceptual hashes coupled with machine learning. Once the fingerprints for new content is generated, the system can compare those fingerprints to fingerprints of known content to determine if the new content contains prohibited content.

Abstract: A device and method perform Simultaneous Localization and Mapping (SLAM). The device includes at least one processor configured to perform the SLAM method, which includes the following operations. Preprocess, in a first processing stage, a received data sequence including multiple images recorded by a camera and sensor readings from multiple sensors in order to obtain a frame sequence. Each frame of the frame sequence includes a visual feature set related to one of the images at a determined time instance and sensor readings from that time instance. Sequentially process, in a second processing stage, each frame of the frame sequence based on the visual feature set and the sensor readings included in that frame in order to generate a sequence mapping graph. Merge, in a third processing stage, the sequence mapping graph with at least one other graph, in order to generate or update a full graph.

Abstract: A method, system, and computer program product for correcting the contrast levels of a medical image of a vascular system is described. One of the methods includes identifying a global reference contrast level. The method includes for each image location which represents a location within the vascular system, determining a corrected contrast level by multiplying the original contrast level of that location by the ratio of the global reference contrast level divided by a local reference contrast level.

Abstract: The present invention relates to an image processing device and method whereby deterioration of effects of filter processing due to local control of filter processing when encoding or decoding can be suppressed. A boundary control flag generating unit of a control information generating unit generates boundary control flags based on system specification information which a system specification managing unit manages. A control unit of an adaptive filter processing unit determines a processing method for filter processing to be performed as to pixels nearby a slice boundary following the value of the boundary control flag. For example, selection is made to perform filter processing straddling slices or to perform filter processing closed at the present slice. The present invention can be applied to an image processing device, for example.

Abstract: The present invention relates to a method and apparatus for facilitating identification, particularly where images are of relatively low resolution and also where images may be taken by a moving imaging device, such as a mobile camera. In order to identify persons from images of persons, it is usually necessary to obtain a relatively high-resolution image from which face recognition data can be obtained. The face recognition data can be then matched against a database of many persons face recognition data obtained by an equivalent registration process. Where the images are of relatively low resolution, however, it is difficult to obtain reliable matching. The system and method described here, obtain a number of images of a person to be identified and consolidate data obtained from the images to increase the accuracy of identification. This system can identify persons even where relatively low resolution images and mobile cameras.

Abstract: A method and a system for image reconstruction are provided. The method may include acquiring raw image data, wherein the raw image data may include a plurality of frequency domain undersampled image data samples. The method may include generating a first reconstruction result based on the raw image data using a first reconstruction method, and generating a second reconstruction result based on the raw image data using a second reconstruction method. The method may further include fusing the first reconstruction result and the second reconstruction result, and generating a reconstructed image based on a result of the fusion.

Abstract: A method and system for determining a position of an object with respect to a semi-transparent container, by: receiving, from a camera attached to a semi-transparent container, an image of a scene in proximity to the semi-transparent container; detecting, by an object detector, a new object present in the image, else marking the image as being an image indicative of a current state of the semi-transparent container; cropping, by a computer processor, the image to produce a cropped object image; analyzing, by the computer processor, the cropped object image; and assigning, by the computer processor, a confidence score in a numeric range to the cropped object image, wherein a score at one end of the range indicates the new object corresponding to the cropped object image is inside the semi-transparent container, and wherein a score at the other end of the range indicates the new object is outside.

Abstract: An image processing apparatus 10 includes an image generation model construction unit 11 configured to construct an image generation model for, using image data in which a part of a person is lost as an input image, generating an image in which the lost part of the person has been interpolated. The image generation model construction unit 11 constructs the image generation model by machine learning in which a first sample image including an image of a person, a second sample image in which a part of the person in the first sample image is lost, and human body information specifying a posture of the person in the first sample image are used.

Abstract: A monitoring device conducts a process of obtaining pieces of distance data representing distances to a plurality of physical objects present in a monitoring region, from measurement results of the monitoring region, and using the distance data as current distance data; obtaining past distance data from the measurement results, and converting the past distance data into comparison distance data; calculating difference values between the current distance data and the comparison distance data, and extracting changed regions whose difference values are greater than or equal to a threshold; creating an image obtained by transforming a frontal viewpoint image such that a viewpoint of the measurement of the monitoring region is moved, the frontal viewpoint image being based on the current distance data and the changed regions; and identifying the plurality of physical objects present in the monitoring region, on the basis of the frontal viewpoint image and the coordinate transformed image.

Abstract: A system and method for reducing streak artifacts in a multiplanar reconstruction image are provided. The method may include: retrieving a first image volume from image data, the first image volume including multiple images, at least one of which includes a streak artifact including multiple streaks of a streak width along a first direction; determining a second image volume and a third image volume based on the first image volume; determining an initial error image volume based on the second image volume and the third image volume; determining a revised error image volume based on the initial error image volume; smoothing the revised error image volume to generate a final error image volume; correcting the first image volume according to the final error image volume; and, generating, based on the corrected first image volume, a corrected image volume.